Isolation and characterization of a transcription pre-initiation complex component, factor IIB, from plants


Material Information

Isolation and characterization of a transcription pre-initiation complex component, factor IIB, from plants
Physical Description:
vii, 133 leaves : ill. ; 29 cm.
Baldwin, Donald Adelphi
Publication Date:


Subjects / Keywords:
Plant genetics   ( lcsh )
Plant physiology   ( lcsh )
Microbiology and Cell Science thesis, Ph. D   ( lcsh )
Dissertations, Academic -- Microbiology and Cell Science -- UF   ( lcsh )
bibliography   ( marcgt )
non-fiction   ( marcgt )


Thesis (Ph. D.)--University of Florida, 1997.
Includes bibliographical references (leaves 114-132).
Statement of Responsibility:
by Donald Adelphi Baldwin.
General Note:
General Note:

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 028001181
oclc - 37795295
System ID:

This item is only available as the following downloads:

Full Text







With highest regards I acknowledge those whose personal and professional

support sustained this work: Dr. Bill Gurley, for wisely putting the focus not so much on

finding the right answers as on searching for the right questions; Drs. Rob Ferl, Don

McCarty, Robert Schmidt, and Francis Davis for their expertise, lab resources, and

valuable advice as committee members; past and present members of the Gurley

laboratory, especially Dr. Eva Czarnecka-Verner; and the students and faculty of the

Department of Microbiology and Cell Science. With deep gratitude I thank Roger and

Evelyn Baldwin, Rev. Verda Aegerter, and Gabriela Romano for filling my mind with the

curiosity of science, my soul with the truth of poetry, and my heart with the fire of the




ACKNOWLEDGMENTS .....................................................ii

LIST OF ABBREVIATIONS................................ .......................v



1 IN TRODU CTION ........................ ................. ..............................

The Pre-Initiation Complex of RNA Polymerase II.............................................
Regulation of Transcription...............................................6
A ctivators ....................................................................... ....................... 7
C o-activators........................... .................................... ........................... 9
R epressors.............................. ............................................................ 11
In itiatio n ..................................................................... ............... ........... 13
E longation ................................................................... ............ ............. 17
R e-initiation ...................................................... ................ ......... 18
Transcription Factor IIB ..........................................................20


Literature Review ........................................ ............................................. 25
M materials and M methods ....................................................... .................................26
cDNA Library Screening for Soybean and Arabidopsis TFIIB.......................26
Southern and Northern Blot Analysis of the Arabidopsis TFIB Clone..........28
Primer Extension Mapping of Arabidopsis mRNA 5'-Terminus ....................29
Phylogeny Analysis of TFIIB cDNAs ......................... ......... ............... 30
R esu lts ..................................................................................... ........................ 30
D discussion ....................................................................... ................................. 40

OF ARABIDOPSIS TRANSCRIPTION FACTOR IIB...................................45

L literature R review ............................................................... ...................................45
Materials and Methods ................... .. ..... .........................50


Expression and Purification of Recombinant Proteins...............................50
Electrophoretic Mobility Shift Assays........................................................ 51
Fluorescence Polarim etry .................................................. ........................ 52
In vitro Transcription................................................ ........................... 53
Yeast Plasmid Shuffle for TFIIB....................................... ........................ 54
R results ............... .................................................................................................. 54
Discussion.................. ..................................................................... 75

WITH TRANSCRIPTIONAL REGULATORS......................................80

Literature Review..................................................................................80
M materials and M ethods ............................................................. .......................85
Yeast Transcription and Two-Hybrid Assays..................................... ............ 85
In vitro Interaction Assays................................................. ........................87
Western Blot Detection ofAtTFIIB1 ........................................................89
Results ...................................................................... ........................ 89
D iscussion...................................... ................................................... 100

5 CONCLUSION.................... ........................................................... 104

LIST OF REFERENCES.................. .........................................................114

BIOGRAPHICAL SKETCH............................................................................ 133



A anisotropy nm nanometer
AD activation domain nM nanomolar
AdMLP adenovirus major late promoter P polarization
P-gal beta-galactosidase PCR polymerase chain reaction
bp base pair PIC pre-initiation complex
BSA bovine serum albumin pmol picomole
bZIP basic region/leucine zipper pol II RNA polymerase II
CaMV cauliflower mosaic virus RAP RNA pol II associated protein
CBP CREB binding protein SDS sodium dodecyl sulfate
CMV cytomegalovirus sec second
cpm counts per minute SGBF2 soybean G-box binding factor 2
CREB cAMP response element binding TAF TBP-associated factor
CTAB cetyltrimethylammonium bromide TBP TATA-binding protein
CTD carboxy-terminal domain TFIIB transcription factor for pol II B
DBD DNA-binding domain TRa, p thyroid hormone receptor a or 0
EMSA electrophoretic mobility shift assay TX100 TRITON X-100
5-FOA 5-fluoroorotic acid VDR vitamin D receptor
Ftz fushi taratzu VPI viviparous 1
GS4b glutathione-Sepharose resin 4b VPl6 herpes virus viral protein 16
GST glutathione S-transferase
HeLa a human cancer cell culture
hr hour
I intensity of fluorescence
Inr initiator element
IPTG isopropyl p-D-thiogalactoside
ka rate of association
kd rate of dissociation
Kd equilibrium dissociation constant
kDa kilodalton
gig microgram
min minute
ml mililiter
mM milimolar
NF-KcB nuclear factor kappa B
ng nanogram

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



Donald Adelphi Baldwin

August 1997

Chairman: William B. Gurley, Ph.D.
Major Department: Microbiology and Cell Science

Initiation of eukaryotic transcription requires formation of a pre-initiation

complex which contains multiple protein components that direct RNA polymerase II to

promoter DNA sequences and converts the enzyme to a mode competent for

transcription. While most of the proteins contained in this complex have been identified

in animal, insect and yeast systems, clones for the corresponding plant genes have

generally not been characterized. This work reports the isolation of cDNA clones for

transcription factor IIB (TFIIB) from Arabidopsis thaliana and soybean. TFIIB is known

to play an important part in early events leading to formation of the pre-initiation

complex and, therefore, may have roles as a target for regulation of transcription as well

as a bridge between regulatory proteins, complex components, and RNA polymerase II.

Sequence comparisons between the plant TFIIB clones and versions from other species

show a high degree of homology and suggest significant functional conservation. An

investigation of the functional properties of the Arabidopsis TFIIB protein found that this

factor could replace its human counterpart in a HeLa in vitro transcription assay.

Additional in vitro experiments were conducted to measure TFIIB binding affinity for

another plant pre-initiation complex member, the TATA-binding protein, and DNA.

Arabidopsis TFIIB had an equilibrium binding constant of 7.1 nM for formation of this

ternary complex and was also capable of associating with TATA-binding proteins from

humans and yeast. The observed inter-species binding failed to predict ability to

substitute in transcription, however, as replacement of the yeast homolog with

Arabidopsis TFIIB produced no viable cells. Many transcriptional regulators have

binding affinity for TFIIB, and this has led to the hypothesis that the protein is a target of

regulatory mechanisms. Arabidopsis TFIIB was shown to bind to the herpes virus VP16

acidic domains, the maize acidic activator VPl, and a soybean proline-rich transcription

factor SGBF2. TFIIB binding reactions with all three factors were not affected by high

salt concentrations, but could be disrupted to various degrees by detergents. The

demonstration that core regions of Arabidopsis TFIIB are involved in these interactions is

consistent with similar mapping experiments for TFIIB homologs from other species.


The Pre-Initiation Complex of RNA Polvmerase II

Eukaryotic RNA polymerase II (pol II) is the nuclear enzyme which transcribes

DNA to RNA for genes encoding proteins, and consists of 8-14 subunits with total mass

greater than 500 kilodaltons (kDa). Despite this complexity, the enzyme has no inherent

ability to initiate transcription in a gene-specific manner. This is the function of a second

multi-subunit complex, containing at least 25 polypeptides and a mass over 750 kDa,

which assembles on the DNA template within the promoter region and at the start site for

transcription. The pre-initiation complex (PIC) contains both sequence-specific and non-

specific DNA-binding proteins as well as many factors that do not contact DNA.

Interactions among these factors and with pol II localize the polymerase to the start site,

and enzymatic activities within the PIC then modify pol II to an active state and open the

template double helix for RNA initiation.

Chromatographic separation of PIC components necessary for in vitro

transcription established the order of assembly: fraction D (TFIID, transcription factor for

pol II), which binds first, contains the TATA-binding protein (TBP) and its associated

factors, followed by TFIIA, TFIIB, pol II, the two proteins in TFIIF, and factors in

fractions E and H (Buratowski et al, 1989; Flores et at, 1991; Flores et al, 1992; Killeen

and Greenblatt, 1992; Maldonado et al., 1990). X-ray crystallography of TBP (Nikolov et

al., 1992) reveals a saddle shaped structure which binds in a sequence specific manner to

the DNA minor groove of the TATA box, an element commonly found in pol II

promoters around 25 base pairs (bp) upstream of the start site. Mutation of the TATA

box sequence in some instances can prevent TBP binding without repressing

transcription. These experiments often show an alteration in the position of the start site,

and when combined with the observation that some promoters do not contain a TATA

box, led to the characterization of the initiator (Inr) element (reviewed in Kraus et al.,

1996). This sequence, present in TATA-less and some TATA-containing promoters, is

located around the start site and directs formation of the PIC without sequence-specific

binding of TBP. The functional flexibility of TBP reflects its utilization in PICs for RNA

polymerases I and n1; much of this plasticity may be due to the TBP associated factors

(TAFs) which co-purify with TBP in fraction D. While the composition varies among

species, the TAFs generally comprise a group of around ten proteins ranging from 18 to

250 kDa, and their presence in in vitro footprinting reactions with TBP extends the

protected region upstream of the TATA box and downstream past the start site (Verrijzer

and Tjian, 1996). When only an Inr element is present, the TAFs are necessary for

formation of the DNA-TFIID complex. Sub-stoichiometric levels of some TAFs in

TFIID compared to TBP suggest there may be several versions of the TBP-TAF complex

in vivo, and these could generate specificity for different promoter architectures within

and among the three polymerase systems.

TFIIA contains two or three proteins (Roeder, 1991) that associate with the N-

terminal stirrup of TBP and with DNA, thus clamping TBP to its binding site and perhaps

stabilizing the interaction by counteracting negative factors in TFIID (Triezenberg, 1995).

The presence of TFIIA is not always required for in vitro transcription and further study is

needed to completely define its role. TFIIA stabilization of TBP may only be necessary

in certain promoter or TAF contexts, or in instances where other regulatory factors exert

their influence on transcription via TFIIA interaction. Like TFIIA, TFIIB clamps TBP to

the promoter but through association with the C-terminal stirrup, stabilizing the

interaction on the opposite DNA face. Unlike TFIIA, factor B is a single 35 kDa

polypeptide required for all pol II transcription. Alone, TFIIB has no affinity for DNA

but when bound to the TATA-TBP complex there are contacts with the phosphodiester

backbone as revealed by the crystal structure of the ternary complex (Nikolov et al.,

1995). TBP places a severe kink in the DNA helix, and in the presence of TFIIA and

TAFs, some of which contain histone-like domains, the proximal promoter region seems

to be wrapped around the nascent PIC, positioning the start site near the region where

TFIIB binds TBP (Oelgeschlager et al., 1996). This correlates with previous observations

that mutations in TFIIB can alter the position of the start site and that TFIIB may provide

a bridging function between the TATA region and pol II.

RNA polymerase II has high in vitro binding affinity for preformed

DNA:TFIID:TFIIB (DB) or DAB complexes, establishing a closed promoter complex,

but conversion to the open promoter form required for transcription initiation does not

proceed without factors from fractions F, E, and H. TFIIF contains two polypeptides,

RAP30 and RAP74 (Greenblatt, 1991), which associate as a complex that binds pol II

before entry into the PIC. RAP30 has a region of homology to the bacterial transcription

factor sigma 70 as well as a DNA binding domain (Tan et al., 1994), features consistent

with one of the proposed roles for TFIIF as an inhibitor of non-specific DNA binding by

pol II and a primary site of interaction between the polymerase and the PIC-DNA.

RAP74 can be photocross-linked with template nucleotides adjacent and upstream to the

start site (Robert et al., 1996) and may promote melting of the double stranded DNA prior

to RNA initiation (Pan and Greenblatt, 1994; Robert et al., 1996). De-phosphorylation of

pol II is also influenced by RAP74. The largest subunit of pol II contains a number of

tandem repeats forming a region called the carboxy-terminal domain (CTD) which

associates with multiple members of the PIC and other transcription factors (holoenzyme

components, described below) when unphosphorylated (Emili and Ingles, 1995; Lu et al.,

1991). Disruption of this binding, especially to TBP, requires extensive phosphorylation

by a CTD kinase in the assembled PIC and is necessary for release of pol II from the

promoter and elongation of the transcript (Laybourn and Dahmus, 1990). CTD

phosphatase activity occurs, presumably after termination of transcription, to restore the

CTD to a form competent for entry into another PIC. The phosphatase apparently docks

to a site on pol II other than the CTD, and its activity is stimulated by RAP74 (Chambers

et al., 1995). The same study suggested that this reaction may be inhibited by TFIB after

pol II entry into the PIC.

The CTD kinase is thought to be a part of TFIIH, a nine subunit complex that

contains DNA-dependent ATPase-helicases and a cyclin-dependent kinase activating

kinase activity (Feaver et al., 1994). Pol II phosphorylation is probably a progressive

event, with initial modification of the hypophosphorylated CTD (Laybourn and Dahmus,

1990) resulting in reduced affinity for the PIC and later kinase activity converting the

enzyme to the form required for efficient elongation; this progression can be blocked in

vivo by anti-TFIIH antibodies and kinase inhibitors which specifically inhibit elongation

but not initiation (Yankulov et al., 1996). The helicase activities of TFIIH, along with

TFIIE, are required for formation of the open promoter complex in a two-step process

that can be distinguished by changes in DNA topology. Potassium permanganate probing

of single-stranded thymidines indicates ATP-dependent melting of the region -9 to +1

nucleotides (relative to the start site) before initiation. After formation of the first RNA

phosphodiester bond, the single-stranded region expands to +8 (Holstege et al., 1996).

The initiation process is complete upon promoter clearance by pol II, but the enzyme in

some cases becomes stalled after incorporating 10-30 nucleotides of RNA (O'Brien et al.,

1994). TFIIF remains associated with pol II and may, in conjunction with other factors,

assist in the conversion to full elongation mode (Chang et al., 1993). All the other

components of the PIC, except the factors in TFIID, dissociate from the promoter, leaving

only the DNA-TBP:TAF complex as a starting point for re-initiation (Zawel et al., 1995).

Much of the work described above characterized transcription in vitro, allowing

careful dissection of the factors necessary for pol II initiation. These systems, however,

suffered from an inability to boost production of RNA transcripts in the presence of

activator proteins known to enhance transcription by binding to promoter regions

upstream of the TATA box. Clearly, the general factors could be assembled for a basal

level of transcription but other proteins required for enhanced RNA production, or

activated transcription, were missing. Two lines of research in yeast have provided

further insight. Partial deletion of the CTD produces several conditional mutant

phenotypes that can be overcome by four different dominant suppressors. These proteins,

the SRBs (suppressor of RNA polymerase B, Koleske and Young, 1994), bind to the

CTD and co-purify with a large complex containing pol II and many of the general

factors, including TFIIB but not TFIID, which are pre-associated with the enzyme despite

the absence of promoter DNA. The SRBs were also present in an enhanced complex of

pol II isolated by purification of a "mediator" fraction that supported activated

transcription (Kim et al., 1994). This form of pol II, termed the holoenzyme, additionally

contains other proteins previously known to be important for transcriptional regulation,

including the SWI/SNF complex which disrupts nucleosome structure to open chromatin

for promoter accessibility (Wilson et al., 1996). The combination of gentle cell

disruption and affinity purification or immuno-precipitation to isolate the yeast

holoenzyme was successfully used to prepare a similar holoenzyme from rat liver nuclear

extracts (Ossipow et al., 1995). This preparation was initially thought to contain all the

general PIC factors, but refinement of the technique suggests TFIID and TFIIB are not

part of the mammalian holoenzyme (Cujec et al., 1997).

