Maldi Tandem Mass Spectrometric Techniques for the Analysis and Imaging of Lignocellulosic Materials

MISSING IMAGE

Material Information

Title:
Maldi Tandem Mass Spectrometric Techniques for the Analysis and Imaging of Lignocellulosic Materials
Physical Description:
1 online resource (211 p.)
Language:
english
Creator:
Lunsford,Kyle Ann
Publisher:
University of Florida
Place of Publication:
Gainesville, Fla.
Publication Date:

Thesis/Dissertation Information

Degree:
Doctorate ( Ph.D.)
Degree Grantor:
University of Florida
Degree Disciplines:
Chemistry
Committee Chair:
Yost, Richard A
Committee Members:
Smith, Benjamin W
Polfer, Nicolas Camille
Powell, David H
Peter, Gary F

Subjects

Subjects / Keywords:
lignocellulosic -- maldi -- mass
Chemistry -- Dissertations, Academic -- UF
Genre:
Chemistry thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract:
The conversion of cellulosic material to ethanol begins with a dilute acid pretreatment to remove hemicellulose and alter the lignin distribution. Despite successful conversions, spatial changes of chemical composition throughout pretreatment are not well characterized. Developing nondestructive analytical methods to visualize changes that occur in biomass throughout the pretreatment processes could help increase efficient of biomass conversion into biofuels. MALDI-mass spectrometric imaging (MS imaging) in a powerful analytical technique that displays a distribution of target molecules in intact tissue. MALDI-TOF (time-of-flight)-MS is commonly used to characterize polysaccharides, but tandem MS capabilities of a TOF are limited. The linear ion trap (LIT) mass analyzer offers tandem MS capabilities, which can provide structural information and monosaccharide composition of polysaccharides. A MALDI-LIT tandem MS method was developed for to characterize three plant-related standards, microcrystalline cellulose (MCC), Birch xylan and Spruce lignin. The standard analysis determined the ions that are formed from MALDI and provide characteristic fragmentation pathways of these standard compound classes. MALDI-MS imaging of intact Populus tissue sections illustrated a nearly even ion signal across the whole tissue. In contrast, tandem MS imaging was able to differentiate between isobaric ions to provide sensitive and selective than single-stage MS imaging of these compound classes. The tandem MS images showed localization of cellulose in the secondary xylem and phloem fiber cells, which are both known to have thickened secondary cell walls. High correlation between MALDI tandem MS imaging, fluorescence imaging and ToF-SIMS imaging was observed. In addition, ToF-SIMS imaging provided a high spatial resolution, chemically selective images to characterize the material further. Principal component analysis (PCA) of MALDI-MS analysis of standards and ToF-SIMS imaging determined ions characteristic of lignified and non-lignified tissue. In conclusion, this dissertation reports a comprehensive analysis of lignocellulosic material (LCM) using three different techniques, MALDI-LIT tandem MS imaging, fluorescence microscopy and ToF-SIMS imaging. Combining the information obtained from these techniques provided a more complete analysis of the tissue sections. Furthermore, the tools for analysis of LCM throughout a pretreatment analysis have been developed.
General Note:
In the series University of Florida Digital Collections.
General Note:
Includes vita.
Bibliography:
Includes bibliographical references.
Source of Description:
Description based on online resource; title from PDF title page.
Source of Description:
This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility:
by Kyle Ann Lunsford.
Thesis:
Thesis (Ph.D.)--University of Florida, 2011.
Local:
Adviser: Yost, Richard A.

Record Information

Source Institution:
UFRGP
Rights Management:
Applicable rights reserved.
Classification:
lcc - LD1780 2011
System ID:
UFE0043118:00001


This item is only available as the following downloads:


Full Text

PAGE 1

1 MALDI TANDEM MASS SPECTROMETRIC TEC H NIQUES FOR THE ANALYSIS AND IMAGING OF LIGNOCELLULOSIC MATERIALS By KYLE ANN LUNSFORD A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2011

PAGE 2

2 2011 Kyle Ann Lunsford

PAGE 3

3 To Mom, Dad Quin and Warren

PAGE 4

4 ACKNOWLEDGMENTS I would like to thank several people for their support, advice and gui dance that ultimately le d to obtaining a PhD. First, I would like to thank my research advisor, Dr. Yost. I cannot thank him enough not only for accepting me into the group, but for also More importantly, Dr. Yost has managed to make me leave graduate school more passionate about science than when I began. I would also like to have been able to accomplish as much as I did without his willingness to help me understand plants as well as to provid e wood samples. With his help, I became more confident in my resea rch abilities, as well as being able explain mass spectrometry s In addition, I would like to thank the Yost Group members, past and present. The Yost group has always been there to support me, to help me grow as a scientist to help me understand the underlying science of my experiments and to guide me in the right direction A special thanks to Dr. Dan and Rob for sharing an office with me for two years O ur in depth conversations (both on and off topic) hard time s each helped me to understand mass spectrometry and my research better I would also like to thank the Bio I maging Mass Spectrometry laboratory at the FOM Institute AMOLF for allowing me to visit for 5 weeks. Specifically, Dr. Ron Heere n for allowing me to visit his lab and providing insight into ToF SIMS imaging, Andras Kiss for helping me perform the ToF SIMS analysis and Gert Eijkel for his help with statistics of ToF SIMS imaging data (and for allowing me to use his ChemomeTricks pro gram).

PAGE 5

5 A true acknowledgements section would not be complete without tha nking my family, Mom, D ad, and Qu i n I would not be here without my parents paving a path for me to go to college, and ultimately graduate school. I cannot begin to express my gratitu de for giving me every opportunity in the world I realize that they did not have to give me anything but instead they chose to give me everything. I hope I have shown my appreciation throughout the years and that this document helps make my case! And to my brother Quin, who I would like to thank for always supporting me in my decisions and helping me get through school and for being the best (and most protective) big brother out there. Last and certainly not least, I would like to thank my husband, Warren who has provided an irreplaceable support system. from a shy reserved scientist, into a scientist (and woman) more confident than I ever thought that I could be I look forward to starting the next stag e of our life and scientific career s together.

PAGE 6

6 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF TABLES ................................ ................................ ................................ .......... 10 LIST OF FIGURES ................................ ................................ ................................ ........ 11 ABSTRACT ................................ ................................ ................................ ................... 14 CHAPTER 1 BACKGROUND ................................ ................................ ................................ ...... 16 Biofuel Research ................................ ................................ ................................ .... 16 Bioethanol ................................ ................................ ................................ ........ 16 Lignocellulosic Ethanol ................................ ................................ ..................... 17 Lignocellulosic Pretreatment ................................ ................................ ............ 19 Anal ytical Approaches for LCM Analysis ................................ .......................... 20 Mass Spectrometry of Polysaccharides ................................ ................................ .. 21 Early MS of Polysaccharides ................................ ................................ ............ 21 Tandem MS of Polysaccharides ................................ ................................ ....... 23 Inst rumentation ................................ ................................ ................................ ....... 25 Matrix Assisted Laser Desorption/Ionization ................................ .................... 25 Ionization mechanism ................................ ................................ ................ 25 MALDI matrix considerations ................................ ................................ ..... 26 Seco ndary Ion Mass Spectrometry ................................ ................................ .. 27 Mass Analyzers ................................ ................................ ................................ 28 Figures of merit ................................ ................................ .......................... 29 Linear ion traps ................................ ................................ .......................... 29 Tandem MS with L IT ................................ ................................ .................. 31 Time of flight ................................ ................................ .............................. 32 Tandem MS with ToF ................................ ................................ ................. 34 Mass Spectrometric Imaging ................................ ................................ .................. 35 Microprobe Mass Spectrometric Imaging of Tissue ................................ .......... 36 Tissue Sectioning ................................ ................................ ............................. 36 MALDI Matrix Application ................................ ................................ ................. 37 Mass Spe ctrometric Imaging Ionization Sources ................................ .............. 39 SIMS ................................ ................................ ................................ .......... 39 DESI ................................ ................................ ................................ .......... 39 MALDI ................................ ................................ ................................ ........ 40 Mass Spectrometric Imaging Mass Analyzers ................................ .................. 41 Time of flight for imaging ................................ ................................ ........... 41 Linear ion traps for MS imaging ................................ ................................ 42 Orbitraps for MS imaging ................................ ................................ ........... 44

PAGE 7

7 MS Image Generation ................................ ................................ ...................... 45 Overview of Dissertation ................................ ................................ ......................... 46 2 MALDI LINEAR ION TRAP TANDEM MS TECHNIQUES FOR THE ANALYSIS AND CHARACTERIZATION OF LIGNOCELLULOSC STANDARDS: POTE NTIAL FOR MALDI MS ANALYSIS OF INTACT TISSUE ............................. 58 Introduction ................................ ................................ ................................ ............. 58 Experimental ................................ ................................ ................................ ........... 59 Chemicals ................................ ................................ ................................ ......... 59 Preparation of MALDI Matrices ................................ ................................ ........ 59 MALDI Sample Plate Preparation ................................ ................................ ..... 60 Untreated and Holocellulose Tissue ................................ ................................ 60 Instrumentation ................................ ................................ ................................ 61 Results and Discus sion ................................ ................................ ........................... 62 Glucan ................................ ................................ ................................ ........... 62 Choosing the optimal matrix ................................ ................................ ....... 62 Opti mization of MALDI matrix additives ................................ ..................... 64 Tandem MS characterization of glucan ................................ .................. 64 Microcrystalline Cellulose (MCC) ................................ ................................ ..... 66 MALDI MS of MCC ................................ ................................ .................... 67 Tandem MS of MCC ................................ ................................ .................. 69 Xylan ................................ ................................ ................................ ................ 71 MS of Birch xylan ................................ ................................ ....................... 71 Tandem MS of Birch xylan ................................ ................................ ......... 72 Lignin ................................ ................................ ................................ ................ 73 MALDI MS of lignin ................................ ................................ .................... 73 MALDI MS analysis of lignin in untreated and holocellulose Populus tissue ................................ ................................ ................................ ...... 75 Summary ................................ ................................ ................................ ................ 75 glucan and MCC ................................ ................................ ........................... 76 Hemicellulose ................................ ................................ ................................ ... 77 Lignin ................................ ................................ ................................ ................ 77 3 DIRECT MATRIX ASSISTED L ASER DESORPTION/IONZATION MASS SPECTROMETRIC IMAGING OF CELLULOSE AND HEMICELLULOSE IN POPULUS TISSUE ................................ ................................ ................................ 98 Introduction ................................ ................................ ................................ ............. 98 Experimental ................................ ................................ ................................ ......... 100 Instrumentation and Data Analysis ................................ ................................ 100 MALDI Matrix ................................ ................................ ................................ .. 100 Wood Samples ................................ ................................ ............................... 100 MALDI Matrix Coating ................................ ................................ .................... 102 Results and Discussion ................................ ................................ ......................... 102 Pine Wood Samples ................................ ................................ ....................... 102 Tandem MS of Y oung Populus Stem Tissue ................................ .................. 103

PAGE 8

8 MS and Tandem MS Imaging of Young Populus Stem Tissue ....................... 104 Conclusions ................................ ................................ ................................ .......... 108 4 IMAGING OF POPULUS TISSUE BY FLUORESCENECE MICROSCOPY ........ 117 Overview ................................ ................................ ................................ ............... 117 Microscopy ................................ ................................ ................................ ..... 117 Fluorescence Microsc opy ................................ ................................ ............... 118 Experimental ................................ ................................ ................................ ......... 119 TBO Stain ................................ ................................ ................................ ....... 119 Fluorescence Stains ................................ ................................ ....................... 119 Sample Preparation and Instrumentation ................................ ....................... 120 Results and Discussion ................................ ................................ ......................... 121 TBO Stain ................................ ................................ ................................ ....... 121 CW Fluorescence Microscopy ................................ ................................ ........ 121 Acri dine Orange Fluorescence Microscopy ................................ .................... 124 Conclusions ................................ ................................ ................................ .......... 127 5 HIGH SPATIAL RESOLUTION IMAGING OF POPULUS TISSUE USING TOF SIMS ................................ ................................ ................................ ..................... 137 Overview ................................ ................................ ................................ ............... 137 ToF SIMS Microscope Im aging ................................ ................................ ...... 137 Ionization ................................ ................................ ................................ ........ 138 Mass Analysis and Detection ................................ ................................ ......... 139 Expe rimental ................................ ................................ ................................ ......... 140 Instrumentation ................................ ................................ ............................... 140 Sample Preparation ................................ ................................ ........................ 140 Imaging Experiments ................................ ................................ ...................... 141 Results and Discussion ................................ ................................ ......................... 142 Standards ................................ ................................ ................................ ....... 142 Populus Tissue ................................ ................................ ............................... 143 Positive ion mode ................................ ................................ ..................... 144 Comparison of ToF SIMS with MALDI LIT MS ................................ ........ 148 Negative Ion Mode ................................ ................................ ................... 149 Conclusions ................................ ................................ ................................ .......... 150 6 MULTIVARIATE ANALYSIS OF MALDI MS AND TOF SIMS IMAGING DATA ... 162 Overview ................................ ................................ ................................ ............... 162 Principle Component Analysis ................................ ................................ ........ 163 PCA of Lignocellulosic Material Standards ................................ ........................... 166 Software ................................ ................................ ................................ ......... 166 Data Preprocessing ................................ ................................ ........................ 166 PCA Results ................................ ................................ ................................ ... 168 PCA and Hierarchical Cluster Analysis of ToF SIMS Imaging Data ..................... 170 Hierarchical Clustering ................................ ................................ ................... 171

PAGE 9

9 Software and Data Preprocessing ................................ ................................ .. 172 Imaging PCA Results ................................ ................................ ..................... 172 Conclusions ................................ ................................ ................................ .......... 174 7 SUMMARY AND FUTURE DIRECTIONS ................................ ............................ 188 APPENDIX A DIFFICULTIES OF MALDI TIME OF FLIGHT ION MOBILITY SPECTROMETRY OF PLANT RELATED STANDARDS AND POPULUS TISSUE ................................ ................................ ................................ ................. 193 Overview ................................ ................................ ................................ ............... 193 Experimental ................................ ................................ ................................ ......... 193 Standard Analysis ................................ ................................ .......................... 193 Instrumentation and Traveling Wave Ion Mobility Parameters ....................... 193 Results and Discussion ................................ ................................ ......................... 194 LIST OF R EFERENCES ................................ ................................ ............................. 202 BIOGRAPHICAL SKETCH ................................ ................................ .......................... 211

PAGE 10

1 0 LIST OF TABLES Table page 1 1 Figures of merit for common mass analyzers ................................ ..................... 54 2 1 List of hy pothesized cellulose ion m/z values to monitor in wood tissue. ............ 97

PAGE 11

11 LIST OF FIGURES Figure page 1 1 Schematic of biomass conversion into ethanol.. ................................ ................. 49 1 2 Cartoon illustration of cellulose and starch . ................................ ....................... 50 1 3 Cartoon illustration of plant cell and zoomed in por tion of cell wall cross section. ................................ ................................ ................................ ............... 51 1 4 Domon and Costello nomenclature of polysaccharide fragmentation ................ 52 1 5 Cartoon schematic of the MALDI process. ................................ ......................... 53 1 6 Schematic of linear ion tr ap adapted from Schwartz et al. ................................ .. 55 1 7 Mathieu stability diagram of the LIT adapted from Douglas et al. ....................... 56 1 8 Workflow for MS imaging experiment. ................................ ................................ 57 2 1 Cartoon representation of cellulose microfibril . ................................ .................. 79 2 2 MALDI MS spectra of glucan standard using Different matrices. .................... 80 2 3 MS Spectra of glucan standard using DHB MALDI matrix with different salt additives.. ................................ ................................ ................................ ........... 81 2 4 MS 2 Spectrum of [Glc 4 +Na] + ................................ ................................ .............. 82 2 5 MS 2 spectra of different cationized glucan ions.. ................................ ............. 83 2 6 Optical images of MALDI sample spots. ................................ ............................. 84 2 7 Comparison of MCC MS by varying DHB concentrations. ................................ .. 85 2 8 Plot of MALDI matrix and cellulose ion intensity versus laser energy. ................ 86 2 9 MS spectrum of MCC using o ptimized experimental parameters. ...................... 87 2 10 MS 2 spectrum of m/z 1157. ................................ ................................ ................ 88 2 11 MS 2 MS 4 analysis of MCC. ................................ ................................ ................ 89 2 12 Common hemicellulose structures. ................................ ................................ ..... 90 2 13 MS spectra of Birch Xylan extract. ................................ ................................ ...... 91 2 14 MS 2 analyses of xylans. ................................ ................................ ...................... 92

PAGE 12

12 2 15 Common building blocks of lignin. ................................ ................................ ...... 93 2 16 MALDI and LDI MS analysis of Spruce lignin extract. ................................ ........ 94 2 17 MS 2 analysis of m/z 181. ................................ ................................ .................... 95 2 18 Optical and MALDI tandem MS image of untreated and holocellulose Populus tissue. ................................ ................................ ................................ ... 96 3 1 Optical image and average MS from Pine wood sample. ................................ 110 3 2 Comparison of spectra from different regions of tissue. ................................ ... 111 3 3 MS 2 of m/z 1319 from wood tissue ................................ ............................. 112 3 4 MS 2 of m/z from wood tissue ................................ ................................ 113 3 5 Optical and MS images of cellulose ion in Populus stem.. ............................... 114 3 6 Optical and MS images of hemicellulose ion in Populus stem .......................... 115 3 7 Optical and MS images of lignin ion in Populus stem. ................................ ...... 116 4 1 Jablonski Diagram representing energy level transitions involved in fluorescence emission. ................................ ................................ ..................... 129 4 2 Schematic of general fluorescence experimente ................................ ............. 130 4 3 Schematic of fluorescence microscopy. ................................ ........................... 131 4 4 Chemical structures of fluorescence stains used. ................................ ............. 132 4 5 TBO Stained Populus stem. ................................ ................................ ............. 133 4 6 CW stained and tandem MS images. ................................ ............................... 134 4 7 AO stained Populus stem. ................................ ................................ ................ 135 4 8 Comparison of AO, CW stain and MS image of same Populus wood stem. ..... 136 5 1 Schematic of Physical Electronic s TRIFT II ToF SIMS instrument .................. 152 5 2 Cartoon illustration representing secondary ionization. ................................ .... 153 5 3 Comparison of ToF SIMS imaging modes. ................................ ....................... 154 5 4 ToF SIMS spectrum of Birch Xylan Extract coated with 2 nm gold.. ................. 155 5 5 ToF SIMS spectra from different regions of wood tissue. ................................ 156

PAGE 13

13 5 6 Optical, fluorescence and positive ion ToF SIMS images of phloem and xylem of Populus wood stem. Images show l ocalization of different ions in different regions of wood tissue. ................................ ................................ ....... 157 5 7 Optical, fluorescence and positive ion ToF SIMS images of pith and xylem of Populus wood stem. ................................ ................................ ......................... 158 5 8 Comparison of different imag ing techniques. ................................ ................... 159 5 9 Optical, fluorescence and negative ion ToF SIMS images of phloem and xylem of Populus wood stem. ................................ ................................ ........... 160 5 10 Optical, fluorescence and negative ion ToF SIMS images of pith and xylem of Populus wood stem. ................................ ................................ ..................... 161 6 1 Graphical representation of PCA analysis. ................................ ....................... 176 6 2 I deal PCA plot of three pure standards. ................................ ............................ 177 6 3 Data matrix set up for PCA analysis. ................................ ................................ 178 6 4 PCA Scores plot of MCC, xylan and lignin analysis by MALDI MS. ................. 179 6 5 PC loadings plot compared with MALDI MS spectrum of lignin. ....................... 180 6 6 PC loadings plot compared with MALDI MS spectrum of MCC. ....................... 181 6 7 PC loadings plot compared with MALDI MS spectrum of Birch xylan. .............. 182 6 8 Illustration of hierarchical clustering adapted from Deininger et al. ................... 183 6 9 Fluorescence images compared with PCA images generated by plotting the positi ve and negative loadings of PC1 and PC2 of ToF SIMS imaging data.. .. 184 6 10 PC2 loading plot and corresponding images. ................................ ................... 185 6 11 PCA images of the pith an d xylem.. ................................ ................................ .. 186 6 12 PC1 loading plot and corresponding images. ................................ ................... 187 A 1 Comparison of MADLI MS of wood tissue from two different instruments. ....... 199 A 2 Ion mobility plot of LDI analysis of intact wood tissue sections. ........................ 200 A 3 Mass spectra from the different regions of the io n mobility plot from Figure A 2 ................................ ................................ ................................ ....................... 201

PAGE 14

14 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy MALDI TANDEM MASS SP ECTROMETRIC TE CH NIQUES FOR THE ANALY SIS AND IMAGING OF LIGNOCELL ULOSIC MATERIALS By Kyle Ann Lunsford August 2011 Chair: Richard A. Yost Major: Chemistry The conversion of cellulosic material to ethanol begins with a dilute acid pretreatment to remove hemicellulose and alter the lignin distribution. Despite successful conversion s spatial changes of chemical composition throughout pretreatment are not well characterized. Developing nondestructive analytical methods to visualize changes that occur in biomass throughout the pretreatment processes could help increase efficient of bio mass conversion into biofuels. MALDI mass spectrometric imaging (MS imaging ) in a powerful analytical technique that displays a distribution of target mol ecules in intact tissue. MALDI TOF (time of flight) MS is commonly used to characterize polysaccharide s, but tandem MS capabilities of a TOF are limited. The linear ion trap (LIT) mass analyzer offers tandem MS capabilities, which can provide structural information and monosaccharide composition of polysaccharides. A MALDI LIT tandem MS method was develop ed for to characterize th re e plant related standards, microcrystalline cellulose (MCC), Birch xylan and Spruce lignin. The standard analysis determined the ions that are formed from MALDI and provide characteristic fragmentation pathways of these standard compound classes MALDI MS

PAGE 15

15 imaging of intact Populus tissue section s illustrated a nearly even ion signal across the whole tissue. In contrast, tandem MS imaging was able to differentiate between isobaric ions to provide sensit ive and selective than single stage MS imaging of these compound classes The tandem MS images showed localization of cellulose in the secondary xylem and phloem fiber cells, which are both known to have thickened secondary cell walls. High correlation between MALDI tandem MS imaging fluorescence imaging and ToF SIMS imaging was observed In addition, ToF SIMS imaging provided a high spatial resolution, chemically selective images to characterize the material further. P rincipal component analysis (PCA) of MALDI MS analysis of standar ds and ToF SIMS imaging determined ions characteristic of lignified and non lignified tissue. In conclusion, this dissertation reports a comprehensive analysis of lignocellulosic material (LCM) using three different techniques, MALDI LIT ta ndem MS imaging fluorescence microscopy and ToF SIMS imaging. Combining the information obtained from these techniques provided a more complete analysis of the tissue sections. Furthermore the tools for analysis of LCM throughout a pretreatment analysis have been devel oped.

PAGE 16

16 CHAPTER 1 BACKGROUND Biofuel Research Alternative energy research, specifically biofuel research has gained attention in the recent years due to increasing energy consumption, global climate changes air pollution and the limited availability of petroleum 1 3 For approximately thirty years, the Department of Energy (DOE) has appointed research funds toward developing renewable, domestically produced energy sources. 3 Development has headed the research developing renewable transportation fuels prepared from biomass mater Biofuel is loosely defined and encompasses a wide range o f materials the most common type of biofuel is bioethanol. 4 Bio ethanol Traditionally, b ioe thanol is prepared by the fermentation of sugars from renewable sources, e.g., corn kernels and sugar cane ; however, there are difficulties to overcome for bioethanol to become a viable alternative energy resource For example, ethanol produces thirty percent less energy per gallon than gasoline used today, thus requiring larger volumes and larger storage containers f or fuel than what is used currently 5 Another disadvantage is that the amount of ethanol generated from current corn and sugar sources accounts for merely 10 % of the total energy demand in the US. Although farmers could devote more of their crops to bioethanol, this ultimately removes the produce from the food market causing an increas e in food prices. 6 To overcom e these limitations, scientists are developing method s or biomass (i.e.,

PAGE 17

17 plant materials left over after food process ing) into bioethanol Bioethanol derived from biomass is no different from bioethanol deri ved from traditional met hods. The advantage over traditional bioethanol is that plant material remaining after processing, such as corn stalks, whea t straw, and forest trimmings that is otherwise discarded can now be converted into bioethanol 6, 7 Prior to using lignocellulosic materials (LCMs), bio ethanol was commonly produced from readily available starch a n 1,4 linked polymer of glucose, in corn kernels ( or sucrose from other cr ops such as sugar cane) As Figure 1 1 illustrates conversion into ethanol starts with an enzymatic digest ion of starch into individual glucose molecules. The glucose is then subjected to fermentation by bacteria (e.g., Escherichia ) or fungi (e.g., Sacch aromyces cervisiae ), which converts glucose into ethanol and carbon dioxide. Although t he lignocellulosic ethanol is fermented by the same process the structural and chemical properties of the material inherently make the conversion into ethanol more difficult. Lignocellulosic Ethanol The difference between corn based and lignocellulosic ethanol is the composition of the starting material, as well as the polysaccharide that i s fermented i nto ethanol. LCMs are primarily composed of cellulose (40 50% of entire mass) hemicellulose (20 25%), lignin (20 25%), and extractives (5%) 8 Cellulose is a linear polymer of glucose linked together by glycosidic bonds. Due to the orientation of CH 2 OH group of the glucose monomer, strong hydrogen bonding occurs between the linear strands of cellulose, forming a secondary structure, ( Figure 1 2). The i nteractions between each strand of cellulose forms a strong, micro crystalline structure which increases resistance to enzymatic digestions 9

PAGE 18

18 Hemicellulose is an amorphous, heterogeneous polysaccharide that can vary between species of woods The m ost common hemicellulose s a re a linear, homogenous polysaccharide backbone (e.g., xylan) containing short sugar or acetate branches. 9 Sugars from both 6 carbon (cellulose) and 5 carbon ( hemicellulose ) sugars c an be fermented into fuels ; 10 h owever, the focus of this study is the conversion of cellulose into biofuel Lastly, lignin is a complex three dimensional polymer of lignols, and is not used for bio fuel production. Lign in is referred tree together, and makes cellulose in accessible for enzymatic digestion. 9 As mentioned previously, the method for producing corn based ethanol makes use of the easily accessible starch from corn kernels. Lignocellulosic ethanol uses cellulose, a structural isomer of starch, to convert into ethanol. Although starch and cellulose only differ in the linkage between glucose monomers, linkages allows for higher order inter actions (specifically, hydrogen bonding ) between each cellulose strand this generates microcrystalline microfibrils ( Figure 1 2 ) that d linked glucose. 9 In addition to the differen t linkages, the orga nization of the cellulose within biomass material increases the difficulty of conversion in to fuels. The current model for biomass organization is illustrated Figure 1 3. Briefly, hemicellulose coats the cellulose microfibrils, and l ignin rengthens cellulose that holds cellulose and hemicellulose together. 9 In order to disrupt the interactions between lignin, hemicellulose and cellulose, a pretreatment step prior to enzy matic digestions is necessary for the conversion of cellulose into ethanol. 11 Although the current pre t reat ment process es work,

PAGE 19

19 lignocellulosic ethanol remains a research focu s in o rder to improve the understanding and efficiency of pretreatments, helping to decrease the cost and make lignocellulosic ethanol a viable fuel source. Lignocellulosic Pretreatment The pretreatment e coating from cellulose, leaving behind greater surface area for enzymatic digestion of cellulose to occur. The first step in pretreatment is size reduction which breaks down the biomass into pieces that are compatible with the pretreatment step this often is performed by a chipping or grinding process. After size reduction, n umerous pretreatment steps have been proposed and explored, including steam explosion (autohydrolysis) 12 ammonia fiber explosion (AFEX), 13 acid hydrolysis, 14 organosolv pretreatment 15 and biological pretreatment. 2 The most common pretreatment method of LCMs is steam explosion. 16 Pieces of LCM (e.g., wood chips) are exposed to high pressure (~0.69 4.83 MPa), saturated steam (~160 260C) for seconds to several minutes, followed by exposure to transform ing the lignin and degrading the hemicellulose, which leaves the cellulose accessible for enzymatic digestion (i.e., en zymatic hydrolysis) 9 Studies have shown that steam explosion pretreated wood chips were 90% enzymatically hydrolyzed, compared to untreated wood chips that were only 15% enzymatically hydrolyzed. 17 Another common pretreatment method is acid hydrolysis, in which the LCM is treated with dilute acids (e.g., sulfuric or hydrochloric) at high temperatures. Dilute acid treatments are effective; however they introduce additional costs for chemicals a nd neutralization, as well as processing difficulties due to the corrosive nature of acids. 14

PAGE 20

20 Although the se pretreatment methods have been successful, all the processes still require size reducti on prior to pretreatment. Size reduction is limiting due to the high energy (and cost) need ed for chipping or grinding, especially for woody material Developing processes that make use of larger starting material could help to improve the overall cost and efficiency of producing lignocellulosic ethanol. In addition spatial changes in the chemical compositions between untreated and pretreated wood chips are not well characterized. 18 Monitoring these spatial changes in chemical composition could provide valuable information to help improve the efficiency and reduce the cost LCM conversion into biofuel Analytical Approaches for LCM Analysis In efforts to monitor the chemical changes throughout a pretreatment process, a variety of analytical approaches have been explored, 19 optical and fluorescence microscopy, 20 magnetic resonance imaging (MRI), 2 1 micro X ray computed tomography ( X Ray CT), 22 and confocal Raman imaging. 23, 24 Optical and fluoresce microscopy have been used to generate high spatial resolution images that visualize polysacch arides within plant cell walls. 25 MRI is used to determin e the moisture (i.e., water) content wit hin plant or wood tissues. 21 In addition X Ray CT provides high resolution images that provide insight into the density of the biomass material 22 Confocal R aman microsco py has been used to generate high resolution images of lignocellulosic tissue, based on scattering light that is characteristic of functional groups on the carbohydrates. 23, 24 Although th ese analytical techniques offer high spatial resolution images, the chemical specificity is often limited. A technique that offers structural analysis with increased molecular selectivity and sensitivity over existing

