LIPOPROTEIN STRUCTURE: APOPROTEIN INTERACTIONS
IN HUMAN PLASMA HIGH DENSITY LIPOPROTEINS
THOMAS ELLIS GROW
A DISSERTATION PRESENTED TO THE GRADUATE COUNCIL OF
THEII UNIVERSITY OF FrLORIDA
IN PARTIAL FULFILLMENT OF THEI REQUIREMENTS FOR THE
DEGREE OF DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
I dedicate this dissertation to my best friend William
A. Patterson and the Krenu Construction Company.
The author would like to express his deep appreciation to
Professor Melvin Fried, his research director, for giving him guidance
and freedom of thought and action during the completion of this work.
The environment of Professor Fried's laboratory was exactly what the
author required for productive work.
The author also wishes to thank his wife, Susan, for her patience,
understanding, and sacrifices made during the last four years.
Special thanks are given to the author's advisory committee and
members of the faculty of the Department of Biochemistry for their
suggestions and criticisms during the execution of this research.
The author also thanks his parents for their understanding during
his years of education.
TABL.E OF CONTENTS
. . i
LIST OF TABLES . . . . . . . . . . . . .
LIST OF FIGURES . . . . . . . . . . . .
KEY TO SYMBOLS AND ABBREVIATIONS NOT DEFINED IN TEXT ......
ABSTRACT . . . . . . . . . . . . . .
INTRODUCTION . . . . . . . . . . . . .
Plasma Lipoproteins ... ... .. .... .........
Apoprotein Structure and Function ..............
Lipoprotein Structure ....................
Alternative Classification of Lipoproteins ..........
Exchange Reactions of Lipoproteins .............
Experimental Rational ....................
Research Objectives ... ..................
MATERIALS AND METHODS .....................
Materials . . . . . . . . . . . . .
Methods . . . . . . . . . . . . . .
RESULTS AND DISCUSSION .....................
Isolation and Purity of Lipoprotein Preparations ......
Crosslinking of Total HDL (HDLT) with DFDNB .........
Effect of Crosslinking on HDL Structure ...........
Crosslinking of HDL2 and HDL3 with DFDNB ..........
Crosslinking of HDLT with DMS and DFDNDPS ..........
Temperature Dependence of Crosslinking. ...........
Interpretation of Crosslinking Results. ... .. .. .. .59
Apoprotein Exchange Between HDL2 and HDL3 . .. .. ... .. .61
Effect of Crosslinking and Monofunctional Labeling
on Apoprotein Exchange. .. .. .. .. .. .. .. ... .66
Effect of Temperature and Dilution on Peptide Exchange .. .69
Rate of Peptide Exchange ... .. ... .. .. .. .. .69
Effect of Bromide lan on Protein Exchange .. .. .. .. .. .80
SDS-Polyacrylamide Gel Analysis of Exchanged Proteins .. .. .80
Interpretations of Protein Exchange Experiments .. . ... .85
Lipid Exchange Between HDL2 and HDL3 in Solution .. .. .. .88
Binding of Labeled HDL to Sepharose .. .. . .. .89
Peptide Exchange Using Sepharose-Bound HDL .. .. .. .. .91
Ultracentrifugal Analysis of Peptide Exchange Using
Sepharose-Bound HDL . .. ... . .. . . . . 91
SDS-PAGE Analysis of Exchanged Peptides Using
Sepharose-Bound RDL . .. .. .. .. ... . 96
Incorporation of Labeled Lipid into RDL2 and HDL3 ... .. 103
Lipid Exchange Between HDL3 and Sepharose-HDLj .. .. .. 106
Effect of Temperature and Dilution on Lipid Exchange .. .. 112
Effect of Saturating Cold Lipid on Labeled Lipid Exchange .. 112
Effect of Crosslinking on Lipid Exchange .. .. .. . 118
Effect of Added Organic Solvent on Exchange of Lipids
and Proteins . .. .. .. .. . .. . .. . 118
Interpretations of Sepharose-HDL Peptide and Lipid
Exchange Experiments ... .. .. .. .. . . . 125
CONCLUSIONS AND SPECULATIONS . ... .. .. ... . .. 128
BIBLIOGRAPHY .. .. .. .. .. .. . .. . . . 32
BIOGRAPHICAL SKETCH .... .. .. ... .. .. . .. 137
LIST OF TABLES
Bifunctional Reagents ...........
Amino Acid Analysis of 46,000 Dalton
Crosslinked Product from RDLT Cross-
linked with DFDNB... .........
Effect of Amount of CNBr Activation
on the Amount of Binding of HDL3 by
Sepharose 48 ........
Peptide Exchange Using Sepharose-Bound HDL
Removal of Labeled Lipid from Sepharose-
Bound HDL3 by SDS Wash ..........
Effect of Temperature and Dilution on
Lipid Exchange Between S-HDLg and RDL3 ..
Effect of Saturating Cold Lipid on Lipid
Exchange . . . . . . . . .
Effect of Crosslinking on Lipid Exchange .
. . .107
. . ..117
. .. .119
. . .68
. . .71
. . .73
. . .75
LIST OF FIGURES
1 SDS-PAGE of solubilized apoproteins
of IlDLy . . . . . .
2 Separation of HDL2 and HDL3 A d=1.125gm/cm3. ..
3 SDS-PAGE of RDLT crosslinked with DFDNB ......
4 Experimental approach to amino acid analysis
of crosslinked products .............
5 Comparison of the SDS-PAGE methods of
Fairbanks and Lammli for the separation
of highly crosslinked RDLT ............
,S-PAGE of HDL2 crosslinked with DFDNB .......
,S-PAGE of HDL3 crosslinked with DFDNB .......
IS-PAGE of HDLT crosslinked with DMS ........
IS-PAGE of HDLT crosslinked with DMS (5% BME). ...
IS-PAGE of HDLT crosslinked with DFDNDPS ......
S-PAGE of HDLT crosslinked with DFDNDPS
(5% BME). . . . . . . . . . . .
change of apoproteins between [12I]-HDL3
and HDL2 . . . . . . . . . .
change of apoproteins between [ I]-HDL2
. . .
d na HD
14 Effect of crosslinking on apoprotein exchange .
15 Effect of FDNB labeling on apoprotein exchange
16 Effect of crosslinking the unlabeled subclass
on apoprotein exchange ...........
17 Effect of temperature on apoprotein exchange .
18 Effect of temperature and dilution on
apoprotein exchange .. . .. .. . .77
19 Trime course of upoprotein exchanglpe .. .. .. .79
20 Effect of KBr on apoprotein exchange .. .. .. .82
21 Polyacrylamide gel analysis of exchanged apoprotein .84
22 Polyacrylamide gel analysis of exchanged apo-
proteins: Effect of crosslinking with DFDNB. .. .87
23 Suunitexchnge etwen S-125I]HL n D 9
24 Submnit exchange between S-[12I]-HDL2 and HDL2 9
25 Polyacrylamide gel analysis of available
apoproteins from S-HDL .. .. .. .. .. .. 100
26 Polyacrylamide gel analysis of apoproteins
exchanged from S-[125I]-HD) L ..... 102
27 Uptake of labeled lipids by HDL3 .. .. .. .. 105
28 Effect of increasing amounts of cold HDL, on
lipid exchange .. .. .. .. .. .. 109
29 Effect of increasing amounts of cold HDL3 on
lipid exchange .......... 111
30 Rate of lipid exchange between S-HDL3 and HDL .. 114
31 Time course for lipid exchange between
S-HD~g and HDL3 .. .. .. ... .. .. .. 116
32 Effect of ethanol of the exchange of lipids
from S-HDL~ to HDL ........ 122
33 Effect of ethanol on the exchange of lipid
and protein from S-HDL~ to RDL3 .. ... .. .' 124
34 Model for HDL3 HDL2 interconversion .. . ... 131
KEY TO SYMBOLS AND ABBREVIATIONS NOT DEFINED IN TEXT
CD circular dichroism spectroscopy
dpm disintegrations per min.
IR infrared spectroscopy
LCAT Lecithin cholesterol acyl transferase
MW molecular weight
NMR nuclear magnetic resonance
ORD Optical rotory dispersion
Abstract of Dissertation Presented to the Graduate Council
of the University of Florida in Partial Fulfillment of the Requirements
for the Degree of Doctor of Philosophy
LIPOPROTEIN STRUCTURE: APOPROTEIN INTERACTIONS
IN HUMAN PLASMA HIGH DENSITY LIPOPROTEINS
Thomas Ellis Grow
Chairman: Melvin Fried
Major Department: Biochemistry
The modes of association of the major apoprotein -species of human
plasma high density lipoprotein (HDL) was investigated using bifunctional
crosslinking reagents. How these associations help to govern the struc-
ture of the RD)L subclass, HDL2 and HDL was studied by observing both
apoprotein and lipid exchange between HDL particles.
Crosslinking of HDL with a 5 to 10-fold molar excess of difluoro-
dinitrobenzene resulted in the production of high molecular weight
crosslinked products with the major product shown to be the combination
of the two major apoprotein species ApoA-1 and ApoA-2. The composition
of this product was confirmed by amino acid analysis.
Crosslinking with other bifunctional reagents indicatesspecificity
of initial reaction sites as well as some freedom of movement of the
apoproteins within the particle. However, the crosslinking patterns
obtained with the RDL subclasses were identical, suggesting specific
structural properties of HDL2 and HDL3
Apoproteins, labeled with 12i, were shown to exchange between and
among the RIDL subclasses. This bidirectional exchange process was
inhibited by crosslinking with bifunctional reagents and appeared
to depend upon collision complex formation. By the utilization of
Sepharose bound IIDL it could be shown that both free apoprotein mole-
cules, as well as subunits consisting of lipid-apoprotein combinations,
could exchange between HDL2 and HDL3. The exchange appears to involve
a conformational change in the apoproteins during the transfer process.
Radioactive cholesterol, cholesterol palmitate, and phosphatidy1
choline, bound to HDL after adsorption to celite particles, could be
shown to undergo rapid exchange between MDL particles. This exchange
also appears to depend on collision complex formation but was not
inhibited by chemical crosslinking of the apoproteins.
The rate of these exchange processes is significant in the life-
time of the lipoprotein particles in vivo: 2.5%/hour for apoprotein
exchange and >50%/hour for lipid exchange.