Regulation of Transcription

The discovery of the holoenzyme is one turning point in a trend to refine the

model of the PIC, especially as it relates to regulation of transcription. Many regulatory

factors that bind to upstream promoter elements have in vitro binding affinity for

members of the PIC, and an ordered progression of PIC assembly easily fit into a model

whereby different regulators influence each step. Recruitment by an upstream factor

would increase the local concentration of a PIC component, thus accelerating the rate of

assembly, and the presence of more factors recruiting many different parts of the PIC

would give correspondingly higher levels of transcription. A pre-assembled holoenzyme,

however, arrives in one binding event and does not require stepwise recruitment of its

components to build the PIC. Regulation of transcription may therefore involve potential

cooperative interactions with multiple targets in the holoenzyme, rather than progressive

PIC assembly. Promoters usually contain binding sites for several transcription factors

and combinations of these factors often have a synergistic effect, which may be a result of

coopertivity in binding. Interactions between transcription factors and their targets are

thought to be highly cooperative since both components also have affinity for the

promoter, so multiple links between the holoenzyme and the promoter may represent a

synergistic combination of such interactions.

As models for transcriptional regulation are refined, greater emphasis is being

placed on what mechanisms, in addition to recruitment of basal factors, might influence

the various interactions observed among PIC members. The research to understand these

mechanisms continues to benefit from new additions to both categories of interactants

involved. The targets of regulation are being identified as proteins in the classical PIC or

holoenzyme, and the upstream regulators are being characterized as members of groups

with shared functions and/or sequence homologies. The following sections summarize

these functional classes and discuss some of the steps at which they may act.


Transcriptional activators are proteins that exhibit DNA-binding affinity and

increase levels of transcription as a consequence of their DNA association. Indirect

activators such as HMG-2 (Shykind et al, 1995) show broad sequence specificity and

remodel chromatin in distal promoter regions to expose binding sites for other factors or

serve as architectural proteins to stabilize DNA conformation at the PIC. Other indirect

activators recognize specific cis-elements within a promoter but require dimerization with

a second protein to activate transcription (see Co-activators), or distort the DNA to help

bring upstream sites nearer the PIC as is the case with the CCAAT-box binding protein

NF-Y (Ronchi et al, 1995).

Direct activators combine DNA binding domains with one or more protein

interaction motifs that influence transcription. The DNA elements to which they localize

may be near the site of PIC formation or quite far (>1 kb) upstream from the TATA

region, and may even be found following the 3' end of a gene's coding sequence. The

DNA looping required to bring these element-activator complexes into the vicinity of the

PIC has been visualized by atomic force microscopy (Becker et al, 1995) and often

requires stabilization by other factors and cooperative interactions at both the DNA

binding site and to the PIC target. Distal cis-elements in some cases may also utilize

specific spacing patterns relative to nucleosome structure for positioning near the TATA


The protein interaction motifs that confer an ability to boost transcription are

termed activation domains (AD). These peptide sequences are usually separated from the

DNA-binding domain (DBD) and can be isolated and studied as fusions with a well-

characterized DBD to facilitate both in vitro and in vivo functional mapping.

Conservation among some ADs can be observed as a general enrichment for particular

amino acids such as proline, serine or glutamine, or at the level of residue organization

(reviewed in Triezenberg, 1995). Acidic activators, for example, often contain regions

high in glutamic and aspartic acids divided into blocks of conserved length by bulky

hydrophobic residues. The term "acidic activators" is also an example of the potential

pitfalls of AD classification by amino acid composition. Several mutational analyses

have made large scale charge alterations and observed little effect on transcriptional

activation. Changing particular hydrophobic residues or the patterning of aromatic and

hydrophobic residues is more likely to disrupt activation potential (Jackson et al., 1996;

Regier et al., 1993; Sainz et al., 1997).

Secondary structure has been difficult to detect for many ADs, leaving most to be

classified as random coils, unstructured, or acid "blobs." Examples are known which

contain alpha-helices, beta-strands, Zn fingers and leucine zippers, and secondary

structure can sometimes be induced under special conditions (Shen et al., 1996). The c-

myc transactivation domain has no secondary structure when observed by circular

dichroism in aqueous buffer, but measurements in hydrophobic solvent or in the presence

of TBP suggest an alpha-helical conformation (McEwan et al., 1996). Even when

unstructured, many ADs can be functionally divided into subdomains that correspond to

distinct patterns of residue predominance or charge, and these subdomains have often

been shown to make synergistic contributions to transcription activation (Artandi et al.,

1995; Blair et al., 1996).


As clones for activators were isolated, a common method for studying their effects

utilized over-expression of the protein and measurement of transcription from a reporter

gene whose promoter featured multiple binding sites for the activator. In some cases,

over-expression led to repressed levels of transcription rather than activation. This

inhibition of activity, or squelching, was thought to be due to depletion of cofactors by

excess activator not bound to DNA (Gill and Ptashne, 1988; Martin et al., 1990; Meyer

et al., 1989). Subsequent purification by activator affinity chromatography identified co-

activators as factors capable of increasing transcription by acting as adapters between

DNA-bound regulators and the PIC. Cellular co-activators can have broad specificity for

enhancement of several classes of activation domains, or can be limited to a narrow

subset of promoter interactions. For example, human PC4 is a serine-rich co-activator

that interacts with representatives of the acidic, proline, and glutamine classes of

activators, and targets TFIIA in a phosphorylation-dependent manner (Ge and Roeder,

1994). In contrast, the human thyroid hormone receptor a (TRa) forms a complex with

specific co-activators only in response to binding of the hormone ligand (Fondell et al.,

1996). Several members of the PIC also fit the definition of co-activators. Some TAFs

and various holoenzyme components such as the mediator fraction are dispensable for

basal transcription but are required for activation by upstream factors.

Viruses seem to be a rich source of co-activators. The adenovirus ElA factor has

no DBD but activates a number of promoters, both cellular and viral, and contains a zinc

binding domain that interacts with TBP or TAFs (Folkers and van der Saag, 1995).

Herpes simplex viral protein VP16 is a common model for acidic activation domains, two

of which are located in the C-terminal 80 amino acids. VP16 activates transcription of

the viral immediate early genes by interacting with the cellular factor Oct-1, which

recognizes binding sites within these viral promoters. The protein arrives as a part of the

virion tegument (Elliott et al., 1995) and, after nuclear localization, activates transcription

through one or more interactions with TFIIB, TBP, TAFs, TFIIH, and PC4 (reviewed in

Ghosh et al., 1996). In addition to this unusually large number of potential targets for

direct activation, as measured by in vitro binding affinity, VP16 has an anti-repressor

activity that epitope maps outside the acidic domains and may be due to dislocation of the

histone HI subunit (Lyons and Chambon, 1995).


The number of identified transcriptional repressors continues to rapidly increase,

and regulation by repression may be as important a mechanism as activator-dependent

promoter control. Repressors exhibit the same range of interactions as activators and can

act not just to prevent activated transcription but also to suppress promoters below their

normal basal level. Unliganded TRao occupies its normal promoter cis-element and

prevents transcription by direct contact with TBP, preventing entry of TFIIA or B into the

PIC (Fondell et al, 1996). Repression is relieved upon hormone binding; the ligand-

binding domain is also the region contacting TBP for repression. ElA from adenovirus

has a repression domain, distinct from the co-activation region, which also binds TBP and

prevents transcription from some cellular promoters. Repression can be overcome in

vitro by adding extra TFIIB, suggesting the Ela repressor blocks TBP-TFIIB interaction

(Song et al., 1997). Other repressors target TFIIB directly, TFIIA, or pol II itself (Gu et

al., 1995; Kirov et al., 1996; Lee et al., 1996). Some repressors are bi-functional and

activate transcription when bound to an upstream location within a promoter but repress

when located near the TATA box (Dostatni et al., 1991). Conversely, the Drosophila

Kruppel protein binds as a monomer to TATA proximal sites and activates through TFIIB

contacts, but as a dimer binds to distal sites and represses transcription through TFIIE

interactions (Sauer et al., 1995). Regulators may actively repress by protein-protein

interaction with PIC components, or cause passive repression by blocking a DNA

sequence normally bound by an activator (Mailly et al., 1996).

Squelching experiments similar to those for co-activators have identified co-

repressors. Over-expression of the N-terminal half of the ligand binding domain from

thyroid hormone receptor P prevents repression by the wild type factor, presumably by

depleting a necessary co-repressor (Nawaz et al., 1995; Tong et al., 1996). Homologs of

the Drl protein from yeast and HeLa cells are DNA-binding repressors that can be

counteracted by acidic or glutamine-rich activators, but not by proline domain activators.

Drl repression is boosted by heterodimerization with a co-repressor, DRAP1, which

increases Drl affinity for TBP at the TATA box and blocks TFIIA and TFIIB binding

(Mermelstein et al., 1996). The repression domain of DRAP1 is proline-rich, and that of

Drl contains multiple glutamine and alanine residues.

Other mechanisms for repressing transcription do not involve separate repressor

proteins as described above but should be noted. The simple lack of an activator, perhaps

by increased proteolysis or down-regulated expression of its promoter, is undoubtedly a

very common method of gene repression, as is silencing by chromatin structure and

promoter methylation. Some activators contain intrapeptide repression domains.

Examples include heat shock transcription factors (Bonner et al., 1992; Sorger, 1990) and

ATF-2, an activator that requires an inducer to disrupt masking of the proline/Zn finger

AD by interactions with the basic region-leucine zipper DBD (Li and Green, 1996).

Alternate mRNA splicing can determine a transcription factor's function, as is the case

for TFE3, a regulator ofimmunoglobulin gene promoters. The repressor TFE3S is a

shorter isoform containing a C-terminal proline domain, and alternate splicing adds an N-

terminal acidic domain which shows synergistic activation when combined with the

proline region (Artandi et al., 1995). Tissue specific splicing-in of a proline-rich exon

converts aNAC from a protein that associates with polypeptides exiting ribosomes to a

DNA-binding activator of transcription in muscle cells (Yotov and St-Amaud, 1996).

Activators and repressors share many general characteristics such as methods for

promoter localization, amino acid composition of interaction domains, and targeting of

multiple members of the PIC. As the list of known transcription factors and their

mutations continues to grow, clues to the question of what activation (or repression)

means biochemically are being uncovered through examinations of what steps in pol II

transcription they influence.


If activators increase the local concentration of a necessary factor at the promoter

by recruitment, a simplistic test would be to artificially raise the overall concentration of

that factor and thus bypass the need for recruitment. Such tests to determine whether

TBP is limiting for pol II transcription in vivo have shown boosts in reporter activity after

TBP plasmid transfections into monkey COS and Drosophila Schneider cells. In COS

cells, a promoter dependent on the activating nuclear factor KB (NF-KB) was unaffected

by additional TBP in the absence of the trans-activator, but cotransfection with NF-KB

and TBP produced 20-fold greater transcription than with activator alone (Schmitz et al.,

1995). Several wild type Drosophila promoters, as well as a minimal TATA construct,

were also enhanced by TBP over-expression, and the fold induction was inversely related

to the promoter's normal strength (Colgan and Manley, 1992). Binding of TBP therefore

appeared to be a rate-limiting step, a conclusion supported by experiments that produced

activated transcription by tethering TBP to a promoter via fusion to specific DNA binding

domains (Chatterjee and Struhl, 1995; Klages and Strubin, 1995; Xiao et al., 1995).

Indirect recruitment of TBP through its TAFs also occurs. The Drosophila proteins

Bicoid and Hunchback bind TAF110 and TAF60, respectively, and each alone can

weakly promote DNA-TFIID association. When both are present, their binding affinity is

synergistically increased and a multiplicative increase in transcription is observed (Sauer

et al., 1995; Sauer et al, 1996).

A series of observations utilizing DNA footprinting revealed there is more to

activation through TBP than its recruitment. Sequences adjacent to the TATA box

showed extended protection in the presence of an activator (Horikoshi et al, 1988), an

effect now known to be due to the TAFs. Kinetic analyses with and without activators

revealed that TBP-TAF binding to DNA and the extension of the footprint could be

assayed sequentially, and that TBP mutants can be generated which are defective for

activation but bind the activator, DNA, TFIIA, and TFIIB normally (Chi et al., 1995;

Hoopes et al., 1992; Stargell and Struhl, 1996). Success with a technically difficult

approach to measure activated in vitro transcription from the same DNA template

population used for footprinting showed that with saturating concentrations of TFIID and

TFIIA, the Epstein-Barr viral activator Zebra still enhanced transcription despite

constitutive occupancy of the TATA box (Chi and Carey, 1996). The enhancement

persisted even when the activator was removed by oligonucleotide competition before

addition of the remaining PIC components, indicating that Zebra acts at the TFIID

isomerization step. This isomerization, which follows TFIID-DNA binding but precedes

entry of TFIIB, is apparently what produces the extended footprint and reflects stabilized

TAF-DNA interaction and a change in conformation that favors PIC completion. Zebra

also has in vitro binding affinity for TFIIB, yet TFIIB rapidly entered the isomerized

TFIID complex in the absence of activator.

Many other activator proteins have affinity for TFIIB. The investigation of

whether binding events involving TFIIB are relevant to activated transcription started

much the same as that for TBP. Over-expression of TFIIB in COS cells had no effect on

basal or NF-KB activated promoter activity (Paal et al., 1997; Schmitz et al., 1995). In

separate Drosophila experiments, TFIIB over-expression increased transcription only 1.1-

or 2-fold for a Fushi tarazu (Ftz) activated reporter, and basal transcription in the absence

of Ftz was either unaffected or increased 2-fold (Colgin et al, 1995; Colgan et al., 1993).

TFIIB would thus seem to not be limiting in vivo, reducing the potential need for its

recruitment during activation. It has been noted that both TFIIB and TBP have charged

regions carrying basic residues, and binding of activators to TFIIB may merely be the

result of affinity for similar regions in the real target, TBP. As more activators are

isolated, however, it has become apparent that there are classes with affinity for TFIIB

only, both TBP and TFIIB, or TBP alone. Also, there are several domains within TFIIB

involved in these interactions, some with characteristics quite different from TBP.

Studies with VP16, one of the first and strongest activators characterized,

continue to play a large role in the question of TFIIB involvement during activated

transcription. VP16 not only binds TBP and TFIIB in vitro, but also TBP-associated

factor TAF32, TFIIH, and the co-activator PC4 (Ghosh et al., 1996). TFIIB mutations

that disrupt interaction with VP16 also prevent in vivo activation by VP16, but minimally

affect basal transcription (Roberts et al., 1993). An attempt to confirm these observations

gave mixed results (Gupta et al., 1996), and the promiscuous interactions of VP16 may

make it an overly complex activator for studying TFIIB specifically. Activation by the

glutamine domain of Ftz can be blocked by co-expression with truncated TFIIB, and the

degree of squelch is much greater than the effect on basal transcription (Colgan et al.,

1995). Activation can be restored by introducing a mutation into the truncated TFIIB

which blocks Ftz binding, or by adding wild-type TFIIB to compete the squelch. These

and other squelch and rescue assays provide in vivo correlation of in vitro binding for

activators with TFIIB, but questions remain regarding whether the TFIIB mutant truly

does not affect basal PIC formation, and whether interaction with the over-expressed

mutant might be artificially interfering with the actual target.

Progress toward understanding the role of TFIIB in activation will require

additional mutation studies, for both activators and TFIIB, to attempt to specifically

observe TFIIB related events. One interesting approach has been the creation of altered

specificity mutations in TBP and TFIIB (Tansey and Herr, 1997). This system starts with

a point mutant of TBP that does not bind TATA boxes, but instead recognizes a TGTAA

binding site. A second TBP mutation in the TFIIB interaction domain swaps an arginine

for a glutamic acid that normally forms an electrostatic contact with an arginine in TFIIB.

Since this change places two basic residues in close proximity, the TBP-TFIIB complex

does not form and transcription from a TGTAA promoter reporter is prevented. Activity

is restored by a mutation of TFIIB, replacing the critical arginine with a glutamic acid.

The serial altered specificity approach utilizes transcription that is independent of

endogenous TBP and TFIIB, allowing mutant scans of both proteins, as well as activators,

that can reveal differences in the ways TBP and TFIIB are targeted for activation.