PAGE 21

21 techniques is mass spectrometry (MS) 26 applying MS to biomass analysis is the focus of this dissertation. Mass Spectrometry of Polysaccharides MS is a common analytical technique used to determine the presence of a compound in a sample and can be use for quantitative measurements. In MS the analyte is transferred into the gas phase, ionized and introduced into the mass spectrometer for mass analys is and detection. Mass analysis measurements are based on the mass to charge ratio s ( m/z ) of the ions generated ; thus if the charge of the ion is known, the mass o f the analyte can be determined; f urthermore, tandem MS can be implemented to elucidate ion structure. The major advantag es that mass spectrometry offers over other analytical techniques is the che mical selectivity the wide variety of analytes that can be analyzed and the ability to perform two or more stages of MS for structural analysis. 27 MS encompasses a wide variety of techniques characterized by the type of ionization source and /or the mass analyzer used which will be discussed in detail in this chapter. Early MS of Polysaccharides MS of polysaccharides has been performed for several decades, with application papers dating back to the 1970 s. 28 Early polysaccharide MS experiments were complex as they involved hydrolysis to break polysaccha rides down into smaller oligosaccharides derivatizations ( such as methylation, acetylation, trimethylsilulyation ) and separations ( such as gas chromatography GC), prior to MS analyses. 26,28, 29 The need for hydrolysis and separations limited MS of polysaccharides to di tri and tetra saccharides In ad dition, polysaccharide analyses were time consuming, in that hydrolysis and derivatizations w ere performed over several days. 29

PAGE 22

22 Although GC/MS is still used for polysaccharide analyses the need for hydrolysis limits the application to extracts or isolated compounds from tissues (i.e., not whole tissue sections) MS of po lysaccharides without prior hydrolysis started after the introduction of fast atom bombardment (FAB) ionization in 1981. 28, 30 FAB was developed to overcome the limitation that required samp les to be introduced into the gas phase prior to ionization. Briefly, analytes are mixed with a mat rix and the matrix is bombard ed with Ar atoms or ions at 2 8 keV to generate analyte ions 31 FAB became the ionization source of choice for oligosaccharide analyses, as it was able to ionize these compounds without hydrolysis or derivatization, as well as generate [M+H] + ions in positive ion mode and [M H] ions in negative ion mode. 30, 32 In 1984, electrospray ionization (ESI) was introduced as a soft ionization technique (i.e., limited in source fragmentation) to transfer large solution phase molecules in to intact gas phase ions 33 and ESI quickly became an ionization source use d for MS analysis of polysaccharides. 32 ESI generates ions by passing a solution through a needle under a high electric field resulting in an aerosol of charged droplets. These droplets undergo evaporation and columbic fission until the charge resides on the dissolved analyte. 32, 33 Despite the widespread application, native polysaccharides are not ionized easily using ESI due to the lack of an acidic or basic functional group, so native polysaccharides must be derivati zed by methylati on or acetylation, prior to ESI MS analysis. ESI also requires analytes to be in solution, which is often d ifficult for polysaccharides, specifically plant related materials, such as celluloses and hemicellulos es. In order to perform ESI on the insoluble materials, derivatization is necessary. 32

PAGE 23

23 The above ionization so urces are adequate for polysaccharide analyses, but ESI and FAB are not amendable for MS imaging applications. The i onization sources that are most commonly used for MS imaging and will be presented in this dissertation are matrix assisted laser desorption /ionization (MALDI) and secondary ion mass spectrometry (SIMS ) and will be discussed in the MS imaging section of Chapter 1 MALDI MS offers several advantages over FAB MS and ESI MS for t he analysis of polysaccharides: no sample derivatization is needed (i.e., the p olysaccharide is analyzed in it s native form ), MALDI MS does not require the analyte to be in solution prior to ionization, and MALDI is a soft ionization source, so less in source fragmentation and higher mo lecular weight ions (t hus more intact ions) are observed compared to FAB MS In addition, MALDI offers increased ionization efficiency of carbohydrates over FAB, which increases the sensitivity of the analysis by approximately 1 0 100 ti mes. 34 Furthermore MALDI typically generates singly charged ions, as opposed to multiply charged ions that are formed in ESI which helps reduce the complexity of the mass spectrum. Tandem MS of Polysaccharides Tandem MS is a techn ique in which mass selected ions are introduced into second (or n) stages of mass analysis. Following isolation of the ion dissociation is induced using a dissociation techniques include electron transfer dissociation (ETD), electron capture dissociation (ECD), infrared multiple photon dissociation (IRMPD) and collision induced dissociation (CID) 35 the most commonly used dissociation techniques for polysaccharides. 27 For CID, i ons of interest (precursor ions) undergo energetic collision s with a neutral background gas, dissociating the precursor ions into product ions which are

PAGE 24

24 then mass analyzed. The pr oduct ions observed in the tandem mass spectrum help to identi fy the precursor ion structure, making tandem MS an indispensible tool for peptide polysaccharide ( most commonly N linked glycan s) structural analysis 26 Early tandem MS experiments focused on glycoconjugate structural analysis which present s a more complex analysis than peptides due to a high order of branching Fragmentation of highly branched glycoconjugates generate more complex product ion spectra were more complex than peptide product ion spectra. The groundwork for tandem MS analysis of polysaccharides was performed using source fragmentation and low ionization efficiencies. 32 These experiments determined that low energy CID of polysaccharides favorably broke glycosidic bonds over cross r ing cleavages In the late 1980s, a polysaccharide CID fragmentation nomenclature was introduced by Domon and Costello 36 and is still used t oday (Figure 1 4) Furthermore, tandem MS studies has reveal ed polysaccharide structural information such as stereochemistry of individual sugar residues, 37 linkage positions, 38 as well as branching structures. 26 36 The introduction of softer ionization sources, e.g., MALDI and ESI, has made tandem MS necessary for most polysaccharide analyses since in source fragmentation is reduced P olysaccharides are typically composed of similar sugar monomers so numerous structural isomers possible for each nominal m/z Tandem MS can determine basic polysaccharide structure (e.g., linear or branched) and can potentially distinguish the linkages between each sugar. Laine et al. demonstrated that isomeric polysaccharides composed of the same monosaccharides but branched differently are easily distinguished using tandem MS. 38 Furthermore, it has been determined that

PAGE 25

25 different ions, e.g., [M+H] + versus [M+Na] + fragment differently during low energy CID, which could help to increase structural information about a polysaccharide. 34 Instrumentation Matrix Assisted Laser Desorption/Ionization MALDI was introduced in the late 1980 s and enabled the ionization of analytes (up to ~100,000 Da) in their native form, without the need for separations or derivatizations. 39, 40 Some of the first reports of MALDI included the analysis of a phytochemical tetrasaccharide, stachyose, showing the viability of MALDI for the analysis of native polysaccharides. 41 Since the late 1980 s MALDI of carbohydrates has become the method of choice for the analyses of branched glycans, as well as linear polysaccharides. 34 Ionization m echanism The ionization mechanism of MA LDI is illustrated in Figure 1 5. MALDI generates ions by the interaction between a pulsed las er, a MALDI matrix and the analyte. The MALDI matrix solution is applied atop the analyte and allowed to co crystallize the laser is then fi red at the MALDI matri x/analyte sample The MALDI matrix, typically an organic acid, is chosen based on the absorp tivity at the wavelength of light emitted by the laser. 39 The MALDI matrix absorbs energy from the laser, which causes a rapid heating that desorbs both MALDI matrix and analyte into the gas phase, i.e., the positive/negative ions exist simultaneously within the MALDI plume and io nization is believed to occur through g a s phase reactions between excited molecule s and ions; h owever, the exact mechanism is still up for debate D ifferent MALDI ionization theories are discussed in detail in several review papers 42 48

PAGE 26

26 MALDI matrix c onsiderations Prior to any MALDI analysis, a MALDI matrix must be chosen. As mentioned previously, the MALDI matrix absorb s the laser energy for desorption and subsequent ionization; thus, the MALDI matrix must have a strong absorpt ivity at the wavelength of the MAL DI laser. Commonly, pulsed UV lasers, such as N 2 (337 nm) or frequency tripled Nd:YAG (355 nm), are used, but IR MALDI has also been reported. 49 Theoretically, any compound that absorbs at the laser wavelength could serve as a MALDI matrix, so exploring all the option s would be nearly impossible. MALDI matrices are typically chosen based on the characteristics of the analyte compounds. 44 For example, a MALDI matrix that works well for polysaccharides might not work well for protein analysis. Several MALDI matrices have been reported for the analysis of polysaccharides, but 2,5 cyannohydroxycinnamic acid and sin n ip inic acid are among the most commonly used and reported. 34 After the MALDI matrix is chosen other parameters to consider include solvent system, matrix additives and MALDI matrix to analyte ratio. The solvent sy stem of the MALDI matrix and the analyte p lays an important role in MALDI. The MALDI matrix and analyte are hypothesized to co cry stallize, thus it is important to use a solvent that allows for a saturated MALDI matrix solution, and is capable of dissolvin g the analyte 43 In addition, the solvent should have high vapor pressure to rapidly evaporate and produce small crystals, as this results in a more uniform coating of matrix Typical solvents for MALDI include methanol, ethanol, water, acetonitrile and mixtures of these MALDI typically produces [M+H] + ions, which may not be the most desirable for analysis (e.g., [M+H] + ions could provide uninformative fragmentation durin g ta ndem MS

PAGE 27

27 experiments). Salts or other compounds can be added to the MALDI matrix solution to promote cationization that wo uld otherwise not be observed. Since sodiated specie s can fragment differently than protonated ions, a sodium salt ( e.g., sodium acetat e) could be added to the MALDI matrix solution to generate more [M+Na] + ions than [M+H] + ions. Another important parameter for MALDI analysis is the matrix to analyte molar ratio. Early experiments showed that there must be a large excess of MALDI matrix m olecule present for the best ionization to occur 44 Common matrix to analyte ratios are 500:1 10,000: 1. 27, 43 Secondary Ion Mass Spectrometry SIMS is a surface analysis technique similar to FAB that has been adapted for biological analyses. 50, 51 For SIMS, a primary ion beam is focused onto a surface sample, and the kinetic energ y of the primary ions is transferred to the surface atoms and molecules through 52 The energy of surface atoms and molecules exceed s the surface binding energy and are desorbed into the gas ph ase. Most of atoms/molecules desorbed into the gas phase are neutral; however, some ions (secondary ions) are generated and analyzed using MS. 53 Typical primary ion beams include Ga + In + Au + Ar + Xe + and Cs + 54 Two types of SIMS can be performed based on the primary ion doses. Static SIMS is a less invasive technique that only penetrates ~0.1% of the surface monolayer static SIMS sources have low primary ion doses (< 10 13 ions per cm 2 ) and low flux (10 pA 5 nA). 53 Dynamic SIMS is more invasive and penetrates ~ 40 nm into the sample surface dynamic SIMS sources have high primary ion doses (> 10 13 ions per cm 2 ) with high flux es (A). 55 Most SIMS experiments are performed under the static SIMS limits

PAGE 28

28 A difficulty of SIMS is the rapid, non linear decrease of secondary ion yield with increasing m/z value, which limits SIMS experiments to small molecule or fragment analyses. 54 Recently, polyatomic primary ion sources (as opp os ed to atomic ion sources) have been shown to improve the secondary ion emission (SIE) yield, increas ing the sensitivity of SIMS experiments and decreasing the sample damage. Some of the polyatomic primary ion sources that have been reported include SF 5 + SF 6 glycerol, Au 3 + Au 4 + and C 60 + 52, 55 Mass Analyzers Once ions are formed, the ions are guided toward a mass analyzer. Many different mass analyzers are available on commercial or homebuilt instruments and each mass analyzer offers advantages and disadvantages over the other. The most comm on mass analyzers include magnetic and electric sectors, Fourier transform ion cyclotron resonance (FT ICR), quadrupoles, triple quadrupoles, three dimensional quadrupole ion trap, time of flight (ToF), linear ion trap (LIT) and orbitraps Magnetic and ele ctric sectors have become less poplar due to the high cost and limited tandem MS abilities. FT ICRs, ToF s and orbitraps offer high mass resolution but FT ICR requires high vacuum and a cryogenic magnetic that is costly to purchase and maintain. O rbitraps are a new technology resulting in similar re solving power and sensitivity to FT ICR; however, initial costs are high ToF are a cheaper alternative for high resolving power, but the tandem MS capabilities are limited Ion traps, three dimensional and lin ear offer the best tandem MS capabilities, but are limited in mass range and mass resolution. 27,28, 35

PAGE 29

29 Figures of m erit In order to compare different mass analyzers, there are figures of merit that are considered, including mass accuracy, resolving power, m/z range, scan speeds, tandem MS capabilities and cost. Table 1 1 compares important figures of merit between typical mass analyzers. 54 56, 57 Mass accuracy is describe as the error in the m/z measurement, typically express in parts per million (ppm). The mass accuracy is determined by E quation 1 1 and considered accurately know n High mass accuracy mass analyzers are necessary to determine exact elemental composition and such information is especially useful for database searches. The greater the mass accuracy, the fewer possible compounds allowing for more reliable ion identification. Compared with the other mass analyzers, FT ICR mass spectrometers and orbitraps provide best mass accuracy ( 1 1) Resolving power is a property of the mass analyzer and is determined by dividing the m/z value of an ion ( ) by the m/z values of ions that can be separated ( ), shown E quation 1 2. A higher resolving power results in greater the sep aration between two ions at a certain m/z value 27 which can distinguish between two ions with the same nominal m/z without the need for tandem MS. 35 ( 1 2 ) Linear ion t raps LITs became widely used in the early 2000 s for their tandem MS capabilities. LITs are composed of four hyperbolic, parallel rods divided into three sections creating a

PAGE 30

30 quadrupolar electric fiel d, as illustrated in Figure 1 6 58 The ions are trapped in two dimensions in the center section of the LIT u sing radial RF vol tages on the parallel rods and DC biased plates 58 60 Stability governed by the solution to the reduced Mathieu Equations ( Equation s 1 3 and 1 4 ) 58, 60, 61 ( 1 3 ) ( 1 4 ) w here is the charge of the ion, is the DC potential on the rods, is the mass of the ion, is the radius of the ion trap and is the angular frequency of the RF voltage. Mass analysis in a LIT can be performed several ways ; a common method is mass selective instability scanning with resonance ejection with a q value set to 59, 62 Mass only mode which is achieved by setting the DC c omponent, to 0V and the trap can be described only in terms of the Mathieu parameter, shown in the Mathieu Stability diagram ( Figure 1 7 ) 59 Once ions are insi de the trap, an RF excitation vol tage applied across an opposite pair of rods is increased When the RF frequency is equal to the secular frequency of an ion at particular m/z the ion is excited and axially ejected through slits in these rods. A mass spectrum is recorded by scanning through the oscillation frequenc ies to eject and detect ions step wise from the trap 59, 62 The LIT s advantage over other MS imaging mass analyzers is the tandem MS capabilities. Tandem MS experiments in a LIT are performed tandem in time, 63 so there

PAGE 31

31 is no limitation on the stages of MS that can be performed (other than the number of ions within the trap ). Disadv an ta g es of the LIT compare to other MS imaging mass analyzers include limited mass range ( vs. ToF) and limited mass spectral resolution ( vs. FT ICR and orbitrap). C ommercial LIT instruments typically have an upper mass range of m/z 4000 but studies have s hown that this can be extended. 64 Under standard parameters, the LIT offer s unit resolu tion. S lower scan speeds can be used to obtain higher mass resolution (e.g., zoom scan and ultra zoom scan); however, this is not practical for an entire imaging experiment. The LIT has been interfaced with the high resolution orbitrap, and its use for imaging has been discussed. 65 Tandem MS with LIT LITs perform tandem in ti me MS experiments, which is adva ntages since they can perform MS n stages of mass spectrometry. 63 Tandem in time denotes that the processes involved in tandem MS (ion accumulation, ion selection, dissociation and mass analyses) occur in the same volume, sequential ly in time This process can theoretically be repeated n times, but is often limited by ion abundance and dissociation efficiency. 35 T andem MS in a LIT ion trap involves a user set m/z isolation w indow, where only ions within the m/z window are stable inside the LIT Typically, an isolation width of 1.2 1.5 amu is used to trap the precursor ion efficiently without the presence of unwanted ions. After trappin g, the ions are excited using an ex t his incr eases the kinetic energy of ions and induces collisions with the neutral background gas, typically He The collisions cause dissociation typically along the lowest energy pathway, and is termed low energy CID. Low energy CID is one of the most comm on ly used dissociation technique specifically in LITs. 27

PAGE 32

32 MS/MS or MS 2 spectra are recorded by scanning product ion out of the trap following di ssociation of the precursor ion this step can be performed multiple times within the LIT For example, after the first precursor ion is fragmented, one of the product ions can be mass selected and CID is performed again The product ions are then scanned o ut for a MS 3 spectrum. The number of stages is typically limited by the number of ions capable of being stored initially within the trap, as well as the dissociation efficiency. A disadvantage of tandem in As sho wn in Figure 1 7, if precursor ions are stored at for CID, product ions that fall at a value gre ater than the low mass cutoff ( ) will not be stored. 66 Time of f light ToF mass analyzers were introduced in 1946 and first commercial ized in 1955 for gas chromatography/MS ( GC/MS ) analys e s 27 The advantage of ToF over other mass analyzers during that time was the spectral acquisition rate ToF mass a nalyzers are relatively simple, as they are composed of a field free region, also called a drift tube, and a detector. The mass analyzer works based on the principle that ions introduce d into the field free region at the same time and with the same kinetic energy will be separated based on their m/z values (i.e., lighter ions will travel more rapidly and heavier ions will travel more slowly ). 67 After ionization, ions are accelerated toward the field free region. The energy ( ) imparted upon an ion of mass , through a voltage, is converted into the kinetic energy ( ) and expressed as ( 1 5 )

PAGE 33

33 Where is the number of electron charges, ( assuming zero ki netic energy prior acceleration). Equation 1 5 is rearranged to solve for velocity. ( 1 6 ) The velocity is determined by dividing the distance the ion travels, by the time it takes the ion to travel that distance Through a series of substitution and rearrangements, the ToF equation reduces to ( 1 7 ) Thus, the ToF of an ion is proportional to the square root of the m/z value In addition to both acceleration time , and ionization time, must be accounted for the total time of flight, ( 1 8 ) Early ToF instruments suffered from poor mass resolution because the speed of the detector electronics was not adequate to distinguish small differences in flight times. The electronics were improved in mid 1990s which allow ed for high resolution ToF mass analysis 67 Furthermore e arly ToF mass analyzers had a linear geometry, i.e., the ionization source, drift tube and detector were aligned on the same axis. A r eflector geometry was later introduced to ToF mass analyzer s to improve the mass resolution; i ons travel through a refle ctor until they reach zero kin e tic energy, at which time they reverse directions. Ions with higher kinetic energy fly deeper into the reflector section than less energetic ions, so the reflector serves to correct the kinetic energy distribution of the ions this dramatically increases t he resolving power of ToF 27

PAGE 34

34 Since ToF requires all ions to enter the drift tube simultaneously (with the same kinetic energy), a pulsed ionization source is necessary Although ToF provides high mass resolution the tandem MS capabilities, which are need for polysaccharide analyses, are limited and described in detail below Tandem MS with ToF In contrast to LITs, tandem MS experiments using a ToF analyzer is performed tandem in space. For a tandem in space an alysis, a precursor ion is mass selected, moved into a different region of space for dissociation and the product ions are moved to another region of space for analysis. The most common tandem in space mass spectrometer is the triple quadrupole, which is c omposed of a mass analyzer quadrupole ( Q 1 ) a collision cell (q 2 ) and a product ion mass analyzer ( Q 3 ). 63 T andem mass spectrometers using a ToF are typically hy b r id instruments, meaning two different mass analyzers a r e used tandem in space (e.g., quadrupole ToF, QIT ToF) ; i n the se instruments, the quadrupole and QIT perform MS1 and the ToF performs MS 2 More recently, ToF ToF instrument s have been introduced with the primary application to rapid protein identification for proteomic analyses. 35 ToF ToF instruments are composed of an ion source (most commonly MALDI), directly followed by one ToF, a gated precursor ion selector, a collision cell, a second ToF and a detector T he first ToF is used to separate the packets of ions into approximately 1 3 m/z value groups for passage into the collision cell using a timed gate. The se precursor ions are decelerated (to a user determined value ) to reduce the ion s kinetic energy prior to entering the collision cell Th e gas within the collision cell acts as the collision gas to induce CID The energy of the collisions i s controlled by electric fields, the mass of the collision gas (the mass of the collision gas alters the ene rgy imparted during collisions) and the

PAGE 35

35 pressure of the collision gas. The pressure of the collision cell is typically ~10 4 10 5 Torr and is typically composed of N 2 He, N e Ar, Kr or Xe. 35 After exiting the co llision cell, the ions are reaccelerated for the second ToF analysis, typically a reflectron ToF. ToF ToF analyses offer advant ages, such fast analysis times that allow for high throughput analyses. Also, ToF ToF allows for control over the CID energy an d furthermore allows for high energy CID, which is unavailable for ion trap instruments. Unfortunately, precursor ion resolution is typically poor and the kinetic energy spread of the product ions produced by CID decreases the resolving power of the second ToF However, the second ToF is still able to provide high mass accuracies ( although lower than a single stage ToF experiment ) 35 Since, both tandem in time and tandem in space MS methods offer advantages depending on the analysis using both methods should provide the most comprehensive analysis of polysaccharides. Mass Spectrometric Imaging MS imaging is a powerful analytical technique that combines the chemical selectivity of MS with two dimensional spat ial information of a sample surface. 54 MS imaging can be performed in two different modes, microscope imaging and microprobe imaging. For microscope MS imaging the spatial position of ions (relative to each ot her) generated from a surface is conserved and detected by a position sensitive detector. The intensity of the ions is then plotted versus the spatial position at which they were detected to generate the MS image Early microscope MS imaging was performed using ion optical collection system coupled to a SIMS ionization source 68, 69 For microprobe MS imaging, the sample is mounted on an x y stage and the sample (or source) is moved to analyze the region of interest. The ion intensities from the mass spectra are recorded at each analytical region and plotted versus the position to generat e the MS

PAGE 36

36 image. 70 Early microprobe imaging experiments were performed using laser microprobe mass analysis (LAMMA), which coupled laser desorption ionization (LDI) to a ToF mass analyzer. 70, 71 Microprobe Mass Spectrometric Imaging of Tissue The field of microprobe MS imaging has rapidly expanded with the introduction of new ionization sources capable of transferring large biomolecules in to the gas phase. A general workflow for a MALDI microprobe MS imaging of tissue (Figure 1 8) starts with slic ing tissue sections into thin ( ~10 20 m thick ) sections. The tissue is then mounted onto a microscope slide and coated with MALDI matrix. The slide is introduced int o the instrument and the plate is rastered underneath the laser beam at a user determine d raster step size (~50 100 m). A mass spectrum is recorded at each laser stop an d ion intensities are plotted versus the positi on to generate an MS image. 54 The spatial resolution of the MS image is dete rmined by the laser spot size discussed below and r aster step size Tissue Sectioning The quality of the MS image obtained is highly dependent upon with the quality of the sample preparation. 72 There are several considerations to address during the tissue sectioning to ensure high quality MS images. For MALDI MS imaging it is important that the sample surface is free of contaminant s that could ionize easily and compete with analy t e ionization, resulting in analyte ion suppression. Tissue s prepared for optical and fluorescence microscopy are routinely embedded in an optimal cutting temperature (OCT) compound for rigidity during slici ng; however, these embedding media can smear across the sample surface and compromise the chemical images obtained. 54 Instead animal tissue is flash frozen and cryosectioned to obtain thin sections without usi ng

PAGE 37

37 embedding media for MS imaging experiments In addition to embedding media, common surface contaminants are introduced from oils on the cryotome blade. 72 It is important to wash the sectioning blade with organic solvent thoroughly prior to sectioning tissues for analysis. Another common technique used to remove ion interferences is washing. 72 Washing the tissue helps to remove contamination that can occ ur during sectioning tissue and excess salts. Animal tissue s typically have high concentrations of natural cations, such as sodium and potassium. Excess salts can interfere with adequate MALDI matrix crystallization, thus reducing the ionization efficiency from atop the tissue section. 73 H igh abundance of salts can also compete for charges during the ionization process, thus reducing the ionization efficiency of analyte compounds. For MALDI MS imaging m ethanol and ethan ol are typ ical washing solvents; however, other solvents or detergents can be used based on the analyte. 54 The limitation of washes is the solubility of the analyte, as it is important that the analyte is not washed away. After the tissue section is mounted onto the glass slide and the appropriate washes are performed, the tissue is coated with MALDI matrix. MALDI Matrix Application The MALDI matrix choice, solvent choice and the application method are all im portant parameters to consider for optimal MS imaging experiments. It is important that the MALDI matrix/analyte co crystals are smaller than the laser spot size to obtain the best possible spatial resolution. 54 As discussed previously in this chapter, the MALDI matrix is chosen based on the classification of analytes, but some matrice s ( cyannohydroxycinnamic acid CHCA), generate smaller crystals than other matrices, ( 2,4 dihydroxybenzoic acid DHB) If both matrices result in efficient ionization of the

PAGE 38

38 analyte it is be beneficial to use CHCA over DHB to generate smaller, more uniform crystals. Another factor that affects M ALDI mat rix crystal size is the solvent. The MALDI matrix solvent is determined based on matrix and analyte solubility Since MALDI involves the co crystallization of the matrix and analyte, it is important that the solvent us ed can dissolve the analyte. T he rate of evaporation of the solvent is also important for generating uniform MALDI matrix crystals. More rapid evaporation g enerates smaller more uniform MALDI matrix crystals that are ideal for MS imaging experiments The most commonly used solvents include a water methanol mixture (30:70, 5 0:50, or 70:30) or organic solvent mixed with 0.1% trifluoroacetic acid (TFA) in DI water. Typical organic solvents used are acetonitrile (ACN), methanol, ethanol, acetone and chloroform. 72 The MALDI matrix coatin g method perhaps has the largest effect on the MS image quality, 72 as it is crucial that the MALDI matrix is applied evenly over the entire tissue section. Some the methods used for coating include electrospray dep osition, neub u lization, artistic airbrushing and ink jet printing. 74 Electrospray deposition generates uniform coating of small crystals; however, electrospray deposition does not work well on insulating samp le mounts, such as glass slides A Meinhard nebulizer can be used to generate small droplets of MALDI matrix to coat the tissue. This technique is discussed i n more detail in the experimental sections of C hapter 2. Despite variety of reported techniques for M ALDI matrix deposition, it is important to note that MALDI matrix coating is more of an art than a science ; d etermining the exact amount of matrix to apply to could vary from sample to sample Typically, coating with a nebulizer uses

PAGE 39

39 ~ 8 mL of MALDI matrix solution but it is possible to need only 6 mL or as much as 12 mL, depending on the sample. Mass Spectrometric Imaging Ionization Sources MS imaging experiments are performed using variety of different ionization sources, such as MALDI, SIMS and desorption electrospray ionization (DESI) and each method offers advantages depending on the analyte and sample type. 54 SIMS SIMS is most commonly combined with ToF, referred to as ToF SIMS More specifically, t he ToF SIMS ion microscope is a unique instrum ent developed for microscope MS imaging and is discussed in more detail in Chapter 5. An advantage of SIMS ionization source, is that no sample preparation is needed (e.g., no matrix is need) for tissue section analysis and the primary ion source diameter is small (~ 1 m), thus, high spatial resolution images are generated 75, 76 Previously, liquid metal ion gun (LMIG) Ga + or In + sources were commonly employed for imaging applications 75 but have recently been replaced with polyatomic primary ion sources, e.g., Au 3 + Au 4 + and C 60 + The polyatomic primary ion sourc es have shown to improve the poor desorptio n efficiency of the previous LMIG. 55 P olyatomic primary ion sources have also increased the applications of ToF SIMS MS imaging for analysis o f large compounds, such as lipids in tissue sections ( ~ m/z 600 800). 54 DESI DESI is an atmospheric pressure soft ionization source developed in 2004. 77 DESI generates ions by aiming an electrospray toward a sample surface at a particular angle. T he charged droplets from the electrospray interact with the sample causing desorption/ionization of the analyte 78 Soon after its introduction, DESI was use d as an