It was concluded that ApoA-1 and ApoA-2 must lie very close to one
another in the intact lipoprotein particle. The reproducibility of
the crosslinking experiments on HDL obtained from several donors suggests
that in the intact HDL particle the apoprotein components may be found
in a rather fixed spatial orientation to one another. It is also
suggested that there is a dynamic relationship between HDL2 and HDL .
Even though these two subclasses can exist as stable separate entities,
when they are present together in solution, significant interaction
between particles may occur. Exchange of lipids and apoproteins occurs
between KDL2-HDL2 and HDL -HDL as well as between HDL2 and HDL3 mole-
It is suggested that the interconversion of RDL2 and HDL3 may be
determined by the availability of certain lipids.
Plasma lipoproteins play an important role in mammalian energy
metabolism and also serve as the major transport particles for many
physiologically important molecular species. It has been suggested
that they are causally related to atherosclerosis since they are the
carriers for those substances deposited in the atheromatous plaque.
Circulatory system pathology found in hyperlipoproteinemics may result
from quite small alterations in lipoprotein structure; differences that
may be due to altered protein-protein or protein-lipid interaction.
Human serum contains four major types of lipoproteins: chylo-
microns (CHYL), very low density lipoproteins (VLDL), low density
lipoproteins (LDL), and high density lipoproteins (HDL). Although
these four classes appear to-differ in both lipid and protein compo-
sition as well as function, there are distinct interrelationships
between the particles. The presence of one class directly influences
the other classes (1).
Chylomicrons, formed in the intestinal mucosa and released into
the blood from the major lymphatic duct, function as the carrier of
many lipid types. Lypolytic enzymes remove lipids from these particles~
as they circulate. VLDL and LDL appear to be closely related as LDL
may be the result of VLDL delipidation (2). LDL functions as the
major cholesterol carrier in blood. HDL can be divided into two
subgroups; HDL2 and RDL3. HDL may be important in cholesterol
deposition disease (3).
The isolation of lipoproteins from serum or plasma is relatively
simple and may be accomplished by flotation in the ultracentrifuge.
As the density (d) of serum lipoproteins ranges from approximately
0.95 gm/cm3 to 1.21 gm/cm3, centrifugation of serumo, the density of
which has been raised to 1.21 gm/cm3 by the addition of salt, will
result in the lipoproteins floating to the top of the centrifuge tube
while the other proteins sink (4). Each lipoprotein class is charac-
terized by its own density range for isolation: CHYL and VLDL < 1.006,
LDL 1.006 1.063, and HDL 1.063 1.21. By stepwise increases in
density (4), or by density gradient centrifugation (5), these lipo-
protein classes may be prepared in relatively pure form. Although
other methods such as molecular sieve chromotography (6) and chemical
precipitation (7) have been described, the ultracentrifugal floatation
method remains the most widely used.
The protein and lipid compositions of these lipoprotein classes
have been studied and the following have been reported (1,8) (all
percentages on a dry weight basis).
HDL, whose molecular weight ranges from 180,000 to 350,000,
consists of approximately 50% protein, 22% phospholipids, 3% free
cholesterol, 14% cholesterol esters, and 8% triglycerides. Of the
phospholipid portion, 70% is phosphatidylcholine 14% is spingomyelin.,
5% phosphatidy1 serine and phosphatidyl ethanolamine, and the remain-
der other minor phospholipids. The protein portion of the RDL particle
consists of two major polypeptides, ApoA-1 (ApoGln-I) and ApoA-2
(ApoGln-II). The amino acid sequences of both polypeptides are known
(9,10,11) and ApoA-2 is known to be a dimer of two identical subunits
connected by a single disulfide bridge at position 6 (10). Another
group of smaller proteins, designated ApoC or the C-peptides, comprises
approximately 5% of the total HDL protein. At least three C-peptides,
C-I, C-II, and C-III, are recognized in HDL (2). The protein of HDL
contains 3-4% covalently bound carbohydrate including glucosamine,
fructose, galactose, mannose, and sialic acid (8). In addition to the
C-peptides, KDL also may contain the "thin-line" protein and "arginine
rich" protein as minor components (1).
LDL particles are larger than HDL with molecular weights ranging
from 2 to 3 million. They also contain a larger percentage of lipid
than HDL (75-78%); consisting of 36-42% cholesterol esters, 21-23%
phospholipids, 8-9% free cholesterol, and 6-8% triglycerides. Other
lipids may be present in smaller amounts. The proteins make up 20-22%
of the particle. The major polypeptide is denoted apoprotein B (ApoB),
a large protein of molecular weight 225,000. There may also be very
small amounts of the C-poptides described for liDL. Again, 3-5% carbo-
hydrate is found covalently associated with the protein.
VLDL particles cover a wide range of densities and molecular
weights (3 128 x 106) with at least two maxima at densities of 0.980
and 0.958 gm/cm3. They contain 90% lipid with an average composition
of 50-60% triglycerides, 10-12% free cholesterol, 4-6% cholesterol
esters, and 18-20% phospholipids. Protein represents only 8-12% of
which 40% is apoprotein B and the remaining 60% is divided among the
three C-peptides described for HDL. A small portion of ApoA may also
be present and it has been demonstrated that different fractions may
contain different amounts of apoproteins (2).
Chyl contain mostly lipid with only 1-2% protein. These very
large particles have molecular weights of 5 4300 x 108. The protein
consists of all types of previously described apoproteins found in the
other lipoprotein classes. It has been shown that human lymph ch-ylo-
microns contain approximately 66% ApoC, 22% ApoB, and 12% ApoA. The
lipids are predominately triglycerides (80%) with only about 8%
phospholipids and 5% cholesterol (1,2).
It is important to note the overlapping protein composition in
all of these classes. It is presumed that the specific protein compo-
sition of each is important in determining its total lipid content and
the specific type of particle that is formed. These proteins should
play a very important role in determining the function of the particle
as well (1).
Apoprotein Structure and Function
ApoC-I has been sequenced (12) and is a single polypeptide chain
of 57 amino acid residues (MW 8000). ApoC-I is known to function in
lipid binding as well as LCAT activation (13). Lipoprotein lipase
also appears to be activated by ApoC-I (14).
ApoC-II is the largest of the C-peptides withi 100 amino acid
residues and a molecular weight of about 12,500 (15). It also binds
lipids and is a potent activator of lipoprotein lipase (16).
ApoC-III, a single polypeptide of 79 amino acid residues has also
been sequenced (17). Attached to threonine-74,ApoC-III contains an
oligosaccharide moiety consisting of galactose, galactosamine, and
sialic acid (17). Although no direct physiological role is yet known,
ApoC-III has been used often for the study of lipid-protein inter-
The characterization of ApoB has led to many conflicting statements
concerning the number and size of the polypeptide subunits. Due to
technical problems associated with isolation of a soluble apoprotein,
a variety of physical methods have indicated molecular weights ranging
from 8,000 to 275,000 (22). However,.most investigators support the
idea of a large (275,000) molecular weight peptide. These problems
have also hindered studies on specific functions of ApoB and as yet
no definite physiological role other than lipid binding has been dis-
ApoA is a term used to describe the two most extensively studied
apolipoproteins, ApoA-1 and ApoA-2. ApoA-1, a single polypeptide of
molecular weight 28,330, contains 245 amino acid residues and has been
sequenced (8,23,24). Although the binding of phospholipids by ApoA-1
has been studied by several investigators, conflicting results on lipid
binding in the absence of ApoA-2 have been reported (25,26,27). These
differences in the extent of lipid binding by ApoA-1 may be due to
self-association of the polypeptides that may mask lipid binding sites
(28). Self-association has been demonstrated by both ultracentrifugal
and chemical techniques (1),
ApoA-2 is a polypeptide dimer of two identical chains with a
molecular weight of 17,500 (29). Each chain contains 77 amino acid
residues. The d-isuffide bond occursu between thec single cysteine
residues that occur at position 6 (29). The disulfide linked dimer
appears to be characteristic of only man and chimpanzee; the dimer form
apparently is not crucial for ApoA-2's lipid carrying role (1). ApoA-2
does bind lipid strongly and may increase the binding capacity of
ApaA-1 as well (2).
Although lipid binding is the only role as yet assigned to ApoA-2,
ApoA-1 appears to have several functions other than lipid binding in
the HDL particle. ApoA-1.serves as an activator of LCAT (30), may be
an acceptor of cholesterol and cholesterol ester during VTLDL catabo-
lism (2), and appears to have some function in regulating the lipid
content of membranes (31,32).
The study of lipoprotein structure has been approached utilizing
three basic types of techniques: 1) spectroscopic methods (IR, CD,
ORD, X-ray, neutron scattering, NMR, and fluorescence); 2) physical
techniques (Ultracentrifugation and electron microscopy); and 3) chem-
ical methods (chemical modification, amino acid analysis, bifunctional
reagents, reassembly). The actual composition of lipoprotein particles
has been fairly well described, as indicated above, and only in the
case of L~DL is there significant disagreement on Apoprotein size and
composition (1,8). However, when three-dimensional structure is
considered, little is known at present regarding specific protein-
protein or protein-lipid spatial relationships or interactions in
intact lipoprotein particles.
VLDL and Chyl have proved difficult to study due to the signifi-
cant turbidity of their solutions. This fact has limited the use of
spectroscopic methods and as a result' only the techniques of ultra-
centrifugation, electron microscopy, and ge1 filtration have been
used. This, coupled with the fact that these classes consist of a
spectrum of a~ wide variety of sizes with differing compositions, has
prevented all but the most basic of structural studies. Ultracentri-
fugation techniques indicate the following: Chyl MW 5-4300 x 108
daltons, diameter 1200-11000 A, hydrated density 0.93gm/cm VLDL-
MW 3-128 x 106 daltons, diameter 300-700 A, hydrated density 0.93
1.0gm/cm3 (1). The only structural determinations made thus far for
these particles are based on thec surface activity of the protein
components which seems to indicate thaot some portion of the proteins
are located on the surface of the particle (2), The information now
available has led to the suggestion of a lipid-core model for the
triglyceride richChyland VLDL particles. This model assumes that the
surface of the particle is occupied by phospholipids, cholesterol, and
proteins, while the central core consists of the less polar and more
hydrophobic triglycerides and cholesterol esters (33). The quantities
of protein, phiospholipid and cholesterol in these particles appear to
be sufficient to cover the surface of the particles as predicted (34).