Less progress has been made regarding other members of the PIC as regulatory

targets for the initiation step. Protein interactions between regulators and pol I, TFIIF,

TFIIE, TFIIH, and the mediator subunits have been observed. It is not known whether

these represent redundant recruitment binding sites for the holoenzyme or if they have

more specialized purposes. The largest subunit of TFIID, TAF250, has been shown to

contain a protein kinase activity capable of autophosphorylation within each of two

separate serine kinase domains. The combination of both domains in the context of the

TFIID complex results in specific phosphorylation of the TFIIF component RAP74

(Dikstein et al., 1996). Similar modifications or conformation changes of basal factors

and pol II itself may be subject to regulation during transcription initiation. The relative

importance of these mechanisms and those involving TBP and TFIIB probably depends

on a number of factors including proximal promoter sequences, the types and number of

upstream regulatory elements, and the host species. Yeast, for example, has TAF

homologs that are apparently dispensable for transcription activation (Moqtaderi et al.,

1996; Walker et al., 1996) and instead relies on the mediator fraction, while Drosophila

and human systems are dependent upon TAFs for activation (Burley and Roeder, 1996).


As described above, modification of the pol I CTD may be an important part of

releasing the enzyme from the PIC for elongation. Other factors may assist with this

release. Yeast SUB1, a homolog of the co-activator PC4, binds to TFIIB in a manner

mutually exclusive to the TBP-TFIIB interaction and increased expression of SUB1

stimulates transcription in vivo (Knaus et al., 1996). It was hypothesized that SUB 1 acts

as a release factor capable of suppressing TFIIB mutations that prevent proper

dissociation of the PIC for pol II elongation.

SUB1 and the HIV activator Tat primarily stimulate elongation, either through

interactions at the PIC or with pol II directly, but other classes of activators can enhance

both initiation and elongation. Multi-functional factors may concurrently or sequentially

contact initial members of the PIC to promote complex formation and then induce

conversion to elongation through interactions with other components such as TFIIH.

Activators p53, E2F1, and VP16 can influence both steps, and they all bind several PIC

subunits including TFIIH. A quadruple point mutant of VP16 has been described which

no longer activates elongation but retains wild type activity for initiation enhancement

(Blau et al., 1996), and a different mutation suppresses the initiation effect but retains

activity for elongation (Ghosh et al., 1996). Other acidic ADs also activate initiation and

elongation, but the effect on elongation requires more copies of the AD to be present at

the promoter than are needed to boost initiation (Blair et al., 1996).


Some quantitative comparisons of basal and activated in vitro transcription

indicate that activators do not generally increase the rate of PIC formation or pol II

elongation, but increase the number of DNA templates occupied by PICs (Yamazaki et

al., 1990). In the case of the human activator Spl, the overall equilibrium constant for

PIC formation is also increased by 10-30 fold (Yean and Gralla, 1996). Since multiple

rounds of transcription can occur in vitro, the resulting boost in RNA production is thus a

function of not just more template copies being utilized but also enhanced levels of re-

initiation, which is probably more relevant to activation of low copy number promoters in

vivo. Pol II escape from the PIC dissociates TFIIB and the other factors, but not TFID,

so the various activator interactions with these basal factors as described above may be

more important for re-initiation than the initial PIC assembly.

A unique study utilizing a synthetic dimerizing ligand demonstrated the

importance of re-initiation for promoter activation (Ho et al., 1996). In yeast and human

cells, two proteins were expressed, one containing an AD and the other a DBD, as fusions

to a ligand binding domain. The cells were then incubated with the lipid-soluble dimeric

ligand to link the DBD and AD, activating transcription from a reporter promoter.

Addition of excess monomeric ligand quickly out-competed the dimer for ligand binding

domains and prevented maintenance of the triple complex. Nuclear run-on analysis

indicated that while the reconstituted activator directed initiation, its continued presence

during re-initiation was necessary to produce RNA levels typical of activated


Before characterization of the PIC, research in transcriptional analysis

concentrated on promoter sequences and the regulatory factors that bound those elements.

One puzzle that became immediately apparent was the redundant use of DNA sequences

acting as binding sites in promoters with different patterns of expression. Clearly not all

specificity was at the level of what individual elements were recognized by particular

factors, but also what combinations of elements were present and how the proteins that

occupied them interacted. The complexity of the PIC now adds additional levels of

specificity as transcription factors can target multiple patterns of individual proteins

within the PIC as well as multiple steps during initiation and early elongation. TFIIB has

received much attention as one of the central links between TFIID and pol II or

holoenzyme, and for its involvement in initiation (or re-initiation) and release for

elongation. It will be interesting to see if the multiple roles of TFIIB represent another

level at which promoter activity and specificity can be fine-tuned.

Transcription Factor IIB

X-ray crystallography and nuclear magnetic resonance spectroscopy have

established partial structures for TFIIB. The C-terminal two-thirds of the sequence

contains imperfect repeats, and in solution they form a bi-lobed structure separated by a

cleft and joined by a short random coil linker (Bagby et al, 1995). Each repeat contains

five a-helices, and the C-terminus contains a short sixth a-helix (Fig. 1-1A). Structural

instability in the N-terminal region of human TFIIB has prevented its inclusion in NMR

and crystal structures, but the first 49 residues of a clone from a thermophilic

Archaebacterium have been shown to form a Zn ribbon, which is the combination of a Zn

finger and three anti-parallel p-sheets (Zhu et al., 1996). Sequence homology among

archaeal and eukaryotic TFIIB clones is high and structural conservation is expected.

The C-terminal repeats, termed the TFIIB core, have been co-crystallized with

TBP bound to a TATA element (Nikolov et al., 1995). TFIIB has numerous electrostatic

interactions, hydrogen bonds, and van der Waals contacts with TBP and the DNA

phosphodiester backbone (Fig. 1-1B), and is oriented with the C-terminus upstream of the

TATA box and the Zn ribbon region downstream near the transcription start site (Leuther

et al., 1996). Other PIC proteins interact with TFIIB, and the interaction domains have

been mapped by in vitro binding with TFIIB deletions and point mutations. RAP30

binding is disrupted by changes in the Zn region and helix Al, and can be displaced by

N Al B1 C1 D1 E1 A2 B2 C2 D2E2
Zn [ conserved j repeat I j repeat 2
____ i______________i-------------i -_____________

RAP30, RAP74
RAP30, RAP74




pol II large subunit

D I El ......C2

, .......... El C2 E2
downstream TATA upstream

El E2
TRa "'- cJUN TAP




, E2

Figure 1-1. Transcription Factor IIB and its interaction domains.
A. General features of TFIIB include a Zn ribbon at the amino terminus, a region of high
conservation, and two structural repeats each containing five alpha helices. B. TFIIB
domains are shown that interact with components of the PIC. Clusters of residues important
for binding are indicated with their helical label or, when between helices, as a dotted line.
DNA and TBP contacts are from the crystal structure (Nikolov et al., 1995) which utilized a
truncated TFIIB; additional contacts are probably present in the N-terminal regions.
C. TFIIB domains that interact with various activators are found in the zinc ribbon and core

RAP74 (Fang and Burton, 1996). Drosophila TAF40 interacts with the linker-helix A2

region (Hori et a., 1995) and pol II with the TFIIB core. An acidic domain near the CTD

in the largest subunit of pol II seems to direct interactions with TFIIB, and those contacts

can be competed by the acidic activation motifs of VP16 (Berroteran et a., 1994; Xiao et

a., 1994).

Similar mapping techniques have been used with transcriptional activators, as

shown in Fig. 1-1C. The Zn ribbon and adjacent conserved domain, which contains the

sequences with highest homology among TFIIB clones, are the binding sites for Ftz, the

vitamin D receptor (VDR), and thyroid hormone receptor P (TRp-C terminal)

(Baniahmad et al., 1993; Colgan et a., 1995; Masuyama et aL, 1997). The core region

binds a number of activators including c-Jun homodimers, the cellular co-activator TAP

associated with HIV expression, NF-KB, hepatitis viral activator X (HBx), and the yeast

activator of Adh2 gene expression ADR1 (Chiang et al., 1996; Franklin et al., 1995; Lin

etal., 1997; Schmitz et a., 1995; Yu et a., 1995). VP16 interaction requires an intact

first repeat of TFIIB and can be repressed by point mutations in helix El (Roberts et al.,


Roberts and Green (1994) also mapped an interesting intramolecular interaction in

TFIIB. The conserved region binds to residues somewhere between the start of the linker

and the end of repeat 2, and this interaction is disrupted by VP16. The resulting VP16-

induced changes in TFIIB can be detected by altered patterns of protease cleavage

products, leading to the suggestion that intra-peptide binding reduces TFIIB affinity for

the PIC and a conformation change must occur for proper interaction and transcription

initiation. This may explain the results of a recent study in yeast in which VP16

continued to activate despite a TBP mutation that prevented interaction with TFIIB (Lee

and Struhl, 1997). VP16 may be acting to stabilize TFIID and catalyze the TFIIB

conformation change, but does not enhance or even require TBP-TFIIB interaction to

activate transcription. It would also seem to suggest that the presence of this activator

can overcome some TBP point mutations that should severely disrupt PIC formation.

Other yeast TFIIB work has characterized additional functions. Two residues in

the conserved domain that are identical among all TFIIB clones were identified as

important for start site selection in a mutant screen (Pinto et al, 1994). These glutamic

acid and arginine residues are thought to form a salt bridge, because while individual

mutations shifted the mRNA start site for several promoters, a charge swap partially

restored the normal phenotype. The same mutants confer cold sensitivity that is

suppressed by SUB1, and it was hypothesized that normal TFIIB release from TBP

requires the ability to form the intra-peptide salt bridge (Knaus et al, 1996). Human

TFIIB will not functionally replace the yeast homolog in vivo unless helix B1 from the

yeast clone is substituted for the same helix in the human protein (Shaw et al., 1996).

Four point mutations were generated to change residues within the yeast BI helix to the

human versions, and the resulting construct is functionally impaired, has a temperature-

sensitive phenotype, and fails to respond to activators at some promoters (Shaw et al.,

1997). A screen for intragenic suppressors of this clone identified separate point mutants

in helix Cl and at two residues in the conserved region that individually reversed the

defect in transcriptional activation.

Increasingly detailed evidence continues to accumulate regarding the roles of

TFIIB in both basal and activated transcription. Over 37 different transcriptional

regulators are now known to interact with TFIIB in vitro, and methodology continues to

develop for precisely measuring their affinities for TFIIB and other PIC components and

to relate in vitro binding to testable in vivo systems. Unfortunately, these biochemical

and genetic models are being generated without input from a biologically and

economically important kingdom, the plants. Research in plant transcriptional regulation

has remained strong in promoter analysis and for characterization of signal transduction

pathways that lead to transcriptional control, but the PIC targets of that regulation are

lacking. When the work described in the following chapters was initiated, clones for TBP

from Arabidopsis, maize, potato, and wheat had been isolated (Apsit et al., 1993; Gasch

et al., 1990; Holdsworth et al., 1992; Kawata et al., 1992; Vogel et al., 1993) as well as

subunits of pol II from Arabidopsis and soybean (Dietrich et al., 1990). It therefore

seemed reasonable to next address the question of whether a TFIIB homolog exists in

plants, and if so, what inferences could be made about plant pol II transcription by

comparison of the factor with other eukaryotic homologs. Additionally, the availability

of a number of plant transcriptional regulators makes rapid screening for TFIIB

interactions possible and might provide new resources for investigating activation



Literature Review

Protein sequencing of purified TFIIB and library screening using derived

oligonucleotides allowed isolation of cDNA clones, first from baker's yeast (yTFIIB)

(Pinto et al., 1992) and then from human cell extracts (hTFIIB) (Ha et al., 1991; Malik et

at, 1991). Sequence information suggested the presence of two regions with a repeated

structural motif in the C-terminal two-thirds of the protein (core domain), and a possible

Zn binding domain near the N-terminus. These features, as well as a highly conserved

region adjacent to the Zn domain, are preserved in subsequent cDNA clones from

Drosophila (Yamashita et al., 1992), rat (Tsuboi et al, 1992), Xenopus (Hisatake et al.,

1991), Kluyveromyces (Na and Hampsey, 1993), and two Archaebacteria (Creti et al.,

1993; Qureshi etal., 1995).

When co-crystallized with Arabidopsis TBP and a consensus TATA box, hTFIIB

contacts one stirrup of the saddle shaped TBP and also has clear interactions with the

DNA phosphodiester backbone (Nikolov et al, 1995). These interactions with TBP are

expected to be similar for homologs from other species since the degree of amino acid

conservation with hTFIIB is quite high, ranging from 79% identity with Drosophila to

94% and 99% identity among vertebrates (frog and rat). Yeast TFIIB, however, shows

only 35% identity with the human version and contains 32 additional amino acids

compared to other eukaryotes. Archaeal clones are 33% identical to hTFIIB and have an

extended N-terminal region.

Until now, no plant versions of TFIIB were available for comparison with other

eukaryotic homologs. Plant TBP clones are around 83% identical to the corresponding

region of human TBP, a remarkable degree of homology that is maintained in yeast and

Drosophila (Hemandez, 1993). All archaeal and metazoan versions of TFIIB and TBP

appear to be present as single copy genes, but there are two copies of TBP in Arabidopsis,

maize and wheat, and protein sequences within each respective pair are 95%, 99% and

92% identical. This chapter describes the isolation and general characterization of cDNA

clones for TFIIB from Arabidopsis and soybean. During analysis of these clones a partial

cDNA with TFIIB homology was submitted to the EMBL database as part of an ongoing

project to identify expressed sequence tags from Arabidopsis (Desprez et al., 1994). This

clone was kindly provided by the Arabidopsis Biological Resource Center at Ohio State

University for sequence comparison.

Materials and Methods

cDNA Library Screening for Soybean and Arabidopsis TFIIB

Putative TFIIB cDNA clones were obtained by screening cDNA libraries with a

synthetic oligomer derived from a highly conserved region of the TFIIB protein. To

generate this probe, amino acid sequences for human, fruit fly, and yeast TFIIB were

aligned using the CLUSTAL W analysis program (Thompson et al., 1994), and a block of

conserved residues near the end of repeat 1 was identified (amino acid residues 166 to

181, human). A 48 nucleotide oligomer was designed based on the DNA consensus

sequence for this conserved region and incorporated allowances for the soybean codon

bias (Murray et al., 1989). This oligomer, 5' ATTGCTTGCAGACAAGAAGGAGTT-

CCAAGAACTTTCAAGGAAATTTGC 3', was 32P end-labeled with T4 poly-nucleotide

kinase to a specific activity of 3.2x108 cpm/pg and used for hybridization screening with

standard methods. Over 750,000 plaques from a Glycine max v. Resnik cDNA library in

Xgtl 1(5'-Stretch, Clontech) were probed on duplicate nitrocellulose filter lifts in 1X SSC

buffer (0.15 M NaCI, 0.015 M Na3 citrate, pH 7.0) at 420 C. One clone gave a strong

hybridization signal after two successive plaque purifications. The 800 bp insert DNA

was PCR amplified with ,gtl 1 primers using Vent polymerase (New England Biolabs)

and subcloned to pUC 18. Sequenase (DNA Sequencing Kit, version 2.0; United States

Biochemical) dideoxy sequencing reactions were performed on a nested series of

deletions created with DNA exonuclease III (Ausubel et al., 1993), and all sequences

reported are the result of at least two reactions on both top and bottom strand templates.

Since this clone was an incomplete coding sequence, an internal PstIVHindIII fragment

was labeled by nick translation to re-probe 1.2x106 plaques of the cDNA library. A

second cDNA library prepared from 2,4-D treated soybean (Glycine max (L.) Merr. cv.

Wayne) plumules in Xgtl0 was provided by Dr. Gretchen Hagen (Dietrich etal., 1990),

and the first round of screening 150,000 plaques with the internal fragment probe

produced a single full length clone.

The soybean partial clone was also used to probe an Arabidopsis thaliana v.

Colombia cDNA library in Xgtl (5'-Stretch, Clontech). After screening 660,000

plaques, eight were plaque purified twice and subcloned. The longest TFIIB isolate

lacked sequences on both 5' and 3' ends but served as a probe for library re-screening of

300,000 plaques. Four additional clones were purified and ligated into pUC19, and the

three longest were sequenced. Sequencing was also performed on the expressed sequence

tag cloned into pBluescript SK-, ABRC DNA stock number ATTS3421.

Southern and Northern Blot Analysis of the Arabidopsis TFIIB Clone

The Arabidopsis TFIIB probe was generated from a pUC19 construct of the

library isolate using PCR amplification directed by primers complementary to the 5' and

3' ends of the coding sequence. This 960 bp DNA was labeled by nick translation (Nick

Translation System, Promega Corp.) to greater than lx108 cpm/lg and denatured by

boiling for 10 min immediately before addition to the hybridization reactions.