PAGE 40

40 MS imaging ionization source by mou nt ing the sample on an XYZ stage and moving the sample relative to the station ary DESI probe generating a 9 x 19 pixel image. 79 A unique feature of DESI imaging is that the ionization source is continuous (as opposed to pulsed like SIMS and MA LDI ) the sample plate is typically rastered at a velocity of 200 m per second and mass spectra are recorded every 0.67 seconds. 78 A major advantage of DESI over other ionization techniques (e.g., MALDI and SIMS) for MS imaging is that little to no matrix or sample preparation is required, i.e., after slicing, the tissue can be directly analyzed using DESI MS. This also helps to eliminate irreproducibility of ion signals introduced by inconsistent (bu t n ecessary) sample preparation, especially for MALDI. Another advantage is that the sample is probed at ambient pressure. Disadvantages of DESI include poor spatial resolution and limited analyte s Since DESI uses solution spray of droplets it is difficult to determine a aser or a primary ion beam, but it is known that the DESI spray is larger (~ 200 m) than a MALDI laser (~75 m) or primary ion beam (< 1 m), and as a result, DESI offers lower spatial resolution. 78 DESI is also limited to lower mass ions, but has recently been effective for lipid imaging. 80 MALDI MALDI has perhaps been the most widely used ionization source f or MS imaging experiments due to capabilities of ionizing intact biomolecules, including both small molecules (for drug and metabolomics analys e s ) and large molecules (for protein analysis). The spatial resolu tion offered by MALDI MS imaging is still less than ToF SIMS imaging but improvements have been made and the spatial resolution of MALDI MS images has been reported down to 10 m. 54

PAGE 41

41 Mass Spectrometric Imaging Mass Analyze rs The majority of MS imaging experiments are p erformed using a ToF mass analyzer, as they offer higher sensitivity, higher mass resolution and higher repetition rates over quadrupole and ion trap mass analyzers. 76 In addition, the pulsed nature of MS imaging i onization sources such as MALDI and SIMS make coupling to ToF mass analyzers easy. However, the disadvantage of ToF analysis are limited tandem MS capabilities discussed previously ; this is perhaps less important for SIMS analyses due to the high degree of in source fragmentation 75, 76 Time of flight for i maging ToF imaging experiments are performed either using linear or orthogon al extraction ToF L inear extraction experiments accelerate the ions directly into the mass analyzer. For orthogonal acceleration, the ions are accelerated orthogonally to the direction of mass analysis. L inear extraction ToF is more commonly employed for MALDI imaging applications due to the higher sensitivity offered To obtain the highest mass resolution, reflectron ToFs are often employed. It has also been shown that including a short time delay after MALDI prior to ion extract ion into the ToF reduce s kinetic energy broadening through collision al coo ling. Reducing the kinetic energy distribution ultimately improves the mass resolution and sensitivity of the experiment. 81 Delayed extraction is not necessary for SIMS experiments, as the distribution of kinetic energy of ions generated by SIMS is inherently narrow. 82 Another parameter for MALDI MS imaging experiments to consider is the number of laser shots needed to obtain one mass spectra. For a MALDI ToF imaging experiment, typically 100 200 laser shots are used at each laser stop on the tissue ;

PAGE 42

42 thus a large number of laser shots is used for one imaging experiment C onsider an MS imaging experiment for a 1 mm x 1 mm square (1000 m x 1000 m). U sing 100 laser shots per stop and 50 m raster step size there are 20 laser stops across the square and 20 laser stops down the square, resulting in 400 laser stops for the experiment. If 100 laser shots and eat laser stop, 40,000 laser shots are used to image the image the 1 mm square. In order to make the experiments feasible, a high repetition laser with a long lifetime along with high speed data acquisition is needed For these reasons, a solid state lase r (Nd:YAG) is often used for MALDI ToF experiments. In addition, the sample surface must be considered for ToF experiments To perform ToF analyses, ions must be generated from the surface with the same kinetic energy (mass resolution is lowered as the kinetic energy distribution is broadened) Thus, samples must be prepared on conductive surfaces such as stainless steel plates or indium tin oxide (ITO) coated glass slides C onductive surface s d o not limit the analyses, but this could cause complications in sample preparation for samples that do not adhere directly to the conductive surface. Linear ion t raps for MS imaging Due to the complexity of intact biological tissues, the need for MS imagi ng coupled with tandem MS has been recognized, and a MALDI source was coupled to a quadruple ion trap 83 and later a linear ion trap 84 for the analysis of intact tissue sections Due to the large m/z range offered, proteins were often the compounds of interest for MALDI ToF MS imaging analyses In contrast, the limited mass range of ion traps introduced a new area of imaging small molecules, particularly lipids and small drug molecules and their metabolites for MALDI ion trap MS imaging 84

PAGE 43

43 An advantage LITs offer over ToFs is the cap ability for tandem MS experiments. Since no separations are performed prior to MS imaging analysis, tandem MS helps to reduce spectral complexity, confirm identity of ion s and aid in unknown ion identification by isolating and dissociating analyte ion s Another advantage of the LIT is the small amount of laser shots needed to generate a mass spectrum, c ompared to T oF mass analyzers. For a LIT MS imaging experiment, typically 1 3 laser shots are required at each laser stop (compared to the 100 200) for ToF experiments. Thus for the same 1 mm square, using 50 m raster step sizes and 3 laser shots per stop, only 1200 laser shots are needed (compared to the 40,000 for ToF). Furthermore, the 1200 laser shots are capable of generating MS, MS 2 or even MS 3 images. In addition to prolonging the lifetime of the laser, using fewer laser shots allows for repeat experiments of the same tissue section. The high number of laser shots needed for MALDI ToF typically allow s for only one analysis per tissue section. It has been shown that s everal MALDI LIT MS imaging experiments (up to 30) can be performed on the same section of tissue. Thus, after collecting MS spectra, subsequent scans can record MS 2 or MS 3 spectra of different analytes for more selective analyses. 85 It is also important to note the ion source differences between the MALDI ToF MS and MALDI LIT MS MALDI ToF MS imaging experiments are typically performed at high vacuum (~10 6 Torr ), which requires samples to be completely dry prior to analysis. Since biological tissues are primarily composed of water, it can take approximately 2 hours to dry a 10 m thick sample. In addition, MALDI at higher vacuum (lower pressure) often induces in source fragmentation, particularly with phospholipids 84 It has been shown that increasing the pressure of the MALDI source allows for collisional

PAGE 44

44 cooling, which reduces the in source fragmentation observed. 86 The MALDI LIT source is operated at intermediate pressure, typically 10 2 to 1 Torr, thus less in source fragmentation is observed for these experiments. Another advantage of MADLI LIT MS imaging is that a conductive surface is not needed for sample mounting 85 Instead, samples can be mounted on common microscope slides used for other techniques, such as histological and fluorescence staining. Furthermore, the se common glass microscope slides are less expensive than the conductive ITO coated slides and stainless steel slides. Stainless steel sample plates must be washed and re used; however, washing the surface m ight not be adequate to remove all contaminants f rom the surface, so cross contamination could be observed for frequently used sample plates. Instead, new glass slides can be used for each sample for a n unadulterated sample surface and allowing samples for sample to be archived for future analyses. Orbitraps for MS imaging The n ewest mass analyzer used for MS imaging is the orbitrap which was introduced in 2000 87 and coupled to an ESI source in 2003. 88 The orbitrap was designed as a more practical alternative to FT ICR, i.e., offer the similar high mass resolution as FT ICR without the need for a cry ogenically coole d, superconducting electromagnet. Orbitraps use o rbital trapping fields to trap the ions. While ions are stored in the trap, they undergo axial oscillation, which are dependent on the m/z of the ion. The axial oscillations are measured using a Fourier transform of the image current detected from the motion of the ions yielding mass resolving p ower around 150,000. 88 The orbitrap has been integrated into a LIT instrument design for a hybrid linear ion trap/orbitr ap mass analysis. The advantage of combining these two mass analyzers

PAGE 45

45 is that the tandem MS cap abilities of the linear ion trap can be used prior to high resolution mass analysis by the orbitrap. The high resolution of the orbitrap is useful in distinguishing between isobar ic ions such as two different lipids at the same nominal m/z value 65 It is evident that the orbitrap offers advantages of both tandem MS and high resolution imaging; however the technology is still new and the applications are starting to be realized MS Image Generation MS imaging experiments record a mass spectrum and position at each laser stop this generates large data sets for whole tissue imaging experiments (the size is affected by the raster step size). Typically, commercial instruments provide imaging software, which plot s the ion intensity versus the position to generate an MS image. A smoothing/interpolation technique is typically used to generate MS images so that the any imaging techni que very simply, interpolation uses the ion intensity at the data points to predict the ion intensity between data points, thus smoothing out the square pixels. There are different types of interpolation that can be performed, but triangulation with linear interpolation is comm only used for imaging software. 89 This technique is a complex mathematical algorithm that generates triangles using three o riginal data points to interpolate the data between the original data points. 89 resolution) of the MALDI MS image s is controlled by the raster step size. Normally, the first image generated in MS imaging data processing is the total ion current (TIC) image, showing the TIC at each laser stop. Although this image does not show any specific compound, the TIC image displays where ion signal was observed from the tissue section To generate com p ound specific images, the software allows the

PAGE 46

46 use r to choose an ion of interest or a ra nge of ions ( e.g., m/z 100 1 amu or m/z 600 800) to plot. The software then plots the intensity of the user determined range versus the position. The compound specific image allows for evaluation of the intensity of a certain compound over the sample, bu t there are considerations for MALDI that could affect the image observed. Due to the variability of the MALDI MS ion signal (and perhaps non uniform sample coating), the variation of ion signal intensity in some regions could arise from the quality of MALDI matrix crystallization. Methods used to correct for the variable MALDI signal include normalizing to the TIC, normalizing to the base peak intensity or normalizing to a MALDI matrix ion signal Although these methods are commonly used, it is diffic ult to determine the best overall method. To obtain more control of image generation MS imaging data can be exported from the instr ument software to produce a spreadsheet with three columns of data, x position y position and ion intensity From this spre adsheet the data can be import ed into different imaging or graphing software to generate an image map and manipulate the data as needed. Overview of Dissertation Improvements in current LCM analysis methods are necessary to impro ve the efficiency of conv ersion into biofuels. The purpose of this research was to develop new methods to analyze and image LCMs to provide a more comprehensive understanding of the composition and organization of compounds within LCMs Prior to imaging intact tissue sections, plan t related standards were first characterized. Chapte r 2 reports the development of a MALDI tandem MS method for the analysis of microcrystalline cellulose (MCC), Birch xylan extract and Spruce lignin

PAGE 47

47 extract. Standard characterization identified ions to analyze in wood tissue, and determined neutral losses to monitor for tandem MS experiments of intact wood tissue After standard characterization, Chapter 3 reports a MALDI tandem MS imaging method developed for the analysis of intact Populus stem cross sections The ions identified in Chapter 2 were monitored in tissue providing MS and MS 2 ion images of cellulose, hemicellulose and lignin ions identified previously. T o further validate the imaging method and obtain higher spatial resolution images, flu orescence microscopy and ToF SIMS image experiments were performed. Chapter 4 reports the use of a polychromic dye, calcofluour white and acridine orange as a way to provide more information about cellulose and lignin localization in wood tissue. The flu orescence microscopy images showed high correlation with the MALDI tandem MS images generated. Furthermore, Chapter 5 reports the adaptation of ToF SIMS methods for the analysis of Populus tissue. The high spatial resolution images offered complementary in formation to fluorescence microscopy and MALDI tandem MS images. One of the difficulties of ToF SIMS is that the high in source fragmentation limits ion identification. In order to help identify ions that were observed in ToF SIMS experiments, multivariate analyses were performed. Chapter 6 reports PCA and cluster analysis of ToF SIMS image data sets. Moreover, the multivariate analysis provided a list of ions that could distinguish between lignified and non lignified tissues. Chapter 7 provides a conclusio n and insight into future experiments that could be used to improve the analysis of LCMs.

PAGE 48

48 Appendix A was added to report experiments performed on a MALDI traveling wave ion mobility (TWIM)ToF mass spectrometry at the FOM Institute on Matter and Physics (AM OLF) The MALDI experiments performed on the MALDI TWIM ToF mass spectrometer were not comparable to experiments performed on the MALDI LIT mass spectrometer. However, LDI was performed and Bucky balls were observed from intact wood tissue.

PAGE 49

49 Fig ure 1 1. Schematic of b iomass c onversion into e thanol A) Steps involved converting corn kernel s into ethanol B) Steps involved with converted lignocellulosic material s such as cornhusks into ethanol

PAGE 50

50 Figure 1 2. Cartoon i llustration of c ellulose and s tarch. A) Structure of cellulose showing 1,4 linkages between glucose monomers. B) Structure of starch, showing 1,4 ages between glucose monomers. C) Composition of cellulose microfibrils. linkages allow for strong i nteractions between cellulose strands, creating microcrystalline structures called microfibrils.

PAGE 51

51 Figure 1 3. Cartoon illustration of plant cell and zoomed in portion of cell wall cross section. Plant cells have a primary cell wall, in addition to a sec ondary cell wall that is split into three different regions (depicted by three different shades of colors). The zoomed in figure shows the hypothesized organization of cellulose microfibrils, hemicellulose and lignin within the secondary cell wall. The hem icellulose coats the cellulose microfibrils, and is link ed together with the lignin.

PAGE 52

52 Figure 1 4. Domon and Costello nomenclature o end. Note that for cellulose, the ends are identical so Y/C ions and Z/B i ons cannot be distinguished.

PAGE 53

53 Figure 1 5. Cartoon schematic of the MALDI process. MALDI matrix and an a lyte solution are applied to a MALDI sample plate and allowed to co crystallize. A laser irradiates the sample surface and the MALDI matrix absorbs the energy desorbing both matrix and analytes into the gas phase (i.e., MALDI plume). The ionization is believed to occur in the MALDI plume through proton transfer reactions.

PAGE 54

54 Table 1 1. Fi gures of merit for common mass analyzers. 54, 56 Mass Analyzer Quadrupole Ion Traps Time of Flight Time of Flight Reflectron Magnetic Sector FT ICR Quadrupole ToF Orbitrap 57 Mass Accuracy 100 ppm 100 ppm 200 ppm 10 ppm < 5 ppm < 5 ppm 10 ppm 1 2 ppm Resolving Power 4000 4000 8000 15,000 30,000 10 0 ,000 10,000 6 0,000 Typical m/z Range 4000 4000 >300,000 10,000 10,000 10,000 10,000 5 ,000 Scan Speed 1 s 1 s 1 ms 1 ms 1 s 1 s 1 s 1 s Tandem MS MS 2 (triple quad) MS n MS MS 2 MS 2 MS n MS 2 MS n Cost Low Low Low Med Hig h High Med High

PAGE 55

55 Figure 1 6 Schematic of linear ion trap adapted from Schwartz, J. C.; et al. J. Am. Soc. Mass Spectrom. 2002 13 659 669. Ions enter the trap along the z to detector t hrough slits in the center section along the positive and negative x axis.

PAGE 56

56 Figure 1 7. Mathieu stabi lity diagram of the LIT adapted from Douglas, D. J. et al. Mass Spectrom. Rev. 2005 24 1 29. Typically, LITs are operated in RF only mode, thus ions are stable along the q x axis wit h a low mass cut off q x = 0.9. Ions are scanned out in q space from low m ass to high as, represented by the circles

PAGE 57

57 Figure 1 8 Workflow for MS imaging experiment. The tissue is sectioned and mounted onto a microscope slide. The tissue is then coated with MALDI matrix and inserted into the instrument. The tissue is raste red beneath the plotted versus the position to generate a MS image.

PAGE 58

58 CHAPTER 2 MALDI LINEAR ION TRA P TANDEM MS TECHNIQU ES FOR THE ANALYSIS AND CHARACTERIZATION OF LIGNOCELLULOS I C STANDARDS: POTENTI AL FOR MALDI MS ANALYSIS OF INTAC T TISSUE Introduction Biofuel production is a process that converts crops, such as corn and sugar cane, to fuel. However, inefficient processing and increas ing food prices have led scientists to a more practical solution investigation. This process is termed cellulosic ethanol and has become a promising new idea for renewable energy. 1, 5 5 Cellulosic ethanol is similar to corn based ethanol in that glucose monomers are fermented to generate carbon dioxide and ethanol; however, the conf ormation of the glucose monomers of cellu lose is different from starch (Figure 1 2). 91 The conformational change results in recalcitrance to enzymatic digestion and requires a pretreatment step prior to fermentation. Pretre atments are typically a high temperature acid baths, which weaken and break interactions between cellulose chains, ultimately making enzymatic digestion possible. Although current pretreatments work, the process is still inefficient and not completely unde rstood. 92 A molecular understanding of the pretreatment could help increase efficiency of ethanol production and make cellulosic ethanol a more practical energy alternative. Mass spectrometry is an analytical tool with high chemical selectivity that could be applied toward understanding chemicals changes throughout a pretreatment process. Advances in ionization sources have increased the capabilities of mass spectrometric (MS) analyses of plant materials, particular ly, matrix assisted laser desorption/ionization

PAGE 59

59 (MALDI). 34 MALDI creates gas phase ions from large biomolecules, (e.g., oligosaccharides and carbohydrates), for MS analysis. 39 Although MALDI is typically coupled to a TOF MS for large biomolecule analysis, tandem MS is limited, thus the tandem MS capabilities of the linear ion trap (LIT) can overcome these limitations. Tandem MS provides more structural information and increases confidence of ion identification. 85, 93 Furthermore, a MALDI tandem MS method for biomass materials is necessary to lay the groundwork for the MS imaging of intact biomass materials. This c hapter reports MALDI tandem MS characterizations of glucan, microcrystalline cellulose (MCC), Birch xylan, and Spruce lignin as standards to m/z values as well as characteristic fragmentation pathways for ion identification in MS analysis of LCMs. Experimental Chemicals MALDI matrices, dihydroxybenzoic acid (DHB), trihydroxyacetophenone (THAP), cyannohydroxycinnamic acid (CHCA) and 3 Aminoquiniline (3AQ) were purchased from Acros Organics (Geel, Belgium). The glucan was purchased from Fischer Scientific (San Jose, CA). The MCC (~20 m) and Birch xylan were purchased from Sigma Aldrich (St. Louis, MO). Spruce lignin extract and Kraft pulp were obtained from School of Forest Resources and Conservat ion. Preparation of MALDI Matrices The various MALDI matrices were dissolved in different solvents, including water, acetonitrile, methanol, acetone, and mixtures of the solvents. Details of exact compositions will be reported in the results section. The m atrices were prepared at varying concentrations (~5 30 mg/mL) in the different solvents. Literature reports

PAGE 60

60 carbohydrates favor cationization 34 (as opposed to protonation) thus different cations, in the form of sa lts, were added to the MALDI matrix solution to determine the highest ionization efficiency The salts tested were lithium chloride, sodium acetate, potassium iodide, cesium iodide and ammonium chloride and were dissolved to a final concentration of 100 mM in the MALDI matrix. MALDI Sample Plate Preparation The standards were suspended in water at a concentration between 5 and 25 mg/mL and spotted using a modi fied dried droplet method As opposed to mixing the standard and MALDI matrix solutions together, 1 L of the standard was pipetted onto the stainless steel MALDI sample plate, immediately followed by 1 L of the MALDI matrix solution. Since the solvents used did not evaporate quickly, this allowed for adequate time of interaction between the MALDI matrix and the analyte. Untreated and Holocellulose Tissue For lignin analyses, radial slices of untreated Populus wood tissue (~20 m) were prepared using a Leica 2010 R sliding mic rotome (Wetzla, Germany). The slices were mounted on tape CryoJane Tape (Instrumedics Inc., Richmond, IL ), and the tape was mounted onto a glass slide using scotch tape for MALDI MS analyses. Holocellulose tissue was prepared by removing from Populus woo d tissue using a sodium hypochlorite 94 treatment. Radial sections of the holocellulose were prepared for MALDI MS using the process described above. The tissues were coated with 25 mg/mL DHB+NaOAc aqueous solution u sing a Meinhard nebulizer. The nebulizer passes the MALDI matrix solution through a small hole, with a nitrogen flow gas to create small droplets of MALDI matrix solution. The MALDI matrix in solution phase interact with the analytes, and as the solvent

PAGE 61

61 ev aporates, small, relatively uniform MALDI matrix crys tals coat the top of the tissue The MALDI matrix solution wets the tissue and it is necessary to allow the tissue to dry completely between coatings rapid evaporation (thus smaller crystals) was promoted by a stream of warm air. Tissues are coated by spraying the solution for 30 seconds, followed drying for 4 minutes; this is repeated until about 8 mL of the MALDI matrix solution is used. Instrumentation The dissertation researched spanned an upgr ade to the instrument used E arly studies were performed on a Thermo MALDI LTQ, whereas later studies were performed on a Thermo MALDI LTQ XL both equipped with a 337 nm 60 Hz N 2 laser Although the instrument used was simply replaced by the updated versi on, it is important to note the small differences betwee n the two instrumental designs the source pressu re, laser set up and laser parameter control The Thermo MALDI LTQ source operates at a pressure of 170 mTorr and used fiber optics to transfer the lase r light to the MALDI plate. en ias a percentage of maximum power T he Thermo MALDI LTQ XL (upgraded instrument) source operates at a pressure of 70 mTorr and use s lenses to direct the laser light toward t No major differences between the spectra obtained from the two diff erent instruments were observed, and a more thorough discussion of the differences are discussed in the MCC section of thi s chapter. For standard analyses, three laser shots were used per laser stop and approximately 50 scans were averaged to generate one spectrum. The laser energy used was varied based on matrix, analyte and nature of MS analys is (i.e., MS versus

PAGE 62

62 tandem MS), and typically fell between 20 and 35 J. Collision induced dissociation (CID) was performed in the LIT with an isolation width of 1.5 m/z and the CID parameters varied between 80 120 (arbitrary units normalized to 400). Data files were analyze d with Qual Browser (v.2.0.7, Thermo Fischer Scientific) Results and Discussion The results are divided into the differen t standard categories, glucan, MCC, Birch xylan, Spruce lignin extract and Kraft pulp All experiments were performed on the Thermo MALDI LTQ XL unless otherwise specified. Glucan Cellulose is a 1,4 linked polymer of glucose, usually with a degree of polymerization ~ 10,000 and can be present in two different forms, crystalline or amorphous. As discussed previously, the orientati on of the CH 2 OH group allows for strong interactions between different polymers of cellulose. Crystalline cellulose, as the name suggests, has highly order hydrogen bonding between strands of cellulose, which creates a rigid secondary structure. The amorp hous regions of cellulose have less ordered bonding between cellulose polymers, generating a weaker secondary structure, as depicted in the cartoon ( Figure 2 1 ) of the different secondary structures of cellulose. Due to the possibilities of highly ordered secondary structures of MCC, glucan a linked polymer of glucose similar to cellulose, was used as a model for optimizing MALDI parameters for further analyses. Choosing the optimal m atrix Optimizing experimental conditions is based on different parameters, such as the response function and signal to background ratio ( S/B ). In analytical chemistry, the response function refers to the analytical signal, and for optimization, experimental

PAGE 63

63 parameters are adjusted to maximize the analyte signal. 95 However, generating a maximum signal is not the only consideration. If optimizing the analyte signal increases the sum of the background signals from other species ( B ), then the experiment is not optim ized. Both the analyte signal and background signal are needed to determine an optimal experimental parameter. 95 The S/B is important (and difficult to calculate) for MALDI MS analyses sinc e excess of background compounds (MALDI matrix) is added to the sample prior to analyses. Due to the inherent high background in MALDI MS the S/B was the focus for determining optimal experimental parameters. Since MALDI matrix serves to transfer energy a nd ionize the analytes, it is important to start experimental optimization with determining the best MALDI matrix for the analyses. Based on previous studies, MALDI matrices evaluated for glucan (4 mg/mL) were DHB, CHCA, THAP, 3 AQ, sinnapinic acid (SA) and a mixture of 3 AQ and SA 96 However, 3 AQ, SA and the 3 AQ/SA mixture did not ionize glucan and were ruled out as possible MALDI matrices. S tarting concentrations of the matrices were chosen to be 25 mg/mL for D HB, 25 mg/mL for THAP and 5 mg/mL for CHCA The MS spectrum from m/z 200 1000 is displayed in Figure 2 2 A s predicted, ions 162 m/z apart were observed, due to the 162 Da repeating unit of glucose (Glc) in the glucan polymer The spectra clearly show t hat DHB as a MALDI matrix generated the highest analyte signals and the lowest background signals, thus the highest S/B From this, it was determined that 25 mg/mL DHB was the optimal matrix and used for the remainder of the experiments.

PAGE 64

64 Optimization of M ALDI matrix a dditives The most abundant analyte ions of glucan in the MS were sodiated ; relatively less potassiated and no protonated ions were observed (Figure 2 2). This suggests that carbohydrates favor cationization in agreement with previous work 34 In order to increase the analyte signal, it is often beneficial to promote a certain type of cation by adding cations (in the form of salts) to the matrix solution. Figure 2 3 shows the MS of glucan with five different salts, LiCl, NaOAc, KCl, CsI, NH 4 Cl, added to the matrix in equal molar amounts (0.100 mM) Although the anions are different for sodium and cesium, this should have little effect on the availability of the cation, since these ionic compounds co mpletely dissolve in solution. The five spectra display, in all cases, that the sodiated species is formed (even without the addition of extra sodium), which further suggests that the sodiated species are favored over the other cations. Sodium acetate was chosen as the optimal matrix additive and used for the remainder of the experiments. Tandem MS characterization of g lucan After determining the optimal matrix, matrix additive and the m/z of analyte ions formed tandem MS experiments were used to characterize fragmentation. The tandem MS spectra are labeled using Domon and Costello nomenclature. 36 It is important to note that distinguishing between C/Y B/Z and A/X fragment ions is not possible due to identi cal terminal sugar groups of glucan (as well as MCC and xylan), so the C B and A nomenclature will be used for the remainder of the dissertation. The fragment ions are labeled by the type of glycosidic fragment (B, C) as well as the position of the frag ment denoted by subscripts. Further, cross ring cleavages are denoted by the letter A, where

PAGE 65

65 the subscripts refer to the sugar residue and the superscripts refer to the position of the bonds breaking. The MS 2 689 ( [Glc 4 +Na] + ) ion, spectrum using CID 85 is displayed in Figure 2 4 The most abundant fragment ion observed is the C 3 fragment which is a result of a neutral loss (NL) of 162 ( Glc) Cross ring cleavages, NL 60, 90 and 120, were also observed corresponding to 0,2 A 4 0,3 A 4 and 2,4 A 4 respectively The NL of 90 could also be due to 1,4 A 4 but is more likely 0,3 A 4 since the oxygen carbon bond is more likely to break than a carbon carbon bond. The B series ions are observed, B 4 ( m/z 671), B 3 ( m/z 509) and B 2 ( m/z 347), which are 18 m/z units lower than the C series ions. Tandem MS of glucan provided diagnostic fragment ions, which will be useful in the identification of compounds within wood tissue. Although NaOAc was chosen as the optimal matrix additive, it is possible that other cationized ions offer more diagnostic fragmentations. Tandem MS of lithiated, sodiated and potassiated ions were explored to test the effects of cation size on fragmentation as well as determining the cation that provides the most diagnostic fragmentation. Figure 2 5 displays the MS 2 spectra of m/z 673, 689, and 705, corresponding to [Glc 4 +Li] + [Glc 4 +Na] + and [Glc 4 +K] + respectively. The most abundant fragment for all three ions is the C 3 ion (NL 162) and no other fragment ions were observed for the lithi a ted and potassiated ions. Cross ring cleavages occur during CID of the sodiated species which offers more structural information about the ion from the MS 2 spectrum Since it was already determined that NaOAc was the optimal matrix additive, these results support that choice. It is interesting to note that there seems to be no trend in

PAGE 66

66 fragmentation according the cation size; however, this experiment only analyzed the four sugar polymer. Since the cations are different sizes, the length of the polymer could affect the strength of the adducts, thus affecting the fragmenta tion pathways. Microcrystalline Cellulose (MCC) After the MALDI parameters were optimized with glucan, MCC was characterized using MALDI tand em MS Although glucan and MCC have similar properties, there are some differences that made MCC analyses more difficult. The major difference is the solubility glucan is soluble in common matrix solvents and MCC is not soluble, with out the addition co ncentrated acid s or base s which break glycosidic bonds. Having an analyte that is not soluble is difficult for MALDI MS since the proposed ionization mechanisms rely on analyte/matrix co crystallization. Despite the inability to dissolve MCC in solvents, MALDI MS analyses were successfully performed Figure 2 6 compares an ideal MALDI sample spot where the analyte and MALDI matrix co crystallize completely to MCC and DHB crystals. Note the small, homogenous crystal sizes of the ideal spot, as opposed to the heterogeneous larger crystals of t he MCC and DHB MALDI sample spo t Since the MCC is not in solution, the DHB crystals tended to gravitate away from the M CC, which caused large background s ignal when spectra are ave rage. Despite complications the sample preparation developed for these analyses allowed for quality spectra that were readily analyzed.