LDL structure has been studied by many of the different techniques
mentioned earlier (2,8,35,36,37,38). Data from electron microscopy
and X-ray diffraction indicate a spherical particle with neither well
defined nor significant subunit structure (2). Information from
analytical ultracentrifugation, as well as X-ray and electron micro-
scopy, give the following information: MIW 2.3-2.7 x 106 daltons,
hydrated density 1.028-1.034gm/cm3, diameter 216-220 A. In distinction
to Chyl and VLDL, LDL is rather uniform in size and composition. The
protein component appears to have about 25% a-helix, 37% B-structure,
and 37% random coil as determined by CD (38). Although this structure
appears to be temperature dependant, a recent report indicates the CD
spectrum may be greatly complicated by the lipids present in the
particle (39). Both NMR and fluorescence studies indicate that the
protein, although near the surface of the LDL particle, is highly
associated with the lipid components of the molecule (36,37). It has
also been suggested that the core of LDL is mostly composed of hydro-
carbon chains while the outer layer of the particle is sparsely
occupied by proteins that emerge from the lipid core (35); a model
similar to the lipid-core model for VLDL. Other investigators,
utilizing electron microscopy and X-ray scattering, have suggested a
significant degree of subunit structure that includes many protein
subunits (20-60 with MW 8,000 or 27,000) arranged with icosahiedral
symmetry. The protein subunits give a symmetrical appearance with
the surface area occupied by both protein and phospholipid (40,41).
These subunit models do not take into consideration recent findings
that indicate the major protein component consists of only two poly-
peptides of molecular weight 250,000 or greater (42).
HDL has been studied most extensively, probably because it is
the easiest of the particles to work with. As noted above, its pro-
tein composition is well known and the major polypeptides have been
sequenced. RDL can be divided into two subclasses, HDL2 and HDL ,
which appear to differ mainly in their molecular weights (HDL2
184,000) (1).~ X-ray studies have indicated that RDL consists of two
regions of differing electron densities: an outer shell and an inner
core. The inner core has a radius of approximately 43 A in HDL2
(37 A in HDL3) while the outer shell radius is 14 A (11 A in HDL3).
The electron density of the outer shell indicates that is consists of
polar lipid headgroups and proteins (43). The size of a fully extended
phospholipid headgroup is about 11 A, a value that fits in well with
the X-ray data. Fluorescence techniques yield information on porcine
HDL3 consistent with a peripheral location of the protein component,
with the suggestion that as much as 80% of the protein is located at
or near the surface (37). This hypothesis is further supported by the
'finding of Scanu that 90% of the lysine residues of the apoprotein
components of human HDL is accessible to succinylating agents (2).
This would indicate that most of the protein is near the surface as
there are many lysines rather evenly distributed throughout the pro-
CD of the protein portions of HDL indicate 70% a-helix, 11%
B-structure, and 19% random coil (38). The removal or alteration of
11pid components greatly affects the structure of HDL apoproteins.
Delipidation lowers the helicity to 52%. Delipidation followed by
relipidation with egg phosphatidyl choline changes a-helicity to 64%
while thle addition of cholesterol further increases the helicity to
70% (38). These observations indicate strong lipid-protein interactions.
Two basic models of HDnL that are consistent with spectroscopic and
chemical data have been proposed. Assman,et al. and Assman and Brewer
(22,44) have suggested a model similar to the membrane structure pro-
posed by Singer and Nicolson (45) in which the HDL protein is depicted
as an "iceberg" floating in a "sea of lipid." This model takes into
account the fact that portions of the helical regions of the apoproteins
are two-sided; that is, one side of the helix is polar while the other
is non-polar (46). This amphipathic organization of the helix could
allow simultaneous interaction of the helix with lipid, protein, and
the aqueous medium. Whether or not the helices are oriented perpen-
dicular or parallel to the surface of the particle isthe_ essential
difference between this model of Assman,et al. and Assman and Brewer
(22,44) and that of Jackson, et al. (25). Jackson has suggested that
the long axes of the helical regions are oriented perpendicularly to
the fatty acy1 chains of the phospholipids. In this model, carbon
atoms 2-4 of the fatty acy1 chains could interact with the non-polar
side of the helix, allowing the hydrophyllic portions to interact with
the medium or other protein chains. There is evidence to suggest that
little electrostatic interaction occurs between phospholipid polar
headgroups and the charged amino acid residues of intact lipo-
Little is known of the protein-protein interactions that may occur
in intact lipoproteins. The only studies reported have been carried
out on human HDL apoproteins in vitro. It has been shown that if one
mixes the RDL apoproteins A-1 and A-2 together in equimolar amounts,
they appear to associate into an aggregate of molecular weight
46,000 (28). This A-1;A-2 complex will bind less lipid than would
be expected, indicating that some of the lipid binding sites are
blocked by protein-protein -interaction. Other studies indicate that
in the absence of lipids and other apoproteins, both A-1 and A-2 will
self-associate (49-53). Although such studies help to understand the
types of interactions that may occur, only work done on the intact
particle can shed light upon the actual associations and interactions
that govern the final structure and function of the lipoproteins.
Recent evidence from Friedberg and Reynolds (54) suggests that
A-1 and A-2 always exist in a molar ratio of 2 A-1 polypeptides to
1 A-2 polypeptide dimer in human HDL. Their study also suggests that
this ratio of 2:1 is constant from individual to individual as well as
from subclass to subclass (HD~L2 to HDL3). This would indicate that
perhaps, for ApoA-1 and ApoA-2, a specific polypeptide composition is
found in HDL2 (4 A-1 + 2 A-2) as well as in HDL3 (2 A-1 + 1 A-2).
It is evident that much is yet unknown concerning the structure
of lipoproteins and that experimental evidence derived from studies
utilizing intact particles and techniques that allow precise definition
of spatial orientation is needed.
Alternative Classification of Lipoproteins
It should be noted at this point that another meth~od of classifi-
cation for plasma lipoproteins has been suggested by Kostner and
Alaupovic (55). Based on immunochemical evidence, they have suggested
that lipoproteins occur in separate families, the composition of which
is determined by the presence of specific apoproteins. The three
families they describe, LP-A, LP-B, and LP-C, represent particles
containing ApoA-1 and/or ApoA-2 (LP-A), ApoB (LP-B), and ApoC-I-III
(LP-C). The important difference between the two methods of classi-
ficatiohn is that, according~ to Kostner and Alaupovic, a physically
defined density class, such as HDL, may contain not only particles
with just ApoA-1 and ApoA-2 but also particles with only ApoB and
only ApoC. One should realize, however, that HDL isolated as a density
class will contain mostly LP-A and LP-C with a very small amount of
LP-B (2%). This LP-B is only found associated with the hDL2 density
subclass suggesting the possibility of contamination by remnant LDL
particles. Even the presence of particles containing only LP-C could
be an artifact resulting from the generation of incomplete or altered
particles during the extensive ultracentrifugation required for the
isolation of the HDL density class. In any case, the actual differ-
ence between HDL as a density class and LP-A as described by these
investigators would only be the presence of the small amount of C
peptides generally believed to be associated with the HDL molecule.
Although the possibility of lipoprotein families is a question that
merits consideration, the results of the experiments described herein
would apply to either model of the lipoprotein particles.
Exchange Reactions of Lipoproteins
Plasma lipoproteins are known to be a very dynamic population of
macromolecules. They are continuously being synthesized and degraded
at a high rate. The amount of any individual lipoprotein class present
at a given time is determined by the balance between synthesis and
catabolism. The individual components of lipoproteins, however, are
not degraded or synthesized at the same rate. This is due to the fact
that both lipid and protein components can be exchanged between indi-
vidual lipoprotein molecules as well as between different classes of
Phospholipids appear to exchange between all classes of serum
lipoproteins at rates dependent upon their relative concentrations in
the individual classes (56,57). When labeled phospholipids are incor-
porated into an individual class of lipoproteins and this class is
allowed to interact with other lipoprotein classes in vivo or in vitro,
an exchange reaction with equilibrium, reached in 4 to 5 hours, is
observed (58). This exchange is independent of any protein exchange
and occurs at significantly different rates for different phospho-
lipid types. Lysalecithin exchanges most rapidly, followed by lecithin
and sphingomyelin (57,59). No significant differences are observed for
subclasses of an individual phospholipid class (59). This exchange can
be increased in vitra by the presence of a phospholipid exchange
Cholesterol is known to exchange rapidly between all classes of
human serulm lipoproteins (61). This rapid exchange occurs both in vivo
and in vitro wFth equilibrium attained after 2 to 6 hours~ (62,63).
Cholesterol ester exchange between plasma lipoproteins in vivo and
in vitra has been both suggested and disputed (62,64,65). There is
evidence that seems to indicate that cholesterol enters can be trans-
ferred from LDL to other lipoproteins in vitro (66).
The exchange of other lipid types has been reported for human as
well as other animal lipoproteins. Triglycerides exchange between
different classes of lipoproteins (67) as do unesterified fatty acids
(68) and a-tocopherol (69)..
Although the exchange of lipids does not appear to depend upon
the simultaneous exchange of protein, it has been shown that apoprotein
exchange does occur. Only ApoC of HDL and VLDL has been shown to
exchange in vitro and in vivo (70,71). This exchange is bidirectional
and not like the apparent transfer of protein from VLDL to LDL (2).
Mechanisms of Exchange
The movement of either lipid or protein from one lipoprotein class
to another can represent an exchange, a transfer, or the combination
of the two processes. It is evident that both processes depend upon
the lipoproteins existing in a dynamic state.
The in vitro exchange of lipids has been the most studied and the
process appears to be a physicochemical one (72). Exchange could occur
by at least two mechanisms. First, it has been suggested that exchange
occurs when individual lipid molecules escape the lipoprotein particle
and enter the aqueous medium. They are then picked up by other lipo-
proteins or membranes (73,74). This model for exchange lacks strong
evidence in its support but may be the mechanism for certain specific
lipid types (74).
An alternate mechanism involves the formation of collision.complexes
between lipoproteins (75). This would allow for diffusion of lipid
and/or protein molecules between the particles and at the same time
would not require the thermodynamically unfavorable dissociation of
hydrophobic molecules into an aqueous medium.-
As both mechanisms have been supported with experimental evidence,
it is important that each be considered as a possible explanation for
any exchange reaction. Experiments that will belp distinguish
between the two mechanisms would include the observation of the effect
of factors that influence collision rates between molecules, i.e.
temperature, dilution, etc.