Arabidopsis (Columbia) genomic DNA was isolated by the CTAB extraction

method (Rogers and Bendich, 1994), and 10 pg were used for each restriction digestion

containing 30 units each ofXbaI or BglI individually, or BglII+EcoRI. After

electrophoresis with markers in a 0.7% agarose gel and denaturation, the DNA was

blotted to Hybond nylon membrane (Amersham) by downward capillary transfer

(Turboblotter, Schleicher and Schuell) in 20X SSC buffer and then immobilized by UV

crosslinking. Blots were prehybridized in 5 ml of APH buffer (5X SSC, 5X Denhardt's

solution, 1% SDS, 100 plg/ml sheared denatured herring sperm DNA) for 15 min at 650 C

(Ausubel et al., 1993), followed by addition of probe to 1xl06 cpm/ml for overnight

hybridization at the same temperature in a rotisserie incubation oven. Two 50 ml washes

were performed at each of four stringency levels: 2X SSC/0.1% SDS at room

temperature; and 0.2X SSC/0.1% SDS at room temperature, 420 C, and 600 C. Blots

were autoradiographed for three days at -700 C with amplification screens on Kodak XR-

Blue X-ray film.

Arabidopsis total RNA was isolated by phenol extraction/LiCl precipitation

(Pawlowski et al., 1994) from four-week old whole plants grown at room temperature

(control) or incubated at 370 C for two hours before extraction (heat shock). mRNA was

purified using an oligo(dT) cellulose spin column (5'-3' Inc.) according to the

manufacturer's instructions. Poly(A)+ RNA (5 jig) from control and heat shock

treatments was electrophoresed beside RNA molecular weight markers (0.24 9.4 kb,

GibcoBRL) in a 1.0% agarose/formaldehyde gel and blotted as described above.

Membranes were prehybridized in 5 ml FPH buffer (5X SSC, 5X Denhardt's solution,

50% formamide, 1% SDS, 100 j.g/ml sheared denatured herring sperm DNA) for 15 min

at 420 C, followed by addition of probe to lxl06 cpm/ml for overnight hybridization at

420 C. Washes at three stringency levels up to 420 C were performed as above, and the

filters were autoradiographed overnight.

Primer Extension Mapping of Arabidopsis mRNA 5'-Terminus

Arabidopsis poly(A)+ RNA (2 pg) was annealed to an oligomer of bottom strand

sequence from 89 to 66 bp downstream from the first AUG: 5' ACCACACTCGGAGC-

AAAGGGTATC 3'. The primer was 32P end-labeled with T4 polynucleotide kinase and

0.3 pg was added to each reaction. Annealing in 2X reverse transcriptase buffer for two

hours at 650 C was followed by slow cooling to room temperature. The reaction was

diluted 1:2 with DTT and mixed dNTPs, and 400 units of SuperScript II reverse

transcriptase (GibcoBRL) were added and incubated 2.5 hr at 420 C. The reaction was

stopped by adding loading buffer, heated to 650 C for 5 min, and electrophoresed on an

8.0% denaturing polyacrylamide gel. A dideoxy-adenosine sequencing reaction using the

same primer and the pUC19-TFIIB plasmid as template was included as a size marker,

and the gel was autoradiographed overnight.

Phvlogenv Analysis of TFIIB cDNAs

DNA coding sequences for TFIIB from soybean, Arabidopsis, human, frog, fruit

fly, yeast, and the thermophilic Archaebacterium Pyrococcus woesei were compared

using the PAUP program (Swofford, 1991) after alignment by the CLUSTAL W

algorithm. Bootstrap analysis was conducted in three separate trials with 100 repetitions

each at the 90% confidence level, and trees were generated by midpoint rooting.


After isolating a number of truncated soybean TFIIB clones, a 1226 bp sequence

was recovered from the plumule cDNA library which seemed to be complete, but lacked

nine codons at the 5' end compared to homologs from other species. Repeated attempts

with various protocols for 5' RACE PCR amplification generated no clones containing

additional coding or leader sequences. For comparison, a second plant TFIIB was cloned

from Arabidopsis using the soybean TFIIB cDNA (GmTFIIB) as a screening probe. The

largest Arabidopsis cDNA isolated was sequenced to reveal a 939 bp open reading frame.

The predicted translations from both sequences produce 34.2 kDa proteins, and amino

acid alignment shows 86% identity and 93% conservation of residues with functional

similarity between Arabidopsis TFIIB1 (AtTFIIBI) and GmTFIIB. As indicated by

Figure 2-1. Comparison of TFIIB sequences from plants, metazoans, fungi and
Soybean (G. max) and Arabidopsis (A. thaliana) amino acid sequences for TFIIB were
aligned with homologs from human, frog (X. laevis), fruit fly (D. melanogaster), yeast
(S. cerevisiae), and the Archaebacterium Pyrococcus woesei by Clustal W analysis.
Asterisks denote identical residues and dots conservative substitutions among the
eukaryotic clones. Locations for a-helices A-E within the core repeats are indicated, as
is the zinc ribbon domain and the four residues which coordinate Zn binding. Structural
boundaries are according to Nikolov et al. (1995) and helix labels are from Bagby et al.
(1995). The P. woesei isolate has an additional 23 amino acids at the N-terminus not

Zn binding, p-sheet



Bl Cl D1 El

A2 B2 C2

D2 E2

11il11l 1ii il II I II II 11 1 1 1i1111i 11I

I lII II III ii ii J i I I I I II I ii 1

Il 11[1111 Ii Illl II ii II II iliil ii

ii I I I III llil ii li Iii I1 II

I I 1 I Ii I I I I I II III ii

ii ii 11 1 I iii ii Ili

II I i I l i i ii H HI III I I 1 II I11111i

1II I1 1 I'I I [I II II 1111111 1 1 1 1 1 1 1 ll Il I

l1 l1 lI 1 111111 I I I I I I 1111111 l Il1




lii ii II II ii H I I I I I l ll II I I I II

Figure 2-2. DNA homology of plant clones for TFIIB.
Alignment of coding regions from cDNAs for AtTFIIB1 (top) and GmTFIIB (bottom)
indicates 77% conservation.

I 11111ii IIi llil i i II l ill II

i111 1ii 1 1 II i 1 i 11 i 1 i 1 1 II 1llli 1



I I I 1 1 1 li I I Ii ll 1 1 II I I I I I I I I IH I I

11 1 11 1 1 11 I I I I II II i 1 111111 Ii M l I ii

S I II I I I I l l 1 ii I I 1 II I II1 i l l

HIi I I I II Gii I Il i i lli ii

Figure 2-2 continued

CLUSTAL alignment (Fig. 2-1), both plant factors show extensive homology to versions

of the protein found in other eukaryotes and in Archaebacteria. AtTFIIB1 is about 46%

identical and 62% conserved to the human, fruit fly, and frog homologs; and is 33%

identical and 54% conserved to yeast. DNA sequences for the coding regions of both

plant genes are 77% identical (Fig. 2-2) and are available at GenBank accession numbers

U31096 (AtTFIIB1) and U31097 (GmTFIIB). Subsequent sequencing of the expressed

sequence tag from Arabidopsis indicates a second gene copy, AtTFIIB2, may be present

for this protein. The isolate lacks 11 codons at the 5' end compared to AtTFIIBl, and the

remaining open reading frame encodes a protein 87% identical, 93% conserved to

AtTFIIBl (Fig. 2-1).

The cDNA for GmTFIIB contains only 50 bp of sequences upstream from the first

ATG, and since the reading frame remains open in this region, the possibility of

additional codons could not be ruled out. Analysis of the AtTFIIB1 cDNA, however,

suggests that plant TFIIB genes are indeed slightly shorter at the amino terminus because

the first ATG codon in the open reading frame of AtTFIIB1 aligns with the putative start

codon in GmTFIIB. In contrast to GmTFIIB, the predicted leader sequence of AtTFIIB1

contains a stop codon 29 bp upstream from the first methionine which serves to further

delimit the reading frame. It was unclear from the sequence analysis if this stop codon

was a part of the AtTFIIB1 untranslated leader since the first isolate for AtTFIIBJ carries

an artifact from library construction. The resulting clone is a fusion with the 60S

ribosomal protein L29 such that the 5' leader regions are joined and the open reading

frames are in opposite orientations. To measure whether the true 5' untranslated sequence

-189 -

-169 -

-149 -

-117 -

1 2

Figure 2-3. Primer extension ofAtTFIIB mRNA.
Sequence markers were produced from the pUC19-AtTFIIB1 plasmid using a bottom
strand primer, and positions upstream from the first ATG are indicated (lane 1). Reverse
transcription (lane 2) from the same primer annealed to Arabidopsis mRNA produced a
cDNA with an endpoint approximately 180 nucleotides upstream from the first AUG.

ofAtTFIIB1 is long enough to include the observed stop codon, a primer extension

reaction was conducted using Arabidopsis mRNA. The 5' terminus of the AtTFIIB1

mRNA was shown to be 180 nucleotides upstream of the putative start codon (Fig. 2-3),

indicating that the cDNA may actually contain the complete N-terminus of the protein

since the mRNA leader is more than long enough to include the in-frame stop codon.

Concurrent with this experiment, two additional independent clones for AtTFIIB were

recovered from the library and sequenced; neither contained the artificial fusion and both

confirmed the sequence of the 5' untranslated leader including the stop codon.

Genomic blots were probed with the coding region of the AtTFIIB1 cDNA in

order to obtain an estimate of copy number and to confirm the plant origin of the clone.

The cDNA contains one BgllI restriction site in the 5' leader sequence and no internal

Xbal sites. A single band hybridizing at high stringency to the cDNA probe was observed

when genomic DNA was digested withXbal (Fig. 2-4A). Two bands resulted from BgllI

digestion, one of which is shorter than the cDNA and probably resulted from the presence

of a second site within an intron. An EcoRI restriction site is present in the coding region

90 bp upstream from the stop codon. Double digests with EcoRI and Bgll are predicted

to produce a 920 bp fragment if no introns are present. Since the genomic fragments

detected were approximately 1600 bp and 550 bp in size, a minimum estimate for intron

sequences in AtTFIIB would be 1230 bp. Moderate stringency washing revealed

additional hybridizing bands which may be the result of AtTFIIB2, or other sequences

with homology to TFIIB such as TFIIIB (Fig. 2-4B).

Northern blots were conducted to demonstrate that the AtTFIIB1 gene is

expressed, determine the size of the AtTFIIBI transcript, and to complement ongoing

1 2 3


1 2 3

Figure 2-4. Southern analysis of AtTFIIBl.
A. Arabidopsis genomic DNA (5 ptg per reaction) was restriction digested with Xbal
(lane 1), BglII (lane 2), or EcoRI + BglII (lane 4), blotted and probed with the coding
region of AtTFIIB1 cDNA, and washed at 600 C. B. Genomic DNA (10 pg) was
digested and probed as in (A.), but blots were washed at 420 C. Fragment length
estimates in kilobase pairs are indicated.

1 2

Figure 2-5. AtTFIIB1 Northern analysis.
Arabidopsis poly(A)+ RNA (5 ig per lane) was blotted and then probed with the coding
region of AtTFllB cDNA. The mRNA was isolated from control (lane 1) and heat-
shocked (lane 2) whole plants.

S121 A. thaliana 1

-142 G. max

S107 human
72 109 X laevis

D. melanogaster
36 P. woesei
419 S. cerevisiae

Figure 2-6. TFIIB phylogram.
DNA sequences from the coding regions of clones from Figure 1 were compared by
parsimony analysis. Length units indicate the number of nucleotide changes between
shared branch points.

studies related to heat stress in plants. Equal amounts of Arabidopsis poly(A)+ RNA

from control and heat shocked plants were analyzed (Fig. 2-5). A transcript of

approximately 1400 nucleotides was detected, with no apparent change in abundance

after heat stress. After the hybridization reaction, the membrane was washed at high

stringency for RNA blots (0.2X SSC/0.1% SDS, 420 C) and similar high stringency

washes for the Southern blots showed no cross-hybridization between AtTFIIB1 and

AtTFIIB2 with this probe. Confirmation of the AtTFIIBI expression pattern and

comparison to the mRNA length of AtTFIIB2 will require additional Northern blots using

specific probes derived from the untranslated cDNA regions of each clone.


Plant versions of TFIIB retain the overall structural organization observed among

animal, fungal, and Archaebacterial homologs previously characterized. The zinc binding

domain, adjacent conserved region, and core repeats show similar lengths and

organization within the protein, no doubt reflecting an ancient common origin for the

basal transcription mechanism (Ouzounis and Sander, 1992). Four cysteines, or three

cysteines and one histidine, coordinate metal ion binding in the Zn ribbon, and the

adjacent conserved regions show several blocks of amino acid identity. A portion of this

conserved region in hTFIIB was postulated to form an intramolecular association with the

second repeat, perhaps folding the protein into a conformation that is altered upon

binding with acidic activators (Roberts and Green, 1994). Whether this also occurs in

plants is to be determined. Mutations in this domain alter the mRNA start site in yeast

(Pinto et al, 1994), and the two residues affected by those mutations are identical in the

three plant clones and in all other homologs. Another functional domain characterized in

yeast is helix B 1, which when swapped between yeast and human homologs confers

hTFIIB with the ability to support yeast cell growth (Shaw et al., 1996). Plant sequences

for this helix are more like those of human than yeast, with eight of 16 residues conserved

with yTFIIB and 12 conserved with hTFIIB.

Phylogenetic analysis confirmed previous reports (Ouzounis and Sander, 1992;

Qureshi et at, 1995) that TFIIB homologs sort into equidistant groups corresponding to

taxonomic kingdoms. Our parsimony tree places human, Xenopus, and Drosophila

clones in one branch (metazoan) and plants, yeast, and Archaebacteria on three separate

branches (Fig. 2-6). It is interesting to note that the number of differences detected

between two dicots, Arabidopsis and soybean, is nearly the same as that between

vertebrates and insects. AtTFIIB2 was not included in the tree because its coding region

is incomplete, but a pair-wise comparison with the partial cDNA sequence indicated it is

76% homologous to AtTFIIB1, and AtTFIIBI shows 77% homology with GmTFIIB.

The two full-length TFIIB proteins from plants show a large degree of similarity

in amino acid sequences throughout the length of the protein. This pattern is also

characteristic among other species when TFIIB comparisons are made within the same

kingdom. In comparisons between kingdoms, regions of conservation emerge which are

most evident in three areas: blocks of identity within the 47 amino acids C-terminal to the

Zn domain (residues 30 to 76 of plant TFIIBs), and the two repeat motifs. One of the

regions of variability between kingdoms is the linker between the two core repeats. The

linker is a random coil (Bagby et al., 1995) and varies in length from 17 amino acids in

human, rat, frog and fly to 23 in plants and 29 in yeast. In the Archaebacterium

Sulfolobus shibatae, the linker is from 9 to 18 residues in length depending on the

location of the core repeat boundaries (Qureshi et al., 1995). It is not known if the

presence of additional residues in the plant linker, compared to hTFIIB, affects the

conformation or distance between repeats, or merely reflects further looping out of the

polypeptide chain. It would seem, however, that binding interactions with TBP and DNA

tolerate differences in linker length since the ternary co-crystal structure was successfully

resolved using Arabidopsis TBP (AtTBP) and hTFIIB (Nikolov et al., 1995). Close

interactions were detected between the two proteins at D207 and L208 in the hTFIIB

linker. These residues are identical in AtTFIIBI and show conservative substitution in

GmTFIIB when aligned by the CLUSTAL analysis.

Nikolov et al. (1995) identified a number of van der Waals contacts, salt bridges,

and hydrogen bonds in the DNA-AtTBP-hTFIIB interaction. Of the 27 hTFIIB residues

involved in these contacts, 21 are identical and three conserved in the AtTFIIBI sequence

and 19 are identical, five conserved in GmTFIIB. The three variant amino acid residues

occur near helix C2 and within helix El. hTFIIB residue G247 near helix C2 contacts the

DNA backbone, and in plants this position is occupied by arginine, suggesting that a

charge interaction with the DNA may occur.