PAGE 67

67 MALDI MS of MCC The MCC standard was suspended in water at ~4 mg/mL and a range of concentrations from 5 25 mg/mL DHB N aOAc was used to determine the optimal MALDI matrix concentration (Figure 2 7) L iterature suggests the optimal MALDI matrix to analyte molar ratio from 1,000 10,000; however, the molarity of MCC of is difficult to determine. For example, MCC is a polymer, with a wide range of sugar chain lengths, thus only an average molecular weight can be calculated Furthermore MCC is not soluble, so determining the molarity of the MCC standard solution is problematic. Instead of the calculating the molar ratios, diffe rent concentrations of the MALDI matrix were tested with a constant concentration of suspended MCC to determine the optimal concentrations for MALDI MS MALDI MS of MCC generated intense ions 162 amu apart, similar to the glucan, and the highest intensity analyte ions with the lowest MALDI matrix ion background signal occurred between m/z 500 2000. Since the analyte ions signal to background is a major concern for MALDI MS analysis, this mass range was used for the majority of the studies. Similar to the glucan, the analyte ions observed from the MS of MCC were sodiated, however, the most intense ions were also dehydrated and represented in the form [Glc n H 2 O+Na] + Figure 2 7 illustrates that 20 mg/mL DHB has a low analyte ion S/B and was ruled out as the optimal matrix concentration. The other three spectra are similar The 10 mg/mL DHB spectrum has the highest analyte signal, but the ion b ackground signal is also higher The 5 mg/mL DHB spectru m produced the best S/B ratio, so it was used for the remainder of the MCC standard experiments. Several different approaches were used to try to improve sample spot quality, for example, ethylenediaminetetraacetic acid (EDTA) was added to the solution. A dding

PAGE 68

68 EDTA to the MALDI matrix solution aggregated the MCC with DHB crystals (Figure 2 6). x ion background signal w ere observed It is interesting to note the ion intensity distributions of the MCC analyte io ns. MS of polymers use the ion intensity distribution to determine molecular weight of polymer as well as the degree of polymerization (DP) However, this might not be the case for the MALDI LIT MS analyses of MCC. MCC is known to have a DP ~10,000, which far exceeds the mass range of the LIT. Despite the high mass of the analyte, ions in the range m/z 500 2000 were observed suggesting that smaller polymers are naturally present within the MCC or in source fragmentation of larger polymers occurs. Since MALDI uses a laser it is likely the energy put into the analyte could induce in source fragmentation, which makes the m/z di stribution of analyte ions not necessarily representative of the MW distributions of MCC. The stability of analyte ions within the ion trap is also important to consider as this could also affect the ion intensities. Figure 2 7 displays that the most inte nse ions occur around m/z 1157, which is close to the middle of the mass range scanned. If the ion intensity distribution were a property of the MCC, then shifting the mass range should have little to no effect on the ion intensity. However, when the mass range is increased to m/z 500 3000, the distribution shifts to make the ions around m/z 2000 the most intense. This is not problematic, as the length of the cellulose is not a focus of this study, instead, determining the presence and spatial distribution of cellulose within a tissue section is the goal of these experiments..

PAGE 69

69 In addition to optimizing the MALDI matrix concentration, the effect of varying the laser energy was also explored. Figure 2 8 illustrates the effect of the laser energy on MALDI matri x ion and MCC ion intensity in the mass spectrum MCC was spotted on a MALDI sample plate, immediately foll owed by 5 mg/mL DHB +NaOAc and EDTA. Three laser shots were used and 25 spectra were averaged and the standard deviation of the mean was calculated for each data point The laser energy was increased from 5 50 J in five J increments The plot illustrates t hat from 5 20 J, the MALDI matrix ion intensity is significantly larger than the MCC analyte ion intensity, and 25 30 J, and the intensities are similar. Laser energies greater than 30 J show a larger intensity analyte ion and lower MALDI matrix ion intensity, which is optimal for the experiments. Figure 2 9 displays the MS spectrum of MCC after all parameters were optimized, showing very little MALDI matrix ion signal. It is also interesting to note that the analyte ion intensity steadily increases over the range of laser energies tested; however, the MALDI matrix analyte ion signal shows more variability, and there is a decrease in ion signal around 20 J. Although the scope of this research was not the MALDI mechanism, these data suggest that the ionization mechanism for MCC differs from typical MALDI experiments. Tandem MS of MCC T andem MS is often necessary for in vivo tissue analyses, so c haracterizing fragmentation of standards is important. The most abundant ions in the MALDI MS of MCC were the sodiated dehydrated specie s m/z 1157, ( [Glc 7 H 2 O+Na] + ) ion. For tandem MS experiments, m/z 1157 ions were isolated and fragmented. Figure 2 10 displays the MS 2 spectrum low er abundance of background ions compared with the MS spectrum The most abundant fragment ions were the B series ions resulting from

PAGE 70

70 sequential NLs of 162 (Glc) due to glycosidic bond cleavages In addition, NL of 18, a non sp ecific loss of water, NL of 60, 0,2 A 7 cross ring cleavage and NL of 144, a C 6 fragment were also observed in the MS spectrum. The difference between CID of sodiated MCC vers us the sodiated dehydrated MCC also was explored. Figure 2 11 displays four stages of MS of the sodiated species m/z 1175, [Glc 7 +Na] + The most intense fragment of the sodiated species is the 0,2 A 7 cross ring cleavage, which is different from the sodiated dehydrated species (most intense fragment is the C 6 ion). In addition, the sodiated species shows both B and C series fragment ions resulting from NL of 162 and 180, indicative of NL of the glucose repe ating unit and glucose molecules. Further, the fragment ion m/z 1115, resulting from a cross ring cleavage, was isolated and fragmented for MS 3 ( m/z 1175 1115 m/z 1115 resulted in NL of 60, possibly another cross ring cleavage, as well as NL of 120, which results in the C 6 fragment the other C fragment ions are observed as well. The MS 4 of m/z 1175 1115 1055 results in NL of 60, likely a cross ring cleavage, to form C series ions. Additionally, NL of 162 from m/z 1055 are also observed, which is indicative that there is still an intact terminal glucose mon omer Tandem MS analyses of MCC show that there are diagnostic fragment ions for both a sodiated and sodiated, dehydrated species to monitor in wood tissue analyses. If the most abundant fragment ion is a NL of 60, than it can be assumed that the ion is sodiated, and if the most abundant fragment is a NL of 162, than the ion can be assumed to be sodiated, dehydrated. One of the most important advantages of tandem MS, is that analyte ions can be isolate d from a high background ion signal to increase

PAGE 71

71 the analyte ion S/B, which is very important for in vivo analy ses where the background ion signal is hypothesized to be very high. Increasing the S/B improves the sensitivity of th e experiment, and isolating the analyte ion helps to improve the selectivity of the experiments. Xylan Xylan is a classification of hemicellulose that is a polymer of xylose, a five carbon sugar with MW of 150 amu The major difference between xylan and M CC (and glucan), is that xylans can be branched an d/or substituted. Figure 2 12 illustrates the common forms of xylan, linear, glucuronic acid (GlcA) substituted, and 4 O methyl glucoronic acid (O MeGlcA) substituted as well as the mass differences from the linear xylan For example, the linear Xyl 4 polymer has a mass of 546 amu (each xylose,Xyl, repeating unit has a mass of 132). Glucuronic acid has a mass of 194 amu, 44 amu more than a xylose monomer, thus the Xyl 3 GlcA polymer is 44 amu more than Xyl 4 Since xylan can be branched and substituted, there is a possibility of more ions, thus more complex mass spectra. Although xylan is more soluble in water than MCC, the same parameters that were optimized for MCC were used for xylan. MS of Birch x ylan Approximately 4 mg/mL Birch xylan was dissolved in water and 1 L was spotted on a MALDI sample plate, immediately followed by 1 L of 15 mg/mL DHB with NaOAc. The MS of 23 averaged spectra is displayed in the inset of Figure 2 13 which shows intense ions 132 amu apart (mass of repeating Xyl residue) and intense background ions in between the intense ions. The most intense ions were identified as sodiated, dehydrated linear xylan ions, represented in the form [Xyl n H 2 O+Na] + where n is the number of Xyl re sidues. Focusing on a smaller portion of the spectrum (Figure 2 13)

PAGE 72

72 illustrates a repeating pattern of ions between the linear xylan ions that were i dentified as substituted xylans, [Xyl 6 GlcA H 2 O+Na] + ( m/z 991) and [Xyl 6 MeGlcA H 2 O+Na] + ( m/z 1005). One of the difficulties of xylan analysis is trying to determine an accurate analyte S/B MCC is a linear polymer, so only one m/z was observed for each length of the polymer it was determined the ions in between the analyte signal were the MALDI matrix ba ckground ions. In contrast, x ylan s can be substituted and branched, so ions in between the linear xylans are not necessarily MALDI matrix background. Instead, the ions are the different variations of the polymer, making the calculation of analyte S/B diffi cult, thus optimization was based primarily on analyte signal of the linear polymers. Tandem MS of Birch x ylan Tandem MS is necessary for Birch xylan analyses due to the abundance of analyte ions at nearly every m/z value Tandem MS isolates on e xylan ion for analyses, which helps to not only positively identify the ion but also to increase the S/B for the analysis. Figure 2 14 shows three tandem MS spectra of the linear, GlcA substituted, and MeGlcA substituted sodiated, dehydrated species m/ z 947, 991, and 100 5, respectively. The major fragment ions from CID of the [Xyl 7 H 2 O+Na] + similar to MCC, are B 7 4 ions, resulting from NL of 132, a Xyl repeating unit ( Figure 2 14 ) Another major fragment is the C 6 ion, the hydrated counterpart to the B 6 ion. Tandem MS is capable of distinguishing the monomers that make up the polymer; however, there are several different structures (i.e., linear or branched) that are possible for a particular m/z value The ion at m/z 947 is only composed of xylose sugars; however, determining the location of the branching or substitution on the polymer backbone is more difficult.

PAGE 73

73 Tandem MS of m/z 991 [Xyl 6 GlcA H 2 O+Na] + shows abundant NL of 44, carboxylic acid from the GlcA suga r ( m/z 947) NL of 176, GlcA ( m/z 815) and NL 132, xylose residue ( m/z 859). In addition, NL of 264, due to losses of two Xyl residues a nd NL of 308, due to losses of Xyl and GlcA were also observed. This again demonstrates that tandem MS is capable of determining the sugar composition of polysaccharides; however, the exact location of the GlcA along the sugar cannot be determined from tandem MS al one. The tandem MS spectrum of m/z 1005 [Xyl 6 MeGlcA H 2 O+Na] + is very simila r to the tandem MS of m/z 991.The difference between the two is 14 amu, due to the methyl group on the MeGlcA (as opposed to the GlcA). Lignin Lignin is a complex, three dimensional polymer composed of a variety of different aromatic mono lignols connected with a variety of different linkages Lignin is found the cellulose intact, providing mechanical strength for the plant. Lignin is derived from three common monolignols, p coumaryl alcohol, coniferyl alcohol and sinapyl alcohol (Figure 2 15 ) that undergo extensive condensation polymerization this results in a variety of different linkages and a degree of polymerization around 10,000. 97 MALDI MS of l ignin Mass spectrometric characterization of lignin is difficult due to the variety of building blocks, strong inter and intra molecular interactions and no standard materials (lignin used for this research was a spruce lignin extract obtained from the School of Forest Resource and Conservation). Since lignin monomers encompass a variety of different compounds (thus molecular weights) MS analysis results in ions at nearly every m/z value without a repeating patt ern (as observed in cellulose and

PAGE 74

74 hemicellulose). This complicates the spectra, making it difficult to identify and characterize Lignin analyses are commonly performed using pyrolysis mass spectrometry. Pyrolysis MS bre aks the lignin polymer apart into the building blocks, and the most common ions monitored in these analyses are vinyl syringol and vinyl guaiacol, also referred to as S and G lignin. 98 The vinyl syringol to vinyl guaiacol ratio (S:G) varies depending on hard or soft wood, as well as different species. Most wood use for this research is Populus which has an S:G ratio greater than one, indicating more S lignin is present in the tissue. 99 The monolignols are similar in structure and chemical properties to common MALDI matrices, specifically the aromatic benzene ri ngs Since it is likely that lignin absorbs the wavelength of the laser used for MALDI (337 nm), lignin analyses wi th and without MALDI matrix were performed (Figure 2 16) As Figure 2 16 displays, MALDI matrix is not necessary to generate ions ; h owever, both analyses are complicated, and using MALDI MS could be difficult. In ord er to find a diagnostic ion for lignin analyses tandem mass spectrometry is necessary. As a starting point m/z 181 ([M+H] + of vinyl syringyl) was isolated for MS 2 analysis to determine if tandem MS can identify a diagnostic ion for lignin in wood tissue. Characterizing low molecular weight ions using MALDI can be complicated due to interferences from the MALDI matrix (specifically, DHB, 154 Da). Figure 2 17 displays MS 2 of m/z 181 from DHB+NaOAc (alone) and un treated Populus wood tissue without MALDI matrix. The two spectra show common fragment ions m/z 163, 153 and 137, so these were not chosen as diagnostic ions of lignin. The most abundant ion,

PAGE 75

75 separated from the other fragment ions observed from the wood t issue was m/z 121, resulting from loss es of both OCH 3 groups (NL of 60) from the vinyl syringol. MALDI MS analysis of lignin in untreated and h olocellulose Populus t issue Radial slices of both untreated and holocellulose tissue was placed on the same slide for MALDI tandem MS Imaging analysis. A thin layer of DHB+NaOAc crystals was applied atop the tissues using a Mienhard nebulizer. Details of MALDI matrix coatings are found in the experimental section. Figure 2 18 displays the optical an d MS 2 images of untreated (top) and holocellulose (bottom) tissues. MS 2 of m/z 181 was performed in the rectangle outlined in blue in the optical image, using 100 m laser step sizes. The intensity of the fragment ion from m/z 181 121 was plotted for the M S 2 image and displays intense ion signal off the untreated wood, and little ion signal from the holocellulose. There were areas of signal from m/z 181 121 observed from the holocellulose tissue, which could result from lignin being left behind after the sodium hypochlorite treatment. I n addition, Figure 2 18 confirms that the ion signal from MS 2 181 121 is not due to MALDI matrix alone. The section in between the two tissue sections (the red region in the optical image) is an area of exposed tape, coated with the MALDI matrix. If the ion signal MS 2 181 121 was only coming from the DHB MALDI matrix, than the region between the tissues would show high ion intensity, but this was not observed. It was determined that MS 2 181 121 is a good diagnostic ion for the presence of lignin within wood tissue, and will be used in future experiments. Summary The results presented in this chapter laid the groundwork to develop a MALDI tandem MS imaging technique for LCM. Due to the complexity of the compounds, as

PAGE 76

76 well as the organization of the compounds, that comprise LCM, it was necessary to first optimize the experimental parameters using standard materials. g lucan and MCC Experiments with glucan were used to determine the optimal MALDI matrix, which includ ed the matrix, solvent as well as additives, for the analyses of LCMs. The highest S/B was observed using DHB + NaOAc dissolved in water. MALDI MS of glucan resulted in the formation of sodiated specie s. The optimal MALDI parameters determined for gluca n were applied for the analysis of MCC The MALDI matrix concentration and laser energy were optimized using the MCC. The plot of laser energy versus a MALDI matrix ion and a cellulose ion illustrated that higher laser energies are needed (> 20 J) to obta in higher analyte signal (as the MALDI matrix ion signal remains similar). Despite the difficulties in obtaining uniform MALDI sample spots, MALDI MS spectra of MCC were obtained, with a high S/B ratio. The MALDI MS spectra displayed ions 162 m/z apart, wh ich is consistent with the molecular weight of the glucose repeating unit of the polymer. The ions were identified as sodiated, dehydrated polymers of glucose and repr esented as [Glc n H 2 O+Na] + where n represents the number of glucose monomers. A list of m/ z values of cellulose to monitor in wood tissue is presented in Table 2 1 In addition to MS, MCC was characterized using tandem MS. A single cellulose ion, m/z 1 157 was isolated and fragmented using collision induced dissociat ion The fragmentation observed after dissociation of the sodiated, dehydrated specie s showed abundant B series fragment ions resulting from glycosidic bond cleavages Cross ring cleavages were also observed; however, they were in low abundance relative to the B series ions.

PAGE 77

77 Hemicellulose Hemicellulose characterization was more difficult compared to the analysis of MCC H emicellulose encompasses linear and branched, as well as substituted polysaccharides. Since cellulose only contains glucose, the molecular weight between each degree of polymer remains the same (162 Da). However, the xylan backbone of hemicellulose, is commonly substituted with GlcA and MeGlcA this generates several different polymer patterns of ions as well as increases the number of analyte ions observe (thus the spectra complexity). The ions that were observed were the sodiated, dehydrated species s in the form [Xyl n H 2 O+Na] + where n represents the number of xylose residues. Tandem MS was used to characterize and identify xylan, GlcA substituted xylan and Me GlcA substituted xylan. The tandem MS of all three compounds displayed NLs of 132, due to glycosidic cleavages of xylose residues. The GlcA substituted was identified by a NL of 176 (GlcA) and MeGlcA was identified by a NL of 190 (MeGlcA). Although the exa ct composition of the monomers that are in the polymers are known, the structure (e.g., branching and where it is branched) is not able to be determined from tandem MS alone. However, this is not vital to the imaging of LCMs. Lignin The MS characterizati on of lignin was difficult since there are no standard materials commercially available and the inherently high heterogeneity of the polymer. Although lignin can be ionized without MALDI matrix, the MS spectra with and without the MALDI matrix look similar and are complex. Ions at nearly every m/z with similar intensities are observed, making it difficult to choose a diagnostic ion for monitoring in

PAGE 78

78 wood tissue. Since lignin has been shown to have an S:G ratio greater than 1, vinyl syringol ( m/z 181) could provide insight into lignin composition in wood tissue. Tandem MS of m/z 181 of MALDI matrix was compared to Populus wood tiss u e, and it was determined that m/z 181 121 is an abundant fragment in the wood tissue, but not the MALDI matrix. To d etermine of tandem MS of m/z 181 121 could be diagnostic of lignin in wood tissue, analysis of untreated wood tissue was compared to holocellu lose tissue The tandem MS image of MS 2 of m/z 181 121 illustrated abundant ion signal from untreated tissue, and less abundant ion signal form the holocellulose, elucidating the m/z 181 as a diagnostic ion for lignin in wood tissue.

PAGE 79

79 Figure 2 1 Cartoon representation of a cellulose microfibril. Each microfibril is composed of several individual cellulose polymers. Both amorphous and crystalline regions of the microfibril exist, depending on the strength of interactions between cellulose polymers.

PAGE 80

80 Figure 2 2. MALDI MS spectra of glucan stan dard using Different matrices. A) 2,5 cyannohydroxycinammic acid (CHCA) C) trihydroxyacetophenone (THAP). The most intense ions observed for the DHB were the [Glc n +Na] + ions, where n represents the number of glucose monomer s. Note that sodiated ions signals were more intense than the potassiated ion signals and no protonated ions were observed. Also, no analyte ions were observed for CHCA. The highest analyte ion signal to background ion signal was observed for DHB and was d etermined to be the optimal matrix.

PAGE 81

81 Figure 2 3. MS Spectra of glucan standard using DHB MALDI matrix with different salt additives. A) Lithium chloride B) Sodium acetate C) Potassium chloride. D) Cesium iodide E) Ammonium Acetate. Spectra illustrat es that regardless of s alt added to the MALDI matrix, sodiated specie s are preferentially formed

PAGE 82

82 Figure 2 4. MS 2 Spectrum of [Glc 4 +Na] + Collision induced dissociation of sodiated glucan ion labeled with Domon and Costello nomenclature of oligosaccharide fragmentation. The most abundant fragment ion, C 3 results from a glycosidic cleavage, a neutral loss of 162 Da. This corresponds with the mass of a glucose monomer. Also, a NL of 60, due to a 0,2 A 4 cross ring cleavage is observed. Minor losses include other cross ring cleavages, as well as the B 3 fragment ion.

PAGE 83

83 Figure 2 5. MS 2 spectra of different cationized glucan ions. A) Lith i ated ion shows NL of 162. B) sodiated species shows NL 162 and cro ss ring cleavage C) potassiated ion shows NL of 162. The sodiated species offers increased fragmentation, which can aid in unknown ion identification.

PAGE 84

84 Figure 2 6. Optical images of MALDI sample spots. A) DHB and sodium acetate crystals, illustrating an ideal MALDI sample spot. For optimal MALDI analyses, the MALDI matrix crystals should be uniform. B) MALDI sample spot of MCC and DHB crystals. The s ampl e spot show s inhomogeneous crystals, which causes poor reproducibility off mass spectra between laser s h ots C) MALDI sample spot of MCC, DHB and EDTA. EDTA caused an reproducibility within that region.

PAGE 85

85 Figure 2 7. Comparison of MCC MS by varying DHB concentrations. A) 5 mg/mL DHB. B) 10 mg/mL DHB. C) 15 mg/mL DHB. D) 20 mg/mL DHB. Analyte ions labeled are in the form [Glc n H 2 O+Na] + It was determined that 5 mg/mL D HB provided the highest signal to background for cellulose ions.

PAGE 86

86 Figure 2 8. Plot of MALDI matrix and cellulose ion intensity versus laser energy. The plot illustrates the analyte ion intensity ( m/z 1157) steadily increases with increasing laser energy. Above 30 J laser energy, the analyte to matrix ion signal is the highest, which is ideal for experiments. The error bars represent the standard deviation of the mean.

PAGE 87

87 Figure 2 9. MS spectrum of MC C using optimized experimental parameters. Ions labeled are in the form [Glc n H 2 O+Na] + The spectrum shows v n al Ions observed in between the majors ions are a repeating pattern, possible due to in source fragmentation.

PAGE 88

88 Figure 2 10. MS 2 spectrum of m/z 1157. Similar to the glucan standard, MS 2 of m/z 1157 ([Glc 7 H 2 O+Na] + results in cross ring cleave and sequential NLs of 162,due to glycosidic bond cleavages.

PAGE 89

89 Figure 2 11. MS 2 MS 4 analysis of MCC. A) MS 2 B) MS 3 C) MS 4 The MS 2 of m/z 1175 ([Glc 7 +Na] + ) shows an abundant NL of 60, due to a cross ring cleavage (top). The MS 3 of 1175 1115, shows NL of 60, as well as NL of 120 (middle) MS 4 of 1175 1115 1055 shows NL of 60, followed by NL of 162 were also observed.

PAGE 90

90 Figure 2 12. Common hemicellulose structures. A) Generic Xylan. B) Linear xylan. C) Glucuronic acid branched xylan. D) O methyl glucuronic acid branched xylan. The most c ommon substitutions include glucuronic acid and methyl glucuronic acid, which increa se the mass of a linear xylan by 44 and 58 Da, respectively

PAGE 91

91 Figure 2 13. MS spectra of Birch Xylan extract. A) m/z 500 2000, with intense ions 132 amu (Xyl residue) apart, which were identified as linear xylan ions in the form, [Xyl n H 2 O+Na] + Abundan t ions in between the linear xylan ions are observed as a result of substituted xylans. B) m/z 945 1250, showing the predicted substituted ions were observed.

PAGE 92

92 Figure 2 14 MS 2 analyses of xylans. A) MS 2 of linear xylan shows sequential NL of 132, xylose residues B) MS 2 of GlcA substituted xylan. NL of 176 is indicative of GlcA substitution. NL s of 132 are also observed. C) MS 2 of MeGlcA substituted xylan. NL of 190 is indicative of MeGlcA substitution

PAGE 93

93 Figure 2 15. Common building blocks of lignin. Lignin is a complex, three dimensional polymer of monolignols, including coumaryl alcohol, coniferyl alcohol and sinapyl alcohol. The structure of monolignols is similar to common MALDI matrices.

PAGE 94

94 Figure 2 16. MALDI and LDI MS analysis of Spruce lignin extract. A) DHB+NaOAc for MALDI MS analysis of lignin extract Abundant ions a t nearly every m/z value are observed. B) MS without using MAL DI matrix (LDI), showing lignin analyses can be performed without MAL DI matrix. This is likely due to the highly aromatic monolignols absorbing at the wavelength of the N 2 laser

PAGE 95

95 Figure 2 17. MS 2 analysis of m/z 181 A) DHB+NaOAc. B) U ntreated wood tissue without MALDI matrix MS 2 of DHB displays three major fragment ions The bottom spectrum preliminarily identified m/z as vinyl syringol (a monolignol), with predicted fragment structures. Fragment m/z 121 could provide a diagnostic fragment for lignin.

PAGE 96

96 Figure 2 18. Optical a nd MALDI tandem MS image of untreated and holocellulose Populus tissue. Lignin is removed from holocellulose. MS 2 analysis of m/z 181 121 illustrates ion signal only from untreated Populus tissue.

PAGE 97

97 Table 2 1. List of hypothesized cellulose ion m/z values to monitor in wood tissue. Number of Glucose Residues Molar Mass (g/mol) Hypothesized m/z [ Glc n H 2 O+Na] + 3 504 509 4 666 671 5 828 833 6 990 995 7 1152 1157 8 1314 1319 9 1476 1481 10 1638 1643 11 1801 1805 12 1962 1967

PAGE 98

98 CHAPTER 3 DIRECT MATRIX ASSISTED LASER DESOR PTION/IONZATION MASS SPECTROMETRIC IMAGIN G OF CELLULOSE AND H EMICELLULOSE IN POPULUS TISSUE Introduction In recent years, alternative energy research has garnered significant attention, particularly research focused on bi biomass. 3, 100 102 Unlike the simple sugars and starch of sugar cane and corn grain, the complex arr angement of cellulose, hemicellulose and li gnin in plants naturally resist enzymatic digestion and limits bioconversion of biomass into biofuel. 100, 103 In bioconversion methods, a pretr eatment step is performed to overcome some of this natural recalcitrance by increasing accessibility of the cellulose to cellulase digestion, particularly at high enzyme loadings. 100 Despite the success of current pretreatment methods, most are conducted with size reduced materials to remove anatomical differences. Due to high energies (and cost) required for size reduction, particularly for woody biomass, the use of larger fragments (e.g., wood chips) is preferred for commercial processes, 3, 101 but the spatial changes in chemical composition in pretreated wood chips are not well characterized. 102 The ability to map spatial changes in chemical compositions within wood tissue should provide valuable information that can be used to improve the efficiency of lignocellulosic bioconversion to biofuel. 103 A variety of techniques are used to image lignocellulosic material, including optical microscopy with stains, ultraviolet microscopy, 20 magnetic resonance imaging (MRI), 21 micro X ray computed tomography (X Ray CT), 22 confocal Raman microscopy, 23, 24 scanning electron and transmission electron microscopy. Although these techniques provide structural information, their molecular selectivity is quite limited. One approach

PAGE 99

99 that helps to overcome this lack of molecular specificity is mass spectrometry, specifically, time of flight secondary ionization mass spectrometry (ToF SIMS) 53 and matrix assisted laser desorption/ionization time of flight mass spectrometry (MALDI ToF MS). 104 In addition to molecular selectivity, ToF SIMS and MALDI MS are capabl e of MS imaging experiments. 53,85,105, 106 Despite widespread applications reported for these techniques in animal studies, 54 ToF SIMS and MALDI MS have only recently been used to analyze lignocellulosic compounds. 107 112 SIMS analyzes secondary ions emitted from a surface, which are characteristic of compounds within a sample. 53, 105 ToF SIMS images of wood tissue and switchgrass have obtained s patial resolution down to 1 m, the highest spatial resolution offered by MS imaging. 112 Although ToF SIMS imaging provides direct surface analysis and high spatial resolution, extensive analyte fragmentation during ioni zation limits the chemical specificity. 26 Because many plant cell wall carbohydrates are complex and composed of closely related building blocks, e.g., glucose (C 6 H 12 O 6 ), xylose (C 5 H 10 O 5 ), glucuronic acid (C 6 H 10 O 7 ), and methyl glucuronic acid (C 7 H 12 O 7 ), SIMS ionization of different carbohydrates often results in non specific fragmentation, complicating interpretation. MALDI is a soft ionization method used to generate biomolecular ions for mass spectrometry 113 and MALDI ToF MS has been applied toward lignocellulosic biomass analyses; 111 however, the molecular specificity of ToF MS is limited. Due to the similarities of lignocellulosic composit ion compounds, ions of different analytes are observed at the same nominal mass to charge ratio ( m/z ) (isobaric ions). Tandem mass spectrometry (MS/MS) can overcome this difficulty by dissociating precursor ions and using fragmentation to distinguish betwe en isobaric ions, thereby increasing confidence

PAGE 100

100 in ion identification. 65, 85 Herein, we report direct imaging of cellulose and hemicellulose in intact wood tissue using MALDI LIT tandem MS. Experimental Ins trumentation and Data Analysis All MS experiments were performed using a Thermo MALDI LTQ XL mass spectrometer (Thermo Fisher Scientific, San Jose, CA) equipped with a nitrogen laser (LTB Lasertechnik Berlin, 337 nm, 60 Hz) with approximately 100 m diameter laser spot size. The samples were analyzed at int ermediate pressure (70 mTorr). Mass spectra were analyzed using Qual Browser v.2.0.7 and MS images were generated with Image Quest v.1.0.1 (Thermo Fisher Scientific, San Jose, CA). MALDI Matrix DHB (99% pure) was purchased from Acros Organics (Geel, Belgium). Although the concentration used for the standards was 5 mg/mL, higher MALD matrix concentrations are needed for i ntact tissue section analysis. The MALDI was prepared at 25 mg/mL DHB dissolved in 0.05 mM aqueous sodium acetate (NaOAc). Wood S amples One of the difficulties of intact wood tissue analysis is mounting the sample onto the sample plate (or glass slide). Animal tissues are sections while still frozen, and can be thaw mounted onto a glass slide with the gentle heat of a finger. However, this cannot be done with wood tissue section. The wood tissue does not stick to the microscope slide and also has a tendency to curl on itself. Also, if the wood did stick to the microscope slide, applying the matrix re wets the sample, thus removing it from t he sample plate and curling the edges.