Although only lipid transfer mechanisms have been postulated,
the same types of models should be considered when observing protein
transfer or exchange.
In studying either protein structure or protein-protein inter-
actions occurrences of natural crosslinking due to disulfide bridges,
peptide linkages between lysine E-NH2 and glutamic or asparatic acid
y or B-COOH, dr carbohydrate bridges have been valuable in under-
standing the final active structure that is being considered, e.g.,
insulin, ribonuclease, collagen. As such crosslinks are often not
found where one might like to find them, the introduction of stable
covalent bridges or crosslinks can aid in the study of protein
structure (See Ref. (76) for a recent review of bifunctional reagents).
Bifunctional reagents are capable of producing three types of cross-
linked products: intramolecularly crosslinked proteins, intermolec-
ularly crosslinked homopolymers (identical subunits), and inter-
molecularly crosslinked protein complexes nonidenticall subunits or
different proteins). Such reagents have been used in a variety of ways
to determine such things as the spatial geometry of membrane components
(77-81), the mapping of ribosome proteins (82), and the subunit
structure of oligomeric proteins (83).
There are two basic types of crosslinking reagents: those that
are cleavable and those that are not. The advantage of cleavable
crosslinkers is that one may isolate the crosslinked product, cleave
the newly formed bridges, and identify the proteins that were cross-
linked. This technique is particularly useful when studying the
crosslinking of multicomponent systems such as membranes.
Table 1 lists a selection of different types of crosslinkers and
gives certain specific characteristics. The selection of reagents
with differing lengths, degrees of hydrophobic character,or reactive
group specificity can yield products that differ because of the specific
structure or physical environment of the proteins that are crosslinked.
For example, in studying the human erythrocyte membrane, Moxley and
Tsai were able to produce different crosslinking patterns by varying
only the length of the bifunctional reagent used (81).
As all lipoprotein particles are thought to contain two or more
polypeptide chains and since the structure and function of these
particles may depend to a great extent upon specific protein composition
and/or protein-protein inlteractions, investigation wlthl btfunctiolnal
reagents could prove to be a useful method to define some of these
Because the function and characteristics of individual lipoprotein
classes differ, and these.differences ultimately must be related to
structure, studies dealing with the interactions and associations of
Bifunctional Crosslinking Reagents
STRUCTURE REACTS WITH
N ztCI NiallI Lysine, Tyrosine
3,C-O--(C-(CH, I C-O-CH,
2. Dinitro Difluoro-
3. Difluoro Dinitro- F ~ F
Diphenyl sulfone ( >-0
4. N, N-Di (Bromoacetyl)
Pd + HOAc
5. Dimethyl Dithiobis
HzC -C-2-(CH,);-5- -(cH,a- c- o-CH,
the characteristic apoproteins of a given lipoprotein class should
prove of value in understanding how these characteristics are deter-
mined by the structure.
KDL, the best characterized of the lipoproteins, was used in the
experiments described in this report. Its small size, stable structure,
ease of isolation and handling, as well as its variety of polypeptides
made it the model of choice. Although no definitive physiological
role has yet been assigned to HDL, certain other characteristics make
this an interesting model for study. C-peptides such as are found in
HDL are known activators of LCAT and lipoprotein lipase, and Apo-1 is
known to activate LCAT specific for certain substrates (1). Reynolds
and Simon (28) have reported altered lipid binding upon associations
of the apoproteins of HDL. Perhaps this interaction is a mechanism for
regulation of lipid binding. As the different lipoprotein classes
appear closely related and even share many of the same components, both
lipid and protein, the actual association of the different protein
components may play a very important role in regulating both structure
and function for the different lipoprotein particles.
The purpose of this research was to gain insight into the structure
of intact serum lipoprotein particles specifically, the high density
lipoprotein particle. This was approached by attempting to understand
the interactions and associations of the individual apoprotein compo-
nents of these macromolecules. To aid in organizing the information
gathered, the following objectives were set forth:
1. To determine if bifunctional reagents are capable of inducing
crosslinks between individual apoproteins in intact HDL
2. To characterize and identify any crosslinked products pro-
3. To evaluate the result of the use of crosslinking reagents of
differing length and hydrophobic or hydrophyllic character
4. To determine and explain any possible differences in the
crosslinking patterns produced from HDL2 vs. RDL3
5. To discover if the apoproteins and lipids of intact HDL
exchange between the separate HDL subclasses
6. To investigate any possible effects of chemical crosslinking
on such exchange processes
7. To gain a greater understanding of the relationships and
structures of the two HDL subclasses, HDL2 and HDL .
MATERIALS AND METHODS
1,5-difloro, 2,4-dinitrobenzene (DFDNB) and sodium dodecyl sulfate
(SDS) were purchased from Sigma Chemical Co., St. Louis, Missouri.
4,4'-difluoro 3,3' -dinitradiphenylsulfone (DFDNPS); dimethyl suber-
imidate dihydrochlorida (DMA); and dimethyl 3,3'-dithiobispropionim-
idate dihydrocholoride (DTBP) were purchased from Pierce, Rockford,
Illinois. Reagents for acrylamide gel electrophoresis were obtained
from BioRad Laboratories, Los Angeles, California. All other chemicals
used were reagent grade and purchased from common suppliers.
Radioisotopes and Scintillation Materials
[12IJ-Nal; [choline-methyl-14C]-phosphatidy1 choline, specific
activity 50mCi/mmole; [11-12 3H]-cholesterol, specific activity
60Ci/mmole; [11-12 3H]-cholesterol palmitate, specific activity 294mCi/
mmole; Aquasol-2; and mini-vials were all purchased from New England
Nuclear, Boston, Massachusetts.
Whole human blood was obtained fresh from healthy male donors or
as outdated blood units from the Civitan Regional Blood Center, Gaines-
ville, Florida. In no instances were the results of any repeated
experiments different depending upon the source of blood.
Isolation of Lipoproteins
Whole blood was centrifuged for 20 min at 2000 RP:M using an IEC
Model SBV centrifuge to remove red cells. The clear serum or plasma
was removed by decanting and the density then raised to 1.063gm/cm3
by the addition of solid KBr. This solution was centrifuged for
20-24 hrs.at 42,000 using a Beckman Ti60 rotor in a Beckman model L2-65B
ultracentrifuge. After centrifugation, the top 3-5 ml containing
CHYL, VLDL, and LDL were removed by aspiration and discarded. The
remaining solution was adjusted to density = 1.21gm/cm3 by the addition
of more solid KBr. The serum was again centrifuged for 24 hrs. at
42,000 RPM as above. The top 2-3 ml of each tube were then removed
using a Pasteur pipet and pooled. When total HDL (HDLt) was desired,
the pooled lipoprotein solution was further purified by another centri-
fugation step. This was carried out as above after the crude HDL
solution was adjusted to density = 1.063gm/cm3 by the addition of 0.15M
NaC1 containing 10 ME EDTA and 0.02% NaN3. Each tube was underlayered
with 5-10 ml of KBr solution of density = 1.21gm/cm~ before centri-
fugation. After centrifugation, the top 5 ml of each tube (LDL
contamination) was removed by aspiration and discarded. The yellow
band in the tube located above the 1.21 density layer was removed using
a Pasteur pipe and saved as purified total RDL. This HDL solution was
immediately dialyzed against 2 x 4 liters of 0.15M NaCl containing
10 ME EDTA and 0.02% NaN3, and stored at 4'C until use.
When the separate RDL subclasses, HDL2 and HDL were desired, the
following procedure was used. After the three ultracentrifugation
steps described above, the total HDL was dialyzed against a Kllr-NaC1
solution of density = 1.125gm/lcm3 contalining 0.15M Nact, 10-51 ULA
and 0.02% NaN3 The solution was placed in centrifuge tubes and
underlayered with 5 ml. of KBr solution, density = 1.21gm/cm3; then
centrifuged for 36-48 hrs. at 45,000 RPM using a Ti60 rotor. This
step resulted in the formation of two separate yellow bands; one at
the top of the tube, and one just above the 1.21 plug. The top band
was taken as HDL2 and the lower as HDL The bands were removed using
a Pasteur pipet and the region between the bands discarded. The HDL2
and HDL3 solutions were then dialyzed against 2 x 4 liters of 0.15M
NaC1 containing 10-5% EDTA and 0.02% NaN3, and stored at 4"C until
use. The purity of HDL preparations was checked by SDS gel electro-
phoresis (see next section for reference and details). In no prepara-
tion did the contamination by other proteins (usually serum albumin)
exceed 1-2% and usually no contamination could be seen at all unless
the gels were greatly overloaded.
SDS polyacrylamide gel electrophoresis (SDS-PAGE)
Polyacrylamide gel electrophoresis was carried out using a Model
300B or 301 BioRad Electrophoresis Cell powered by a BioRad Model 400
Power Supply. TLwo types of gel formulation systems were used. First,
gels were prepared using the procedure of Lammli (84). The acrylamide
concentration of the separating gel was adjusted to 10-12% by alteration
of the ratio of acrylamide to water during gel formulation. SDS concen-
tration was maintained at 0.1% and for 5 x 100rmm gels electrophoresis
was carried out at 1.5ma/gel. Bromophenol Blue was used as a tracking
dye and electrophoresis continued until the dye band was approximately
Icm from the bottom of the gel. Gels were fixed in 12.5% TCA: 40%
ethanol: 7% acetic acid, stained with 0.05% Coomassie Blue in 10%
ethanol: 7%. acetic acid, and destained using a Bio~ad Model 172A2
diffusion destainer. Gels were stored in 7% acetic acid.
For reasons to be discussed later, the gel system was changed
during the course of the work to that of Fairbanks, et al. (85). A
7.8% acrylamide gel with 1.0% SDS was prepared and run at 8mal/gel for
5 x 10Dam gels. Pyronin Y was used as a tracking dye. After electro-
phoresis the gels were fixed in 40% propanol: 7.5% acetic, and stained
with 0.05% Coomassie Blue in 40% propanol: 7.5% acetic acid. After
destaining as above the gels were stored in 7.5% acetic acid.