The other two points of variance between plant and metazoan proteins occur in

helix El, which has been shown to have several roles in the function of TFIIB, including

binding with activators and TBP. A helical wheel projection of this domain (Fig 2-7A)

indicates that the three residues that interact with the C-terminal stirrup of AtTBP

(hTFIIB K188, G192, and F195) cluster to one face along one-fourth of the helical

surface. Both plant clones have substituted F195 with lysine, changing the nature of this


12 5 Y28D282
1292 Y293

S288 E2 helix
T284 V283

Required for VP16 binding

| DNA contact i


O DNA contact

TBP contact

185 195 200

VSR- I "1I l-: 1 -F i : L [ Fl L


El helix


STBP contact
0 DNA contact

280 290


E2 helix

E DNA contact

Figure 2-7. Interactions with and comparisons between the El and E2 a-helices of TFIIB.
A. A helical wheel representation of the amphipathic El helix of hTFIIB shows sites of
contact with TBP and the DNA backbone occur on two adjacent faces. B. hTFIIB
mutants (residue pairs joined by brackets) that disrupt VP16 binding occur at sites that
overlap DNA contact points within and flanking the El helix (Roberts et al., 1993).
Four of the six residues interacting with DNA or TBP are conserved between human and
plant clones. C. The helical wheel projection for hTFIIB helix E2 shows DNA contact
sites map to opposite faces. D. DNA-binding residues in the E2 helix region are
conserved. Contact points with TBP and DNA were determined for hTFIIB by Nikolov
et al (1995).

position from hydrophobic non-polar to basic. While this hTFIIB residue forms a van der

Waals contact with a glutamic acid in TBP, the corresponding plant interaction is perhaps

electrostatic instead. Three more residues contact the DNA backbone (hTFIIB K189,

R193, and K196), and occupy a second face of the helix. Double mutation of two of

these, K189 and R193, disrupts binding of the acidic activator VP16 (Roberts et al.,

1993), which indicates a somewhat unusual overlap of functions on the same helical face.

In plants, K196 is changed to a glutamic acid (GmTFIIB) or aspartic acid (AtTFIIB1)

(Fig. 2-7B), altering the charge from basic to acidic and decreasing the likelihood of

interaction at this residue with either DNA or an acidic activator domain. In both yeast

versions, this position is occupied by polar uncharged amino acids. Since acidic

activators function in both plants and yeast, the two invariant basic residues along the El

helical face must be sufficient for activation of transcription, in combination with other

sites of interaction that may utilize not just charge but also the bulky hydrophobic

residues often observed within acidic domains. VP16, for example, also binds hTFIIB

helix E2, the analog to El in the second repeat which has no interaction with TBP but

five contacts with DNA. Two of these residues are basic and are located on one helical

face, while the remaining three are uncharged and on the opposite side, a pattern that is

identical in the plant clones (Fig. 2-7C and D). This opposite face contains central

hydrophilic residues surrounded by hydrophobic groups, some of which may facilitate

VP16 binding via the hydrophobic amino acids flanking its acidic domains. Functional

conservation of this TFIIB region appears high since eight of nine other residues also

within helix E2, but not in contact with DNA, are identical or conserved between

metazoans and plants.


Literature Review

Newly cloned homologs of TFIIB are often functionally characterized by testing

their ability to interact with a DNA-TBP complex and to support transcription in a

reaction lacking endogenous TFIIB. Interaction with DNA-TBP has generally been

measured by the electrophoresis mobility shift assay (EMSA) (Ha et al., 1993), in which a

radiolabeled DNA probe is allowed to interact in solution with the binding protein(s) and

the complex is electrophoresed in a native polyacrylamide gel. Probe interacting with

protein will exhibit slower mobility compared to free DNA, and binding is stabilized

upon exposure to the electric field by caging effects of the polyacrylamide matrix (Fried

and Crothers, 1981; Garner and Revzin, 1981). EMSAs for the DNA-TBP and DNA-

TBP-TFIIB interactions have given variable results, depending on the EMSA conditions

and the source and purity of the TBP protein.

Initial binding studies with the yeast TFIID fraction required TFIIA to produce a

stable interaction (Buratowski et al., 1989). The subsequent use of recombinant yeast

TBP (yTBP) allowed its characterization in the absence of TAFs or other TFIID proteins,

and demonstrated TATA interaction without a requirement for TFIIA (Horikoshi et al.,

1989; Kao et al., 1990; Peterson et al., 1990). Recombinant human TBP (hTBP) directs

formation of a more stable complex with TFIIA and TFIIB than does yTBP (Maldonado

et al., 1990). Under conditions optimized for TFIIB binding, there is sometimes no DNA

interaction with yTBP alone (Buratowski and Zhou, 1993), and hTBP often forms a

DNA-TFIIB triple complex that does not resolve as a separate band but instead is an

extended region of enhanced probe binding (Thompson et a., 1995; Yamashita et al.,

1993). The presence of TFIIB in these cases can be confirmed by addition of antibodies

raised against TFIIB, producing a supershifted signal present only when TFIIB has

interacted with the labeled probe through TBP (Maldonado et a., 1990; Thompson et a.,

1995). EMSAs have been used to estimate the apparent equilibrium dissociation constant

(Kd) of yTBP bound to a TATA probe as 2 nM and the Kd of potato TBP as 5 nM, and the

rates of association and dissociation were described as slow (Hahn et al., 1989;

Holdsworth et a., 1992).

The EMSA technique has been combined with DNase I protection assays to more

accurately measure TBP binding kinetics. DNA binding proteins block access to the

sequences they bind, preventing DNase I digestion in that region and generating a

footprint that can be resolved by electrophoresis on sequencing gels. Several reactions

can be run in parallel to titrate the amount of protein required to occupy a binding

element, and this approach was used to estimate the yTBP Kd as 3 nM (Hoopes et al.,

1992). It was also observed that the rate of dissociation (kd) remained slow but the rate of

association (ka) was quite rapid. The differences in ka between this and previous

experiments were attributed to the use of higher TBP concentrations. DNase I

footprinting of yTBP showed a 2.5-fold increase in TATA affinity when TFIIB was added

and no effect with the addition of TFIIA (Imbalzano et al., 1994). Altering the binding

buffer to create sub-optimal conditions for yTBP, however, produced a 10-fold increase

in binding affinity when either hTFIB or yeast TFIIA was included. No estimates for the

Kd of TFIIB or TFIIA binding to a DNA-TBP complex have been reported.

An alternative assay for protein-DNA binding utilizes fluorescence anisotropy

(Lundblad et al, 1996). An oligonucleotide labeled with a fluorescein tag is exposed to

polarized light at the excitation wavelength, and if no molecular motion occurs all the

resulting fluorescence emission will remain polarized and is detected in the original

plane. In solution, however, the DNA rapidly rotates, carrying the excited tag out of the

original plane of polarization before emission and reducing the amount of detected

fluorescence. Since the rate of rotation is related to the molecular volume, interaction

between the oligomer and a DNA-binding protein will result in a complex with increased

volume, slower rotation, and less signal leaving the plane of excitation (Cantor and

Schimmel, 1980). In practice, anisotropy (A) is measured by detecting fluorescence

intensities (I) in both the excitation plane (III, polarizing filter for detection set parallel to

excitation filter) and perpendicular to that plane (IL). The result of calculating

II + 2I1

expresses the rotational depolarization of the measured fluorescence; the denominator

indicates total emission intensity and the mathematical maximum for A is 0.4 (no

molecular rotation) (Cantor and Schimmel, 1980; Jameson and Sawyer, 1995). Several

DNA-transcription factor complexes have been analyzed by fluorescence anisotropy,

including the binding of CREB to its cAMP-responsive enhancer element and the

subsequent association of a co-activator, CBP (Lundblad et al., 1996). Advantages of this

method compared to standard EMSAs include having all reactants at equilibrium in

solution rather than in a gel matrix and being able to rapidly measure multiple interaction

reactions in real time under various buffer conditions.

A second functional test for TFIIB is the ability to support in vitro transcription.

Multi-column chromatography protocols are used to fractionate nuclear extracts, and

when the fractions are appropriately re-mixed transcription can be detected, often using a

specialized template or primer extension to ensure the assay measures RNA specifically

initiated from the pol II promoter. A functional recombinant TFIIB protein should

substitute for fraction B after E. coli expression and purification (Tsuboi et a., 1992;

Wampler and Kadonaga, 1992; Yamashita et al., 1992). Similar experiments can be

performed on crude nuclear extracts that have been specifically depleted for the factor of

interest using immobilized antibodies or, as was the case for hTFIIB, a column of

immobilized VP16 activation domain (Roberts et al., 1993).

Besides a few activation studies in monkey and Drosophila cells (Colghn et al.,

1995; Paal et al., 1997; Schmitz et al., 1995), most in vivo characterizations of TFIIB

have been in yeast. Mutations affecting the transcription start site have been mapped to

the conserved region of yTFIIB adjacent to the Zn domain (Pinto et a., 1994), and

suppressors of these mutants have been used to identify interactions with pol II

(Berroteran et al., 1994; Sun et al., 1996; Xiao et al., 1994) and other cellular factors

(Sun and Hampsey, 1996). Yeast plasmid shuffle experiments in which chromosome-

encoded yTFIIB is replaced with a plasmid-bome copy have detected differences between

the K. lactis and S. cerevisiae homologs (Na and Hampsey, 1993) and indicate that rat

and hTFIIB are unable to substitute for yTFIIB (Shaw et al., 1996; Tschochner et al.,

1992). Promoter-specific differences have also been observed with in vitro comparisons

of Drosophila and hTFIIB (Wampler and Kadonaga, 1992).

A series of yeast/human TFIIB hybrids was constructed to map the domain

responsible for incompatibility in vivo, and species specificity was localized to helix B1,

the second a-helix in TFIIB repeat 1 (Shaw et al, 1996). Gain-of-function experiments

with an inactive yTFIIB carrying a substituted human B1 helix confirmed that four

residues within the helix are critical for proper function; their reversal to the yeast

versions restored activity (Shaw et al., 1997). Point mutations were generated to change

yTFIIB to the hTFIIB sequence at those four positions and the resulting constructs

displayed reduced viability, temperature-sensitivity, and reduced activation at some

promoters. Since cell viability was not abolished (as occurs upon substitution by wild

type hTFIIB) other regions of TFIIB must contribute to inter-species functionality and

transcription activation, a conclusion supported by the recovery of intragenic mutations

outside helix Bl which suppress the point mutants described.

This chapter investigates the ability of AtTFIIB 1 to interact with complexes

derived from Arabidopsis, human and yeast TBP using EMSAs and fluorescence

anisotropy. An estimate of the equilibrium binding constant for AtTFIIB1 interaction

with the DNA-AtTBP complex is made, as well as measurements of AtTBP on- and off-

rates with and without AtTFIIB 1. Confirmation that AtTFIIB1 supports transcription is

provided by in vitro reactions, and the HeLa origin of these reactions suggests

compatibility with human transcription factors but not with the yeast system where

plasmid shuffle tests produced non-viable cells.

Materials and Methods

Expression and Purification of Recombinant Proteins

A PCR amplification of AtTFIIB1 was performed to attach a Kpn2I site before the

first methionine and a Sail site in place of the stop codon using the primers 5AtIIB: 5' T-


ACTTGACAGGT 3'. The PCR product was blunt-end ligated into the SmaI site of

pUC19, and the resulting plasmid pAtTFIIB 1 was digested with SstI and Sall. The

resulting fragment was cloned into plasmid pET24b (Novagen) at those sites to produce a

fusion protein carrying the T4 epitope at the N-terminus and a six-histidine tag at the C-

terminus. Overnight cultures ofE. coli BL21(DE3) were diluted 1:200 in LB medium

with 0.01 mg/ml kanamycin and incubated at 370 C with shaking for 2.5 hr. After

addition of 0.1 mM isopropyl P-D-thiogalactoside (IPTG), incubation continued with

vigorous shaking at room temperature for 3-4 hr. Bacterial pellets were collected and

sonicated in Ni binding buffer (5 mM imidazole, 0.5 M NaCI, 20 mM Tris-HCl pH 7.9)

with protease inhibitors (1 mM phenylmethylsulfonyl fluoride, 20 ig/ml pepstatin A, 20

pg/ml leupeptin, 1% aprotinin) and 0.1% NP-40. Aliquots (1.2 ml) of the cleared lysate

were mixed with 100 pl of HisBind resin (50% slurry, Ni charged; Novagen) for 15-30

min at 40 C. Resin batches were pelleted by low-speed centrifugation and washed once

with 1 ml Ni binding buffer and twice with Ni wash buffer (Ni binding buffer with 60

mM imidazole). Fifty to 100 pl of Ni elution buffer (0.3 M imidazole, 0.5 M NaCI, 20

mM Tris-HCI pH 7.9) were added and the sample was rocked for 10-20 min at 40 C.

Eluted proteins were stored at 40 C or flash frozen in liquid nitrogen after addition of

glycerol to 20% for storage at -700 C. pET vectors carrying histidine tagged cDNAs

encoding AtTBP2, hTBP and yTBP (provided by Drs. Nam-Hai Chua and Robert Roeder,

Rockefeller University) were used for TBP expression and purification as described

above for AtTFIIB1. Total protein concentrations were determined with dotMETRIC

colorimetric strips (Geno Technology, Inc.) and compared to standard curves by spot

densitometry using an IS1000 Digital Imaging System (Alpha Innotech Corp.). E. coli

expressed proteins were electrophoresed on SDS-12% polyacrylamide gels along with

Mid-Range Protein Markers (Promega) and visualized by Coomassie blue staining. The

purity of recombinant proteins was determined by band peak integration with the IS 1000

system. Western blot detection of TFIIB proteins is described in Chapter 4.

Electrophoretic Mobility Shift Assays

An oligonucleotide derived from the TATA region of the adenovirus major late

promoter (AdMLP), 5' GGCTATAAAAGGGCTG 3', was 32P end-labeled to 4.6 x 109

cpm/lg with T4 polynucleotide kinase and annealed with a 2-fold excess of bottom-

strand oligomer. Probe (5 x 104 cpm) was added to EMSA reactions containing BS buffer

(10 mM HEPES pH 7.9,4 mM MgC12, 5 mM NH3SO4, 50 mM KC1, 0.2 mM EDTA,

10% glycerol, 5 mM P-mercaptoethanol) and incubated with TBP or TBP+TFIIB at 370 C

for 45 min. Reactions were electrophoresed (125 V at room temperature) on 4%

polyacrylamide (30:1) gels containing 10% glycerol in either TGM buffer (25 mM Tris,

100 mM glycine, 4 mM MgCI2, 1 mM EDTA, 0.5 mM DTT) or 0.5X TBE buffer

supplemented with the same concentrations of Mg, glycerol and DTT.

Fluorescence Polarimetry

A second AdMLP TATA probe was synthesized (CyberSyn) carrying sequences

through the transcription start site and modified with a 3' fluorescein tag on the bottom


competitor DNA was also created by annealing oligonucleotides with the same sequences

but lacking the fluorescein tag. Binding reactions contained BS buffer with no glycerol or

p-mercaptoethanol, 0.5 nM DNA probe, and TBP or TBP+TFIIB in a volume of 0.35 ml

at ambient temperature in the sample chamber (280 C after warm-up). Data were

collected from an SLM/Aminco polarizing luminescence spectrometer (Series 2, Aminco

Bowman) set for excitation at 494 nm, bandpass 4 and detection of emission at 520 nm,

bandpass 16. Data collected under equilibrium conditions are the average of 10 one-

second measurements, and kinetic experiments were measured at five-second intervals for

500-1000 sec. Anisotropy values were calculated after subtraction of background

fluorescence, which was measured in parallel reactions that received identical

experimental manipulation but contained no fluorescein-labeled DNA. The data in Fig.

3-3 are raw signal traces from the spectrometer without background correction and were

collected as polarization units (P) to emphasize this difference. Polarization is calculated


Iii I_
Ii + I1

and the mathematical maximum for P is 0.5. Binding reactions were analyzed by non-

linear regression with curve fitting (SigmaPlot, Jandel Scientific).

In vitro Transcription

HeLa crude nuclear extracts prepared in the laboratory of Dr. J. B. Flanegan

(University of Florida) (Dignam et al., 1983) were incubated on ice for 30 min with IgG

monoclonal antibodies to hTFIIB (Promega) at 400 pg per ml extract. Sepharose linked

to anti-IgG antibody (Sigma) was then added, incubated and centrifuged for 2 min at 500

x g to deplete hTFIIB from the extract. Reactions were assembled containing

transcription buffer (10 mM HEPES pH 7.9, 50 mM KC1, 0.1 mM EDTA, 0.25 mM

DTT, 10% glycerol), 3 mM MgC12, 0.4 mM each ofATP, CTP and GTP, 100 ng template

DNA, and 1 pl of a-32P UTP (high specific activity, 10 mCi/ml, Amersham) in a final

volume of 30 [l. The supercoiled plasmid used as a transcription template carries the

cytomegalovirus (CMV) immediate early promoter fused to a hammerhead ribozyme

sequence (Batt, 1996) so that template derived RNA will be consistently self-cleaved at

the same 3' terminus. Reactions also contained 5 gl of HeLa extract (6.3 mg/ml total

protein) or an equivalent amount (by protein mass) of depleted extract plus recombinant

TFIIB, and were incubated at 300 C for 60 min before adding 175 il of stop mix (0.3 M

Tris-HCI pH 7.4, 0.3 M sodium acetate, 0.5% SDS, 2 mM EDTA, 3 tg/ml tRNA).