PAGE 101

101 rigidity sectioning very difficult. For optical microscopy studies of wood tissue, a tape is mou n ted mounted onto the tape. The tape works well to keep the tissue intact, flat and on the sample slide for analysis. However due to t he molecular selectivity (and sensitivity to contamination of sample surfaces), it was possible that the adhesive from the tape could seep through the wood tissue and ionize more readily over the wood compounds. To avoid contamination, great care was taken to ensure the sample surface remained free of the adhesive of the tape. E xperiments comparing the ion signal obtained from the tape with MALDI matrix versus the ion signal from the wood tissue with MALDI matrix clearly displayed no conta mination problems Thus, it was determined mounted the wood tissue sections onto tape would provide the highest quality wood tissue sections and highest quality mass spectra. Untreated, radial pine wood samples were sectioned to ~ 30 m thickness using a Leica crytome (Inst rumedic s, Inc., Richmond, IL #475214). and mounted onto a tape The tape was then mounted onto a glass microscope slide for analysis. Small wood blocks, 1 cm x 1 cm x 1 cm, were cut from a field grown, three year old stem of Populus deltoidies, and 50 m t hick radial sections were cut on a sliding microtome (Lieca, SM2010R). In a subset of sections, the lignin was removed with sodium hypochlorite treatment 94 to produce holocellulose. For tissue imaging, the lower stem regions from 10 week old greenhouse grown Populus deltoidies X Populus trichocarpa X Populus deltoides hybrids were sectioned with a vibratome (Lieca,

PAGE 102

102 VT100S) to obtain 50 m thick transverse sections. All sections were mounted on CryoJane tape (Instrume dics, Inc., Richmond, IL #475214) prior to analysis. MALDI Matrix C oating The MALDI matrix (DHB) was applied to the t issue with a Meinhard nebulizer the tissue was sprayed for approximately 30 seconds, followed by a 4 minute drying time aided with a warm air stream. This process was repeated until a white crystal layer was observed atop the wood tissue (~8 mL of matrix solution). MS images were genera ted by rastering the tissue underneath the laser in 50 m step sizes, using three laser shots per spot and 35 J and 40 J laser energy for MS and MS 2 analyses, respectively. A 15 x 15 mm rectangle was imaged at a rate of 131 scans per minute for both MS a nd MS 2 The MS 2 experiments on tissue sections were performed using the same CID energy and isolation width as the standard experiments. Results and Discussion Pine Wood Samples The average of 1549 spectra from m/z 500 2000 of a pine wood sample is illustr ated in Figure 3 1. The average spectrum shows abundant ions at nearly every m/z which was likely due to lignin ion signal. In addition, ions 162 m/z units apart, consistent with the m/z values from MCC standard analysis, confirming that cellulose was obs erved from the intact wood tissue. I t is interesting to note that the average spectrum from the first 180 scans versus the last 180 scans is different (Figure 3 2). The first 180 scans shows ions at nearly every m/z value at similar relative abundances the last 180 scans displays more signal from the ions identified as MCC. The different spectra correlate well with the different tissue types observed in the optical image (Figure 3 1), which likely have different composition s of cell ulose, hemicellulose

PAGE 103

103 and lignin. The difference in MS spectra demonstrates that MS imaging is capable of distinguishing between differences in chemical composition. To investigate different tissue types further cross sections, which contain vario us regions of tissue types, were examined. Tandem MS of Young Populus Stem Tissue To confirm the identity, tandem MS of ions observed from wood tissue were compared with the tandem MS of MCC and xylan standards. MS 2 analyses of the four most abundant ions observed from wood tissue, m/z 833, 995, 1157, and 1319 ([Glc 4 7 H 2 O+Na] + ), were compared to MS 2 of the same ions of the MCC standard. The MS 2 spectra from both wood tissue and MCC st andard displayed major fragment ions resulting from glycosidic bond clea vages (sequential NLs of 162), which confirmed cellulose was observed from wood tissue. The MS 2 spectra of m/z 833, 995, and 1157 displayed an NL of 154 (molecular weight of DHB), suggesting a DHB cluster ion at the same nominal m/z, thus m/z 1319 was used for further cellulose analysis on wood tissue. The comparison of MS 2 of 1319 of the MCC standard with the spectrum from untreated wood tissue is displayed in Figure 3 3 As expected, both spectra show similar f ragmentation the major fragmen t ions ( m/z 1157, 995, 833, 671, and 509) result from successive NLs of 162, due to glycosidic bond cleavag es. Although the major fragment ions of MCC and untreated wood tissue are similar, m/z 1319 1301 and 1319 1275, NLs of 18 and 44 respectively, are mo re abundant in the MS 2 spectra from wood tissue. The NL of 18 is likely due to a loss of water and is a non specific loss, i.e., most ions with an OH group can lose water during collision induced dissociation in the ion trap. The NL of 44 from m/z 1319 su ggests the presence of a n acetyl or

PAGE 104

104 carboxylic acid functional group, which is common for carbohydrates other than cellulose present in wood tissue. Thus, the great er abundance of these fragment ions observed in the MS 2 spectrum of untreated wood tissue st rongly suggests the presence of at least two isobaric ions at m/z 1319. Figure 3 4 displays the MS 2 spectra of m/z 1079 fr om purchased Birch xylan and Populus wood tissue Sequential NLs of 132 are observed in the MS 2 of m/z 1079 of xylan standard, result ing from glycosidic bond cleavages between xylose monomers this ion was identified as [Xyl 8 H 2 O+Na] + The MS 2 spectrum of m/ z 1079 from Populus tissue shows m/z 1079 917 (NLs of 162) and 1079 875 (NL 204), in addition to m/ z 1079 847 ( NL of 132).Since these fragment ions are not observed in the standard, at least two isobaric ions are likely present at m/z 1079. Further stages of MS determined that in addition to the linear xylan, another ion is of the form, [(162) 4 (204) 2 +Na] + and preliminarily identif ied as an O acetylgalactoglucomannan (another classification of hemicellulose). However, analyses of additional standards, such as galactoglucomannan are needed to confirm this preliminary identification more positively MS and Tandem MS Imaging of Young P opulus Stem Tissue MALDI LIT MS imaging of the young polar stem was performed on a quarter section of the Populus stem, to ensure all different regions of the tissue section were examined. Optical, MS and MS 2 images of a transverse section of young Populus stem are displayed in Figure 3 5 and Figure 3 6 The white dotted line in the optical image (a) illustrates the region of MS analysis. The pith is located in the center of the stem and is composed of thin, non lignified, primary walled cells. The secondar y xylem extends from the pith to the vascular cambium and is composed of living ray parenchyma, as

PAGE 105

105 well as nonliving, lignified, fiber and vessel cells with thickened secondary walls The vascular cambium is a thin layer (~50 m) of living tissue between t he secondary xylem and secondary phloem where new cells are produced for secondary, or radial, growth. The secondary phloem is composed of living, primary walled parenchyma and companion cells, as well as dead sieve and fiber cells with thickened secondary walls. In contrast to the sieve cells, the fiber walls are lignified. Abundant ion signal at nearly every m/z value w as observed for all the regions; however, different regions displayed characteristic spectra. For example, around the mass spectra from t he pith and the secondary xylem, where most live tissue is found, displayed ions at every m/z value at 50 % or greater relative abundance. On the other hand, abundant ions 162 Da apart (corresponding to ions observed in the MCC standard) were observed above the background ion signal in the secondary xylem and the center of the phloem. These general differences in the observed spectra were attributed to characteristics of the different tissues present in the different regions of Populus stem probed. Although MS analyses are capable of discerning general differences in regions of wood tissue, tandem MS increases the molecular specificity of the experiment to analyze specific analytes in the different regions of wood tissues. The extracted ion MS image of m/z 1319, identified with standards as a cellulose ion [Glc 7 H 2 O+Na] + displays uniform ion signal over all tissue regions of the Populus wood stem (Figure 3 5b ), which is expected for a cellulose ion. Interestingly, the thin, primary walled cells of the pith and intact vascular cambium (arrow) have similar intensity as the secondary xylem and phloem tissues, even though they are known to contain relatively less cellulose.

PAGE 106

106 The MS 2 images of m/z 1319 1275 and 1319 995, resulting from NL of 44 and two sequentia l NLs of 162 (Glc monomers) from the precursor ion, ( m/z 1319) are displayed in Figure 3 5 c and Figure 3 5 d respectively. Figure 3 5 c displays more ion signal in the gap around the pith, lacking cells, and the secondary phloem. On the other hand, the MS 2 image of m/z 1319 995 shows a negative image of m/z 1319 1247. Specifically, m/z 1319 995 illustrates intense ion signal in the center region of the pith with cells, the secondary xylem, and small localized regions in the secondary phloem, which could corr espond to fiber bundles. The other regions, the vascular cambium and secondary phloem without fiber bundles, display less abundant ion signal. Furthermore, the m/z 1319 995 ion signal closest to the pith, which corresponds to older xylem tissue in the sam ple, suggesting this region could have thicker cell walls, thus increased cellulose composition. However, the difference in cellulose ion signal variation due to inhomogeneous crystal formation atop the wood tissue needs to be ruled out. As previously disc ussed, the MS 2 spectrum of Populus wood tissue suggests at least two isobaric ions at m/z 1319. This was confirmed by comparing the MS image of m/z 1319 with the drastically different MS 2 images obtained from plotting m/z 1319 1247 and m/z 1319 995. The comparison of m/z 1079 (standard analysis identified as a linear xylan ion, [Xyl 8 H 2 O+Na] + ) MS with the MS 2 images also exemplifies the necessity of tandem MS for wood tissue analyses. Specifically, tandem MS is needed to distinguish between different hemicellulose ions or other interfering ions at the same nominal m/z The extracted ion MS image of m/z 1079 shows nearly even ion signal over the entire tissue section (Figure 3 6 ). MS 2 images of two fragment ions m/z 947 (NL of 132, 5 carbon

PAGE 107

107 sugar), and m/z 917 (NL of 162, 6 carbon sugar) illustrate different ion intensities in different regions of the stem tissue. The MS 2 image of m/z 1079 947 (Figure 3 6b ) shows higher ion signal localized in the secondary xylem closest to the pith (similar to the cel lulose ion intensity), also consistent with the hypothesis that this region contains thicker cell walls. A localization of m/z 1079 947 is also observed in the secondary phloem and vascular cambium, which is different from the observed cellulose signal. Al though no xylan is located in the vascular cambium, this region is thinner than the spatial resolution of the imaging experiment. The MS 2 image of m/z 1079 917 ( Figure 3 6d) shows localized ion intensity in the region of the secondary xylem closest to the pith, but less intense ion signal is observed in the secondary phloem compared to the MS 2 image of m/z 1079 947. Moreover less signal is observed around the pith, which further demonstrates the need for tandem MS to obtain accurate spatial distributions of ions at a single m/z As discussed in Chapter 2, performing MS 2 on m/z 181 could be used to identify the presence of lignin. Figure 3 7 displays the optical image and MS 2 images of m/z 181 163 (NL 18) and m/z 181 121 (NL 60). The localization of the non specific NL 18 shows ion signal over the entire tissue section. This was expected since NL of 18 is a non selective loss. The image of m/z 181 121 shows ion signal localization in the secondary xylem, the center of the pith and small regions in the sec ondary phloem. The localization of the ion signal is similar to distribution of cellulose. Since lignin is typically present in the secondary cell wall s which contain more cellulose than primary cell walls, it was expected that the lignin ion signal would show a similar distribution to the

PAGE 108

108 cellulose ion signal. Comparing the two different MS 2 images also demonstrates that tandem MS can distinguish between two different ions at the same nominal m/z Lignin compounds are similar in structure (thus m/z ) to MA LDI matrix ions, thus it is necessary to use tandem MS to increase the selectivity of the experiment. Although a complete chemical composition analysis of all the regions of wood tissue was not completed, the experiments show the viability of MALDI MS n imaging of cellulosic tissue. More specifically, MS alone is incapable of providing accurate spatial distributions of different ions at a single m/z Instead, the necessity and advantages of tandem MS analyses of wood tissue are evident after comparing MS 2 images with the MS images. Specifically, characteristic fragment ions resulting from collision induced dissociation of a precursor ion are needed to differentiate between isobaric ions, which are inherent when analyzing a complex tissue, such as Populus wood. The MS 2 images provided more selectivity of cellulose and hemicellulose ion signal, as well as significantly reduce background ion signal compared to the MS spectrum. Furthermore, tandem MS was shown to differentiate between an analytes and inferrin g ions at one nominal m/z Conclusions The direct analysis of cellulose and hemicellulose in Populus tissue using MALDI tandem MS was reported. Plotting different fragment ions from MS 2 experiments resulted in different ion signal localization, thus tandem MS is necessary for separating isobaric species and generated accurate wood tissue MS images. In addition to elucidating isobaric species, tandem MS reduces the background ion signal o f wood tissue, improving the signal to background of imaging experiments. The combination of MS 2 and imaging provides specific chemical mapping in intact wood tissue, and this

PAGE 109

109 MALDI LIT tandem MS method could be applied toward imaging other intact cellulos ic tissues, such as switch grass or sugarcane. Imaging specific compounds within cellulosic tissue will provide further insight into the secondary structure, and furthermore, monitor changes throughout pretreatment for the conversion to ethanol. Fut ure experiments will focus on comparing MALDI LIT MS images to complementary techniques to improve the understanding of chemical changes that occur during the pretreatment process. Furthermore, future experiments will focus on developing a MALDI LIT MS tec hnique to characterize lignin in wood tissue.

PAGE 110

110 Figure 3 1. Optical image and average MS from Pine wood sample. A) The optical image of the radial pine wood sections shows two different regions of tissue (the region imaged outlined in black). B) The MS sp ectra averaged over 1500 scans displays spectra representative of a mixture of lignin standard, with ions at nearly every m/z and MCC standard, ions 162 m/z difference and xylan standard, ions 132 m/z difference.

PAGE 111

111 Figure 3 2. Comparison of spectra from different regions of tissue. The optical image (Figure 3 1) clearly illustrated two different regions of tissue in the area analyzed, which resulted in two different MS spectra. A) The first 180 scans average (top) shows a characteristic spectrum from lig nin, with even intensity of ions at nearly every m/z The ions that appear above the background are likely to do MALDI matrix. B) Th e average of the last 180 scans displays a spectrum more representative of c ellulose, with intense ions 162 amu difference.

PAGE 112

112 Figure 3 3. MS 2 of m/z 1319 MCC standard B) W ood tissue. The comparison of the standard wit h the wood tissue spectrum confirms that cellulose was detected from wood tissue., as the major fragment ions N L of 162, are similar. Fragment ions m/z 1301 (NL 18) and m/z 1275 (NL 44) are observed in higher abundance from the wood tissue analysis, suggested isobaric species at m/z 1319, which can be distinguished using tandem MS.

PAGE 113

113 Figure 3 4. MS 2 of m/z 1079 A) Birch xylan standard B) W o od tissue. The comparison of the standard wit h the wood tissue spectrum confirms that hemicellulose was detected from wood tiss ue, as similar fragment ions (NL 132) were observed in both analyses.. MS 2 of m/z 1079 from wood shows a more complicated spect rum indicative of isobaric ions. The starred ions refer to fragment ions not observed in the standard spectrum and were preliminarily identified as O actylglucomannan, a different classification of hemicellulose

PAGE 114

114 Figure 3 5. Optical and MS images of cell ulose ion in Populus stem. A) Optical image of Populus wood tissue showing the pith (1), secondary xylem (2), vascular cambium (3), and secondary phloem (4). The white outline shows the area of MS and MS 2 analyses. B) MS image of m/z 1319 ([Glc 7 H 2 O+Na] + ) shows ion signal over the whole tissue. C) The MS 2 image of m/z 1319 1275 (NL of 44) displays more abundant ion signal in the region around the pith and the secondary phloem, which is consistent with tissue compositions. D) MS 2 image of m/ z 1319 995, res ulting from two NLs of 162, shows localization in the secondary xylem, closest to the pith, and reduced ion signal in the center of the pith, vascular cambium and secondary phloem.

PAGE 115

115 Figure 3 6 Optical and MS images of hemicellulose ion in Populus stem A) Optical image of Populus wood tissue with the region imaged outlined in white. B) MS image of m/z 1079 (identified as [Xyl 7 H 2 O+Na ]+ ), shows ion signal over the whole tissue section, with more intensity in the seconda ry phloem and around the pith. C) MS 2 image of m/z 1079 947 (NL of 132) displays increased ion intensity in the secondary phloem. D) The MS 2 image of m/z 1079 947 (NL of 162) displayed increased ion signal in the secondary xylem closest to the pith. Comparing the MS 2 images of tw o different fragment ions illustrates two isobaric compounds are present and tandem MS can distinguish between them

PAGE 116

116 Figure 3 7 Optical and MS images of lignin ion in Populus stem A) Optical image of Populus wood tissue with the re gion imaged outlined in white. B) MS 2 image of m/z 181 163 (NL 18) displays ion signal over the whole tissue section. C) MS 2 image of m/z 181 121 (NL of 60) displays increased ion intensity in the secondary xylem and regions of high intensity in the secondary phloem and center of the pith. The transition of m/z 181 121 was identified as lignin ion and shows similar distribution as the cellulose ion in Figure 5 4

PAGE 117

117 CHAPTER 4 IMAGING OF POPULUS TISSUE BY FLUORESCEN ECE MICROSCOPY Overvie w MS imaging provides high chemical selectivity; however the spatial resolution is inadequate to probe individual plant cells. In contrast, microscopy techniques provide images with su perior spatial resolution to MS imaging Fu r thermore, fluorescence imaging can offer selective staining techniques to provide high resolution images Combin ing fluorescence microscopy with MALDI MS imaging could provide a more comprehensive analysis of wood tissue. Moreover, the complementary information of MS imaging and microscopy offers more insight into the spatial po sitioning of chemical compounds than other techniques alone. This chapter discusses fluorescence microscopy analysis of Populus stem cross sections. Furthermore, the fluorescenc e images are correlated with MS imaging to obtain complementary information and aid in method validation Microscopy Microscopy is a technique in biological sciences that uses a microscop e to observe the structure, architecture and anatomy of tissues and/or cells 114 The three main types of microscopy include optical, electron and scanning probe microscopy T his chapter discuss es optical microscopy techniques applied to plant tissue analyses Very simply, optical microsco py is performed by passing light through a series of optics which can generate a magnified (and focused) image of the samp le. Depending on the arrangement of th e optics presence of filters collection of light (e.g., absorbed or scattered light) within in the microscope, different optical microscopy techniques can be

PAGE 118

118 perf ormed. T he most commonly used optical techniques are bright field microscopy, dark field mic roscopy and fluorescence microscopy 114 In addition to obtaining magnified images of samples, s tains and dyes can be implemented for microscopy experiments to contrast of regions of tissue that contain different compounds. F or example, the polychromic dye toluidine blue O (TB O) changes color when exposed to lignified cells versus non lignified cells, 115 which helps to identify different regions of tissue to correlate with ot her techniques (e.g., MS imaging ). In recent years, fluorescence dyes, or fluorophores, have become increasingly popular in plant cell biology. Fluorophores are used similar to traditional dyes, in that a solution is placed atop the tissue and the dye or f luorophore selectively binds to specific class es of compounds ; thus the localization of observed fluorescence is indicative of those particular compounds. To increase the selectivity of the experiment, scientists can add fluorescence tags to molecules des igned to bind to one particular compound as opposed to compound classes (i.e., carbohydrates). 116 Fluorescence Microscopy Fluorescence is performed by irradiat ing a sample with a certain wavelength of ex ) this causes the molecule to enter an excited electronic state The molecule then relaxes to the lowest vibrational state, and then relaxes back down to the ground electron em ) this light is referred to as fluorescence and is characteristic of certain molecule (Figure 4 1) 117 Traditionally fluorescence spectroscopy (Figure 4 2 ) was performed i n solution for quantitative as opposed to qualit ati ve analyses since fluorescence intensity is proportional to concentration and fluorescence can provide low detection limites Today,

PAGE 119

119 fluorescence is has been in corporated into microscopy techniques and has become a widely used method in biological sciences. 118 Fluorescence m icroscopy experiments involve a fluorescence stain, referred to as a fluorophore, that selectively bind s to specific molecule or clas sifications of molecules, e.g., linked glucans. Once the sample is stained, it is placed atop a slide and observed using a fluorescence microscope that is equipped with an excitation UV source. The light emitted from the source can contain several differ ent wavelengths and is typically ex for the fluorophore. The emitted fluorescence from the excited fluorophore is passed through a barrier filter before detection. The fluorescence can be detected by the eye (as in a typical microscope) and /or a digital camera, which can be connected to a computer to capture and save images. A schematic of a fluorescence microscopy experiment is illustrated in Figure 4 3 Experimental TBO Stain A TBO staining technique was modified for Populus stems prepared for MS imaging experiments. A phosphate buffer (pH 6.8) containing 1 M K 2 HPO 4 and 1 M KH 2 PO 4 was prepared in water, and 0.10 g of TBO blue was dissolved in 100 mL of the phosphate buffer solu tion. Fluorescence Stains C alcofluor white (CW) stain was purchased from Sigma Alridch (St. Louis, MO) and is composed of ca lco fluor white M2R (shown in Figure 4 4 also referred to as Fluorescence bright en ing agent 28) (1g/L) and Evans blue (0.5 g/L) the ex is 360 nm em is 460 nm. Evans blue is added to the staining so that the staining solution is

PAGE 120

120 blue (as opposed to colorless) to ensure an even stain across the tissue and to help reduce background fluorescence of tissues. CW has been reported to bind selectively to glucans (including 1,3 and 1,4 linked polysaccharides). 119 120 and was used to identify relative concentrations of cellulose ( 1,4 gluca n) in the wood tissue sections. auto fluoresces ex em ~490 nm. Thus, it is possible to observe fluorescence from wood stems without the addition of a fluorophore. However, the fluoresce nce signal is weak and difficult to observe at higher magnifications. To enhance the fluorescence signal due to lignin in the tissue, the fluorophore acri d in e orange (AO, Figure 4 4 ex em 535 and 630 nm was used. AO has been reported t o interact with lignin inside of cells, thus the presence of lignin can be correlated with the observed green fluorescence. 121 Sample Preparation and Instrumentation Fifty micron cross sections of Populus stems were cut using a sliding microtome and mounted atop CryoJane Tape (Instrumedics Inc., Richmond, IL), and the tape was mounted onto a glass slide. Some wood tissue samples were coated with 25 mg/mL DHB+NaOAc us ing a Meinhard nebulizer for MS imaging experiments. MS imaging experiments were performed using a Thermo MALDI LTQ XL with 50 m raster step sizes and parameters discussed in Chapter 3. TBO staining was p erformed by applying approximately 500 L of the prepared TBO solution atop the 50 m Popul us tissue section for approximately 1 minute The stain was washed off using water and observed under a microscope. Fluorescence microscopy wa s performed before and after MS imaging experiments. For CW, approximately 500 L of the CW stain solution was pipetted atop the Populus cross section. After five minutes, the CW stain was washed off with excess

PAGE 121

121 water and a cover slip was applied on to the stem section to observe the fluorescence. AO stained was performed in a simil ar manner approximately 1 mg/mL i n 50:50 methanol/water solution was pipetted atop the stem and washed off with water after five minutes. The fluorescence was observed and captured using a n Olympu s Mac roscope and high resolution images were captured with an Olympus x70 fluorescence micros cope s. Results and Discussion TBO Stain The TBO stained wood tissue section is illustrated in Figure 4 5. TBO changes colors when the stain interacts with cells of different composition. Specifically, lignified cells appear blue gray and non lignified cell s appear a bright red purple color. Figure 4 5 displays the secondary xylem and small regions within the phloem appear blue gray, which i s indicative of lignified cells ( thus secondary cell wall s) The phloem and intact region of the pith appear a red purp le, illustrating the phloem cells do not contain lignin (i.e., composed of only primary cell walls). The results of the TBO st ain were consistent with the MS imaging results discussed in Chapter 3. Specifically, the blue regions in the phloem confirmed that lignified cells are observed in the phlo em the high correlation between an established staining method and the MS imaging result s demonstrate that the MS imaging method provides high molecular selectivity that can distinguish between different tissue types CW Fluorescence Microscopy The stem used for this experiment was first stained with CW and t he fluorescence images were captured with a fluorescence microscope equipped with a digital camer a As Figure 4 4 ex similar to

PAGE 122

122 wavelength of the MALDI laser on the LTQ XL. It was hypothesized that the CW could act as a MALDI matrix, thus no additional DHB was added atop the stems for the MS imaging e xperiments displayed in this section. Figure 4 6 displays the CW fluorescence stained and MS image of the same Populus wood stem cross section. The CW image shows blue fluorescence in the presences of 1,4 and 1 ,3 linked glucans, which can be correlate d with the localization of cellulose. S everal pictures were taken over the imaged area of the stem and stitched together using photo editing software; therefore, some of the areas of the stem observed in the MS image were not observed in the fluorescence image. The fluorescence image displays that the most intense fluorescence signal comes from the secondary xylem, which is prima rily composed of thickened secondary cell walls. It is also interest ing to note the fluorescence signal observed from the cen ter of the pith, as well as from small regions within the secondary phloem. Although the pith is composed of mostly primary cell wa lls, cellulose is still present but in lower abundance. The intensity of the fluorescence can be correlated to the relative abundance of cellulose between the regions of the tissue. For example, the intense fluorescence observed in the left region of the s tem can be correlated with higher relative concentration of cellulose when compared with the center region of the pith. Figure 4 6 also displays tandem MS images of m/z 1319 (previously identified as a cellulose ion) of th e same stem, after CW staining. M ore specifically, Fi gure 4 6 b shows the TIC of the MS 2 of m/z 1319 and Figure 4 6 c shows the MS 2 of m/z 1319 995 (NL of 324). The TIC ion signal shows the most intense ion signal a round the pith region and in the secondary phloem. CW is believed to act as the MALDI

PAGE 123

123 matrix. Since CW binds selectively to glucans, less intense ion signal would be expected in regions where CW (and glucans) is not present. Comparing this tandem MS image to the CW fluorescence image demonstrates that the most inten se ion signals are localized in the regions with the least intense fluorescence. ( Note, the gap between the secondary phloem and secondary xylem is due to a crack in the tissue that occurred after washing, drying and placing the stem into intermed iate pres sure for MS analysis.) T he comparison of these two images shows that CW can be used to generate ions from the surface on the stem and the CW does not appear to ionize glucans selectively The comparison of Figure 4 6 b and c again demonstrate s the necess ity of tandem MS imaging for intact wood tissue analysis Figure 4 6 c shows the NL of 324 (two Glc residues) from m/z 1319. The ion signal localization is similar to that observed in Chapter 3, where the most intense ion signal is localized in the thick, secondary cell walls of the secondary xylem, the center of the pith, and small localized regions in the secondary phloem. The localization of the ion signal is similar to the fluorescence observed with the selective fluorophore. This comparison helped to v alidate the developed tandem MS imaging method of intact plant tissue, and could be used to generate relative quantitative methods for MALDI tandem MS imaging. Traditionally, MALDI MS is a more qualitative method since the ion intensities observed could be a ffected by several different parameters (i.e., ion intensity does not always correlated with analyte concentration) for example, heterogeneous MALDI matrix crystal and/or topographic differences of the stem surface. Fluorescence microscopy coupled with MALDI MS co uld be used to increase quantita tive information.