Molecular weights were determined from the recorded scans using
added molecular weight markers as well as the location of the apo-
proteins of known molecular weights.
The desired amount of protein solution to be crosslinked was added
to a 13 x 100mm glass tube and 10X concentrated buffer added until a
final concentration of 0.025M buffer was achieved. Crosslinker
dissolved in ethanol, water, or acetone (1-5mg/ml) was then added with
rapid mixing to attain the desired molar ratio of crosslinker to protein.
The amount of organic solvent added never exceeded 5% of the total
volume of the reaction mixture. For specific crosslinkers the follow-
ing buffer and solvent systems were used: DFDNB 0.025M Na2CO3 buffer,
pH 8.8, ethanol: DFDNPS 0.025M Na2CO3 buffer pH 8.8, acetone; DMA,
DMSDTB .02M N2HPO4 buffer pH 7.5, water. The reaction mixtures
were incubated for at least 2 hrs. at room temperature unless otherwise
indicated. The reaction mixtures were used directly without further
treatment as it could been shown that only very small amounts of the
free hydrolysis products of the crosslinkers were formed.
For labeling with FDNB, the monofunctional analog of DFDNB, the
conditions described for DFDNB were used.
Amino acid analysis of specific crosslinked products
To determine the amino acid content of a specific crosslinked
product it was first necessary to separate that product from all the
other products of the crosslinking reaction. This was achieved by
preparative gel electrophoresis of fluorescent labeled proteins. Using
a BioRad Model 301 gel system and 10% Lammli gels as described above,
0.5 to 1.0 ml of HDL solution containing 1-3mg/m1 protein, that had
been made fluorescent by reaction with fluorescamine (86), was applied
to the top of each gel. Electrophoresis was carried out at 6 ma/gel
until sufficient resolution of the separate bands was achieved. The
bands were easily visualized using a long wave UV light and after
electrophoresis were sliced from the gel using a razor blade. Slices
from duplicate gels were combined, crushed with a glass rod, and
mixed with 5 to 10 ml of 0.1% SDS solution. The mixture was incubated
at 370C with shaking for 12 hrs., filtered, and the procedure repeated.
The combined solutions were filtered through a .45p Millipore filter
and dialyzed against 3 x 4 liters of H20. The dialysis bag was
agitated to suspend the fine precipitate and the suspension lyophyllized.
The powdered residue was dissolved in 5 ml of 6N-HC1 with 10pl of
B-mercaptoethano1 and 10pl of 0.5% phenol. The solution was sealed in
glass ampules in vacuo and then hydrolyzed for 24 hrs. at 110*C. The
hydrolyzate was dried by lyophyllization, redissolved in 0.2M citrate
buffer pH 2.2, and standard amino acid analysis run on a Beckman
Model 120C amino acid analyzer. Standard citrate buffers and a
ninhydrin detection system were used.
Calculation of the amino acid composition of then proteins analyzed
was carried out using the formula below.
residues of leu or val predicted/molecule XmoeAA=residues AA
mrmoles leu or val molecule
As the amina acid composition was used as a method of verification
of predicted composition, this type of calculation proved most useful.
Delipidation of lipoproteins
Ten ml of HDL solution to be delipidated wereadded with stirring
to 250 ml of chloroform:methano1 (2:1) at -10"C. The turbid solution
was allowed to stand at -10DC for 30 min, 250 ml of cold ethyl ether
were added and the solution filtered on a Buchner funnel through
Whatman #1 filter paper. The filtrate was washed with 250 ml of cold
ethyl ether and dried with a stream of N2. The precipitate was removed
from the paper and stored at -10*C under N2 until use. This method
is a modification of the procedure of Lux et al. (87).
Cel filtration chromatography
Sephadex 0-200 was allowed to swell in water and then poured into
a 2.5 x 100cm glass column to a height of 90cm. It was eqluilibrated
with 0.1% SDS, 0.02M tris-HC1, 10mM NaN3, 1mM EDTA buffer pH( 8.0,
30mg of delipidated protein dissolved in this buffer was added to the
column. An ascending flow rate of 10ml/hr was used and 150 drop frac-
tions collected. The OD28 was monitored using an ISCO UA-2 UV monitor.
The column was stopped after 120ml of buffer had passed through the
Iodination of lipoproteins
Five ml of the sample to be labeled were dialyzed against 0.5M
glycine buffer pH 9.4, containing 0.15M NaC1, 10 M EDTA, and 0.02%
NaN3; then labeled using the iodine monochloride (IM) method of
MacFarlane (88). The iodine carrier solution (0.5ml) containing
250plCi of [12I]-Nal was added to the HDL solution with rapid mixing.
After 10 min. the unreacted iodine and glycine buffer were removed by
gel filtration chromotography using Sephadex G-25 equilibrated withl
0.15M NaC1 containing 10 M EDTA and 0.02% NaN3. Protein solutions
with a specific activity of 5,000 to 10,000cpm/pg were routinely
obtained using this method. Unlabeled controls were treated as above
except for the addition of the [12I]-iodine carrier solution. It
could be shown that approximately 99% of the radioactivity was asso-
ciated with the apoproteins while only 1% was found in the lipid
Incorporation of labeled lipid into HDL,
Fifty mg of celite (50p1 particles, Johns-Mansville Co.) were mixed
with 1.0ml of chloroform to which 50-100pl of labeled lipid dissolved
in toluene-ethanol (1:1) had been added. After mixing, the solvent vol-
ume was reduced with a stream of N2 and the remainder of the solvent
removed in vacuo. Two ml of the HDL solution to be labeled was added
with gentle mixing and the suspension incubated for the desired time
at 370C. The celite was kept in suspension by occasional swirling of
the tubes. After incubation was complete the suspension was filtered
through a 0.45v Millipore filter to remove the celite. All incubations
were carried out for 2 hrs. unless otherwise indicated.
Exch~ange reactions Incubation of HDL2 and HDL mixtures
Thre desired amoulnts of labeled and unlabeled IlDL2 and HlDL3 in
solution were added directly to 5/8" x 3" cellulose nitrate centrifuge
tubes and incubated for the- time and temperature indicated. SDS-PAGE
of controls was used as a method to detect any possible deterioration
of the RDL during incubation. Unless otherwise indicated, for all
incubations described herein the following volumes and concentrations
were used: controls 200pl containing 300pg of HDL protein, mixtures
400pl containing 300vg labeled HDL protein and 300pg cold EDL protein.
For the chaotropic ion experiments, solid KBr was added directly to
the solution to attain the desired concentration of Br ion.
Ultracentrifugation of HDL, HDL, mixtures
The reseparation of labeled and unlabeled HDL2 HDL3 mixtures was
achieved in the following manner. As the mixture to be resolved had
already been placed directly into centrifuge tubes, sufficient NaBr
solution of density =- 1.3288gm/cm3 was added to the mixture to raise
the density to 1.125gm/cm3. NaBr-NaC1 solution (density = 1.125gm/cm3)
containing 0.15M NnC1, 10-5M EDTA, and 0.02% NaN3 was added to give ai
final volume of 10.Dm1. The solutions were then underlayered with
1.0ml of NaBr-NaCL, density 1.21gm/cm3. The tubes were capped and
placed in a Beckman Ti50 rotor for centrifugation at 42,000 RPM for
After ultracentrifugation all gradients were fractionated using -an
ISCO Model 183 gradient fractionator modified to collect fractions from
the bottom of the tube. Before fractionation each gradient was over-
layered with 1.Dml of 0.15M Nacl solution. 30 drop (0.6ml) fractions
were collected for all experiments.
Activation of Sep~harose by Cyanagen Bromide
Ten ml of packed and well washed Sepharose 4B were placed in a
150ml beaker, 20ml of H20 were added, the pH adjusted to 11.0 with
LM NaOH, and the temperature adjusted to 20'C by the addition of a
small amount of ice. Finely divided CNBr (200mg) were added with
constant stirring and the reaction allowed to proceed until cessation
of proton release was observed by monitoring the pH. The pH was
maintained at 11 during the reaction by the addition of 1M NaOH when
needed and the temperature was adjusted by the addition of ice. When
the reaction was complete, after approximately 15-20 min., 100ml of
ground ice wereadded to the mixture. To Sepharose-ice slurry was then
placed on a coarse glass filter and the Sepharose washed with several
volumes of ice-cold water followed by 200ml of ice-cold 0.1M NaHCO3:
0.5M NaC1 buffer pH 8.0. The washed "activated" Sepharose was used
immediately according to the following procedure.
Coupling of RDL to "Activated" Sepharose
Ten ml of "activated" Sepharose prepared as above or obtained from
Pharmacia were mixed with 10-15ml of 0.1M NaHCO :0.5M Nac1 buffer pH 8.0
in a 100ml beaker. HDL solution (5-10ml of 1-5mg/ml) was added and the
mixture stirred gently for 2-3 hours at 40C. The mixture was then
filtered on a glass filter and washed for 30 min. with several volumes
of IM ethanolamine. The Sepharose-HDL (S-HDL) was then washed with
4 x 100ml of 0.1M NaOAc:0.5M NaCl buffer pH 4.0; 4 x 100ml of 0.1M
NaHCO3:0.5 NaCl buffer pH 8.0; and finally with 2 x 100ml of 0.15M Napl:
10-5M EDTA:0.025% NaN3 The Sepharose prepared in this manner was
stored wet in the final wash solution at 40C.
Exchange reactions using Sepharose-bound HDL
A measured portion of packed Sepharose-HDL labeled with  or
radioactive lipid was pipeted into 13 x 100mm glass tubes and an equal
volume of 0.15M NaCZ was added. Cold HDL was added in solution and
the mixture swirled to suspend the gel. After incubation for the
desired time, with occasional swirling to keep the gel suspended, the
mixture was filtered to remove the gel. The filtrate was then counted
in Aquasol-2 in a liquid scintillation counter for lipid exchange exper-
iments, counted in the gamma counter for peptide exchange experiments,
or subjected to ultracentrifugation or SDS-PAGE. For lipid exchange
experiments the following incubation mixtures and times were used:
2591 of Sepharose-HDLg (3mg HDL/ml gel), 200pl of 0.15M NaC1, 50pl of
HDL3 (4tmg/ml) or HDL2' (3mg/ml) incubated for 15 min. at 37"C. For
peptide exchange experiments larger amounts of gel were used due to the
lower levels of HDL binding to the gel (100pg HDL/ml gel). For most
experiments the conditions used were: 400pl Sepharose-[12I]-HDL
(HDL2 or HDL ), 20011l HDL (HDL2 or HDL ) in solution, 200ul 0.15M MaC1.