Extraction with phenol:chloroform:isoamyl alcohol was followed by ethanol precipitation

and resuspension of the RNA pellet in nuclease-free water. Samples were

electrophoresed on a 6% denaturing polyacrylamide (40:1) gel. Autoradiographs were

quantified by densitometry on the IS 1000 system.

Yeast Plasmid Shuffle for TFIIB

AtTFIIB1 cDNA was subcloned as an NdeI-NotI fragment from pET24b into two

yeast expression vectors. pl210A is a high copy number plasmid driving expression

from the strong yeast Adhl promoter, and pLC1210 is a low copy number version

expressing AtTFIIB1 from the wild type yeast TFIIB promoter. Both plasmids were

provided by Dr. Jun Ma (University of Cincinnati), and in his laboratory Dr. Shruti Shaw

performed the yeast transformations to replace yTFIIB with AtTFIIBI as described (Shaw

etal., 1996).


Proteins produced by E. coli expression and elution from HisBind resin ranged

from 50-80% purity and from 0.2-0.5 tg/gl in concentration (Fig. 3-1A). EMSA gels for

AtTBP and hTBP interaction with the TATA site were best resolved in supplemented

TGM buffer. AtTBP binding was specific for the TATA oligomer probe and stable in the

presence of non-specific competitors (Fig. 3-1B). Addition of AtTFIIB1 to the AtTBP

reaction produced a second band with slower mobility (Fig. 3-1B lane 5) that was not

present after addition of HisBind purified E. coli lysate from a culture not expressing

AtTFIIBI (data not shown). A similar reaction combining hTBP and AtTFIIB1 did not

show a discrete second band or any other obvious changes in hTBP binding (Fig. 3-1C).

Yeast TBP showed poor EMSA resolution in the TGM buffer system, so gels

were run in supplemented 0.5X TBE. Increasing concentrations ofyTBP shifted a greater

proportion of the probe (Fig. 3-2A), and three of these concentrations were tested for

A. MW (kDa)
66.2 -
55.0 -

31.0 -

14.4 -

1 24 5 6

1 2 3 4 5 6


- yTBP



free probe

1 2 3 4 5

Figure 3-1. Purification and interaction of TBP and AtTFIIBl.
A. Coomassie stained recombinant proteins (10 p1) were electrophoresed on a 12% SDS-
polyacrylamide gel. Lanes: 1) 1.5 Vtg of marker, 2) AtTFIIBI, 3) 0.5 ig of marker,
4) AtTBP, 5) hTBP, 6) yTBP. B. EMSAs detect TATA binding by 1 tg of AtTBP (lane
1) which is stable in 1 tg of poly d(I)d(C) + 0.5 Lg oftRNA (lane 2), but specifically
competed by 100-fold excess of unlabeled TATA oligomer (lane 3). AtTBP (0.1 ig)
also binds (lane 4) and shows enhanced interaction and a larger complex with addition of
0.2 ig of AtTFIIB1 (lane 5).


1 2

Figure 3-1 -- continued
C. EMSAs detect similar TATA binding by 0.2 tg of hTBP without (lane 1) or with
(lane 2) 0.4 Vig ofAtTFIIB1.

A. B.

1 2 3 4 5 1 2 3 4 5 6

Figure 3-2. Interaction ofyTBP with the TATA box and AtTFIIBl.
A. Gels in 0.5X TBE resolve the products of EMSA reactions with the TATA probe and
varying amounts ofyTBP. Lanes: 1) 30 ng, 2) 60 ng, 3) 0.12 gg, 4) 0.24 lg, 5) 0.48 pg.
B. Lanes 1-3 are replicates of those in (A.), lanes 4-6 are replicates with 60 ng (lane 4),
0.12 gg (lane 5), and 0.24 gg (lane 6) of AtTFIIB1 added.

AtTFIIBl binding. With a limiting quantity of yTBP insufficient to shift a detectable

amount of probe, addition of AtTFIIBl enhanced yTBP binding significantly (Fig. 3-2B,

lanes 1 and 4) and with a wider range of mobility. At higher yTBP concentrations

binding appeared to be reduced rather than enhanced, but with a consistent reduction in

the amount of free probe detected.

The EMSA analysis suggested AtTFIIB1 does have affinity for AtTBP and yTBP

proteins, but variable conditions were required and produced results which were difficult

to quantify. Fluorescence polarimetry was therefore tested to determine whether changes

in anisotropy could be detected for TBP binding to the TATA box and for subsequent

binding of AtTFIIB1. Fig. 3-3 is a trace of typical polarization data obtained in an

experiment monitoring AtTBP binding to fluorescein-labeled TATA box DNA. Free

probe rotates quickly and showed little polarization upon addition to the cuvette at 65 see,

but introducing AtTBP at 100 sec immediately formed a larger volume complex that

slowed rotation and increased polarization to more than 0.3 P (Fig. 3-3A). A 10-fold

excess of unlabeled competitor TATA reduced the binding signal (Fig. 3-3B, 500 sec),

and 100-fold excess of competitor DNA further reduced binding to a level near that of

free probe (700 sec). A parallel reaction to test for non-specific binding produced a

maximum signal around 0.25 P and was not affected by 100-fold excess poly d(I)d(C)

(Fig. 3-3C).

TBP binding reactions were measured at equilibrium, corrected for background

fluorescence and plotted in anisotropy units for comparisons of AtTFIIB1 interaction.

Where indicated, total F is the total fluorescence emitted and should remain at a steady

level for different treatments within an experiment. Variation of this control indicates an


0.34 1---------- ------------------------------' "- ---

P 0.2-
0.1 -

0 100 200 300 400 500
Time (sec)



P 0.2 -
0 --ii I I I I L n i i I T in i i

500 600 700 800 900 1000
Time (sec)



P 0.2 -

0 100 200 300 400 500
Time (sec)

Figure 3-3. Fluorescence polarization vs. time for AtTBP interaction with the TATA box.
A. Fluorescein-tagged TATA probe (1 nM) exhibited 0.1 polarization units (P) when added
to the binding reaction at 65 sec, and addition of 0.5 gg of AtTBP at 100 sec created a
complex with over 0.3 P. B. Addition of 10-fold excess unlabeled TATA competitor to
the reaction in (A.) at 500 sec reduced the binding signal to 0.2 P, and a further reduction
occurred when the competitor was increased to 100-fold excess over probe at 700 sec.
C. A reaction similar to (A.) reached equilibrium at 0.25 P before addition of poly d(I)-
d(C) to 100 nM (arrow).


A 0.15

0 8 16 24 32 40 60

Figure 3-4. Fluorescence anisotropy of the TATA box with AtTFIIBl.
TATA probe (0.5 nM) was combined with the indicated concentrations of AtTFIIB and
incubated 15 min. Average anisotropy with standard deviation often measurements is

alteration of fluorescence intensity from the fluorescein tag that may indicate large scale

conformation changes or interference from excess protein levels. The results of a second

experimental control are shown in Fig. 3-4 in which AtTFIIBl alone showed no binding

affinity for the TATA probe. Figs. 3-5 and 3-6 show data for three TBP homologs

obtained from progressive assembly of two parallel binding reactions, measured after 10

min incubations of each assembly step to ensure the components are at equilibrium.

Assembly steps 1-3 were in a double-volume reaction and represent free probe (1), probe-

TBP (2), and probe-TBP plus 100-fold excess competitor DNA (3). The addition of

competitor oligomer incorporated most free TBP into TATA complexes so that any

further change in anisotropy was due to TFIIB binding and not simply an equilibrium

shift toward more TBP binding DNA. The reactions were split following step 3 to test

AtTFIIB1 and control treatments. Fig. 3-5 shows that addition of equal or 2-fold amounts

of AtTFIIB 1 relative to AtTBP slightly increased the anisotropy and a 3-fold excess of

AtTFIIB1 showed a significant anisotropy increase, suggesting that TFIIB was binding to

the TATA-TBP complex. The parallel reaction received equal or excess amounts of BSA

rather than AtTFIIBl. The results (Fig. 3-5 open bars) indicated that the increased

anisotropy observed with AtTFIIB1 was specific for ternary complexes and was not due

to general TBP stabilization by higher protein concentrations. In fact, extra non-specific

protein appears to de-stabilize TBP binding, an effect previously reported for much

higher BSA concentrations in yTBP footprint experiments (Imbalzano et al., 1994). It is

unclear why BSA induced less anisotropy than observed for free probe in this experiment.

In Fig. 3-6, AtTFIIB1 slightly increased anisotropy with hTBP but not to the degree seen






1 2 3 4 5 6 7



a 6


2 3. 0o *o 0O 0 *0 D

1 2 3 4 5 6 7

Figure 3-5. Equilibrium anisotropy of TATA-AtTBP-AtTFIIB interaction.
A. A double-volume reaction containing 0.5 nM TATA probe was incubated with 15 nM
AtTBP, followed by 50 nM competitor oligomer. Anisotropy (A) was measured for free
probe (1), probe+AtTBP (2), and probe+AtTBP+competitor (3). The reaction was then
equally divided into two cuvettes. Sequential aliquots of AtTFIIB1 were added to one
treatment (filled bars) to final concentrations of 20 nM (4), 40 nM (5), and 60 nM (6).
The second reaction (open bars) received no AtTFIIB1, and BSA was added to final
concentrations of 20 nM (4), 60 nM (5), 120 nM (6), and 240 nM (7). Each assembly
step was incubated for 10 min before collecting 10 measurements of A. Average A with
standard deviations are shown. B. Total fluorescence intensity for the AtTFIIB I reaction
(filled diamonds) and BSA reaction (open squares) remained constant. All protein
concentrations reported for AtTBP and AtTFIIB in figures for this chapter are adjusted
for binding activity, as assayed in Fig. 3-10.



0.2 -


1 2 3 4 5 6 7 8






1 2 3 4 5 6

Figure 3-6. Equilibrium anisotropy of AtTFIIB1 interactions with hTBP and yTBP.
A. Reactions were assembled as in Fig. 3-5 substituting hTBP for AtTBP. AtTFIIBI
was added (filled bars) to final concentrations of 20 nM (4), 40 nM (5), 60 nM (6), and
80 nM (7). The BSA series (open bars) received 60 nM (4), 120 nM (5), 180 nM (6),
240 nM (7), and 300 nM (8) non-specific protein. B. yTBP was combined with
AtTFIIB1 (filled bars) at concentrations listed in Fig. 3-5, or with BSA (open bars) at
60 nM (4), 120 nM (5), and 180 nM (6).

with AtTBP. yTBP appears to have high affinity for AtTFIIB1, but BSA also

significantly increased anisotropy with increasing concentration. No change in total

fluorescence was observed in these experiments (Fig. 3-4 and data not shown).

To determine whether AtTFIIB affects the Kd for TBP-DNA binding in these

reactions, binding curves were measured and are plotted in Figs. 3-7 to 3-9. Increasing

concentrations of TBP with and without a 3-fold excess of AtTFIIBl were added to 0.5

nM TATA probe and allowed to reach equilibrium. The inflection points of the resulting

curves are the apparent Kd values. The presence of AtTFIIB1 raised anisotropy values

throughout the curve and increased AtTBP affinity for probe DNA by 1.9-fold (Fig. 3-7).

AtTFIIB1 also increased anisotropy with hTBP (Fig. 3-8), but the Kd appears unchanged.

The fitted curves for yTBP binding were linear, rather than sigmoidal, which is usually

interpreted as a non-specific interaction (Fig. 3-9). Attempts to detect non-specific DNA

binding by poly d(I)d(C) competition showed no effect on yTBP or hTBP (data not


A more detailed investigation ofAtTFIIBl influences on AtTBP was conducted

by measuring the on- and off-rates for DNA binding. A plot of binding vs. time indicated

that AtTBP quickly associated with the TATA box with the probe half occupied in less

than five seconds (Fig. 3-10A). Repeating the assay with AtTFIIB1 present had little

effect, and BSA reduced overall binding but not the rate of association. Dissociation,

however, was altered by AtTFIIB1 (Fig. 3-10B and C) which obviously decreased the

AtTBP off-rate after addition of 200-fold excess competitor DNA. The kd for each fitted

curve was calculated by plotting the natural log of (Aobs Ai )/( A Amin) vs. time,


40 o .0 0 M.

0.2 -


0.1 -
0 0




0 I I ...i I I
1.0 10 100 1000
AtTBP (nM)

Figure 3-7. TATA binding titration of AtTBP with and without AtTFIIB1.
Increasing concentrations of AtTBP were bound to 0.5 nM TATA probe and allowed to
reach equilibrium in the absence (open circles) or presence (filled circles) of AtTFIIBI at
three-fold higher concentrations. The average and standard deviation of ten anisotropy
measurements per reaction are indicated (middle panel), as well as total fluorescence
(top). Fitted curves were generated by non-linear regression (bottom), and residuals from
the equation for each curve are shown as vertical error bars.


1 .

1.0 10 100 1000
hTBP (nM)

Figure 3-8. TATA binding titration of hTBP with and without AtTFIIB1.
Increasing concentrations of hTBP were bound to 0.5 nM TATA probe and allowed to
reach equilibrium in the absence (open circles) or presence (filled circles) of AtTFIIB 1 at
three-fold higher concentrations. Data are plotted as described in Fig. 3-7.

1.0 10 100 1000
yTBP (nM)

Figure 3-9. TATA binding titration of yTBP with and without AtTFIIBl.
Increasing concentrations ofyTBP were bound to 0.5 nM TATA probe and allowed to
reach equilibrium in the absence (open circles) or presence (filled circles) ofAtTFIIB1 at
three-fold higher concentrations. Data are plotted as described in Fig. 3-7.

-T 8 _____________


0.1 +BSA


0 200 400

Time (sec)

4 -------------------



0 200 400

Time (sec)

Figure 3-10. Rates of AtTBP association and dissociation from the TATA box with and
without AtTFIIB1.
A. A reaction containing 0.5 nM TATA probe was adjusted to 47 nM AtTBP and
anisotropy was measured at five second intervals. The assay was then repeated in the
presence of 84 nM AtTFIIB1 or 84 nM BSA. Curve-fitting equations for AtTBP and
AtTBP+AtTFIIBI were normalized to the same plateau to compare curve shapes.
B. AtTBP (35 nM) was pre-equilibrated with 0.5 nM TATA probe, or probe and 75 nM
AtTFIIBl or BSA, before addition of competitor oligomer to 100 nM at time 0.



Time (sec)

Figure 3-10 -- continued
C. Fitted curves for the data collected in (B.) were calculated after normalization to
equalize the binding signals at 0 sec.

- I .. ... I

[ t

where Aob s the anisotropy at each time point and Amax and Ami, are the maximum and

minimum values observed during the assay (Beacon Applications Guide, 1995). The

absolute value of the slope of the resulting line is the kd which for AtTBP alone was 1.01

x 10.2 sec-' with a half-time (tin) of 69 sec. Addition of AtTFIIB1 reduced the kd to 6.9 x

10"3 sec-' (t/2 = 100 sec), a 1.5-fold decrease in the off-rate.

In order to estimate the apparent Kd for AtTFIIB1 binding to the DNA-AtTBP

complex, the binding activity of the E. coli expressed proteins was first determined. Probe

(0.5 nM) was mixed with 50 nM TATA box competitor and AtTBP was added until the

anisotropy signal plateaued, indicating saturation of all TATA binding sites (Fig. 3-1 1A).