PAGE 124

124 For example, the most intense fluorescence signal is observed in the left and right hand regions of the xylem with a region of less intense fluorescence in the bottom center of the secondary xy lem and around the pith region The relative intensities from MALDI tandem MS imaging are opposite t he most intense ion signal is observed in the bottom region of the secondary xylem and around the pith region and less intense ion signal is observed in the right and left sides of the xylem. However, this information could still be used to help increase quantitative information obtained from MALDI MS experiments. Although no absolute quantitation was obtained, the difference in ion intensity correlated with the difference in the fluorescence signal. Although the correlation was inversely related (i.e., higher fluorescence correlated with lower ion signal), this suggests that the variation in the ion signal is a result of relative changes in cellulose concentr ation and/or accessibility which could help to improve the understanding of pretreatment processes for cellulosic ethanol. Ac ri dine Orange Fl uor e scence Microscopy P rimary and secondary cell walls both contain cellulose (cellulose is more concentrated in the secondary cell wall ), thus distinguishing regions of primary versus secondary cell walls could be difficult using a cellulose fluorophore alone. Lignin, however, is more representative of the secondary cell wall, since primary cell walls do not contain any lignin AO was used to stain lignin and the fluorescence images are displayed in Figure 4 7 The green fluorescence signal is observed across the secondary xylem and intense regions in the phloem. This confirms the regions are lignified phloem fiber c ells that contain secondary cell walls and higher concentrations of cellulose compared with the rest of the phloem

PAGE 125

12 5 It is also interesting to note that the fluorescence within the xylem is not an even sign al (i.e., there are regions of more intense fluore scence ). These areas were observed with h igher magnification and displayed in Figure 4 7 b c and demonstrated smaller structures within the xylem, specifically the xylem vessels and the cell walls, were able to be observed. Xylem vessels are long tubes that extend the length of the stem and are la rger in diameter than the plant cells, approximately 30 m. As previously mentioned, the secondary cell walls are identified by green fluorescence of the AO and Figure 4 7c illustrates the ce lls are approximatel y 12 m in diameter and the cell wall approximately 2 m thick. Figure 4 7c clearly shows the difference between the regions of high fluorescence intensity, h igh lignin content (bottom left ) and the lower fluorescence intensity, low lignin content (top ri ght). The cell walls in the bottom left region shows more regular ly shaped, right region are less regular ly appear l shape s are other characteristics of tension wood include thicker secondary cell walls, increased cellulose concentration and lower lignin conce ntration, which is consistent with the observed fluorescence of the cellulose cell ular composition (i.e., cellulose, hemicellulose, lignin concentrations) changes to alleviate the stress induced on the pla nt. The fluorescence image of the same ste m stained with CW is displayed in Figure 4 8 to contrast the two fluorescence signals. In the regions less intense AO fluorescence signal is observed, greater CW signal is observed (relative to the signal

PAGE 126

126 across each respective tissue) these regions are in dicated by the arrows in Figure 4 8. Contrasting the fluorescence signals of CW and AO signal helps to identify these regions of tension wood. It is also interesting to note the difference in the CW fluorescence signal in the phloem between Figure 4 6 a and Figure 4 8 b. Figure 4 6 shows little CW fluorescence signal in the phloem, relative to the lignified phloem fiber cells. In contrast, Figure 4 8 b shows similar CW signal in both the phloem and the lignified fiber cells ; t his difference in fluorescen ce signal could be explained by the age of the stem The stem displayed in Figure 4 8 is younger than the stem in Figure 4 6, which could explain the differences in cellulose content within the phloem. The differences observed between these stem demonstrat e that different techniques are needed for full characterization of LCM tissue. The advantage of analyzing regions of t ension wood is that the regions of cellulose and lignin concentrations differences can be observed within one tissue section, and thus c an be analyzed in one analytical scan. These analyses could help lead to develop quantitat ion methods using MS imaging ; h owever, three main difficulties need to be overcome. Since tension wood is a natural reaction to stress, there is not a way to guarante e that every section prepared (or every stem analyzed) will contain tension wood areas Also, the regions of tension wood are difficult to identify without fluorescence stains, so prob ing the correct regions with MS imaging could be difficult. Lastly tens ion wood areas within these stem sections are small, which is difficult to analyze with the current spatial resolution limitations of MALDI MS imaging ( Figure 4 8c )

PAGE 127

127 Figure 4 8c shows the MS 2 image of m/z 1319 833 stem after CW and AO s taining and shows increased cellulose signal in the top region of the stem. This could possibly be correlat ed with the region of tension wood observed; however, the quality of the MS after both fluorescence stains were applied to the tissue was poor. Also, t he region of tension wood displayed in Figure 4 8b is approximately 200 m wide. The laser s pot size is > 75 m in diameter. Without large over sampling, only four laser s t ops would probe the region of tension wood, limiting the analysis of that area A higher spatial resolution technique is needed to probe these regions and to show more structural features of LCMs. In addition, the cellulose ion signal was observed in the phloem region, which is consistent with the CW fluorescence microscopy signal. It i s important to note that the older stem (Figure 4 6) displayed little CW fluorescence signal in the phloem and was also observed in the MALDI MS image of a cellulose ion. These results further validate the method of MALDI MS imaging for the analysis of int act tissue. Furthermore, these images demonstrate that the selectivity of MALDI tandem MS imaging is needed for accurate analysis of LCMs. Conclusions This chapter discussed the ease and viability of obtain ing high spatial resolution images of intact wood tissue sections using fluorescence microscopy The staining protocols are easy to follow and offer more insight into the chemical composition the traditional microscopy or polychromic dyes. In addition, it was shown that CW can be quantitative method for the analysis of cellulose within intact LCM tissue sections.

PAGE 128

128 Correlating fluorescence microscopy with MALDI MS imaging validate d the developed MALDI MS imaging method and illustrated the need to improve spatial resolution of MALDI MS imaging experiments Coupling fluorescence staining with MALDI LIT MS imaging was shown to produce a more comprehensive analysis of the Populus stem.

PAGE 129

129 Figure 4 1. Jablonski d iagram rep resenting energy level transitions involved in fluorescence emission. The molecule absorbs a certain wavelength of light, which raises the energy to an excited electronic state. The molecule undergoes vibration relaxation, and electronic relaxation by the emission of light (fluorescence). The energy of the emission is always less than the energy of absorption.

PAGE 130

130 Figure 4 2. Schematic of a general fluorescence experiment The light from an emission source is passed through and excitation filter, and only t he excitation wavelength reaches the sample. The fluorescence emitted from the sample passes through a barrier filter before detection.

PAGE 131

131 Figure 4 3. Schematic of fluorescence microscopy. The excitation source is reflected toward the sample by a dichroic beam splitter. The fluorescence emitted from the sample is passed through objective lens for magnification and lastly to the detector (eyes or digital camera)

PAGE 132

132 Figure 4 4. Chemical structures of fluorescence stains used. A) C alcofluor white B ) A cridine orange. Note the highly aromatic structures and wavelength of excitation is similar to the N 2 MALDI laser.

PAGE 133

133 Figure 4 5. TBO Stained Populus stem. TBO is a polychro matic stain used to determine lignified cells. In the presence of lignin, TBO appea rs a blue gray color and in the absence of lignin, appears a red purple color

PAGE 134

134 Figure 4 6. CW stained and tandem MS images. A) The CW image shows fluorescence signal localized in the secondary xylem a nd in regions within the phloem. B) The MS 2 image shows ion signal was not affected by the CW stain. C) The MS 2 1319 995 (NL 324) displays high correlation between ion signal and CW fluorescence.

PAGE 135

135 Figure 4 7. AO stained Populus stem. AO fluoresces in the present of lignin and is indicative of secondary cell walls. A) AO signal is localized in the secondary xylem and intense regions in the phloem, indicative of the phloem fiber bundles B) High magnification reveals areas of high and lower fluorescence inte nsity and was determined to be rep resentative of tension wood. C) Further magnification illustrates the irregularly shaped cell walls of the tension wood.

PAGE 136

136 Figure 4 8. Comparison of AO, CW stain and MS image of same Populus wood stem. A) AO fluorescence. B) CW fluorescence. C) MS 2 1319 833 image. Contrast of AO and CW illustrates th e lower AO fluorescence signal (identified as a region of tension wood) shows increased CW signal MS image illustrates increased cellulose ion signal at the top of the stem; however, the spatial resolution is not adequate for the analysis of tension wood regions. Also, the AO and CW stain appeared to reduce the quality of MS spectra from the stained w ood tissue.

PAGE 137

137 CHAPTER 5 HIGH SPATIAL RESOLUTION IMAGING OF POPULUS TISSUE USING TOF SIMS Overview As in any field of study, using one analysis technique could l imi t the information obtained about the sample For example, MS imaging techniques that pro vide h igh molecular selectivity typically lack spatial resolution and vice versa. Thus, performing a variety of different techniques on the same sample offers complementary information to allow for a more comprehensive understanding of the sample. This chapter focuses on employing c omplementary ima ging techniques to verify and compare with the developed matrix assisted laser desorption/ionization linear ion trap MALDI LIT tandem MS imaging method for the analysis of lignocellulosic materials ( LCMs ) T hese techni ques include ToF SIMS and fluorescence microscopy ToF SIMS Microscope Imaging As previously describe d in Chapter 1, ToF SIMS is a sensitive surface analysis technique that offers superior spatial resolution over other MS imaging techniques Thus, t he goals of these experiments were to obtain higher spatial resolution ToF SIMS images and to correlate the MALDI MS analyte ion signal with the SIMS analyte ion signal The two major differences between MA LD I imaging and ToF SIMS imaging are the ionization m e chanism and the image generation MALDI relies on interaction s between analyte s and the MALDI matrix during laser desorption to generate analyte ions The MALDI matrix is designed to absorb most of the laser energy, which makes MALDI a soft ionization te chnique capable of introducing intact b iomolecules into the gas phase. SIMS generates secondary analyte ions by bombarding the sample with a high energy primary ion beam the high energy causes

PAGE 138

138 intense fragmen tation, which produces complex spectra of ions t ypically lower than m/z 1000. In addition to the different ionization sources, t he ToF SIMS instrument used in this chapter is operated in the microscope imaging mode (versus the microprob e imaging of MALDI experiments) I nstead of rastering the sample beneath the laser ( in the case of MALDI LIT experiments) the spatial distribution of ions generated from the surface is conserved through the mass analyzer focused through ion optics and directed toward a position sen sitive detector. T he spatial resolut ion of the images generated is determined by the magnification of the microscop e the quality of the ion optics and the resolution of the position sensitive detector. 54 A general schematic of the Ph ysical Elect ronics TRIFT II (Chanhessen, MN) instrument used for these experiments is displayed in Figure 5 1 The instrument is operated under high vacuum (10 10 torr) to increase the mean free path, thus mass resolution, of the ToF SIMS experiments. Ionization Liquid metal ion guns (LMIGs) are the most common primary ion source for SIMS experiments. The gun operates by heating the metal (gold for the experiments reported) until it reaches the liquid phase. The liquid metal is passed through a needle induced with a large electric field, as the liquid moves further down the needle, the electric field strength increases and ionization is induced. Parameters controlled to start the LMIG include the heater and the extractor voltage ( tips of LMIG are fragile so great care must be taken to not damage the ion gun needle ) The ion current at the tip is monitored for proper operation and must be stable before measurements can be obtained. The gold primary ions (Au + ) are guided through deflector plates and a focusing lens t oward the sample surface. 76

PAGE 139

139 The point size (i.e., sample spot) of a SIMS source is much smaller than the MALDI laser spot diameter. Atomic ion sources (e.g., Au + ) routinely produce 50 nm spot sizes and have been reported smaller 122 The impact of the primary ions c auses a from the primary ions to the surface which releases ions, neutrals and clu s ters within 10 of the surface (assuming static SIMS limits, Figure 5 2 ). T he damage cross section measures the surface area damage d by the primary ion beam and is an important parameter of SIMS. If significant surface damage is induced, the measurement is no longer of the pristine sample surface; instead, the secondary ions generated will be characteristic of the damaged surface. The primary ion dose can be altered to change the amount of surface damage induced on the sample. Typically, low primary ion doses ( < 10 13 io ns per cm 2 ) are used for static SIMS experiments ; thus it can be assumed that the secondary ions are generated from the monolayer. 54 M ass A nalysis and D etection The secondary ions generated from the surface are electrostically extracted through an immersion lens followed by two transfer lenses these magnify the secondary ion profile prior to entering the mass spectrometer. The magnified ion packet is then focused through three, identical 90 electrostatic analyzers (ESA s), which help to compensate for the kinetic energy distribution. Illustrated in Figure 5 1, the secondary ions also pass through a contrast diaphragm to limit the lateral energy distribution and an energy slit to limit the axial energy distribution. 76 The secondary ion image is then projected onto the position sensitive detector.

PAGE 140

140 The TRIFT II instrument uses a dual microchannel plate (DMCP) for detection. As the secondary ions hit the first MCP detector electrons are generated and amplified for detection and the time of detection is used for ToF mass analysis. The image is generated by setting a time (mass) window for collection, and each mass spectrum is c orrelated to the position of the detector. The detector is a 256 x 256 array detector and the counts for each ion at each position are summed to generate the ion image and mass spectrum at each location. 76 Experime ntal Instrumentation Experiments reported in this chapter were carried out at the FOM Institute AMOLF, ToF SIMS ex periments were performed on a T RIFT II by Physical Electronics (Cha nhessen, MN) within the static SIMS limit (< 10 13 ions cm 2 ) The primary ion source w as a LMIG of Au + Fluorescence microscopy was performed using a Leica DMRX microsc ope equipped with a Nikon DXM1200 camera and filters from Chroma Technology. Sample P reparation M icrocrystalline cellulose (MCC) Birch xylan and S pruce lignin were prepared as described in Chapter 2 and 1 L of the suspensions were dropped onto an ITO coated glass slide and allowed to dry. Half of the sample spots were coated with 1 nm of gold using a sputter coater (SC7640 Quo rum Technologies, Newhaven, UK) to determine the best sample preparation. 10 week old Populus hybrids were grown under high nitrogen c onditions and portions of the stem were subjected to carbohydrate and lignin analyses. App roximately

PAGE 141

141 5 0 m thick cross sections were sliced on a sliding microtome and mounted onto a glass slide with a cover slip. The wood sections were then floated in acetone, transferred onto double sided conductive tape and mounted onto an ITO coated glass slide. For method development studies, different sample coa tings were used. Four s tems were prepared as follows: a) no gold, no DHB, b) DHB only, c) DHB and gold and d) gold only. The DHB was applied using the ImageP rep an automated MALDI matrix coating device (Bruker Daltronics, Billerica Mass.) with 25 mg/mL DHB dissolved in 50/50 methanol:water. A 1 nm thick layer of gold was deposited onto the surface using a sputter coater (SC7640 Quorum Technol ogies, Newhaven, UK). I maging E xperiment s Mosaic imaging experiments were set up to determine the optimal sample preparation. Mosaic imag es are generated by piecing together tiles, which have a set resolution of 256 x 256 pixels. The dimensions of each til e are user determined and were set to 8 7 .5 x 87.5 m for these experiments The spatial resolution can be approximated by solving for the dimensions of 1 imaging pixel using the ratio 5 1 However, due to the large amount of data sets, under the normal settings, the software reduces the 256 pixels into 8 pixels ; thus the actual spatial resolution observed from the softwa re is lower than the physically measurement The full resolution images can be observed using different software, but needs a high performance computer as some data files can take days to convert and op en.

PAGE 142

142 The spatial resolution and mass resolution can al so be altered by operating in ions are bunched together in shorter pulses to generate more ions per time. This shorter ion pulse results in a smaller kinetic energy distribution and higher mass resolution. Unbunched mode increases the ion pulse time, thus increasing kinetic energy distribution, lower ing the mass resolution, but increasing the spatial resolution dramatically ( Figure 5 3 ) Results and Discussion ToF SIMS spectra are more complex compared the MALDI spectra, due to extensive fragmentation of the high energy ionization source. Also, ions observed from MALDI experimen ts are not expected in SIMS since cationization (or adduction) may not occur due to the high energy ionization Thus, no comparisons between ions observed in MALDI MS and ToF SIMS were made. Another difficulty of ToF SIMS analyses is the sensitivity to su rface contaminants. Despite great caution taken for sample preparation, phthalates and silicones are commonly observed in ToF SIMS spectra more specifically poly(dimethysiloxzane) (PDMS). Since PDMS is a polymer, a series of ions (as opposed to just one ion) can be monitored for the presence of PDMS and the commonly observed series of ions include s m/z values 28, 43, 59, 73, 147, 207 and 221. 123 Standards ToF SIMS analyses were performed on MCC, Birch xylan extract and Spruce lignin extract. The best spectra were obtained from coating the standards with a 2 nm layer of gold prior to analysis. An example of ToF SIMS spectrum recorded from the birch xylan standard is displayed in Figure 5 4 and demonstrates the intense

PAGE 143

143 fragmentation characteristic of SIMS. Some predicted ion identifications were made based on the known composition of xylans; however, positive identifications could not be assigned One of the difficulties of adding gold to th e sample surface prior to analysis is that the gold ions and cluster ions observed become the base peaks of the MS. Despite this obvious disadvantage, no analyte ionization was observed without the gold coating. The ToF SIMS analysis of the three different standards (MCC, Birch xylan extract,Spruce lignin extract) resulted in similar mass spectra, demonstrating the difficulties of ToF SIMS as an exploratory method. Sin c e these standards are all similar in composition (i.e., all contain carbon, hydrogen and oxygen), the m/z values of many fragment ions are likely to be isobaric making it difficult to differentiate between different compounds at the same nominal mass. Populus Tissue T he optimal sample preparation method for wood tissue was determined by comparing the spectra obtained from four different sample preparation methods, a) no gold coating, b) DHB coating, c) DHB and gold coating and d) gold coating This comparison revealed that 1 nm of gold coating gave opti mal spectra of wood tissue. Briefly, the intensity of ions was increased and higher mass ions were observed all wood tissue samples reported in this chapter were coated with a 1 nm layer of gold prior to ToF SIMS analysis (the gold layer was increased to 2 nm for standard analysis). The focus of this section was to compare/contrast different regions within on e tissue section in order to determine characteristi c ions of a particular region, e.g., regions of lignified versus non lignified cells.

PAGE 144

144 Positive ion mode Pos itive ion ToF SIMS mass spectra from regions of interest are displayed in Figure 5 5 The regions of interest are identified by the color of the corresponding MS spectrum. Briefly, red is from the phloem, green is from the lignified phloem fiber c ells and blue is from the secondary xylem. The spectra from the three different regions of the tissue are all similar; however some differences were observed. For example, in the top spectrum, the red line (phloem) is more abundant at m/z value 23 and 39, which are identified as Na + and K + respectively. The salts are interesting to observe in intact wood tissue because the localization of salts can be correlated with regions that are more characteristics of living cells, i.e, only primary cell walls and the vascular cambium, unlike the dead cells of the secondary xylem. The similarities observed in the ToF SIMS spectra between the different regions of interest result from the similarities of the carbohydrates that compose the primary and secon dary cell walls which tend to fragment similarly. Significant differences are observed, specifically, m/z value 2 65 2 66 and 607; however, some of the differences could be subtle an d more in depth analysis to distinguish between the regions is required F or example, a multivariate analysis technique could be used to determine ions that are significantly different, thus representative of certain tissue regions ; this approach will be discussed in Chapter 6. The optical, autofluorescence, calcofluor white ( C W ) total ion current (TIC) and extracted ion images of the secondary phloem, vascular cambium and secondary xylem regi ons are displayed in Figure 5 6 The autofluorescence image (Figure 5 6 b) identifies the regions of lignified cell walls, specifically located in the secondary xylem and the phloem fiber cells the lack of green fluorescence observed in the secondary phloem

PAGE 145

145 identifies the regions containing only primary cell walls. The CW white image (Figure 5 6 c) displays that cellulose is present in both the secondary xylem and phloem. The fluorescence signal appears less intense in the secondary xylem, which could be a result of the interaction of the cellulose with lignin. The fluorescence images are used as markers for the different regions and are use ful in ion identification. The TIC image (Figure 5 6 d) illustrates the higher spatial resolut ion when compared with MALDI MS imaging and shows that ion signal is observed over the whole tissu e section The image of m/z 39, K + (Figure 5 6 e) shows intense ion signal in the bottom part of the phloem which is more characteristic of living cells and less intense signal in the secondary xylem, which is more characteristic of dead cells It is also interesting to note the intense K + ion signal observe in the re gion between the secondary xylem and phloem. This is consistent with the location of vascular cambium which is more representative of living cells This demonstrates that ToF SIMS has adequate spatial resolution and selectivity to distinguish between thes e areas, as the spatial resolution of MALDI MS imaging could not. Another ion observed localized in the vascular cambium region (which also showed increased K + signal) was m/z 331, displayed in Figure 5 6 f m/z 331 is consistent with the m/z value of a radical cation on of a classification of compounds called Gibberellins (GA). GA are plant hormone s that has been associated with cambial growth and has been reported to be present in low abundance in healthy stems and aids in the differentiatio n of phloem and xylem cells during wood formation. 124, 125 The ion intensity observed was low, which is expected due to the width of the region (~50 m), as well as the relative abundance of GA e xpected in vascular cambium tissue.

PAGE 146

146 Although the identification cannot be confirmed in the present study, the evidence suggests that m/z 331 could be tentatively identified as GA. Figure 5 6 g displays the ion image of m/z value 85, which has been previousl y identified as a non specific polysaccharide fragment, C 4 H 7 O 2 + 123 and shows relatively even signal across the entire tissue. In contrast, Figure 5 6 h ( m/z 69) and Figure 5 6 i ( m/z 607) display distinct localization in the phloem fiber cells and the secondary xylem, similar to the lignin autofluorescence signal (Figure 5 6 b). The ion, m/z value 69, has been previously reported as both C 4 H 5 O + resulting from a non specific polysaccharide fragment and C 5 H 9 + resulting from a lignin fragment 123 However, the localization of the ion signal suggests that the m/z value 69 is indicative of lignified cells but the presence of ions originating from polysaccharides canno t be ruled out. The extracted ion image demonstrates that more intense ion signal is observed in the presence of lignifi ed cells. These results are also consistent with that MALDI LIT tandem MS analysis that showed localization of lignin ( m/z 181 ) in the p hloem fiber cells and secondary xylem (Figure 3 7). Furthermore m/z 607 shows similar ion signal localization, also suggesting an ion indicative of lignified cells. Although it appears that m/z 607 is more selective (i.e., no ion signal is observed in the phloem, suggesting no isobaric carbohydrate ions), the absolute intensity of m/z 607 is and order of magnitude lower than m/z 69. The lower observed ion signal could be a result of the inherent decrease with m/z value in secondary ion yield of ToF SIMS experiments, but can still be informative about relative lignin composition within the intact tissue section.

PAGE 147

147 Analyzing the phloem and the xylem region using ToF SIMS, and comparing the results to fluorescence microscopy aided in identifying ions r epresentative of different regions of tissue. To help to validate ion identifications reported above the same m/z values were monitored in different sections as well as different areas of the tissue, and showed consistent data. Figure 5 7 displays the op tical, fluorescence microscopy and ion images of the pith and secondary xylem region of the stem, and corresponds to the same ion images displayed in Figure 5 6 Briefly, the auto fluorescence shows more lignified cells in the secondary xylem. The ion image of m/z 39 (K + ) shows localization in the pith, which is more characteristic of living cells. The ion signal of m/z value 331 is localized only in the pith region and was not observed in the secondary xylem these results are still consistent with the iden tification of GA. Since GA is a plant hormone, it would not be expected to be observed in dead cells (e.g., the secondary xylem). The pith, on the other hand, is primarily composed of primary cell walls, and a higher concentration of salts were observed, t hus is expected to have more characteristic s of living cells Figures 5 7 g and Figure 5 7 i demonstrate similar trends to those discussed for Figure 5 6 The ion m/z value 85 shows even ion intensity over the tissue, and identified as a non specific carbohydrate fragment. The ion image of m/z value 69 shows ion signal over the whole tissue, but displays an increase in ion signal where lignified cells are observed in the autofluorescence image. Lastly, the ion image m/z 607 shows highly localiz ed ion signal in the secondary xylem ( lignified cell walls ) Observing the autofluorescence of the lig nin more closely (Figure 5 7 b) shows a gradient of green fluorescence signal the most intense signal is observed closest to the pith and the

PAGE 148

148 least intense sign al s are observed in the bottom of the ion image This same trend can be correlated with the ion signal of m/z value 607 where more intense ions signal is observed in the region nearest the pith and decreased ion signal is observ e further away from the pith. This further suggests that m/z value 607 can identify regions lignified tissue and furthermore provide the relative lignin concentrations Comparison of ToF SIMS with MALDI LIT MS Figure 5 8 compa res microscopy, MALDI tandem MS imaging ToF SIMS imaging and fluorescence microscopy of Populus stem. To make a better comparison, the raster step size of the MALD tandem MS image displayed in Figure 5 8 b was reduced from 50 m (as used in Chapter 3) to 10 m ( which introduces a lar ge amount of oversampling). The tandem MS image of m/z 1319 995, identified as a cellulose ion, displays intense ion signal in the secondary xylem and the lignified phloem fiber cells. Figure 5 8 c shows ToF SIMS image of m/z 69 (different stem), which was identified to be characteristic of cellulose ions and Figure 5 8 d displays a CW fluor escence image of the same stem. Comparing the MALDI MS image with ToF SIMS image illustrates the dramatic increase in spatial resolution achieved with ToF SIMS imaging ( even when a significant amount of oversampling was performed for MALDI imaging). Specifically, the ToF SIMS image can resolve cell walls within the lignified phloem bundles; the MALDI MS image displays a large region of intense ion signal in the phloem fi ber bundles Since biomass research focuses on the thickened secondary cell walls, ToF SIMS is valuable to obtain information about the cell wall thickness; however, identifying only one classification of ions to a particular m/z value is less informative than the chemical information provided by MALDI MS imaging.

PAGE 149

149 Another diffe rence observed between MALDI MS imaging and ToF SIMS imaging is selectivity of the experiment. As discussed in Chapter 3, MALDI generates more intact ions, which helps to provide a more accurate identif ication. In addition, tandem MS imaging provides for selectivity by generating product ion spectra, which can distinguish between isobaric ions ToF SIMS, The small fragment m/z 69 could result from fragmentation of cellulose, hemicell ulose and/or lignin; fluorescence microscopy helps to provide a different approach to provide more positive compound identification. Negative Ion Mode The same tissue sections describe above were also analyzed in negative ion mode ; the results are display ed in Figures 5 9 and 5 10 For negative ion mode analyses, the same ionization parameters are used and negative ions are analyzed by switching the polarity of the ion extraction, ion optics and ESA voltages. Negative ion mode analysis, in general, resulte d in less characteristic ions and lower absolute abundance of the ions observed; however, some ions displayed localization that corresponded with lignified cell walls. As observed in Figures 5 9 c and 5 10 c, ion signal is observe d over the whole tissue and the image quality is similar to positive ion mode. The extracted ion image of m/z value 35 identified as a chloride ion (Cl ) displays ion signal over the whole tissue, but appears to be more abundant in the phloem regions. Since chloride is a common anion of salts, the localization is expected to be similar to the cations, K + displayed in the previous figures. However, the sensitivity of negative ion mode is likely lower than the sensitivity in positive ion mode due to the decrease in TI C Extracted ion images of m/z 80 and 97 (Figures 2 9 e f and 2 10 e f) are localized in the regions of lignified cells t entatively, m/z value 80 could be identified as [ SO 3 ] and

PAGE 150

150 m/z value 97 can be identified as [ H 2 PO 4 ] however the identification cannot verified without MS/MS or accurate mass data In addition, the localization of these particular ions cannot be explained based on known composition of wood tissue in literature searches Negative ion mode analyses were m ore exploratory in nature, and the results were not conclusive. However, despite the lack of identification, the localization of some ion signal was consistent with regions containing lignified cells identif ied by fluorescent microscopy. Further studies s uch as increasing the acquisition time at each tile, could be performed to increase the sensitivity of the experiment to help identify ions and possibly determine other characteristic ions of the different regions of wood tissue. Conclusions This chapter reported the successful analysis of int act Populus tissue sections using a ToF SIMS microscope imaging method. Different sample preparation techniques were used and the optimal spectra were observed when the sample (or standard) was coated with a thin lay er of gold. Since wood tissue is a non conductive surface, the gold coating helped to decrease charge build up on the sample surface, generate ions at the same initial potential energy and increase secondary ion yield of higher mass ions, which ultimately increased the sensitivity of the experiment. E xtracted ion images demonstrate the ability of ToF SIMS to provi de high spatial resolution, chemically selective images of intac t wood tissue sections. Several ions, although not positively identified, were lo calized in different regions of tissue and showed high correlated with fluorescence microscopy t hese ions can be considered characteristic of these tissue regions to be monitored. In addition, further data

PAGE 151

151 processing could help to increase the chemical inf ormation obtained from ToF SIMS imaging. In conclusion, the ability to correlate ToF SIMS images with other imaging techniques will help to provide a more comprehensive analysis of wood tissue. ToF SIMS imaging provides the spatial resolution needed to pr ovide more insight into spatial changes in chemical composition of pretreated wood tissue. Ultimately this technique could help to visualize the changes of LCM throughout pretreatment processing.

PAGE 152

152 Figure 5 1. Schematic of Physical Electronics TRIFT II ToF SIMS instrument. The secondary ions generated from the sample are guided through series of optics and three electrostatic analyzers ( ESA ) The spatial position of the secondary ions is conserved and detected w ith an MCP detector.

PAGE 153

153 Figure 5 2. Cartoon illustration representing secondary ionization. The high energy primary ion beam bombards the sample surface to release secondary particles, including positive/negative ions and neutrals.

PAGE 154

154 Figure 5 3. Compari son of ToF SIMS imaging modes. A) Unbunched. B) Bunched. The u nbunched mode offers superior spatial resolution at the cost of poorer mass resolution

PAGE 155

155 Figure 5 4 ToF SIMS spectrum of Birch Xylan e xtract coated with 2 nm gold. Spectra were sp lit to show lower abundance ion s. SIMS is a high energy ionization source, which results in high fragmentation and thus higher m/z ions are typically lower in abundance, as seen here.

PAGE 156

156 Figure 5 5. ToF SIMS spectra from different regions of interest in wood tissue. Colors correspond to the regions outlined in the optical image. Despite the different regions, most ions are similar showing the difficulties in identifying chemical markers for distinctive regions of wood tissue.

PAGE 157

157 Figure 5 6. O ptical, fluorescence and positive ion ToF SIMS images of phloem and xylem of Populus wood stem. Images show localization of different ions in different regions of wood tissue.

PAGE 158

158 Figure 5 7. Optical, fluorescence and positive ion ToF SIMS images of p ith and xylem of Populus wood stem. Images show localization of different ions in different regions of wood tissue.

PAGE 159

159 Figure 5 8. Comparison of different imaging techniques. A) The microscope image. B) MALDI LIT MS image of a cellulose ion. C) ToF SIMS im age. D) CW fluorescence image Reducing the raster step size to 10 m shows intense ion signal in the region of the lignified phloem fibers. Note A and B are images of the same stem. C and D are the same stem, but different from A and B.

PAGE 160

160 Figure 5 9 O ptical, fluorescence and negative ion ToF SIMS images of phloem and xylem of Populus wood stem. Images show localization of different ions in different regions of wood tissue.

PAGE 161

161 Figure 5 10 Optical, fluorescence and negative ion ToF SIMS ima ges of pith and xylem of Populus wood stem. Images show localization of different ions in different regions of wood tissue.