Incubations were carried out for 5 hours at 370C,
Determination of radioactivity
Al~l counting of  was done using a Nuclear Chicago Model 8725
manual gamma counter. The counting efficiency was approximately 63%
based on a comparison with the efficiency of counting of a'Packard
Auto-Gamma Scintillation Spectrophotometer Model 5130 (76% efficiency).
All samples were counted in 13 x 100mm glass tubes using the same volume
of solution in each tube (0.6ml).
For the measurement of [14C] and [ H], liquid scintillation spectro-
scopy was used. The sample was added to 5.0m1 of Aquasol-2 and the
mixture counted in "mini-vials" purchased from New England Nuclear.
All scintillation counting was carried out using a Beckman Liquid
Scintillation Spectrophotometer. The counting efficiency for [14C
was calculated to be 92% while for ( li] the efficiency was 46%.
All scintillation counting values were corrected for background
and counting efficiency by coincidence counting with commercially
RESULTS AND DISCUSSION
Isolation and Purity of Lipoprotein Preparations
As the procedure for isolation of HDL is well established (4,6),
the two factors of importance considered were the purity of HDLT prepa-
rations with respect to contamination by other proteins, and the effec-
tiveness of separation during the preparation of RDL subclasses, HDL2
and HDL .
Figures la and lb show the scan of SDS gels of a typical prepara-
tion of RDLT. It can be seen that the only major polypeptide bands
observed have molecular weights that correspond to the major apoproteins
known for HDL. As expected, the ApoA-2 peak (MW 17,500) is split into
its monomer form upon the addition of B-mercaptoethano1 (Fig. lb). The
peak is reduced in size due to a significant loss of low molecular
weight proteins during the fixation and staining of the gels (89).
The purity of HDL2 and HDL3 preparations was checked by recentri-
fugation of the separate subclasses at the density of 1.125gm/cm3 used
during isolation. Under these conditions HDL3 should sink toward the
bottom of the centrifuge tube while HDL2 should float at the top of the
solution. Figure 2 shows the fractionation patterns after this recen-
trifugation. It can clearly be seen that both HDL2 and HDL3 preparations
are pure by the physical criterion of ultracentrifugal mobility. All
preparations of RDL2 were shown to be pure by this method. .HDL~ prepa-
rations, however, occasionally showed some slight contamination by HDL2'
Figure 1. SDS-PAGE of SDS solubilized apoproteins of HDLT
A. Pattern obtained in the absence of reducing agent.
B. Pattern observed when 5% B-mercaptoethanol is
included in the preparation.
Figure 2. Distribution of protein after recentrifugation
of purified HDL2 (6) or HDL3 ($) at a density
8 i HDL3 HDL2 *
S200 + a
0 2 4 6 8 10 12 14 16 18 20
This contamination never represented more than 5% of the total protein
present, and, when observed, an additional centrifugation step was
used to remove it. This HDL2 contamination did not appear to be the
result of HDL3 dimer formation; as once removed, no new contamninant
appeared under any conditions of treatment.
Crosslinking of Total HDL (R~DLT) with DFDNB
The selection of DFDNB as the initial crosslinking reagent to be
evaluated was based on several factors. First, it has been widely
used in previous crosslinking studies involving a variety of protein
types (76). Second, its monofunctional analog FDNB has also been widely
used (76). Third, it was believed that its hydrophobic nature would
'aid in its ability to penetrate the lipoprotein particles. Finally,
pure DFDNB and FDNB are available from commercial chemical laboratories
at low cost.
Figure 3 shows the ge1 patterns observed when a 2mg/ml protein
solution of HDLT is crosslinked withr a 5 fold molar excess of DFDNB
and then run on SDS-PAGE. Figure 3a shows the pattern obtained in the
absence of any reducing agent while 3b is observed in the presence of
1% 8-mercaptoethanol. Although a variety of high molecular weight
crosslinked products are produced (Figure 3a), the first product to
appear and in the highest quantity is a polypeptide complex with an
apparent molecular weight of approximately 46,000. In the presence of
B-mercaptoethanol this major peak is shifted to an apparent molecular
weight of 37,000. Considering only ApoA-1 (HW 28,330) and ApoA-2
dimerr MW 17,500) the appearance of a crosslinked product with a
molecular weight of 46,000 could only be the result of one A-1 linked
Figure 3. SDS-PAGE (10% Lammli gels) of SDS solubilized apoproteins
of HDLT crosslinked with a 5-fold molar excess of DFDNB.
(A) DFDNB crosslinked HDLT, no reducing agent.
(B) DFDNB crosslinked HDLT, with 5% B-mercaptoethanol.
to one A-2 unit. This conclusion is supported by Figure 3b which shows
that the addition of the disulfide cleaving agent, 8-mercaptoethanal,
results in the loss of half of the A-2 dimer leaving a product con-
sisting of one A-1 linked to onle A\-2 monomefr. Th~e molecular weight
of such a complex would be approximately 36,000.
As thle estimation of: molecular weights by SDS get clectrophonresis
can result in considerable error when considering proteins which may
have both inter- and intra-peptide crosslinks (90), additional evidence
is desirable before one can conclude that an A-1***A-2 crosslinked
product is actually produced. To approach this problem it is necessary
to separate the individual crosslinked products so they can be charac-
terized. Gel filtration chromotography using Sephadex G-200 containing
1.0% SDS was attempted. Although this procedure was capable of a
reasonable separation of ApoA-1 from ApoA-2, the crosslinked products
were eluted as a broad, even band; each fraction containing a mixture
of similar molecular weight products. This procedure was not capable
of sufficient purification of any single product. The problem was
solved using the experimental approach diagrammed in Figure 4. The
procedure is detailed in the Methods section of this dissertation. The
results of the amino acid analysis of the 46,000 dalton crosslinked
product is shown in Table 2. Comparison to the predicted values for
sleected amlino acids for an A-1***A-2 complex indicates the 46,000
dalton product is likely a complex of one A-1 and one A-2. Certain
amino acid values were not shown because they were components of the
ge1 buffer system, gave unexpected high or low values for purified A-1
or A-2, or were destroyed during hydrolysis. That both A-1 and A-2
HDL (HDL2 or HDL )
Chemical Crosslinking ----- DS PAGE
Prep. SDS PAGE
Slice out protein band
Elution of protein
Amino acid hydrolysis
Figure 4. Experimental approach to amino acid analysis of
Amino Acid Analysis of the 46,000 Dalton Crosslinked Product
from HDL, Crosslinked with DFDNB
aAll values expressed as residues per mole
of protein based on the value
are components of this complex is suggested by two observations. First,
the shift in molecular weight upon addition of reducing agent could
only occur if ApoA-2 was present as no disulfide bonds occur in Apo-1.
Second, there is no histidine in ApoA-2 or isoleucine in ApoA-1, but
both occur in the 46,000 dalton product. As ApoA-1 contains 5 histidine
residues/molecules and ApoA-2 dimer contains 2 isoleucine residues/mol-
ecule, those must be the source of the histidine and isoleucine found
in the complex. Although the C-peptides contain both isoleucine and
histidine, the low total amount of C-peptides in RDL excludes these
polypeptides from a significant contribution to the composition of any
major crosslinked product.
Effect of Crosslinking on HDL Str'ucture
In discussing the meaning of the results of crosslinking experi-
ments it is important to establish that the introduction of these
crosslinks does not significantly alter the structure of the macro-
molecular complex that is being studied. In the case of HDL, it could
be shown that crosslinked HDL retains the physical properties of native
RDL and will migrate in the analytical ultracentrifuge as a single peak
essentially identical to native RDL. Although even very high levels
of crosslinking (100-fold molar excess) do not appear to disrupt the
particles, all conclusions presented are based on data obtained at
mruch lower levels of crosslinking.
Crosslinking of HDL~ and RDL_ with DFDNB
Although the peptide composition of the two HDL subclasses, RDL2
and RDL appears to be the same, the amount of total protein is
different (1). HDL2 should contain approximately 160,000 daltons of
protein/molecule while HDL3 should only contain 80,000 daltons/molecule.
Under conditions of increasing amounts of crosslinker, one might expect
to observe larger molecular weight crosslinked products for HDL2 as
compared to HDL However,. using the Lammli gel system with molar
ratios of crosslinker higher than 5 to 10-fold, one observes only a
smear of stain at the top of the gel for each subclass. This phenomenon
has been noted for other types of crosslinkers (91). For this reason
a different gel system was used. Fairbanks, et al. have noted that an
SDS concentration of 1%, rather than the 0.1% found in the Lammli
system, is necessary to dissociate membrane protein complexes (85).
Figure 5 shows the comparison of the pattern obtained for HDLT cross-
linked with a 20-fold molar excess of DFDNB and then run on the two gel
systems (5a-Lammli, 5b-Fairbanks). The 1% SDS in the Fairbanks gel
system is apparently sufficient to cause total dissociation of even
highly crosslinked lipoprotein particles.
Figures 6 (HDL2) and 7 (RDL ) demonstrate that for both HDL2 and
HDL3 the crosslinking patterns are essentially the same. It is inter-
esting to note that no crosslinked products of apparent molecular weight
products occur even if the level of crosslinker is increased up to
100-fold. For this reason no separation of HDL2 and HDL3 was carried
out for other crosslinking experiments; instead, HDLT was used.
Crosslinking of HDL_ with DMS and DFDNDPS
To investigate the hypothesis that the hydrophobic nature and length
of the bifunctional reagent used should have some effect .0n the extent
and type of crosslinking produced, a comparison of the gel patterns
obtained was made after DMS or DFDNDPS was used to crosslink HDLT. These
Figure 5. Comparison of the SDS-PAGE methods
of Fairbanks (A) and Lammli (B) for
the separation of highly crosslinked
HDLT (25x DFDNB) .
Figure 6. SDS-PAGE (7.8% Fairbanks gels) of SDS solubilized HDL2
crosslinked with a 10-fold molar excess of DFDNB.
Figure 7, SDS-PAGE (7.8% Fairbanks gels) of SDS solubilized HDL3
crosslinked with a 10-fold molar excess of DFDNB.
two reagents have similar distances between the functional groups but
differ greatly in hydrophobic character. DMS is water soluble and car-
ries a positively charged imido group while DFDNDPS is quite insoluble
in water and must be added in an acetone solution. At low levels of
crosslinker (10-fold molar excess) these two reagents appear to
generate significantly different initial crosslinked products (Figures
8-11). Two major differences between DMS and DFDNDPS crosslinking can
be seen. First, the total amount of crosslinking is smaller for the
hydrophyllic reagent DMS. Second, the initial site of reaction for DMS
appears to differ from DFDNDPS with a 65,000 dalton product appearing
first. At higher molar ratios of these reagents the patterns become
very similar with all different products appearing in similar amounts
in each case.
The initial products produced with DFDNDPS crosslinking are
essentially the same as those formed with DFDNB even though the former
reagent is twice as long as the latter.
Temperature Dependence of Crosslinking
When the crosslinking of HDLT with a 10-fold molar excess of
DFDNB is carried out at different temperatures ranging from 40 to 700C,
no significant differences in the crosslinking patterns are produced.
The number and relative amounts of the separate crosslinked products
remains similar at all temperatures studied. The only difference
observed is a reduced amount of crosslinking at the very high temper-
atures, 60-70"C. It should be noted, however, that the rate of
formation of hydrolysis products of this bifunctional reagent increases
significantly at these elevated temperatures. This would lead to
Figure 8. SDS-PAGE (10% Lammli gels) of SDS solubilized HDL,
crosslinked with a 5-fold molar excess of DMS. No
reducing agent present.
Figure 9. SDS-PAGE (10% Lammli gels) of SDS solubilized HDLT
crosslinked with a 5-fold molar excess of DMS.
5% B-mercaptoethanal included.
Figure 10. SDS-PAGE (10% Lammli gels) of SDS solubilized HDL,
crosslinked with a 5-fold molar excess of DFDNDPS.
No reducing agent present.
Figure 11. SDS-PAGE (10% Lammli gels) of SDS solubilized HDL,
crosslinked with a 5-fold molar excess of DFDNDPS.
5% B-mercaptoethanol included.
decreased availability of reagent as well as an increased probability
of monofunctional labeling.
Interpretation of Cross'linking Results
From the data in this section we conclude that at least some of
the two major apoprotein components must lie very close to one another
in the intact lipoproteln particle since DFDNB will only crosslink
groups about 5A apart. If each HDL particle contains both A-1 and A-2
molecules, a number of possible crosslinking products could result.
From our findings with low levels of DFDNB (5-10 fold molar excess)
several observations can be made. First, the major crosslinked product
appears to comprise one A-1 plus one A-2; second, there is no indication
of any A-2 intramolecular crosslinking which might be expected with a
molecule already crosslinked with a disulfide bridge. .There also
seems to be little significant crosslinking-reactive association of two
A-1 molecules as suggested by Scanu (92). Since both A-1 and A-2
contain numerous reactive residues fairly evenly distributed throughout
their sequence (1), any close association of the chains should make
multiple sites available for possible crosslinking.
Experiments using higher molar ratios of crosslinker (20-100X) do
show that most possible combinations of A-1 and A-2 are formed. Still,
even at these high concentrations, no A-2 intramalecular crosslinked
molecules are produced. The C-peptides do not appear to be involved in
the crosslinking reactions. At high levels of reagent the amount of
unreacted C-peptide does not appear to decrease when compared to con-
trols. This suggests little association of C-peptides with A-1 or A-2.
The formation of most of the possible products at high crosslinker
concentrations seems to indicate either a significant degree of mobility
of the apoproteins or significant alteration of the native structure of
the particle. 2Two factors suggest that although some mobility exists,
a high degree of specific association of the apoproteins occurs in the
intact HDL particle. First, the decreased crosslinking observed at
high temperature could be due to increased mobility of the polypeptide
chains. Second, the failure of A-2 to form intramalecular crosslinks
suggests a stable conformation that remains fairly fixed even at these
elevated temperatures. The reproducibility of the crosslinking exper-
iments with respect to order of formation and ratios of products formed,
even when HDL is obtained from varied sources, suggests that in the
intact HDL particle the apoprotein components may be found in a rather
fixed spatial orientation to one another.
That the crosslinking patterns for HDL2 and RDL3 do not differ also
argues for a fixed association of 2A-1 and 1A-2 apoproteins. Even though
RDL2 may have two HDL3 protein subunits (54), these do not appear to
interact. This would lead one to speculate that HDL2 might be a dimer
of 2-RDL3 molecules. This idea will be discussed in detail later in
An additional argument in favor of fixed associations with flexi-
bility of the apoproteins comes from the experiments using the two
reagents DMS and DFDNDPS. These two reagents consistently yield initial
crosslinking products that.differ. This may occur due to different
initial sites of reaction for these reagents. The initial products do
not differ when one compares DFDNB and DFDNDPS, although the latter is
twice as long as DFDNB. This again suggests some flexibility in the
associations of the apoproteins.
From these observations one might suggest a model in which the
apoproteins lie in a fixed, flexible conformation held in place in
part by protein-protein, lipid-protein, and lipid-lipid interactions.
The ability to bind lipids may in turn be governed by-these inter-
actions; the flexibility allowing for differing degrees of binding of
specific "functional" lipids while certain "structural" lipids remain
Aorotein Exchange Between HDL^ and RDL
If the ratio of A-1 to A-2 dimer is 2:1, in both HDL2 and HDL3 as
suggested by Friedberg and Reynolds (54), then the difference between
RDL2 and HDL3 may represent the addition of one A-1:A-2 unit, which
consists of two A-1 plus one A-2 peptide dimer. Because our cross-
linking data suggests a degree of flexibility as well as specific inter-
action within an A-1:A-2 unit but the two units do not appear to cross-
link in intact HDL2, it is necessary to consider possible modes of
association of the apoproteips of the two proposed units in HDL .
To investigate this possible association, as well as the relation-
ship between the apoproteins of the different subclasses of HDL, we
have examined the exchange of labeled apoproteins between HDL2 and HDL3,
in the presence and absence of bifunctional crosslinking reagents.
Figures 12 and 13 show the ultracentrifugal fractionation patterns
obtained for controls (300ug HDL protein in 200pl) of [12I]-labeled
HDL3 and [125I-labeled HDL2 incubated In the absence of any other
protein. The results are expressed as dpm per fraction as it was found
that the radioactivity and protein concentration curves coincided (the
specific activities of control fractions were constant throughout the
Figure 12. Exchange of apoproteins between [12I]-HDL3 and HDL2'
Distribution of label after centrifugation tosaat
HDL2 and HDL3. $ = [125I]-HDL3 control; O =[1 ]HD
incubated with cold HDL2*
16000 + 3
8000 +B 8
0 2 4 6 8 10 12 14 16 18 20
Figure 13. Exchange of apoproteins between [1250-HDL2 and HDL3.
Distribution of label ~ter centrifugation to a Parate
HDL2 and HDL3. $ = [1 I]-HDL2 control; 8 = [1 DI]-HDL2
incubated with cold HDL3
40000 + e
~ + HDL3
8000 + /"a.
0 2 4 6 8 10 12 14 16 18 20
tube). Figures 12 and 13 also illustrate the results obtained after
mixing equal amounts (300pg protein in 200)11) of (12I]-labeled HDL3
and cold HDL2 (Figure 12) or (12Ii-labeled HDL2 and cold HDL3 and
incubating the mixture for 7 brs. at 370C (Figure 13). In Figure 12
it can clearly be seen that A transfer of label from HDL3 to HDL2 has
occurred. Further, the specific activity of the HDL3 fraction was
significantly decreased when compared to controls. In a typical
experiment, the values for a selected fraction in the RDL3 region of
the tube (fraction 6) were: [12I]-HDL3 control = 2850 dpm/pg;
112I]-HDL3 incubated with cold HDL2 = 1680 dpm/pg. Similar results
were obtained when labeled HDL2 was incubated with cold HDL3 (Figure 13).
However, due to the breadth of the HDL3 peak as compared with the HDL2
fractions, the results are not as easily visualized.
Effect of Crosslinking and Monofunctional
Labeling on Apoprotein Exchange
If the labeled HDL3 is crosslinked with DFDNB before incubation
with HDL2, the transfer of labeled protein is reduced depending upon
the extent of crosslinking. Figure 14 shows the three fractionation
patterns obtained with : [12I]-HlDL3 incubated with HDL2 (control);
[125I]-HDL3j crosslinked with a 5-fold molar excess of DFDNB and then
incubated with HDL2; and [12I]-HDL3 incubated with HDL2 following
crossl~inking of the HIDL3 with a 25-fold molar excess of DFDNB. It canl
be seen that increased crosslinking results in decreased exchange.
To rule out the possibility that the monofunctional substitution
of the apoproteins with the crosslinking reagent was the cause of this'
decrease in exchange, FDNB, a monofunctional analog of DFDNB, was used
to modify the proteins. Reaction with concentrations of FDNB, comparable
Figure 14. Effect of crosslinking on the apoprotein exchange between
[125IJ-HDL3 and HDL2
Distribution of label after centrifugation to separate
HDL2 and HDL3. 6 = [125I]-HDL3 incubated with cold HDL2;
$ = [125I] _"L3 crosslinked with a 5-fold molar excess of
DFDNB followed by incubation with HDL2; P = [125I]-HDL3
crosslinked with a 25-fold molar excess of DFDNB then
incubated with HDL2*
soo e e se~e e
4000 + \
0 +********..... ***********************--*+ **.......+....... .. ......4......... ....+
O 2 4 6 8 10 12 14 16
to amounts of DFDNB that greatly inhibited exchange, had little effect
on the exchange process (Figure 15).
When HDL2 is crosslinked before incubation with labeled llDL the
exchange process is also Inhi~bited (Figulre 16). 'The reverse also hioldu
true; that is for crosslinked HDL3 incubated with labeled HDL2, the
exchange is reduced.
Effect of Temperature and Dilution on Peptide Exchange
Both low temperature and dilution have an effect upon the rate of
protein exchange in HDL. If the incubation is carried out at 4.0, very
little exchange will occur in the time period in which significant
exchange occurs at 37"C (Figure 17). Using the amounts of HDL2 and
HDL3 described in the Methods section, but diluting this mixture with
a 10-fold excess (5ml) of 0.15M NaC1, the rate of exchange at 370C is
also significantly reduced (Figure 18). These two observations are
consistent with a mechanism of exchange involving a collisional com-
It is this reduction in exchange rates that permits one to employ
the 48 hour centrifugation requir-ed for reseparation of the HDL2 and
HDL3 mixtures. When equal amounts of labeled HDL3 and cold HDL2 are
mixed, then immediately diluted, cooled to 4.0, and placed in the
ultracentrifuge, essentially no exchange occurs, even after 48 hrs.
Rate of Peptide Exchange
Figure 19 shows a time course plot for the exchange of apoproteins
from [12I]-HDL3 into HDL2. The radioactivity represents the total
transferred into the HDL2 region of the centrifuge tube after subtrac-
tion of the amount in those fractions attributed to HDL Estimation
Figure 15. Effect of monofunctional labeling with FDNB on the exchange
of apoproteins between [125I]-HDL3 and HDL2*
Distribution of label after centrifugation to separate HDL2
and HDL~3. 6 = [125I]-HDL3 reacted with a 25-fold molar
excess of FDNB; 8 = [125I]-HDL3 reacted with a 25-fold molar
excess of FDNB followed by incubation with HDL2*
[12I]-HDL3 25X FDNB
[125 L3 25X FDNB
/ p + HDL2
,0 + t
o +*********+. - -----.:........+......~...+..... ...........+.. .........4.......,.. ......... ... _~
0 2 4 6 8 10 12 14 16 18 20
Figure 16. Effect of crosslinking the unlabeled HDL subclass on
the exchange of apoproteins between -HDL3 and HDL2.
Distribution of label after centrifugation to separate
HDL2 and HDL3' O = 125I]-HDL3 control; 8 = [12b] -HDL3
incubated with HDL2 crosslinked with a 25-fold molar
excess of DFDNB.
rl5]-HDL + HDL2 25X DFDNB
0 +*********+********+********4 *******.+........-t +.................4 c...... ...4.e ....
0 2 4 6 8 10 12 14 16 18 20
Figure 17. Effect of temperature on the apoprotein exchange between
[125I] HD 3 and HDL2
Distribution of label after centrifugation to se rate
HDL2 and HDL3. 9 = [125I]-HDL3 control; af = 12IJ-HDL3
incubated with HDL2 at 4oC. Compare to Figure 12.
12000 + /
0~ * + * * * * . . . . . + . . 4 . . . . o . g
a 0 0 2 4 8 1 1 4 6 18 2
Figure 18. Effect of temperature and dilution on the apoprotein
exchange between [125I]-HDL3 and HDL2*
Distribution of label after centrifugation to separate
HDL andHDLg 6= 125I]-HDL, and HDL, were incubated
at C after dilution with 5ml30f 0.15M2NaCl.
0 +*********+*********+***-*****+ ***. .....+.........'.. ................4........4........4.......
0 2 4 6 8 10 12 14 16 18 20
Figure 19. Time course for the exchange of [ I]-labeled apoproteins
from [125I]-HDL3 into HDL2. 3mg HDL protein/ml each subclass.
/ 3900; +
O 36 72 108 144 180 216 252 288 324 360
of the initial slope at 2300 dpm/hr out of a total of 90,000 dpm gives
an exchange rate of approximately 2.5%/hr. These calculations are based
upon two separate experiments utilizing concentrations of llDL.2 and
HDL3 companrable to plasma levels (3mng/ml eachc subclass). Th'le rate
can be altered by varying the initial concentrations of either or both
Effect of Bromide lon on Protein Exchange
If the chaotropic Br ion is included in the.incubation mixture at
37PC, the amount of exchange in a given time period is greater when
compared to controls with no Br Figure 20 shows the fractionation
pattern obtained when 2M1 KBr is included in the 7 hour incubation solu-
tion as compared to controls. A significant increase in exchange can
be seen. As chaotropic ions are known to effect protein conformation
and association (93), these factors may be involved in the exchange
SDS-Polyacrylamide Gel Analysis of Exchanged Proteins
Gel electrophoresis of the recipient HDL subclass, into which
labeled apoproteins have been transferred, indicates that ApoA-1,
ApoA-2, and the C-peptides are all exchanged. Figure 21 shows the
radioactivity pattern obtained when RDL2, into which labeled apopro-
teins have been transferred by incubation with [ 2I]-HDL3, is run on
SDS-PAGE. The radioactive bands were identified by staining and
scanning duplicate gels.
If the labeled HDL subclass is subjected to limited crosslinking
with a 10-fold molar excess of DFDNB before mixing and incubation
Figure 20. Effect of KBr on the exchange of apoproteins between
[125I]-HDL3 and HDL2
Distribution of label after enetrifugation to separate
HDL2 and HDL3. 6 = [125I]-HDL3 incubated with HDL2;
s = -HDL3 incubated with HDL2 in the presence of
10000 + e *
0 ****+**********+ **** + **********+ ****+ ******** *****
100 2,B / 0 2 1 1 8 2
Figure 21. Polyacrylamide gel analysis of exchanged apoproteins.
SDS-PAGE of HDL2 after incubation with [ I]-HDLa
followed by reseparation. The gel was sliced it
2mm slices and the slices counted.
0 4 8 12 16 20 24 28 32 36 40
GEL SLICE NUMBER
with cold HDL, the labeled apoproteins that are exchanged are those
that were not crosslinked. Figure 22 shows the gel pattern obtained
for crosslinked [12I]-HDL3 alone, and for HDL2 that was labeled by
exchange of apoproteins from crosslinked -HDL Although a
significant portion of the HDL3 apoproteins were crosslinked into
higher molecular weight products, only uncrosslinked A-1 and A-2
appear to be exchanged.
Interpretations of Protein Exchange Experiments
The transfer of labeled apoprotein from one HDL subclass to
another does not proceed with a loss of protein content of the orig-
inally labeled fraction. The specific activity of that fraction
decreases but the protein content remains unchanged. This is con-
sistent with an exchange process in which there is a one to one exchange
of protein of one subclass with the other. Such exchange could occur
by the transfer of single polypeptide apoproteins, i.e., A-1 or A-2;
or by the exchange of complexes of apoproteins such as 2A-1:1A-2.
These experimental results do not permit us to rule out either of these
possibilities. It might be suggested that since crosslinking inhibits
the exchange, single polypeptide chains are involved. However, the
crosslinking may also prevent a conformational change that might be
necessary for the exchange process. That a conformational change
might be involved is suggested by the fact that when high concentra-
tions of the chaotropic ion, Br are included in the incubation mixture,
increases in the rates of exchange are observed. An actual exchange
involving a conformation change is also suggested by the finding that
crosslinking the unlabeled subclass also greatly inhibits the exchange
Figure 22. Polyacrylamide gel analysis of exchanged apoproteins:
Effect of crosslinking with DFDNB.
SDS-PAGE of HDL2 reisolated after incubation with [ IJ-HDL3
crosslinked with DFDNB. Q = [125I]-HDLgL crosslinked with a3
25-fold molar excess of DFDNB; 6 = HDL2 after incubation with
240 + I
*:~ ~ h,
0~~t +* *******.....+ .... ....... s
O 4 8 12 16 20 24 28 32 36 40
GEL SLICE NUMBER
process. If a simple transfer of peptide into the recipient class
occurred, one would not expect crosslinking to have much effect.
If HDL2 is made up of two identical subunits as suggested by
Friedberg and Reynolds (54), the transfer of label might be visualized
in terms of the exchange of RDL3 for one of the HDL2 subunits.
Although this is a reasonable possibility, one must again consider the
possibility of a conformational change to explain the results of the
Even though the two RDL subclasses can exist as stable separate
entities, when they are present together in solution, significant inter-
action between particles may occur. Exchange of apoproteins could
possibly occur between HDL2-HDL2 or HDL3-HDL3, as well. as between HDL2
and RDL3 molecules. If collisional complexes cause exchange and single
polypeptides are capable of exchange, then one would expect HDL -HDL3
exchange to occur. Collisional complex formation should also result
in the rapid exchange of lipids of all types due to their small size
and the fusion required for peptide exchange.
Lipid Exchange Between HDL2 and HDL, in Solution
Whcn IlDL3 Is Ioledcir withl [310j-chleslctcorl, [3Hl]-cholesterol panlm--
tate, or (14C]-phosphatidyl choline, thecn incubated with cold HDL2, as
withi peptide exchange; label is transferred to HDL2. However, even
under conditions of low temperature, dilution, and no incubation, total
exchange occurs during the ultracentrifugal reseparation of R1DL2 and
HDL3. This rapid movement prevents the study of the exchange using the
experimental approach that was possible for peptide exchange. For this
reason an alternative approach was developed.
Binding of Labeled HDL to Sepharose
The rapid rate of lipid exchange places one requirement on the
method employed to study this exchange: the separation of HDL3 and
HDL2 'must be done very quickly. As use of the ultracentrifuge requires
at least 24 hours even at 60,000 rpm, it was decided to approach the
problem by immobilizing one of the subclasses by covalently binding
it to Sepharose. This would allow separation by filtration; a process
that would take less than one minute.
When (12I]-HDL3 was reacted with commercially obtained CNBr-acti-
vated Sepharose 4B (Pharmacia) under the coupling conditions described
in the Methods sections, a Sepharose preparation containing approxi-
mately 0.5mg protein per ml of packed Sepharose was obtained. To
determine what percentage of the protein was bound to the Sepharose, a
portion of the gel was washed with 1% SDS solution. Under these
conditions any protein chains not covalently bound to the gel should
be washed off. Table 3 shows the results of such an experiment. Using
the Pharmacia CNBr-activated Sepharose, no more than 30% of the label
could be removed with SDS. Assuming only 3 polypeptide chains/HDL3
molecule, this indicates that an average of two are linked to the
Sepharose. To obtain a preparation with a lower level of binding, it
was necessary to activate our own Sepharose 4B using the conditions
described in the Methods section. Under these conditions, preparations
with over 50% of the protein removable with SDS could be obtained. The
same conditions also gave acceptable levels of binding for HDL2 prepa-
rations (20-40% bound). For all peptide exchange experiments,
preparations with low levels of binding were used. For lipid exchange