The half-maximal binding point on this curve indicates that 42.7 pmol (955 ng) of AtTBP

was required to occupy 8.8 pmol (222 ng) ofDNA oligomer, or 20.6% specific binding

activity for AtTBP assuming binding occurs as a monomer. This is similar to the activity

measured for recombinant yTBP (Hoopes et at, 1992). A second reaction was performed

with the same DNA concentration, 11 pmol of AtTBP (246 ng), and increasing amounts

of AtTFIIBI (Fig. 3-11B). Since there should have been enough excess DNA present to

occupy most of the active AtTBP, the inflection point of this curve indicates 21.3 pmol

(728 ng) of AtTFIIB1 was required to bind 5.5 pmol of DNA-AtTBP, or a binding

activity of 25.7% for AtTFIIB. This calculation again assumes monomeric binding and

also no significant affinity ofAtTFIIBI for the inactive, non-DNA associated AtTBP

which would reduce the activity observed. There are no reported measurements of TFIIB

affinity for TBP in the absence of TATA DNA using recombinant proteins, but the two

factors do not co-purify from nuclear extracts (Buratowski et al., 1989; Maldonado et al.,

1990). A final titration was conducted with 0.5 nM labeled probe, 8 nM active AtTBP,



A 0.15-

0.1 -




1 10 100 1000 10000

AtTBP (ng)

1 10 100 1000

AtTFIIB (ng)

Figure 3-11. Titration of binding activity for recombinant AtTBP and AtTFIIB.
A. Unlabeled TATA oligomer (50 nM) was mixed with 0.5 nM TATA probe. Average
anisotropy was measured after 10 min incubations with sequentially added aliquots of
AtTBP. Data are shown after curve fitting and residuals are indicated. B. A reaction
containing 50 nM unlabeled TATA oligomer, 0.5 nM TATA probe, and 31 nM AtTBP
(246 ng) was pre-equilibrated before addition of AtTFIIB1 and analysis as in (A.).

L" 6

I 4

Figure 3-12. Determination of equilibrium binding constant for AtTFIIB1 interaction
with DNA-AtTBP.
TATA probe (0.5 nM) and 8 nM AtTBP were at equilibrium before addition of AtTFIIB1
aliquots for additional 10 min incubations. Data were analyzed as in Fig. 3-7.

00 0 0 0

* *.^*


0.1 1.0 10 100 1000


and a range of concentrations of AtTFIIB 1 to generate the binding curve depicted in Fig.

3-12. Under these conditions probe is limiting for complex formation and the inflection

point occurs at the apparent Kd of 7.1 nM for AtTFIIB 1 interaction with DNA-AtTBP.

The ability of AtTFIIB to interact with hTBP and support in vitro transcription

was tested using HeLa nuclear extracts. Transcripts from the CMV promoter-ribozyme

reporter were resolved on a 6% sequencing gel as seven bands one nucleotide apart when

untreated extract was used (Fig. 3-13A, lane 1), and depletion of the extract with anti-

hTFIIB antibodies reduced transcription in lane 2 to 35% relative to the untreated

reaction. The background bacterial proteins present after purification ofE. coli lysates

did not influence the observed transcription (lane 3). Addition of increasing amounts of

recombinant hTFIIB or AtTFIIB 1 to the depleted extract partially restored activity by

increasing accumulation of three transcripts with adjacent start sites. A site-directed

mutagenesis reaction was used to create a single amino acid substitution in AtTFIIB1 at

residue 43 (glutamic acid changed to arginine) in the conserved region. The analogous

mutation in yTFIIB altered the start site at several promoters in vivo (Pinto et al., 1994),

but had no effect on AtTFIIBI in this assay. The values shown for percent relative

transcription for each treatment in Fig. 3-13A were derived from six replicate reactions

that contained 400 ng of the various TFIIB proteins tested. The seemingly impaired

ability of hTFIIB to restore transcription may have been due to carryover of the antibody

used for immuno-depletion, which bound recombinant hTFIIB but not AtTFIIB (Fig. 3-

13B). The anti-IgG resin used to deplete the extract was assumed to be present in

sufficient concentration to prevent such carryover; it is also possible that E. coli

expressed hTFIIB is less active than AtTFIIB1.

1 2 3 4 5 6 7 8 9 10

Il .. .. ,, -.. -i ..- ,


% Relative I, ""' R-!.0 WIN' N15 F!k"'I /1.. .. -
Transcription 100 34.3 35.5 49.0 64.2 57.8
+/- SD 0.75 0.5 3.8 3.8 3.3


1 2 3 4 5 6

Figure 3-13. AtTFIIB1 supports in vitro transcription using HeLa nuclear extracts.
A. In vitro transcription from a CMV promoter produces ribozyme RNA transcripts centered around 209 nucleotides in length. Five tl
of HeLa nuclear extract (lanes 1 and 16) or an aliquot of depleted extract containing an equivalent amount of total protein was used
for each reaction. Extract depleted ofhTFIIB (lane 2) was supplemented with control E. coli lysate (lane 3) or 50, 100, 200, and 400
ng purified hTFIIB (lanes 4-7). The same amounts of purified AtTFIIBI were also tested (lanes 8-11) as well as the AtTFIIBI mutant
E43R (lanes 12-15). Six replicates of the reactions containing 400 ng recombinant protein were performed (not shown), and the levels
of transcription relative to untreated extract are indicated as average percentages with standard deviations. B. In lanes 1-3, the
monoclonal anti-hTFIIB antibody used for immuno-depletion detects recombinant hTFIIB (lane 1) but not AtTFIIBl (lane 2) or its
mutant E43R (lane 3). Lanes 4-6 are a replicate Western blot but probed with antibody against the T7 epitope which is fused to
AtTFIIB1 and its mutant E43R in the pET24b expression vector (see also Fig. 4-6A).

A plasmid substitution experiment was conducted to test the compatibility of

AtTFIIBl in yeast transcription. Constructs designed to produce physiological levels or

over-expression of AtTFIIBl were transformed into yeast cells, and selection with 5-FOA

forced the loss of plasmid-bome yTFIIB. No viable colonies were observed after plasmid

substitution at either expression level.


An accurate comparison of the AtTFIIB1 binding affinity for TBP from

Arabidopsis, humans and yeast was difficult due to the variable EMSA conditions

required to measure interactions and qualitative differences in the appearance of EMSA

results. These differences also seemed to affect the more sensitive assays using

fluorescence anisotropy (Figs. 3-5 and 3-6). AtTFIIBI had high affinity for AtTBP and

yTBP, but the yeast protein exhibited a different pattern of sensitivity to BSA. hTBP

showed less relative binding to AtTFIIBI under the standard assay conditions, but the

interaction seems to be productive given the results from in vitro transcription. Different

characteristics again appear in titrations of TBP-TATA binding. Addition of AtTFIIBl in

all cases gave greater anisotropy than when only TBP was present, but an influence on the

apparent binding constant was only observed with AtTBP in Fig. 3-7. The inability to

generate a normal binding curve with yTBP, even in the absence of TFIIB, indicates that

alternative reaction conditions are needed to properly assay its activity as was the case for

EMSA experiments. The yeast TFIIB protein may be susceptible to aggregation or

precipitation since free probe disappeared without visible conversion to slower mobility

complexes in the yTBP-AtTFIIB1 EMSA reactions. Although not seen in Fig. 3-2B, the

wells of yTBP EMSA gels occasionally showed significant amounts of radiolabeled

probe that was apparently unable to enter the gel matrix.

The fluorescence polarization results suggest the nature of TFIIB species

specificity is not strictly dependent upon the level of affinity for a TATA-TBP complex.

Equilibrium anisotropy measurements predicted higher affinity binding ofAtTFIIB1 to

AtTBP and yTBP than to hTBP, yet the plant protein supported transcription from HeLa

extracts but not in yeast substitutions. A comparison of the B1 helices from TFIIB

homologs shows that the AtTFIIBI sequence (and that of other plant homologs) is more

conserved with human than yeast factors (Fig. 3-14), but AtTFIIBI is equally conserved

with hTFIIB and yTFIIB at the four residues within the helix that were initially implicated

as being important for species specificity (Shaw et a., 1996). Subsequent mutations of

the four residues revealed that yeast K151 is most critical for in vivo function, followed

by equal effects resulting from changing C149 or E152, and mutations at K147 were least

disruptive (Shaw et al., 1997). The identity between plant clones and hTFIIB at positions

analogous to yeast amino acids 151 and 149 would therefore seem to be reflected in the

inability of either TFIIB homolog to substitute into the yeast system. Helix B1 does not

contact DNA or TBP in the ternary crystal structure (Nikolov et al., 1995), so the

interactions important for its function in yeast may occur at a subsequent step in PIC

formation. Shaw et al. observed differential effects at various promoters for the yeast to

human point mutations suggesting decreased ability to interact with some activators.

Fluorescence anisotropy was more useful for quantitation of AtTFIIB effects on

the AtTBP binding curve. Models for PIC assembly start with the TBP-TAF complex


4 2 13
: I I I I I I
Arabidopsis 122 VATIKDRANELYKRLEDQ 139
I 11:1 1 1::: :D1

B. K E137.N130
Y133 K134
A129 BI helix D138
L136 E131

1125 T124
L132 R135

identical to human identical to yeast

Figure 3-14. Comparison of the BI a-helices from TFIIB homologs.
A. Alignment of Bi helix sequences shows plant clones are slightly more conserved
with the human version than the yeast homolog. Residue positions are numbered at the
ends of each sequence, and four residues critical for yeast plasmid substitutions are
numbered in order of influence above the yeast sequence. B. A helical wheel projection
of the B helix from Arabidopsis indicates conserved positions among the four critical
residues do not map to the same face.

binding to the TATA region followed by interaction with TFIIB alone or as a part of a

holoenzyme (Cujec et al., 1997; Kim et al., 1994; Koleske and Young, 1994). This order

of assembly would predict TFIIB has little effect on the TBP ka since TFIB would not yet

be present to influence the rate of association. This is supported by the observation that

the decrease in the Kd of AtTBP for TATA mediated by AtTFIIB1 (Fig. 3-7) can be

almost entirely accounted for by a 1.5-fold decrease in the AtTBP off-rate (Fig. 3-10).

Curve fitting analysis was used to calculate the inflection points for binding assays in Fig.

3-7, and the results indicated the Kd of AtTBP alone for TATA was 34.1 nM. This value

was reduced to 17.5 nM when AtTFIIB1 was present, a 1.9-fold increase in affinity. It is

unclear why the observed AtTBP Kd for TATA was higher than previous measurements

of 2-4 nM for yTBP and 5 nM for potato TBP (Hahn et al., 1989; Holdsworth et al.,

1992; Hoopes et al., 1992). The AdMLP probe used may not contain TATA or flanking

sequences optimal for interaction with AtTBP, but TBP generally shows a fairly wide

range of binding sequence specificity (Hahn et al., 1989; Mukumoto et al., 1993).

An initial estimate of the Kd for AtTFIIBI binding to AtTBP-DNA is 7.1 nM.

Attempts to confirm this value by independently measuring the ka and kd will require

further optimization of the anisotropy assay or use of alternate detectors. The on-rate at

various AtTFIIB 1 concentrations was too fast to accurately observe at five-second

intervals, and reducing the time of detection gave excessive increases in the margin of

error. Measurement of the off-rate in this system requires high concentrations of

competitor, which for AtTFIIBI would be unlabeled and pre-formed DNA-AtTBP.

While addition of large amounts of DNA oligomer did not affect fluorescence

polarimetry, the presence of high protein concentrations can overwhelm the binding

signal with background fluorescence. Proper calculation of these kinetic rates will thus

require more sensitive emission detectors which would allow the use of lower reactant

concentrations, or switching to an immobilized ligand system such as those utilized by

biosensor instruments (BIAcore or Affinity Sensors) to bypass the need for unlabeled

competitor. The signal-to-noise problem may be one source of inaccuracy in the binding

curves obtained for AtTFIIB1; in some cases the total fluorescence at high protein

concentrations began to vary from the normal value (Figs. 3-8 and 3-12). Increased

background fluorescence often prevented binding measurements at high reactant

concentrations, and this translated to fewer observations near the upper asymptote of the

binding plots. While curve fitting usually detected an upper plateau in the binding signal,

a few more data points would help confirm the shape of these curves and their inflection

points. It should also be noted that the AtTFIIBl KI is probably affected in vivo by the

presence of TAFs that alter the DNA topology and TBP conformation (Burley and

Roeder, 1996), and TAFs can provide additional interaction surfaces for TFIIB binding

(Hori et al., 1995).


Literature Review

Plant transcription regulatory factors contain many of the same classes of

activation domains (AD) characterized in other eukaryotes. While a shared predominance

of certain amino acids serves as the basis for such classification, it is becoming clear that

not all members within these groups may activate transcription by the same mechanisms.

Studies to make such distinctions, and identify mechanisms of activation in general, have

progressed in human, yeast, and fruit fly systems, but have just begun in plants. This

investigation will attempt to confirm predicted interactions between a plant pol II PIC

component and two plant transcription factors, one of which contains an acidic AD and

the other a proline-rich region.

Acidic activators are a large group of transcription factors containing a high

proportion of negatively charged residues in their AD (Hope and Struhl, 1986; Ma and

Ptashne, 1987). The AD of one of the archetypes, VP16, is composed of two adjacent

subdomains with 25% acidic and only 4% basic residues. While overall charge is

necessary but not sufficient for activation (Cress and Triezenberg, 1991), the bulky

hydrophobic residues leucine, isoleucine and methionine that are interspersed within the

acidic blocks are functionally critical (Regier et al., 1993). Experiments using wild-type

VP16 or fusions of its AD with DNA-binding domains (DBD) have demonstrated

regulation of transcription via TBP and TFIB binding and PIC assembly (Ingles et al.,

1991; Roberts et al., 1993; Stringer et al, 1990), coactivator binding (Silverman et al.,

1994), enhanced elongation (Ghosh et al., 1996), and anti-repressive histone or chromatin

binding (Bunker and Kingston, 1996; Lyons and Chambon, 1995). VP16 mutants

affecting in vitro binding and in vivo activation have implicated recruitment of TFIIB

(Roberts et al., 1995) as one mechanism of regulation, and also have been used to argue

against TFIIB as a target (Gupta et al., 1996). As a class, acidic activators show affinity

in vitro for multiple members of the PIC and binding often occurs with both TBP and

TFIIB, and sometimes to TFIIB but not to TBP (Chiang et al., 1996; Haviv et al., 1996;

Tong et a., 1995).

Several plant transcription factors have acidic ADs. Factors with homeodomain

(Korfhage et al., 1994), Ap2 (Stockinger et al., 1997), and Myb (Urao et al., 1996)

homologies respond to various stress conditions and contain acidic activation regions.

Proteins containing basic region/leucine zipper (bZIP) motifs, which direct DNA binding

and factor dimerization, are often members of the acidic AD class (Unger et al., 1993;

Weisshaar et al., 1991). Other examples include an embryo specific activator from bean,

PvALF (Bobb et al., 1995) and regulators of the maize anthocyanin pathway such as Lc

(Ludwig et al., 1989) and Cl. Alignment of the Cl AD with VP16 shows similarity in

amino acid sequences. The acidic C-terminus of C1 is localized to promoter elements

through an N-terminal Myb DNA-binding domain, and was randomly mutagenized to

determine what residues contribute to the activation function (Sainz et al, 1997). A

shared leucine is critical for both proteins, but changes at other hydrophobic positions

important for VP16 had little effect on Cl in maize or yeast assays. An aspartate

mutation in C1 also decreased activity, but charge substitutions generally did not affect

transcription. Whether these similarities and differences among acidic activators will be

reflected in the details of their activation mechanisms is to be determined and will be

aided by correlation of reduced activation potential with mutations that impair PIC


The Cl gene promoter is regulated by another acidic activator, VIVIPAROUS 1

(VPI), which also contains a repressor domain and affects transcription of seed

developmental genes (Hoecker et al., 1995; McCarty et al., 1991). A cryptic C-terminal

DNA-binding domain (Suzuki et al., 1997) is adjacent to the negative regulatory region.

The N-terminal AD contains three acidic blocks punctuated by bulky hydrophobic

residues and a region rich in serine and threonine, another marker for some types of ADs

(Triezenberg, 1995). This N-terminal pattern is also found in the dicot homolog of VPI,

ABI3 from Arabidopsis (Parcy et a., 1994). Experiments using an internal deletion

construct and a gain-of-function hybrid fusion confirmed that two of the VPI acidic

domains, linked by the serine/threonine region, comprise a transcriptional AD in plant

cells (McCarty et al., 1991). Activity in yeast was not tested.

A second class of ADs is defined by enrichment for proline and was first observed

in mammalian CTF/NF-1 and AP-2 (Mermod et al., 1989). CTF and other proline

activators have subsequently been shown to interact with TFIIB (Kim and Roeder, 1994;

Malik and Karathanasis, 1996), and the CTF proline domain is organized in a manner

similar to the CTD of the pol II largest subunit (Xiao et al., 1994). Like the CTD, this

protein was shown to bind to TBP in addition to TFIIB. The observed TBP interaction

led to the proposal of an activation mechanism in which the proline domain displaces the

TBP-CTD interaction and thus disrupts a pol II link to the PIC, allowing the enzyme to

progress to the elongation mode (Xiao et al., 1994). Multiple PIC interactions are

common for transcriptional activators, and many show functionally synergistic

combinations of ADs such as TFE3 which contains both an acidic domain and a proline

region (Artandi et al., 1995).