PAGE 162

162 CHAPTER 6 MULTIVARIATE ANALYSI S OF MALDI MS AND TOF SIMS IMAGING DATA Overview The large data sets generated by MS imaging experiments make a complete data analysis difficult i.e., a large portion of the data is discarded or overlooked M ultivariate and/or statistical analyses can overcome this difficulty by analyzing larger portions of the data set i n a timely manner. In addition multivariate analyses helps reduce the data set to explain differences observed in mass spectra in fewer variables. This is especially useful for a more comprehensive analysis of the complex mass spectra observed in MS imagi ng of tissue sections. Using multivariate analysis in chemistry is not a new idea. In the 1980s multivariate analys e s were developed to analyze the complex spectra of pyrolysis mass spectrometry (Py MS) data M any multivariate analysis methods required re ference spectra not feasible for Py MS data analysis of biological samples, such as yeasts and extracts 126 To overcome these limitations, mul tiv ariate methods that do not require calibrations were applied to chemical and biological data sets e.g., factor analysis and principal component analysis (PCA) 127, 128 Factor analysis and PCA are un supervised methods that efficiently describe differences between samples within a data set, which can be used to identify and quantify the components of a mixture. In contrast to supervised methods, unsupervised statistical methods are performed without p rior grouping (or assumptions) of the input data. 129 Multivariate analysis of Py MS data extract ed differences within a data set of related samples ; 127 this is consistent wi th the goal of matrix assisted laser

PAGE 163

163 desorption/ionization mass spectrometr (MALDI MS) and time of flight mass spectrometry ToF MS analysis of intact Populus tissue. This chapter dicusses th e adaptation of multivariate statistical method s for the analysis of MALDI MS data of the plant related standards discussed in Chapter 2. The statistical method provides the groundwork for using MALDI MS to determin e cellulose, hemicellulo se and lignin concentrations in intact lignocellulosic material (LCM) Further, an in house written statistical software program was used to analyze ToF SIMS imaging data to aid in identifying ions that are observed in different regions within an intact tissue section. Princip al Component Analysis PCA has been a p opular unsupervised, mu ltivariate analysis technique for chemists since the 1970 s due to the multivariate data collection inherent in many chemistry experiments. In the broadest scope PCA is considered a pattern recognition technique and specifically within mass spectrometry, PCA compares mass spectra within a data set to determine patterns in the mass spectra that describe the most variance between the samples. 130 Two important parameters of PCA include, (1) the number of princi pal components (PCs) that account for the variance within data set and (2) the scores and loadings of a PC, which is discussed in detail below. 130 Briefly describing the mathematics behind PCA the data is arranged in a data matrix, where the rows represent different samples and columns represent different variables, such as m/z values 131 Using linear algebra and abstract transformation s, PCA descr ibes data matrix as a product of column matrix and row matrix represented by ( 6 1 )

PAGE 164

164 Where is the scores and are the loadings (Figure 6 1) The PCA scores represent the projection of the data set on the determined PCs, and the PCA loadings are used to determine the contribution of the variable on each PC. The number of rows observed in and the number of columns observed in are limited by the number of rows and columns in the original data matrix T he common dimension represents the PCs and are linear combina tions of the original variables c alculated from PCA 130 The P Cs determined are chosen so that PC1 accounts for the most variability, PC2 accounts for the second most variability, etc, which helps with data deconvolution and identification of representative markers. Furthermore, PCA determine s the variance accounted for by a PC and often only 10 PCs (as opposed to hundreds or thousands of variables) can account for > 99% of the variability in the data set 129 PCA results are typically displayed in two (o r three) dimensi on al PCA s cores p plotted for the different PCs 130 An ideal PCA scores plot of three entirely different sam ples (three replicates of each) is displayed in Figure 6 2. PCA analysis should to each other, and should separate the different samples the plot in Figure 6 s are separated by PC1 and A is separated from B and C by PC2. In addition to sc sis PCA loadings plots are generated by plotting the Eigen value (loading) for each vector (representing each vari able) for a particular PC. Loading s can be both positive or negative and the absolute value of the loading represents the contribution to variance of the analysis (e.g, higher absolute value loading scores accounts for more variability)

PAGE 165

165 For example, Figur e 6 2 illustrates that B has a positive loading on PC 1 and a negative loading on PC2; thus it is determined that PC 1 is characteristic of sample B. Plotting that display ions cont ributing to the variance of PC1. These loadings plots should ideally resemble the raw mass spectrum of sample B. 126 These type s of analyses have been reported to determine the composition and concentration of complex mixtures. 126 In the example above if a MALDI MS analysis of an unknown mixture contain A, B and/or C was performed, the position of the unknown mixture data on the Scores Plot sho uld be indicative of the composition. For example, if the unknown mixture is plotted more closely to the C group than it could be concluded that the unknown mixture contains more of compound C than the other compounds if the mixture was composed of 33.3% of each A, B and C, than the mixture would appear In addition to analysis of standards and mixtures, PCA has been applied to both ToF SIMS microscope imaging dat a 132, 133 and more recently MALDI MS imaging data 134 136 Due to the high fragmentation (in the case of ToF SIMS) high complexity of the s pectra of any tissue analysis PCA is often necessary to analyze the MS imaging data to help differentiate regions of tissue. 133, 137 s are generated to visualize the differences between the regions. 133, 136 Once differences are recognized, the loadings can be used to identify characteristic ions. PCA is a complex series of mathematical manipulations and an entire PhD project could be aimed toward develo ping PCA for the analysis of MS imaging data ; t hus, PCA is often performed using purchased software packages and/or open source program s.

PAGE 166

166 Data pre processing, such as mean centering or nor malization is often necessary prior to any statistical analysis, which may or may not be included in the software pac kage T hese steps as well as other pre processing techniques, can alter PCA results dramatically and are important for successful PCA ana lyses. The techniques used for these analysis will be discussed later in the chapter. PCA of L ignocellulosic Material Standards Since lignocellulosic materials ( LCM ) are primarily composed of cellulose, hemicellulose and lignin, PCA experiments aimed to de termine the composition of each classification of compound within a tissue section were designed. Three replicates of a n average of 50 scans lignin extract were used for the experiments reported below. Software PCA performed on MALDI MS data used Tanagra v 1.4.36 (Lyon, France) and open source free DATA MINING software available for download ( http://er ic.univ lyon2.fr/~ricco/tanagra/index.html ). 138 Prior to analysis in the software, the raw data were exported from Qual Browser v. 2.0 (Thermo Scientific, Inc.) into Microsoft Excel to generate the original data matrix (Figure 6 3) and data preprocessing. MetaboAnalyst, 139 an online webser ver metabolomic data analysis system was also explored, but no results from those experiments are reported in this chapter. Dat a P rep rocessing One of the biggest challenges in data preprocessing was manipulating the data to ensure that the m/z values in each sample were in line with each other. When exporting data from Qual Browser, only the m/z values with ion intensities are exporting, leaving gaps where ions are not observed. Since different samples have different ions that are

PAGE 167

167 not observed in the mass spectrum, adjusting the data to have the m/z values in the proper columns to complete the ma trix was difficult. To overcome this difficulty a Macro program was used in Excel to insert the missing values in a series of m/z When inserting m/z values that are not observed in the mass spectra, an intensity value for that m/z value must also be ins erted. One of the solutions is to insert a zero for the missing intensity; however, this could lead to false representations of the PCA data. For example, if there is an ion that is not observed in any of the samples, then the intensity of that certain m/z will be zero and will be highly correlated to each other. In mass spectrometry, zero intensity is not always indicative of the absence of an analyte, thus PCA describing correlation between the samples based on the zero intensity is not a valid result. Di fferent methods have been developed to insert missing values (other than zero) to alleviate this challenge Some of the commo n methods that are used include inserting random values, inserting a small value (half of the minimum value observed) and inserting the mean value of the observed intensities 139 For these experiments, m/z values that were not present in 2/3 of the samples were discarded and small values were used for the intensities of inserted m/z Due to the irreproducibility of ion intensities between different MALDI MS analys e s a normalization procedure is important to help to increase the quality of the PCA analysis, i.e., without normalization the variations determined by PCA could be due to intrinsic MAL DI differences Some normalization methods include : (1) normalizing to the sum of the absolute value of all the variables, (2) normalizing to the sum of the square value of all variables for a given sample and (3) normalizing to the maximum value observed for all variables for the given sample. 139 These experiments normalized

PAGE 168

168 all the intensities to the maximum value observed for a particular sample, e.g., each intensity value of a m/z value was normalized to the average TIC of those spectra. PCA Results The optimized PCA score s plot of three replicate spots of MALDI MS analysis from three different standards, MCC, Birch xylan extract and Spruce lignin extract are displayed in Figure 6 4. As expected, the th ree replicates of each compound grouped closely together, and the three different groups formed a triangle. The plot also demonstrates that the variance of PC1 is due to the differences between lignin and cellulose the score of Birch xylan extract on PC1 i s nearly zero, ind icating PC1 is not affected by B irch xylan extract. The variance of PC2 describes the difference between Birch xylan from MCC and lignin. variables on the PC s d etermined Figures 6 5 through 6 7 displays the comparison between PC loading plots (purple) and the raw mass spectra (black) The PCA loading plots were prepared by extracting the loading values from the Tanagra program and exporting into excel. The value s were then multiplied by the average value for each m/z in order to put the intensities on a similar scale as the mass spectra. 126 The loadings plots wer e prepared using the Microscoft Excel graphing tool and the loading value is plotted versus the m/z so a direct comparison can be made with the mass spectra Figure 6 5 displays the positive loadings on PC1, which should be ch aracteristic of lignin extract based on the scores plot in Figure 6 4. The comparison of the PC1 loading plot (positive y axis) with the MALDI MS of lignin shows high correlation with ions observed at nearly every m/z and the distribution of the ions are similar. It is interesting to note, that in the MALDI MS of lignin, MCC ion s are observed (since this is an extract,

PAGE 169

169 other compounds are present); however, in the PC1 loadings plot, the MCC ions observed are removed by the PCA. This shows the ability of PCA to determine small differences between samples. Sim il arly, Figure 6 6 sho ws the PC1 negative loadings plot, compared with the MALD MS spectrum of MCC. Again, the PC factor spectrum is similar to MALDI MS spectrum, showing abundant ions 162 mass units apart, indicative of a polymer of a six carbon sugar. In addition, the PC1 loa dings plot also shows similar ion abundance distributions. Another example displayed in Figure 6 7, is the Birch xylan MALDI MS spectrum compared with the positive loadings of PC2, which shows similarities. The similarities between the PC loadings plots an d the MALDI MS analysis of the standards show the ability of PCA to determine PCs that are highly correlating with the ions observed in the MALDI MS of LCMs. After the analysis was optimized for the standards, PCA was used to analyze the standards as wel l as MALDI MS spectra of intact wood tissue (data not shown). The PCA scores plot showed that most of the wood samples were grouping close to the lignin. This was not expected since lignin is least abundant classification of compounds in wood tissue (when compared to cellulose and hemicellulose). However, after analyzing the data, the observed results were logical due to several reasons. For m/z which is similar to the wood tissue analysis, even though wood is not primarily composed of lignin. PCA is a n unsupervised, pattern recognition technique, thus it is reasonable that the analysis would group these together. In the future, these experiments will be performed using a supervised techniqu e, such as discrimina nt analysis (DA) or partial least squares DA

PAGE 170

170 (PLS DA) In supervise d techniques, the chemist provides information to the program on the identity of the classifications of compounds, which could help to correct for the intense ion signa l of lignin in LCMs. A nother difficulty with PCA analyses of MALDI MS is the ionization efficiency of each classification of compounds may be very different from each other. For example, hemicelluloses are more easily ionized than cellulose, thus, the ion signal from hemicellulose could be observed higher than cellulose. This would cause the PCA scores of the tissue to be more closely related to the xylan, even if the tissue is compose of more cellulose. Differences in ionizat ion efficiency could be corrected by using standards more representative of the samples, using a supervised method or perhaps modifying the program so that could it consider ionization efficiencies Despite the inability to apply MALDI MS spectra from int act tissue in to the PCA analysis, the ground work has been prepared for a multivariate analysis method for the analysis of LCM. PCA is capable of distinguishing between the different standards with high correlation of the factor loadings to the MALDI MS spe ctra. However, for intact tissue, other multivariate analysis methods could provide a more accurate analysis of components within the mixture. PCA and Hierarchical Cluster Analysis of ToF SIMS Imaging Data Chapter 5 reported localizations of ions from ToF SIMS imaging that were consistent with regions of different wood tissue. However, the raw data from ToF SIMS is difficult to analyze due to the high fragmentation observed from the primary ion gu n and the low ion signal (at higher m/z values) PCA analysis and hierarchical clustering (HC) was performed on the large imaging data sets to help determine ions that are

PAGE 171

171 characteristic of differe nt regions of wood tissue, thus improving the characterization of LCM. Hierarchical Clustering PCA is an unsupervised me thod that effectively displays variations of a large data set in fewer PC images ; 135 however, PCA does not allow for exploration of the data set based on spectral similaritie s, which limits the detail of analysis. HC methods overcome this limitations by generating groupings of similar spectra without prior knowledge of the sample (as in the case of supervised methods, such as DA). Moreover HC allows for interactive exploration of the data set that genera tes images based on mass spectral similarities Clustering method s in combination with PCA has been shown to provide more insight into histological of MS imaging data. 134 136 HC groups together similar mass spectra and builds a dendrogram of nodes from the bottom up ; the more similar spectra are, the closer they are grouped together. An illustration of a simplified HC analysis of five elements is di splayed in Figure 6 8. The first step in HC groups together two elements with the smallest different between them (i.e., most similarities); this group is now considered a new element. The next step groups together the next two elements with the smallest d istance between then, and his grouping becomes a new element. This process is repeated until all the elements have been clustered together. For MS imaging data sets, HC groups together imaging pixels based on the position in the multidimensional PCA space 134 After HC, the images are regenerated by plotting the mass spectra present at the selected node. The image changes as the different branches are selective, which allows for interactive exploration of the imaging data set. 136

PAGE 172

172 Software and Data Preprocessing Raw ToF SIMS imaging data w ere converted from WinCadence (Physical Electronics) into a MATLAB (The Mathworks, Inc.) format using a MATLAB program. The were l oaded into a n in house written statistics software, ChemomeTricks v 0.99993 beta (FOM Inst itute, AMOLF Amsterdam, The Nethe rland s) for PCA and cluster analyses. HC analysis was performed using the first 5 PCs and was used to remove silicone contaminants observed at m/z 7 3 and 1 47 After the silicone ions were removed from the data set, PCA was performed, and PC scores images were generated. Imaging PCA Results The PCA scores images of a Populus wood tissue section are displayed in Figure 6 9 and compared with the ToF SIMS TIC and fluorescence images. The PC scores image suggests that PC2 separate s the lignified from the n on lignified cells. Specifically, the positive scores on PC 2 (PC2+) image correlates well with the lignin autofluorescence image ; thus PC2 could be used to determine the ions that distinguish between lignified and non lignified cells. Figure 6 10 displays the positive and negative PC2 loadings plot (the PC scores images are disp layed next to each loading plot). The PC2+ loadings, which ar e localized in lignified cells are dominated by m/z values 69, 81, 13 3, 344 and 607. Although these ions are not positively id entified, predicted structures are also displayed in Figure 6 10 In addition to determining representative ions of different tissue regions, the PC scores images help to cluster different ions together The low ionization efficiency, especially with hi gher molecular weight ions requires long analyses and large data sets to obtain quality spectra for higher m/z value ion Instea d, PCA can help to identify the

PAGE 173

173 lower intensity, higher mass ions tha t would otherwise be overlooked, as well as distinguish series of ions that could be indicative of diffe rent regions (as opposed to plotting as single ion as in the case with most MS imag ing experiments ). The negative PC2 (PC2 ) loadings plot shows the v ascular cambium and the phloem are dominated by the salts, Na + and K + (Figure 6 10 ). The salts should be more abundant in the living cell region of the vascular cambium, which is observed in the PC image. Also observed in the PC loading plot is m/z 331 with a series of ions similar in m/z values which could be identified as different Gibberellins (GA) in the vascular cambium region. In addition, cellulose ions could be observed in these regions of the tissue, since cellulose is present in the phloem (just at lower abundance). The ions m/z values 102, 130 and 140 have all be reported as degradation products of cellulose during pyrolysis 140 so could be indicative of cellulose ions. It is also important to note the loading values of these predicted cellulose ions are smaller than the salts and the Gibberellins which is to be expected. The positive and negative scores image of PC1 are il lustrated in Figure 6 9 b c The PC1+ scores are localized in the regions where higher cellulose concentrations are expected, specifically, in the xylem a nd the lignified phloem fiber cells. The PC1+ loading plot (data not shown) is dominated by m/z values 130, 140 and 274 as previously mentioned, m/z values 130 and 140 have been reported as possible cellulose fragment ions The ion m/z values 274, is 144 Da higher than m/z 130, which is a commonly observed sugar fragment, thus would be consistent with the cellulose ion identification The m/z value 607 which show localization in lignified cells and dominated the PC2+ loading plot, is also present in the PC1+ loading plot; however, the

PAGE 174

174 relative loading value is lower than the other ions observed. These results further suggest that PC1+ is more characteristic of cellulose ions, and PC2+ is more characteristic of lignin ions Another example of the PCA analysis on ToF SIMS im agi ng is presented in Figure 6 11 The PC1 scores images show an efficient distinction between the pith, non lignified cells, and the xylem, lignified cells. The PC1 + loadings plot (Figure 6 12) is similar to the PC2 + loading plot ( Figure 6 9 ) where m/z v alue s 69 and 607 dominate the spectrum. This helps to further validate the PCA method, as well as validate the ions that have been assigned to a certain classification of compo unds. Furthermore, the PC2 scores image (Figure 6 1 1 ) shows a distinction betwee n the xylem cells and pith, and the PC 1 loading plot is displayed in Figure 6 12 The positive loadings plot is dominated by the m/z value 130, 140 and 274, which are the same ions observed in the previous example. In addition, the ions more characteristic of lignin, m/z value 69 and 607, are observed, but in lower relative abundance. Conclusions The analysis reported above showed the viability of PCA to distinguish between regions of different tissue composition within intact plant tissue. The lignified ve rsus the non lignified cells were efficiently separated. Furthermore the PC loading plots provided series of ions to monitor for different classifications of compounds In general, the non lignified, living regions of tissue were dominated by the Na + and K + ( m/z value 23 and 39) as well as m/z value 331, 389, which were identified as GAs The lignified regions of tissue were dominated by m/z value 69, 81, 334, and 607 these ions provided a localization that was most similar to the lignin autofluoresce nce

PAGE 175

175 Taking a closer look at the PC loadings plots, the m/z values 130, 140 and 102 were present in both the lignified and non lignified regions, which is consistent with known composition of plants Cellulose is present in non lignified cells, but believ ed to be in lower abundance and is consistent with the observations of the low loadings on the PCs characteristic of these cell types. In conclusion, PCA offers insight into the MS analysis of different regions. Due to the high fragmentation and com plex ToF SIMS spectra, PCA helps to reduce the data set and determine differences that can otherwise be overlooked. In addition, PCA can also help to identify between different fragment ions at the same m/z This is important since cellulose, hemicellulose and lignin often result in the same molecular weight fragment ions and p lotting only one particular m/z at a time could result in uninformative, non selective images. Furthermore, PCA could become a useful tool to analyze differences between chemically p retreated and untreated wood samples. Since ions are abundant and nearly every m/z ratio, PCA can identify ions at lower abundance that change as a result of the pretreatment, which can optimistically provide insight to develop a more efficient pretreatmen t process.

PAGE 176

176 Figure 6 1. Graphical representation of PCA analysis. Original data matrix is transformed into two different matrices, called the loadings and the scores matrix.

PAGE 177

177 Figure 6 2. Ideal PCA plot of three pure standards. PCA analyse s of pure standards will help to determine the compon ents and concentrations within unknown mixture s

PAGE 178

178 Figure 6 3. Data matrix set up for PCA analysis. Three replicates of each standard are in the rows and each column represents a different m/z.

PAGE 179

179 Figure 6 4. PCA Scores plot of MCC, xylan and lignin analysis by MALDI MS The three replicates of the same standard group together and PCA analysis forms a triangle, which is consistent with the ideal PCA analysis displayed in Figure 6 2. Data points are located in the center of each square label.

PAGE 180

180 Figure 6 5. PC loadings plot compared with MALDI MS spectrum of lignin A) Positive PC1 loadings plot. B) MALDI MS of lignin. The loadings scores are representative of the MALDI MS of the lignin, which shows PCA is separating based on the correct ions.

PAGE 181

181 Figure 6 6. PC loadings plot compared with MALDI MS spectrum of MCC. A) Negativ e PC1 factor loadings. B) MALDI MS of MCC. The negative loadings show characteristic ions that are observed in the MALDI MS of the pure standard.

PAGE 182

182 Figure 6 7. PC loadings plot compared with MALDI MS spectrum of Birch xylan. A) Positive PC2 factor loadings. B) MALDI MS of Birch x ylan. The PC2 loading show similar abundant ions as the MALDI MS analysis and appears to reduce the number of ions between the linear xylans.

PAGE 183

183 Figure 6 8. Illustration of hierarchical clustering adapted from Deininger, S. et al Proteome Res. 2008 7 5 230 5236. A) Five elements subjected to HC. B) Two closest elements are grouped together and are now considered one element. C) The next two closest elements are grouped together and considered one element. D) The next closest elements are clustered togeth er, forming another cluster. E) Complete HC of the five elements.

PAGE 184

184 Figure 6 9 Fluorescence images compared with PCA images generated by plotting the positive and negative loadings of PC1 ( top) and PC2 (bottom) of ToF SIMS imaging data. PC1 different iate between cellulose ions and other primary cell wall material The PC2 separates effectively between lignified and non lignified cells

PAGE 185

185 Figure 6 10 PC2 loading plot and corresponding images. PC1+ is more characteristic of lignified cells an d provides series of ions that are characteristic of lignin ions. Structures drawn are predicted structures; however, they could not be positively identified. PC2 shows more salt signal and ions that are consistent with the m/z of Gibberellins

PAGE 186

186 Figure 6 11 PCA images of the pith and xylem. PC1 efficiently distinguishes the regions of the pith and the xylem. PC2 identifies signals characteristic of cellulose.

PAGE 187

187 Figure 6 12 PC1 loading plot and corresponding images The positive loadings show more characteristic of cellulose ions

PAGE 188

188 CHAPTER 7 SUMMARY AND FUTURE D IRECTIONS The analysis of LCMs is difficult due to the number and complexity of the compounds as well as the spatial organization of these compounds. This dissertation reports the develop ment of methods that reduce the complexity and offer high chemical selectivity for the analysis of intact LCM s; specifically mass spectrometric imaging (MS imaging ) techniques were utilized in the imaging and characterization of intact Populus wood tissue sections. Prior to intact tissue analysis a matrix assisted laser desorption/ionization linear ion trap mass spectrometric ( MALDI LIT MS ) method was developed to characterize standard materials that were anticipated to be present within the wo od tissue sections microcrystalline cellulose ( MCC ) Birch Xylan and Spruce lignin The MALDI tandem MS method ionized these compounds and determined the m/z values to monitor in the intact tissue section studies. Furthermore, tandem MS was useful in identifying common NLs and common fragment ions observe d for each classification of compounds. Although these standards were adequate for method development, fut ure work should include a large r library of standards specifically, standards that are directly related to tissue type. For example, Populus was used for all intact wood tissue analyses ; however, the hemicellulose standard that was used was from Birch and the lignin standard was from Spruce. It is not anticipated that the hemicellulose and lignin varies greatly between species, but differences may occur, which could affect the intact tissue analy sis In addition, the standard an alyses were more difficult s ince two of the in addition to the compound of interest. It is difficult to overcome this limitation, but perhaps a purification

PAGE 189

189 step, prior to MS analysis, could result in higher purity standards especially in the case of lignin. In addition to standard analyses, a MALDI tandem MS imaging for the analysis of intact LCM s was developed. Other analytical methods used to ionize LCM may suffer higher s patial resolution than MALDI MS imaging but the chem ical selectivity is typically less MS imaging provides a way to display the localization of specific compounds with confidence in compound identification A major advantage of the developed technique over the existing techniques is the abilit y to perform tandem MS experiments. It was demonstrated that tandem MS improved the sensitivity and selectivity of the experiment my isolating a small range of ions, distinguished between isobaric species and identified unknown ions observed from wood tiss ue. Despite the advantages of this technique, future work could improve the data obtained from the analysis. T he major drawback to MALDI MS imaging is that the spatial resolution is inadequate to map distinct features within the wood tissue sections. Futur e experiments will aim to improve t he spatial resolution of the MS imaging experiment by reducing the laser spot size. Reducing the laser spot size is practical in a physical sense, but could lead to MALDI related challenges For example, reducing the laser spot size is present in a single sample spot will be lower and coul d be below the limit of detection of the experiment. This is problematic, especially in MALDI, due to the hi gh MALDI matrix ion signal observed. Fortuantely, MS/MS will provide lower limits of detection. Furthermore, reducing the spot size increases the fluence of the laser and could adversely affect the measurements.

PAGE 190

190 Since this dissertation was not aimed toward reducing the spot size, the limited spatial resolution was circumvented by implementing other analytical methods that offer superior spatial resolution, e.g., fluorescence microscopy and ToF SIMS imaging. Fluorescence microscopy is commonly used for the a nalysis of wood tissues and offers high spatial resolution. However, the fluorescence labels provide less chemical selectivity and are can also be expensive. In this dissertation, fluorescence microscopy was used to validate the MALDI MS imaging method as well as to make comparisons to help increase the comprehensive understanding of the chemical composition of intact wood tissue. Fluorescence microscopy also aided in determining difference s in cellulose concentration within one intact tissue section whi ch could be useful in later developing a semi quantitative method. One of the experiments that could be performed is to develop a way to quantify the cellulose ions observe in MALDI by using CW fluorescence microscopy. For example, the fluor escence signal observed from CW is directly correlated with the abundance of cellulose present in the tissue. Correlating the fluorescence signal with the observed MALDI analyte ion si gnal could prove to be useful calibration relative abunda nce of cellulose (or available cellulose). This information could ultimately be used in the analysis of LCM to determine the accessibility of cellulose before and after a chemical pretreatment. ToF SIMS imaging produced high spatial resolution MS images in which features not resolved in the MALDI MS image were observed. Despite the superior spatial resolution, the identification of the ions was difficult due to extensive fragmentation observed in SIMS spectra, making it difficult to distinguish between diff erent

PAGE 191

191 classifications of polysaccharides. However, ToF SIMS was successful in distinguishing between different regions of tissue, specifically lignified versus non lignified tissues. This is useful in the analysis of LCMs because this differentiation is im portant during pretreatment for LCM conversion into fuel. As analytical chemists, it is important to identify ions that are observed in spectra, but with ToF SIMS, ion identification is limited. However, in LCM analysis the ability to determine different regions, chemically and spatially could provide more insight into the pretreatment process than what is already known. In other words, know ing the exact identification of the compound observed is not necessary to analyze the material Ions that are characteristic of certain compound s were identified, specifically lignin ions. Since one of the biggest difficulties in understanding pretreatment is the spatial changes of lignin in the tissue after pretreatment, ToF SIMS could prove to show chemically sp ecific information, without knowing the exact identity of the ion. Future experiments will be aimed toward analyzing both untreated and pretreated wood tissue samples to help determine spatial changes in the ions identified in this dissertation. MS and mo re specifically, MS imaging generates large data sets with many variables that can easily be overlooked. Multivariate analyses are often useful in analyzing data with a large amount of variables. Princip al component analysis (PCA) was used to analyze pure standa rds, as a way to develop a semi quantitative analysis for intact plant tissue sections. Future experiments should be aimed toward analyzing intact wood tissue section s and perhaps generating internal calibrations of different regions of tissue that are known to having different concentrations of cellulose, hemicellulose and lignin.

PAGE 192

192 Lastly, PCA proved to be an invaluable tool identifying ions or series of ions that are characteristic of different regions of wood tissue Specifically, the lignified and non lign ified cells were easily distinguishable using PCA analysis. In addition, PC loadings were used to generate PC images, which also correlated well with fluorescence microscopy. I n conclusion, this dissertation outlines different analytical techniqu es that can be used for the analysis of LCM. It is clear that one method does not provide all the characterization needed for LCM analysis. Instead, correlating the strengths of each technique proved to provide a more comprehensive understanding of LCM. Ul timately, these can be used to characterize the materials that are useful for processing into fuel sources and aid in improving the efficiency of the overall process.

PAGE 193

193 A PPENDIX A DIFFICULTIES OF MALDI TIME OF FLIGHT ION MOBILITY SPECTROMETRY OF PLANT REL ATED STANDARDS AND POPULUS TISSU E Overview A goal of th e research presented herein was to compare MALDI LI T MS with MALDI quadrupole time of flight (Q ToF) MS of plant related standard and intact tissue. However, generating trends and correlations were difficult due to the unexpected results obtained from MALDI ToF MS experiments. This Appendix briefly reports the experiments performed and results that were obtained from MALDI Q ToF MS and LDI ToF MS analys is performed on the Waters Synapt mass spectrometer Experimental Standard Analysis Standards and MALDI matrices used were prepared as described in Chapter 2. Briefly, the suspensions of the plant related standards were prepared at 4 mg/mL in water. The MALDI matrix, DHB, was prepared at varying concentrations (~0.5 20 mg/mL) to determine the optimal matrix to analyte ratio. The standards were prepared for MALDI analysis by pipetting 1 L of the standard, followed by 1 L of the MALDI matrix onto a stai nless steel MALDI sample plate. Instrumentation and Traveling Wave Ion Mobility Parameters A Waters Synapt G1 ( Waters Corp., Milford, M A ) was used for MALDI Q ToF MS analyses. The MALDI sour ce is equipped with a frequency tripled, solid state, Nd:YAG laser (355 nm) with an ad justable repetition rate up to 2 50 Hz The laser energy was adjusted from about 200 350 au, which were values typically used for experiments with

PAGE 194

194 this instrument. The number of laser shots per spot was adjusted from about 200 300. The MALDI source pressure approximately 10 2 mTorr. The traveling wave velocity and height are parameters that can be altered using the soft ware provided to operate the instrument The wave velocity was 500 m/s and the wave height was 12 volts peak to pe ak Results and Discussion The analys es of the q plant related standards that were discussed in Chapter 2 repeated on the MALDI Q ToF mass spectrometer, were not successful. The spectra obtained showed mostly MALDI matrix (DHB) ions (data not shown). In ad dition, the m/z values that were hypothesized to be present in the samples were isolated within the quadrupole and fragmented for MS/MS but the data again was not comparable to MALDI LIT MS experiments. The major different between the instruments is the MALDI laser source which could explain the discrepancies between the results The MAL D I LTQ XL is equipped with a 60 Hz, N 2 (337 nm) laser and the Synapt is equipped with a 250 Hz, solid state, tripled Nd:YAG laser (355 nm). The biggest differences between the lasers are the repetition rate, laser profile and fluence the effects of these parameters on MALDI ToF MS sensitivity has been discussed in the literature 141 It is r eported that N 2 lasers generate higher total ion intensity at the same laser fluence when compared with the frequency tripled Nd:YAG laser. This ult imately leads to improved MALDI ToF MS sensitivity when using a N 2 laser, assuming all other parameters are equal. In addition to these parameters, there is a small difference in the laser wavelength; however, this is unlikely to cause such a dramatic difference in the spectra. Although these differences

PAGE 195

195 are present, there is not an obvious reason as to why the fluence and different wavelength would have such a dramatic effect on the spectra observed. For intact Populus tissue analyses, the sample preparation discussed in Chapter 3 was followed for analysis on the Synapt mass spectrometer but conductive, double sided tape was used instead for the Q ToF MS analysis. A comparison of the MALDI spectra obtained from the two different instruments from m/z values 500 1300 is d isplayed in Figure A 1. The top spectrum in Figure A 1 displays i ons that were above the backg round in the MALDI Q ToF MS analys i s; however these ions could not be identified. The m/z difference s between the peaks were not consistent across the whole spectrum and did not correspond to the masses of sugar ions that were observed in MALDI LTQ spectra. The ions observed in the MALDI Q ToF MS analysis might have been assigned as MALDI matrix (DHB) cluster ions but the m/z values were not consistent with DHB cluster ions observed with the MALDI LTQ instrument. The ions observed in th e MALDI ToF MS analysis were isolated within the quadrupole and fragmented for MS/MS to aid in ion identification The NLs observe d were not consistent with previous analyses, and were likely due to DHB cluster ions, and thus remaining unidentified despite tandem MS capabilities. One of the main goals of using the Waters Synapt instrument was to perform an ion mobility separation prior to MS analys i s, which was hypothesized to reduce spectral complexity. However, for MALDI analysis of the intact tissue sect ion, the ion mobility plot only had one grouping, thus no separation was observed. Ion mobility p arameters were altered to try to obtain better separation; however, this was still unsuccessful in trying to separate the ions generated from the wood tissue.

PAGE 196

196 After the unsuccessful attempt to compare MALDI MS between the two instruments, LDI (i.e., with no additional MALDI matrix was applied atop the tissue) experiments were performed on the Populus tissue. For these experiments, the laser energy was incr eased and the laser raster step size was decreased from 200 m to 50 m. In contrast to the MALDI analysis, the ion mobility plot from LDI showed separation resulting in four different mobility regions (Figure A 2). Th is ion mobility plots ion intensity at each m/z value vs. the drift time; thus isobaric ions can potentially be separated by their drift times which differ with differences in collisional cross sections The mass spectra from regions outlined in white are displayed in Figure A 3. The top spectrum in Figure A 3 corresponds to the highest drift time mobility line (farthest right in Figure A 2) and displays ions at nearly every m/z value but with a dramatic increase in ion intensity above m/z 700 The center mass spectrum corresponds to the middle mobility line and shows an intense ion at m/z 720 as well as a distinct patter n of ions w ith a mass difference of 24. After further investigation, it is believed that this is due to the formation of carbon clusters (similar to a Buckminsterfullerene 142 ) from the tissue surface. The ions observed at m/z value 720 corresponds to the radical cation of C 60 and the difference between the other ions observed is 24 which the mass of C 2 In one of t he original report s of fullerenes, C 60 clusters were formed by focusing a second harmonic (532 nm) Q switched Nd:YAG laser onto a graphite sample and the ions produced were detected with ToF MS. 142 T he most stable co nfiguration of the clusters observed was C 60 and only even numbered carbon clusters were observed, which explains the difference of 24 mass units. In the bottom spectrum (F igure A 4c) from the left ion mobility regions ions that are 12 m/z

PAGE 197

197 units apart were observed After a closer look at the isotope distribution, these ions were identified as doubly charged carbon clusters ( m=24, z =2, so m/z =12) Since wood tissue is composed of mostly carbon, oxygen and hydrogen, it would be possible that a carboni zation can occur as a result of the rapid heating and increased energy from the Nd:YAG laser used for LDI. It is also interest ing to note that more abundant carbon cluster ion signal is observed when the laser r aster size is reduced; at 200 m raster step sizes little carbon cluster ion signal was observed, but at 50 m step sizes, a significant increase of carbon cluster ion signal was observed. Rastering at 50 m step sizes results in oversampling, increas ing the heat and energy that is imparted on one a rea of tissue which should lead to an increase the formation of the carbon clusters. In order to confirm that the ions observed at m/z 720, and the subsequent ions observed 24 m/z units higher, corresponded to carbon clusters, the wood sample was analyzed with a MALDI FT ICR to obtain high mass resolution data. The 7T FT ICR was equipped with an Nd:YAG laser and a home built sample stage. The LDI analysis produced the same carbon clusters that were observed on the MALDI ToF MS. The recorded m/z value of 720 confirmed that the ion observed contained only carbon atoms, thus confirming that this was a C 60 cluster. Although the MALDI QToF results reported have no immediate implications, it is interesting, nonetheless, to report the laser induced fo rmation of carbon clusters from wood tissue. Furthermore, it would be interesting to perform MS imaging using th ese instruments to plot the localization of the C 60 ion to determine the regions of tissue that produce the most carbon clusters. One of the dif ficulties of converting biomass into fuel

PAGE 198

198 is the need to remove the high oxygen content of the carbohydrates present in wood tissue. Observing the localization of carbon clusters from wood tissue might help to identify the characteristic tissues that could most efficiently produce biofuels.

PAGE 199

199 Figure A 1. Comparison of MADLI MS of wood tissue from two different instruments. A) MALDI Q ToF MS on Waters Synapt. B) MALDI L IT MS on Thermo MALDI LTQ XL The dramatic increase in ion signal above m/z 700 is observed in all spectra.

PAGE 200

200 Figure A 2. Ion mobility plot of LDI analysis of intact wood tissue sections. The different regions outline represent compounds with different drift times, which may be related to the collisional cross section.

PAGE 201

201 Figure A 3. Mass spectra from the different regions of the ion mobility plot from Figure A 2 A) S pectrum that looks similar to LDI spectrum from MALDI LIT. B) I ntense ion at m/z 720, which was identified as C 60 cluster. The ions observed at higher m/z v alue s are 24 mass units apart. C) doubly charged carbon cluster

PAGE 202

202 LIST OF REFERENC E S (1) Himmel, M. E.; Ding, S.; Johnson, D. K.; Adney, W. S .; Nimlos, M. R.; Brady, J. W.; Foust, T. D. Science 2007 315 804 807. (2) Sun, Y.; Cheng, J. Bioresour. Technol. 2002 83 1 11. (3) Wooley, R.; Ruth, M.; Glassner, D.; Sheehan, J. Biotechnol. Prog. 1999 15 794 803. (4) Biofuels Solid, liquid or gas Fuels made f rom Biomass. (accessed December 7, 2010). (5) Service, R. F. Science 2007 315 14 88 1491. (6) Sluiter, J. B.; Ruiz, R. O.; Scarlata, C. J.; Sluiter, A. D.; Templeton, D. W. J. Agric. Food Chem. 2010 58 9043 9053. (7) Zaldivar, J.; Nielsen, J.; Olsson, L. Appl. Microbiol. Biotechnol. 2001 56 17 34. (8) Wyman, C. E. Bioresour. Technol. 1994 50 3 16. (9) Duff, S. J. B.; Murray, W. D. Bioresour. Technol. 1996 55 1 33. (10) Katzen, R.; Fowler, D. E. Appl. Biochem. Biotechnol. 1994 45 6 697 707. (11) Grethlein, H. E.; Converse, A. O. Bioresour. Technol. 1991 36 77 82. (12) Lora, J. H.; Wayman, M. Tappi 1978 61 47 50. (13) Dale, B. E.; Henk, L. L.; Shiang, M. Dev. Ind. Microbiol. 1985 26 223 233. (14) Esteghlalian, A.; Hashimoto, A. G.; Fenske, J. J.; Penner, M. H. Bioresour. Technol. 1997 59 129 136. ( 15) Chum, H. L.; Johnson, D. K.; Black, S.; Baker, J.; Grohmann, K.; Sarkanen, K. V.; Wallace, K.; Schroeder, H. A. Biotechnol. Bioeng. 1988 31 643 649. (16) Mcmillan, J. D. Pretreatment of Lignocellulosic Biomass. In Enzymatic Conversion of Biomass for Fuels Production Himmel, M.E., Baker, J. O., Overend, R.P., Eds.; ACS Symposium Series 566; American Chemical Society: Washington, DC, 1994 pp 292 324. (17) Grous, W. R.; Converse, A. O.; Grethlein, H. E. Enzyme Microb. Technol. 1986 8 274 280.

PAGE 203

203 (18) Lloyd, T. A.; Wyman, C. E. Bioresour. Technol. 2005 96 1967 1977. (19) Chundawat, S. P. S.; Donohoe, B. S.; Sousa, L. d. C.; Elder, T.; Agarwal, U. P.; Lu, F.; Ralph, J.; Himmel, M. E.; Balan, V.; Dale, B. E. Energy Environ. Sci. 2011 4 973 984. (20) Ding, S.; Xu, Q.; Ali, M. K.; Baker, J. O.; Bayer, E. A.; Barak, Y.; Lamed, R.; Sugiyama, J.; Rumbles, G.; Himmel, M. E. BioTechniques 2006 41 435 443. (21) Ishida, N.; Koizumi, M.; Kano, H. Ann. Bot. 2000 86 259 278. (22) Bulcke, J. v. d.; Mas schaele, B.; Dierick, M.; Acker, J. v.; Stevens, M.; Hoorebeke, L. v. Int. Biodeterior. Biodegrad. 2008 61 278 286. (23) Agarwal, U. P. Planta 2006 224 1141 1153. (24) Gierlinger, N.; Schwanninger, M. Plant Physiol. 2006 140 1246 1254. (25) Yarbro ugh, J. M.; Himmel, M. E.; Ding, S. Y. Biotechnology for Biofuels 2009 2 17. (26) Zaia, J. Mass Spectrom. Rev. 2004 23 161 227. (27) Gross, J. H. Mass Spectrometry A Textbook; Springe r Verlag: Berlin, Germany, 2004. (28) de Hoffmann, E.; Stroobant, V. Mass Spectrometry Principles and Applications, Second Edition; John Wiley & Sons, LTD: West Sussex, England, 2002 (29) Fournet, B.; Dhalluin, J. M.; Strecker, G.; Montreuil, J.; Bosso, C.; Defaye, J. Anal. Biochem. 1980 108 35 56. (30) Barber, M.; Bo rdoli, R. S.; Sedgwick, R. D.; Tyler, A. N. J. Chem. Soc. Chem. Commun. 1981 7 325 327. (31) Barber, M.; Bordoli, R. S.; Sedgwick, R. D.; Tyler, A. N. Nature 1981 293 270 275. (32) Reinhold, V. N.; Reinhold, B. B.; Costello, C. E. Anal. Chem. 1995 67 1772 1784. (33) Yamashita, M.; Fenn, J. B. J. Phys. Chem. 1984 88 4451 4459. (34) Harvey, D. J. International Journal of Mass Spectrometry 2003 226 1 35. (35) Watson, J. T.; Sparkman, D. O. Introduction to Mass Spectrometry Instrumentation, Appl ications and Strategies for Data Interpreatation; John Wiley & Sons Ltd.: Chinchester, West Sussex, 2007

PAGE 204

204 (36) Domon, B.; Costello, C. E. Glycoconj. J. 1988 5 397 409. (37) Mueller, D. R.; Domon, B. M.; Blum, W.; Raschdorf, F.; Richter, W. J. Biomed a nd Environ l Mass Spectrom 1988 15 441 446. (38) Laine, R. A.; Pamidimukkala, K. M.; Hall, R. W.; French, A. D.; Matta, K. L. Glycoconj. J. 1988 5 292 292. (39) Karas, M.; Bachmann, D.; Bahr, U.; Hillenkamp, F. Int. J. Mass Spectrom. Ion Process. 19 87 78 53 68. (40) Hillenkamp, F.; Karas, M.; Beavis, R. C.; Chait, B. T. Anal. Chem. 1991 63 A1193 A1202. (41) Stahl, B.; Steup, M.; Karas, M.; Hillenkamp, F. Anal. Chem. 1991 63 1463 1466. (42) Wang, B. H.; Dreisewerd, K.; Bahr, U.; Karas, M.; Hi llenkamp, F. J. Am. Soc. Mass Spectrom. 1993 4 393 398. (43) Dreisewerd, K. Chem. Rev. 2003 103 395 426. (44) Hillenkamp, F.; Karas, M. Int. J. of Mass Spectrom. 2000 200 71 77. (45) Karas, M.; Kruger, R. Chem. Rev. 2003 103 427 439. (46) Knochenmuss, R.; Zenobi, R. Chem. Rev. 2003 103 441 452. (47) Knochenmuss, R.; Stortelder, A.; Breuker, K.; Zenobi, R. J of Mass Spectrom 2000 35 1237 1245. (48) Jaskolla, T. W.; Karas, M. J. Am. Soc. Mass Spectrom. 2011 22 976 988. (49) Berk enkamp, S.; Menzel, C.; Karas, M.; Hillenkamp, F. Rapid Commun. Mass Spectrom. 1997 11 1399 1406. (50) Burns, M. S. Ultramicroscopy 1988 24 269 281. (51) Griffiths, J. Anal. Chem. 2008 80 7194 7197. (52) Walker, A. V. Anal. Chem. 2008 80 8865 88 70. (53) Sodhi, R. N. S. Analyst 2004 129 483 487. (54) McDonnell, L. A.; Heeren, R. M. A. Mass Spectrom. Rev. 2007 26 606 643.

PAGE 205

205 (55) Touboul, D.; Halgand, F.; Brunelle, A.; Kersting, R.; Tallarek, E.; Hagenhoff, B.; Laprevote, O. Anal. Chem. 2004 7 6 1550 1559. (56) Scripps Center for Metabolomics and Mass Spectrometry. http://masspec.scripps.edu/mshistory/whatisms_details.php (accessed January 10, 2011). (57) Scigelova, M.; Makarov, A. Proteomics 2006 6 16 21. (58) March, R. E.; Todd, J. F. J., Eds.; Practical Aspects of Ion Trap Mass Spectrometry; CRC Press: New York, NY, 1995; Vol. 1. (59) Schwartz, J. C.; Senko, M. W.; Syka, J. E. P. J. Am. Soc. Mass Spectrom. 2002 13 659 669. (60) March, R. E.; Todd, J. F. J. Quadrupole Ion Trap Mass Spectrometry; Wiley & Sons, Inc.: Hoboken, New Jersey, 2005; (61) Douglas, D. J.; Frank, A. J.; Mao, D. M. Mass Spectrom. Rev. 2005 24 1 29. (62) Stafford, G. C.; Kelley, P. E.; Syka, J. E. P .; Reynolds, W. E.; Todd, J. F. J. Int. J. Mass Spectrom. Ion Process. 1984 60 85 98. (63) Johnson, J. V.; Yost, R. A.; Kelley, P. E.; Bradford, D. C. Anal. Chem. 1990 62 2162 2172. (64) Magparangalan, D. P.; Garrett, T. J.; Drexler, D. M.; Yost, R. A. Anal. Chem. 2010 82 930 934. (65) Landgraf, R. R.; Conaway, M. C. P.; Garrett, T. J.; Stacpoole, P. W.; Yost, R. A. Anal. Chem. 2009 81 8488 8495. (66) Hager, J. W. Rapid Commun. Mass Spectrom. 2002 16 512 526. (67) de Hoffmann, E.; Stroobant, V. In Mass Spectrometry Principles and Applications; John Wiley & Sons, Ltd.: Chichester, New York, 2002 . (68) Chandra, S.; Chabot, J. F.; Morrison, G. H.; Leopold, A. C. Science 1982 216 1221 1223. (69) Pacholski, M. L.; Winograd, N. Chem. Rev. 1999 99 2977 3006. (70) Denoyer, E.; Vangrieken, R.; Adams, F.; Natusch, D. F. S. Anal. Chem. 1982 54 26 41.

PAGE 206

206 (71) Wechsung, R.; Hillenkamp, F.; Kaufmann, R.; Nitsche, R.; Vogt, H. Mikroskopie 1978 34 47 54. (72) Schwartz, S. A.; Reyzer, M. L.; Caprioli R. M. J. Mass Spectrom. 2003 38 699 708. (73) Kruse, R.; Sweedler, J. V. J. Am. Soc. Mass Spectrom. 2003 14 752 759. (74) Baluya, D. L.; Garrett, T. J.; Yost, R. A. Anal. Chem. 2007 79 6862 6867. (75) Todd, P. J.; Schaaff, T. G.; Chaurand, P.; C aprioli, R. M. J. Mass Spectrom. 2001 36 355 369. (76) Schueler, B. W. Microsc. Microanal. Microstruct. 1992 3 119 139. (77) Takats, Z.; Wiseman, J. M.; Gologan, B.; Cooks, R. G. Science 2004 306 471 473. (78) Ifa, D. R.; Gumaelius, L. M.; Eberlin, L. S.; Manicke, N. E.; Cooks, R. G. Analyst 2007 132 461 467. (79) Kertesz, V.; Ford, M. J.; Van Berkel, G. J. Anal. Chem. 2005 77 7183 7189. (80) Ellis, S. R.; Wu, C.; Deeley, J. M.; Zhu, X.; Truscott, R. J. W.; Panhuis, M. I. H.; Cooks, R. G.; Mitchell, T. W.; Blanksby, S. J. J. Am. Soc. Mass Spectrom. 2010 21 2095 2104. (81) Brown, R. S.; Lennon, J. J. Anal. Chem. 1995 67 1998 2003. (82) Garrison, B. J.; Delcorte, A.; Zhigilei, L.; Itina, T. E.; Kra ntzman, K. D.; Yingling, Y. G.; McQuaw, C. M.; Smiley, E. J.; Winograd, N. Appl. Surf. Sci. 2003 203 69 71. (83) Troendle, F. J.; Reddick, C. D.; Yost, R. A. J. Am. Soc. Mass Spectrom. 1999 10 1315 1321. (84) Garrett, T. J.; Yost, R. A. Anal. Chem. 2 006 78 2465 2469. (85) Garrett, T. J.; Prieto Conaway, M.; Kovtoun, V.; Bui, H.; Izgarian, N.; Stafford, G.; Yost, R. A. Int. J. of Mass Spectrom. 2007 260 166 176. (86) O'Connor, P. B.; Mirgorodskaya, E.; Costello, C. E. J. Am. Soc. Mass Spectrom. 2 002 13 402 407. (87) Makarov, A. Anal. Chem. 2000 72 1156 1162. (88) Hardman, M.; Makarov, A. A. Anal. Chem. 2003 75 1699 1705.

PAGE 207

207 (89) Surfer Surface Mapping System, version 8.05; Golden Software, Inc.: Golden, CO, 2004. (90) Ritter, S. K. Chem. Eng. News 2008 86 59 68. (91) WKU Bio 113 Carbohydrates. http://bioweb.wku.edu/courses/BIOL115/Wyatt/Biochem/Carbos.htm (accessed Aug. 11, 2009). (92) Mosier, N.; Wyman, C.; Dale, B.; Elander, R.; Lee, Y. Y.; Holtzapple, M.; Ladisch, M. Bioresour. Technol. 2005 96 673 686. (93) Reis, A.; Pinto, P.; Evtuguin, D. V.; Neto, C. P.; Domingues, P.; Ferrer Correia, A. J.; Rosario, M.; Domingues, M. Rapid Commun. Mass Spectrom. 2005 19 3589 3599. (94) Yokoyama, T.; Kadla, J. F.; Chang, H. M J. Agric. Food Chem. 2002 50 1040 1044. (95) Ingle, J. D.; Crouch, S. R. Spectrochemical Analysis; Prentice Hall, Inc/: Upper Saddle River, NJ, 1988. (96) Guo, Z.; Lin He Analytical and Bioanalytical Chemistry 2007 387 1939 1944. (97) US Departmen t of Energy Biomass Program: Feedstock Composition Glossary. http://www1.eere.energy.gov/biomass/feedstock_glossary.html (accessed December 20, 2010). (98) Sykes, R.; Kodrzycki, B.; Tuskan, G.; Foutz, K.; Davis, M. Wood Sci. Technol. 2008 42 649 661. ( 99) Bose, S. K.; Francis, R. C.; Govender, M.; Bush, T.; Spark, A. Bioresour. Technol. 2009 100 1628 1633. (100) Ragauskas, A. J.; Williams, C. K.; Davison, B. H.; Britovsek, G.; Cairney, J.; Eckert, C. A.; Frederick, W. J.; Hallett, J. P.; Leak, D. J.; Liotta, C. L.; Mielenz, J. R.; Murphy, R.; Templer, R.; Tschaplinski, T. Science 2006 311 484 489. (101) Lynd, L. R.; Wyman, C. E.; Gerngross, T. U. Biotechnol. Prog. 1999 15 777 793. (102) Wyman, C. E.; Dale, B. E.; Elander, R. T.; Holtzapple, M.; Ladisch, M. R.; Lee, Y. Y. Bioresour. Technol. 2005 96 1959 1966. (103) McCann, M. C.; Carpita, N. C. Curr. Opin. Plant Biol. 2008 11 314 320. (104) Karas, M.; Hillenkamp, F. Anal. Chem. 1988 60 2299 2301.

PAGE 208

208 (105) Belu, A. M.; Graham, D. J.; Castner D. G. Biomaterials 2003 24 3635 3653. (106) Caprioli, R. M.; Farmer, T. B.; Gile, J. Anal. Chem. 1997 69 4751 4760. (107) Cha, S. W.; Song, Z. H.; Nikolau, B. J.; Yeung, E. S. Anal. Chem. 2009 81 2991 3000. (108) Jung, S.; Chen, Y.; Sullards, M. C.; Ragauskas, A. J. Rapid Commun. Mass Spectrom. 2010 24 3230 3236. (109) Jung, S.; Foston, M.; Sullards, M. C.; Ragauskas, A. J. Energy Fuels 2010 24 1347 1357. (110) Tokareva, E. N.; Fardim, P.; Pranovich, A. V.; Fagerholm, H. P.; Daniel, G.; Holmbom, B. Appl. Surf. Sci. 2007 253 7569 7577. (111) Li, Z.; Chu, L.; Sweedler, J. V.; Bohn, P. W. Anal. Chem. 2010 82 2608 2611. (112) Li, Z.; Bohn, P. W.; Sweedler, J. V. Bioresour. Technol. 2010 101 5578 5585 (113) Karas, M.; Bahr, U.; Hillenkamp, F. Int. J. of Mass Spectrom. Ion Processes 1989 92 231 242. (114) Blancaflor, E. B.; Gilroy, S. Am. J. Bot. 2000 87 1547 1560. (115) Obrien, T. P.; Feder, N.; Mccully, M. E. Protoplasma 1964 59 367 373. (116) Lacayo, C. I.; Malkin, A. J.; Holman, H. N.; Chen, L.; Ding, S.; Hwang, M. S.; Thelen, M. P. Plant Physiol. 2010 154 121 133. (117) Skoog, D. A.; West, D. M.; Holler, F. J.; Crouch, S. R. In Fundamentals of Analytical Chemistry; Brooks/Cole Thomso n Learning: Belmont, CA, 2004; pp 1051. (118) Lakowicz, J. R. In Principles of Fluorescence Spectroscopy; Springer Science+Buisness Media, LLC: New York, NY, 2006 (119) Wood, P. J. Carbohydr. Res. 1980 85 271 287. (120) Ram, A. F. J.; Klis, F. M. N at. Protoc. 2006 1 2253 2256. (121) Li, K. C.; Reeve, D. W. J. Wood Chem. Technol. 2004 24 169 181. (122) Slodzian, G.; Daigne, B.; Girard, F.; Boust, F.; Hillion, F. Biol. Cell. 1992 74 43 50.

PAGE 209

209 (123) Goacher, R. E.; Jeremic, D.; Master, E. R. Anal Chem. 2011 83 804 812. (124) Genetics and Genomics of Populus; Jansson, S ., Bhalerao, R P ., Groover, A T Ed s .; Plant Genetic s and Genomics Crops and Models : 8; S pringer, N ew York, NY 2010. (125) Mauseth, J. D. In An Introduction to Plant Biology; Saunders College Publishing, Philadelphia, PA, 1991. (126) Meuzelaar, H. L. C.; Isenhour, T. L., Eds.; Computer Enhanced Analytical Spectroscopy; Plenum Press: New York, NY, 1987 (127) Windig, W.; Haverkamp, J.; Kistemaker, P. G. Anal. Chem. 1983 55 8 1 88. (128) Windig, W.; Meuzelaar, H. L. C. Anal. Chem. 1984 56 2297 2303. (129) Miller, J. N.; Miller, J. C. In Statistic and Chemometrics for Analytical Chemistry; Pearson Education Limited: Essex, England, 2010 (130) Brereton, R. G. In Chemometric s: Data Analysis for the Laboratory and Chemical Plant; John Wiley & Sons, Ltd.: West Sussex, England, 2003 (131) Brereton, R. G. Analyst 1995 120 2313 2336. (132) Altelaar, A. F. M.; Luxembourg, S. L.; McDonnell, L. A.; Piersma, S. R.; Heeren, R. M. A Nat. Protoc. 2007 2 1185 1196. (133) Wu, L.; Felton, J. S.; Wu, K. J. J. Mass Spectrometry Imaging: Principles and Protocols 2010 (134) McCombie, G.; Staab, D.; Stoeckli, M.; Knochenmuss, R. Anal. Chem. 2005 77 6118 6124. (135) Deininger, S.; Ebe rt, M. P.; Fuetterer, A.; Gerhard, M.; Roecken, C. J. Proteome Res. 2008 7 5230 5236. (136) Deininger, S.; Becker, M.; Suckau, D. Mass Spectrometry Imaging: Principles and Protocols 2010 , 385 403. (137) Vaezian, B.; Anderton, C. R.; Kraft, M. L. Anal. Chem. 2010 82 10006 10014. (138) Rakotomalala, R. Proceedings of EGC2005, RNTI E 3 2005 2 697 702. (139) Xia, J.; Psychogios, N.; Young, N.; Wishart, D. S. Nucleic Acids Res. 2009 37 W652 W660.

PAGE 210

210 (140) Pouwels, A. Analytical Pyrolysis Mass Spe ctrometry of Wood Derived Polymer Fractions, FOM Institute for Atomic and Molecular Physics, Amsterdam, The Netherlands, 1989. (141) Qiao, H.; Spicer, V.; Ens, W. Rapid Commun. Mass Spectrom. 2008 22 2779 2790. (142) Kroto, H. W.; Heath, J. R.; Obrien, S. C.; Curl, R. F.; Smalley, R. E. Nature 1985 318 162 163.

PAGE 211

211 BIOGRAPHICAL SKETCH of Quin Lunsford, grew up in Augusta, Georgia, where she completed elementary, middle and high school. In 2007, Kyle graduated from Elon University, a small liberal arts school in Elon, North Carolina, with a B.S. in chemistry (one of four) and minor in mathemati cs. While at Elon, Kyle conducted undergraduate research under Dr. Joel Karty using electrical resistivity experiments to elucidate the mechanism of periodic precipitation reactions. During the summer of 2005 and 2006, Kyle had the unique experience to in tern at the Idaho National Laboratory (INL) with Dr. Jeffrey Giglio at the Material Fuels Complex. At INL, Kyle worked on characterizing nuclear fuel rods using inductively coupled plasma mass spectromet ry (ICP MS). In addition to working at the INL, Kyle Greensboro, NC. Kyle assisted in compounding new flavors, as well as integrating flavors into products, such as doughnuts and hard candies. Both of these experiences pushed Ky le to pursue a graduate degree in chemistry, specifically analytical chemistry Kyle started graduate school at the Univ ersity of Florida fall of 2007 and later joined the Yost group. Throughout her four years, she worked on a Department of Energy funded r esearch project in collaboration with D r. Gary F. Peter in the School of Forest Resources and Conservation. Kyle received her PhD in analytical chemistry in the summer of 2011.