A subgroup of the plant bZIP factors has putative ADs rich in proline. These

include G-box binding factors (GBF) that recognize various forms of a consensus DNA

element commonly found in activated and inducible plant promoters. Arabidopsis GBF1

was shown to direct transcription activation through its proline domain (Schindler et al.,

1992), and the high homology among GBFs suggests this function of the proline region is

conserved. Soybean proteins SGBFI and SGBF2 were isolated as binding factors for G-

box related elements in a promoter regulated by auxin (Hong et al., 1995). They exhibit

typical GBF architecture including the N-terminal proline domain (19% P), a central

nuclear localization sequence, and a C-terminal bZIP region. The soybean factors can

heterodimerize, as can many leucine zipper proteins, and EMSA reactions containing

SGBF1, SGBF2 or the two factors together show distinct DNA binding patterns in each

case (Hong et al., 1995). This plasticity in DNA binding specificity may have been

exploited during evolution to generate promoter variants that selectively utilize GBF

monomers, homo- or heterodimers. Dimer binding to promoter elements is another

mechanism that could result in multiple ADs being present when needed for enhanced

activation potential.

Screening for the interaction targets of ADs usually begins with affinity

chromatography techniques to detect binding between the transcription activator and

components of the PIC. The interaction assay passes one protein reactant over a

Sepharose resin column carrying immobilized fusion proteins of the second reactant, or

combines the two in a low volume batch from which the resin beads can be recovered by

slow centrifugation. Input reactant retained on the resin by interaction with the fusion

protein is then eluted and detected, and the binding, wash and elution conditions can be

manipulated to assess interaction affinities. Such in vitro assays allow rapid screening for

binding and for effects of mutations that can be used to map the interaction domains. The

results obtained help guide the design of in vivo experiments to confirm interactions, an

important second step since in vitro binding events sometimes do not correlate with

actual function (Tansey and Herr, 1995).

Two techniques have been developed to help improve such correlations. The

yeast two-hybrid assay (Chien et al, 1991; Fields and Song, 1989) fuses one potential

reactant protein to a DBD which has specific affinity for a binding element in the

promoter of a reporter gene, and the second reactant protein with a strong yeast AD.

Interaction of the two proteins localizes the AD to the promoter and results in detectable

transcription. Advantages of this system include in vivo conditions for interaction and

eukaryotic protein expression and modification; of course, the latter may result in

improper modification and the proteins of interest must be functionally stable as fusions

with the yeast domains. A second improvement has been the development of very

sensitive protein interaction detectors based on changes in refracted or reflected light.

Surface plasmon resonance, for example, can be used to detect the interaction of a free

ligand with a surface layer containing a second protein, and this event is measured in real

time with continuous buffer flow (Jonsson et al., 1991). The advantage over affinity

chromatography is an ability to quickly determine optimal binding conditions, view the

results of variations in those conditions, and accurately discriminate between different

ligands or mutants by measuring slight changes in binding kinetics. The technique has

been used recently to measure affinities of PIC proteins for two acidic ADs, both of

which interacted with TBP and TFIIB in previous affinity chromatography assays.

Surface plasmon resonance indicated the human NF-KB subunit p65, which contains an

acidic AD similar to VP16, bound TBP with two orders of magnitude greater affinity than

the p65 interaction with TFIIB (Paal et al., 1997). An acidic AD from yeast GAL4,

however, showed only a 3-fold greater affinity for yeast TBP compared to TFIIB, and a

series of AD mutants equally affected binding to both basal factors and displayed a linear

range of binding constants that directly correlated to the level of activated transcription

observed in yeast (Wu et al., 1996).

The experiments described below represent an initial screen for interaction of

AtTFIIBl with the plant acidic activator VP1 and the proline-bZIP factor SGBF2.

Limited success using the in vivo yeast two-hybrid assay was followed by affinity

chromatography under a variety of conditions to probe interaction chemistry. Deletion

constructs of AtTFIIB 1 were also used to map regions of interaction.

Materials and Methods

Yeast Transcription and Two-Hybrid Assays

Yeast plasmids pGBT9 and pGAD424 (Clontech) were used to express fusion

proteins from the constitutive promoter ofADHJ. pGBT9 fuses the GAL4 DBD (amino

acids 1-147) to the N-terminus of expressed proteins, and pGAD424 substitutes the

GAL4 AD (768-881) at the same location. PCR amplification was performed to attach

EcoRI sites to a DNA fragment encoding the activation domain from VP16, amino acids

2076-2309, and this fragment was cloned into the EcoRI sites of both yeast plasmids.

PCR amplification from pEThVP23 (Dr. Don McCarty, University of Florida) of the

entire coding region for VP 1 attached an EcoRI site before the first methionine and

substituted an XhoI site for the stop codon using the primers 5VP1: 5' CCTGAATTCA-


ATCTG 3'. The resulting fragment was ligated to the EcoRI and Sall sites of the yeast

vectors. Deletions of VP1 were constructed in the yeast plasmids using an in-frame

BamHI site at residue 191. An EcoRI-BamHI digestion of the PCR product and ligation

into those sites in the yeast vectors produced the construct VP a encoding the N-terminal

one-third of VP1. A BamHI-PstI fragment from the PCR product encodes the central

portion from residues 192-406 and was ligated into those sites for yeast expression of

VPlb. The C-terminal region (VPlc, residues 413-691) is encoded on a Pvull fragment

of the amplification product and was blunt-end ligated into the Smal site ofpUC19.

Digestion of that construct with EcoRI and XhoI released the VPIc fragment for ligation

into the yeast plasmids at the EcoRI and Sall sites. The plasmid pAtTFIIB1 (Chapter 3)

was digested with Kpn21 and Sall, and the resulting fragment was cloned into the PspAl

and Sall sites of pGBT9.

Activation of transcription in yeast strain PCY2 was measured by assaying P-

galactosidase (P-gal) activity that results from expression of a chromosomal lacZ gene

under control of a promoter containing GAL4 binding sites. Yeast cells transformed with

one or both of the expression vectors were plated on the appropriate auxotrophic selection

media, and five independently transformed colonies were grown for liquid culture 1-gal

assays as described (Miller, 1972). Cell extracts were incubated with o-nitrophenyl p-D-

galactopyranoside for 20 min at 300 C before quenching.

In vitro Interaction Assays

pGEX vectors designed for bacterial expression of proteins fused at the N-

terminus to glutathione S-transferase (GST) were used to create immobilized protein

affinity resins. The EcoRI fragment from pEThVP23 was ligated into pGEX-2TK

(Pharmacia) to produce pGEXVPl, and the BamH1 fragment from pGEXVPl was re-

cloned into pGEX-2TK to create pGEXVPla. Primers incorporating a BamH1 site at the

5' end and an EcoRI site at the 3' end of a cDNA for SGBF-2 (Dr. Tom Guilfoyle,

University of Missouri) were used to PCR amplify the entire coding region and clone it

into pGEX-KG. The primers used were 5GmGBF2: 5' GCTGGATCCATGGGAAA-


GCATTAG 3'. A PCR reaction was also used to attach BamH1 and EcoRI sites to the

VP16 AD for cloning into pGEX-2TK. Stationary phase E. coli BL21 cultures

transformed with pGEXVPla, pGEXSGBF2, or pGEXVPI6 were diluted 1:200 and

incubated 2.5 hr at 370 C before induction with 1 mM IPTG. Incubation continued for 3

hr and then cells were pelleted, resuspended in lX PBS (140 mM NaCI, 2.7 mM KC1, 10

mM Na2HPO4, 1.8 mM KH2 P04) with protease inhibitors (Chapter 3) and 1% Triton X-

100, and sonicated. Cleared lysates were batch purified by binding to glutathione-

Sepharose resin (GS4b, Pharmacia) and washed with IX PBS.

Deletion mutants of AtTFIIB were generated by PCR amplification of the cDNA

using primers that incorporated appropriate restriction sites. Amplified regions were

cloned into pET24b for HisBind purification and Western detection of the T7 epitope

fused to the N-terminus. Purification of intact peptides was achieved for constructs

carrying several AtTFIIB1 domains, including residues 1-215 (Zn ribbon to repeat 1,

AR2), 44-312 (conserved region to repeat 2, AZn), 100-312 (repeats 1 and 2, core), and

100-215 (repeat 1, RI). Expression and purification of these proteins and full length

AtTFIIBI were as described in Chapter 3.

GS4b-immobilized VP16, VPla, SGBF2, or GST alone (10-15 lpg total protein

per assay) was mixed with an equal packed volume of unbound beads to create an affinity

resin mixture suitable for retention of GST fusion proteins through multiple

manipulations. Resin batches were collected by low speed centrifugation into a packed

volume generally between 20-30 l1, and then mixed with 1-2 gig AtTFIIB1 in BS buffer

without glycerol (Chapter 3) in a final volume of 60 gl. Binding reactions were gently

rocked at room temperature for 30 min with occasional additional mixing. Supernatant

fractions collected after binding included flow-through reactantss not bound to resin),

three washes (each with 3X bed volume of interaction buffer), three elutions (3X bed

volume of buffer with various salt and/or detergent concentrations), and bound reactantss

released into buffer by boiling the resin). When multiple elutions were tested in one

binding assay, each sequential step was comprised of three elution fractions per treatment.

Detergents used for elution above and below their critical micelle concentrations (CMC)

included sodium dodecyl sulfate (SDS, CMC 7-10 mM), cetyltrimethylammonium

bromide (CTAB, CMC 1 mM), and TRITON X-100 (TX100, CMC 0.29 mM).

Western Blot Detection of AtTFIIB

One-third of the total volume from each protein fraction was electrophoresed on

12% SDS-polyacrylamide (30:1) gels and transferred to PVDF membranes (Immobilon-

P, Millipore) using a Trans-Blot SD electrophoretic cell (Bio-Rad) according to the

manufacturer's instructions. Blots were washed and incubated in TBST (0.1 M Tris-HCl

pH 7.5, 0.9% NaCI, 0.1% Tween-20) with a 1:10,000 dilution of anti-T7 mouse antibody

(Novagen) for 1 hr at room temperature. TBST washes to remove the primary antibody

were followed by incubation with a 1:10,000 dilution of anti-mouse IgG antibody

conjugated with horseradish peroxidase (Promega) for 45 min at room temperature.

Chemiluminescent detection was performed with the ECL protocol (Amersham) and blots

were exposed to Kodak XR-Blue X-ray film for 15, 30 and 60 sec. Detected bands were

quantified by densitometry with the IS 1000 system.


In preparation for two-hybrid interaction screening, fusion constructs of GAL4

DBD-VP1 were transformed into yeast to determine whether the maize factor was

properly expressed and functionally active. As shown in Table 4-1, full length VP1

activated transcription of the lacZ reporter. The transcriptional activity maps to the N-

terminal 191 amino acids (VPla), and the level of activity is comparable to that from the

positive control, VP16. Cultures transformed with the remaining VP1 regions, VPlb and

VPlc, produced no detectable p-gal. The ADs of VPI and VP16 were then transferred to

the GAL4 AD fusion vector and co-transformed with AtTFIIBl fused to the GAL4 DBD.

The negative control containing AtTFIIBl alone gave no transcription (Table 4-1), but

neither did the putative positive control co-expressing VP16, which is known to interact

with TFIIB in in vitro systems (Roberts et al, 1993). VPla+AtTFIIB1 also showed no

detectable p-gal activity.

Table 4-1. Activation of yeast transcription by fusion proteins containing the GAL4
DBD or the GAL4 AD. Average reporter activity is indicated with standard deviation
GAL4 fusions: p-gal (Miller)
DBD- AD- +/-SD
VP16 1.33 0.33
VP1 0.79 0.09
VPla 1.16 0.14
VPlb 0
VPlc 0

AtTFIIBI was next tested in an in vitro interaction assay. GST fusions with the

ADs of VP16 and VPI, and the entire SGBF2 protein, were expressed inE. coli and

immobilized on GS4b resin. AtTFIIBI was also E. coli expressed, purified via the C-

terminal His tag, and detected in binding reactions by Western blotting using antibodies

against the N-terminal T7 epitope. This AtTFIIBI configuration with short peptide tags

on each terminus was previously shown to be functionally active (Chapter 3). Incubation

of AtTFIIB1 with immobilized VP16, VPla, or SGBF2 generally resulted in retention of

about 40% of the input protein, and the binding was specific for the ADs since no

AtTFIIBI was retained on resin carrying only GST (Fig. 4-1). The interaction with all

1 2 3 4 5 6 7 8 9

10 11 12 13 14 15 16 17 18

Figure 4-1. Western blot detection of AtTFIIB1 from affinity chromatography fractions.
AtTFIIBI was incubated at room temp. with immobilized fusion proteins in buffer
containing 50 mM KC1. After collecting the flow-through fraction and three washes in
binding buffer, three elution fractions were collected for each sequential KCI treatment
described. Protein that remained associated with the resin was released by boiling.
Trace amounts of AtTFIIB 1 were detected in the first wash fractions; only the flow-
through and first elution fractions are shown. One-third of the total fraction volume was
loaded on the gel for blotting. Lanes 1 and 10, 10% input AtTFIIB 1; lanes 2, 6, 11 and
15, flow-through; lanes 3, 7, 12 and 16, elution with 0.5 M KCI; lanes 4, 8, 13 and 17,
elution with 1.0 M KCI; lanes 5, 9, 14 and 18, boiled. The immobilized protein present
in each reaction is indicated over the lane numbers.


1 2 3 4 5

em- .ini


Figure 4-2. Ethylene glycol elution of AtTFIIBI from acidic activation domains.
A. AtTFIIBI (10% input, lane 1) is detected in flow-through (lane 2) and bound (lane 3)
fractions after interaction with immobilized VP16. A replicate reaction was sequentially
eluted with 10% (lane 4), 30% (lane 5), and 50% (lane 6) ethylene glycol in 0.1 M KC1,
but most AtTFIIB1 remains bound (lane 7). A third reaction was eluted with 30%
ethylene glycol in 0.5 M (lane 8) or 1.0 M (lane 9) KCI, lane 10 is the boiled fraction.
B. AtTFIIBI (10% input, lane 1) fractions after interaction with VPla include flow-
through (lane 2), sequential elution with 10% (lane 3) or 50% (lane 4) ethylene glycol in
0.1 M KC1, and boiled (lane 5). C. Results of a duplicate reaction as in B., except
elution buffer was supplemented with 1.0 M KCI.

three ADs was stable in high salt concentrations since no AtTFIIB was detected in

elution fractions containing 0.5 M or 1.0 M KC1.

TFIB binding to the initiator-binding factor YY1 has also been recently shown to

be very salt-stable (Usheva and Shenk, 1996), but TFIIB could be released using an

elution buffer with 50% ethylene glycol, which is known to disrupt hydrophobic

interactions (Thompson et al., 1990). Ten percent ethylene glycol released a slight

fraction of the AtTFIIB1 bound to VP16, and additional sequential elutions with 30% or

50% ethylene glycol had no effect (Fig. 4-2A). High KCl concentrations combined with

ethylene glycol were ineffective, and the same results were obtained with AtTFIIB

binding to VPla (Fig. 4-2B and C). To further investigate the susceptibility of AtTFIIB1

interactions to disruption, three detergents were included in the elution buffers. Each

detergent was tested at concentrations below and above the point at which monomeric

molecules aggregate to form micelles, or the critical micelle concentration (CMC)

(Neugebauer, 1994). As shown in Fig. 4-3A and Fig. 4-4, the anionic detergent SDS

completely eluted AtTFIIBl from immobilized VP16 and VPla while nearly none was

released by the cationic detergent CTAB. A non-ionic reagent, TX100, eluted a portion

of the AtTFIIB1 with each increase in detergent concentration but did not completely

disrupt binding to VP16 or VPla. The contrasting abilities of CTAB and SDS to affect

binding were also seen upon comparison at equimolar concentrations (Fig. 4-3B).

Binding of AtTFIIB1 to SGBF2 showed similar results as with VP16 and VPla.

SDS again completely eluted the bound AtTFIIBl, but CTAB was slightly more effective

for disrupting this interaction compared to the acidic ADs (Fig. 4-5). Detectable

AtTFIIB1 was found in elution fractions at both CTAB concentrations, but the majority

Full Text
xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd