| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
,
Departments of
*
Biochemistry,
Microbiology and Immunology, and
Cell Biology, Albert Einstein College of Medicine, Bronx, NY 10461;
The Jackson Laboratory, Bar Harbor, ME 04609;
¶ Department of Molecular and Cell Biology, University of California, Berkeley, CA; and
||
The La Jolla Institute for Allergy and Immunology, San Diego, CA 92121
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
An equally remarkable facet of the H13 system is that the allelic H13 peptides do not conform to the conventional MHC motif paradigm (3, 4, 5). MHC allelic variation gives rise to differential peptide binding due to the presence of polymorphic residues, which give distinct chemical and size characteristics to six pockets (A-F; Refs. 6 and 7) within the peptide binding groove. Allele-specific binding motifs were initially revealed by sequencing peptides eluted from MHC molecules (8, 9). For example, peptides eluted from Db exhibit XXXXAsnXXXMet/Leu, where X is any amino acid. Both crystallographic and biochemical analyses support the importance of the centrally positioned P5 asparagine (P5Asn) side chain in stabilizing the pDb complex by forming hydrogen bonds (H-bonds) with C pocket residues Gln70, Gln97, and Tyr156 in the Ag binding domain (4, 10). Strikingly, P5 of H13 peptides is glycine (2), which cannot make the canonical contacts or occupy the C pocket of the Db molecule. To understand how H13 peptides interact and stabilize Db in the absence of the P5 anchor residue and how T cells distinguish self from nonself based on such subtle antigenic peptide differences, we describe high resolution crystal structures of allelic H13 peptide/Db complexes, as well as compare them with the structure of a pDb complex formed with a peptide modified to contain the canonical asparagine anchor residue.
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Using a method adopted from Young et al. (10),
Escherichia coli inclusion bodies of
H2-Db and
2-microglobulin were separately denatured in
8.0 M urea and 20 mM Tris, pH 8.0, and were separately mixed with the
synthetically prepared peptides representing the H13 peptides:
H13a, (SSVVGVWYL),
H13b, (SSVIGVWYL), and the P5 analog (SSVVNVWYL)
in a mass ratio of 3:1:0.5 in 6 M urea and 20 mM Tris, pH 8.0. These
mixtures were refolded by dialysis against 10 mM Tris, pH 8.0, at 4°C
in Spectra/Por CE dialysis tubing (MWCO 500) for 48 h. The
dialysate was subsequently centrifuged at 15,000 x g
and concentrated (30,000 MWCO; Amicon, Beverly, MA). The supernatant
was chromatographed in a buffer containing 20 mM Tris and 150 mM NaCl,
pH 8.0, on a Superdex75 column (Pharmacia Biotech, Uppsala, Sweden).
Fractions containing the highest concentration of
pDb complexes were pooled from multiple runs,
dialyzed against water, and concentrated (30,000 MWCO; Centricon,
Bedford, MA). The synthetic peptides were produced by the
Peptide Synthesis Facility of the Albert Einstein College of Medicine
and Research Genetics (Huntsville, AL) using solid-phase F-moc
chemistry, purified by HPLC, and confirmed by mass spectrometric
analysis.
Crystallization of the H2-Db complexes
Crystals were grown using the sparse scan technique (11) with the hanging drop vapor diffusion method. Large single crystals of SVL9/Db were produced in 12% polyethylene glycol 4000 and 0.1 M HEPES, pH 7.0 from drops incubated at 18°C. SIL9/Db and SVNL9/Db crystals were produced in identical conditions except the crystallization buffer also contained 0.2 M sodium acetate. Crystals were observed within 1 day.
Data collection
Crystals were transferred stepwise in reservoir solutions
modified to contain up to 20% glycerol as cryoprotectant buffers, and
flash cooled at 100 K. Data sets were collected at the X9B beamline at
the Albert Einstein Center for Synchrotron Biosciences (National
Synchrotron Light Source, Brookhaven National Laboratory, Upton,
NY) with a MARCCD detector. Data were indexed, scaled, and
merged using the programs DENZO and SCALEPACK (12). All
crystals belong to the monoclinic space group C2 with similar unit cell
dimensions. Data collection and statistics are reported in Table I
. The structure of
H13a peptide SVL9 (SSVVGVWYL) bound to
H2-Db was solved by molecular replacement with
the program AmoRe (13) using the crystal structure of
H2-Db complexed to an influenza virus peptide,
Np9, (ASNENMETM; Ref.10 ; Protein Data Bank identification
code 1hoc) as a model. The 
12 domains forming the Ag binding
cleft, the 3 domain, and
2-microglobulin were separately fitted
by rigid body refinement in XPLOR was subsequently subjected to cycles
of anisotropic and bulk solvent correction, simulated annealing
refinement, individual B-factor refinement using the program CNS
(14) and model adjustment and rebuilding using the program
O (15). The structures of SIL9 and SVNL9 were
solved by Fournier methods using SVL9 and SIL9, respectively, as
starting models. Analysis of the final models show that they have
characteristics of well-refined structural models with low values for
Rcryst and
Rfree and good geometry
(16). Coordinates have been deposited in the Protein Data
Bank (1INQ).
|
Solvent accessibility statistics were calculated with the
program DSSP (17) and the Protein-Protein Interaction
Server (18). [/
] and 
angles were calculated with
the program VADAR. H-bond and van der Waals contacts were calculated
with the programs HBPLUS (19) and CONTACSYM
(20), respectively.
Peptide binding assays
The binding of peptides to Db molecules was assessed by measuring the amount of peptide required to inhibit by 50% the binding of a known radioligand to soluble purified Db, as previously described (21).
H13 minigene expression constructs
To determine the effect of amino acid substitutions in the H13 peptide on T cell recognition, minigene constructs were prepared as described (2, 22). Briefly, the minigene constructs were produced to encode Met-SSV(X)GVWYL, Met-SSVVG(X)WYL, or Met-SSVGVY(X)L in which (X) is a random amino acid substitution of the core SSVVGVWYL H13a peptide.
Cell lines, T cell activation, and cytotoxic assays
Ag-specific responses of T cell hybrids were determined by the
production of -galactosidase (lacZ) activity
(23). The H13a-specific hybrid
30NX/B10Z and the H13b-specific hybrid B/NXZ have
been described (2). In brief, 3–10 x
104 cells were cocultured overnight in duplicate
with 2–5 x 104 LMtk cells cotransfected
with H13 minigene constructs, Db, and B7.2 cDNA.
The peptide/MHC-induced T cell response was assayed as lacZ
activity using the substrate chlorophenol red
-D-galactopyrannoside. The conversion of this
substrate to chlorophenol red was measured at 595 and 655 nm as a
reference wavelength with a 96-well microplate reader (Bio-Rad,
Richmond, CA). To generate bulk anti-H13a
CTLs, B10.CE(30NX)-H13b/Sn mice
(maintained at The Jackson Laboratory, Bar Harbor, ME) were primed
twice with 2 x 107 spleen cells from
H13a male C57BL/10Sn (B10) mice, and were then
restimulated for 5–6 days with spleen cells from 2000 rad-irradiated
female B10 mice in DMEM medium supplemented on day 3 of culture with
10–30 U/ml rIL-2 as described (24). To generate bulk
anti-H13b CTLs, female B10 mice were primed
similarly with cells from
B10.CE(30NX)-H13b/Sn male mice and were
restimulated in vitro with cells from female
CE(30NX)-H13b/Sn mice. Specificity of the
CTLs for naturally presented H13a and
H13b minor H Ags, respectively, was confirmed by
conventional cell-mediated lympholysis analysis of female
B10.CE(30NX)-H13b/Sn and B10 lymphoblast
target cells (data not shown). The standard 51Cr
release cell-mediated lympholysis assay as described previously was
used for assessment of cytotoxic activity against synthetic peptide
pulsed target cells (2). The target cells were T2 cells
transfected to express H2-Db, kindly provided by
P. Cresswell (Yale University, New Haven, CT). Synthetic peptide-pulsed
T2-Db target cells were prepared by adding 5
x 104 51Cr-labeled target cells to V-bottom
microtiter wells carrying varying peptide concentrations for 1 h.
Effector cells were then added at E:T ratio of 10:1 or 20:1 and were
incubated for 4 h. The percentage of specific lysis was calculated
from the amount of 51Cr released into the culture
supernatant and is the mean of triplicate cultures.
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
To investigate the question of how noncanonical peptides
compensate for the lack of a conventional anchor residue, we solved
crystal structures of Db in complex with both
allelic forms of H13: H13a (termed SVL9:
SSVVGVWYL) and H13b (SIL9: SSVIGVWYL; Table I
). In addition, we determined the structure of
Db complexed to a peptide analog of
H13a in which we have imposed a
P5Asn anchor residue (SVNL9: SSVVNVWYL). Omit
electron density maps for the regions corresponding to the H13 peptides
are shown in Fig. 1.
|
and 2), which form H-bonds to C pocket side
chains (Gln70, Gln97, and
Tyr156). These interactions stabilize the
architecture of the groove and enable the anchorless H13 peptides to
assume backbone conformations characteristic of conventionally anchored
peptides.
|
A in which the
H13a peptide, SVL9, is superimposed on a
Db-bound peptide that contains the
P5Asn anchor residue, SEV9 (FAPGNYPAL; Ref.
23). The MHC binding strategy used by the H13 peptides is
distinct from that used by a previously described peptide, p1027
(FAPGVFPYM; Ref. 24), which binds Db
despite the lack of a P5Asn anchor by bulging in
the central region of the peptide to allow the
P6Phe side chain to serve as a new anchor (Fig. 3B). These results demonstrate that noncanonical peptides
are capable of adopting multiple conformations to accommodate MHC
binding, importantly including the conventional conformation as
demonstrated by both allelic forms of H13 (Fig. 3, A and
C).
|
The only difference between the allelic H13 peptides is the
presence of either P4Val or
P4Ile. As shown in Fig. 4, the P4Ile side
chain of SIL9 assumes a similar side chain rotamer conformation as
P4Val in SVL9, with the additional methyl group
(C
1) positioned directly between the 1 and 2 helices of
Db. P4Ile in the SIL9
peptide is more exposed to solvent than P4Val in
SVL9, and is predicted to protrude directly into the central portion of
the TCR-binding interface (Fig. 4A; Refs.
27, 28, 29). The P4Val 
P4Ile substitution results in only minimal
alteration in the P4 side chain at a site readily accessible to TCR
contact.
|
, C and D). Second, a H-bond between
Glu163 and the hydroxyl group of
P1Ser is observed in H13b
(SIL9/pDb), but not in H13a
(SVL9/pDb). This H-bond is also found in SVNL9
(Fig. 4C) and promotes
formation of a salt bridge across the Ag binding cleft from
Lys66 of the 1 helix to Glu163 of the 2
helix (Fig. 4C). The variability observed at
Glu163 in pDb crystal structures
(Fig. 4C) suggests conformational plasticity at this
site. Mapping sites critical for T cell recognition of H13
Analysis of solvent accessibility indicates that among the H13
peptide residues, P4Val/Ile,
P6Val, P7Trp, and
P8Tyr are most accessible to the TCR
(17). The relative contribution of these amino acid side
chains to T cell recognition and activation was analyzed. First,
LMtk- cells expressing Db
cells were transfected with H13 minigenes encoding H13 peptides that
have incorporated random substitutions at solvent accessible residues.
T cell triggering by anti-H13a and
anti-H13b T cell hybridomas was assayed to
determine which mutations were tolerated by the hybridoma TCRs (Fig. 5). The anti-H13b hybridoma tolerated only
Ile at P4 (Fig. 5B), demonstrating a stringent requirement
for the C1 atom of P4Ile in the natural SIL9
ligand. In contrast, the anti-H13a hybridoma
showed a less stringent specificity, as it tolerated multiple
substitutions at P4 (Fig. 5A). We also examined the effect
of substitutions at P6–8, which bulge out of the peptide binding
groove to traverse the hydrophobic ridge in the
Db cleft. Substitutions at P7 and P8, the most
solvent exposed positions of the H13 peptides, were not tolerated,
whereas a wide array of P6 substitutions were tolerated (data not
shown). Therefore, side chains of P4, P7, and P8, but not P6 are
essential components in the binding of both
anti-H13a and
anti-H13b TCRs.
|
The identification of antigenic T cell epitopes and the design of agonist and antagonist ligands have in large part relied on the elaboration of allele-specific binding motifs and the definition of anchor residues (30, 31, 32, 33, 34). Immune responses to tumor Ag peptides lacking anchor residues have been reported to be enhanced by modifications that provide a surrogate anchor (35, 36). However, the extent to which peptides with imposed anchors improve antigenicity remains uncertain. The H13 model provides a sensitive and robust system to investigate this issue. pMHC complexes rendered stable by the imposition of anchor residues would be expected to enhance antigenicity, presumably by increasing cognate pMHC density on the plasma membrane. Alternatively, peptides modified so that they have improved anchors could alter the conformation of the peptide, and in doing so, disrupt TCR binding. To test these possibilities, the effects of adding a central anchor residue to H13 peptides on MHC binding and T cell recognition were analyzed. Analysis of the crystal structures of the anchorless and anchored pDb complexes then allowed for the correlation of functional parameters with structural alterations.
Consistent with a role for the conventional P5 anchor in pMHC
stabilization, replacement of the native P5Gly of
SVL9 with a P5Asn increased its affinity for
Db 
10-fold (Table I). This increased affinity
is consistent with the crystal structure of SVNL9, showing that the
P5Asn side chain participates in prototypic
H-bonds with amino acid side chains of C pocket residues
Gln70, Gln97, and
Tyr156 (Fig. 2D).
To determine the consequence of P5Asn
substitutions on CTL recognition, CTLs generated after reciprocal H13
congenic strain immunization were analyzed. As previously shown
(2), allelic discrimination by normal CTLs occurs within a
remarkably narrow peptide concentration range: 1–100 pM for
anti-H13a CTLs and 1–1000 pM for
anti-H13b CTLs (Fig. 5, C and
D). H13 peptides substituted with
P5Asn anchors were no more active than the
naturally occurring SIL9 and SVL9 peptides in this assay (Fig. 5, C and D).
The P5Asn-substituted peptides actually inhibited
recognition by anti-H13a CTL (Fig. 4C), suggesting that the improved anchors induced structural
changes that modified the complementarity of the
anti-H13a TCR with the pMHC interface to
outside the biologically optimal range. Overall, the imposed
P5Asn anchor pulls the central portion of the
SVNL9 peptide backbone more toward the 2 helix compared with SVL9
and SIL9 (0.6 Å at the C atoms of P4). In addition, the imposed
anchor induced a change in the rotamer conformation of
P4Val (Fig. 4B), the key TCR contact
residue in the H13 response.
It is not clear how TCR binding affects the conformation of the
anchor-imposed SVNL9 peptide. Because structures of TCR complexed to
pMHC have shown that TCR binding can alter the peptide conformation
(37, 38, 39, 40), TCR binding may induce a conformational change
in SVL9. One explanation that may account for the inhibited T cell
response to SVNL9 is that the main chain of SVNL9 is restricted from
undergoing a conformational change induced by TCR binding. Despite
observed and proposed structural changes, it is important to note that
allelic discrimination, i.e. the ability of the CTLs to discriminate
allo-P4Ile from self-P4Val
substitutions, was preserved in the context of the
P5Asn-modified H13 peptides. This is supported by
the ranges of concentrations required for inducing the CTLs, which
demonstrate that self and allo agonist activity was minimally affected
by the P5Asn substitution (Fig. 5, C
and D). In the case of anti-H13a
CTLs, the P5Asn modification did not abrogate
allelic discrimination, but rather acted additively to increase the
concentration required to achieve equivalent levels of cytolysis. These
results suggest that structural changes induced by the
P5Asn substitution are independent of those
involved in allelic discrimination.
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Minor H Ags are allelically variant self peptides that pose a
serious clinical concern in organ and bone marrow transplants, even
under conditions of an MHC match (41). An increasing
number of minor H Ags are being identified at the molecular level, but
none have been examined at the structural level (42). This
characterization of Snell’s classical H13 Ag provides a model for
understanding how the most subtle molecular changes are sensed by T
cells, ultimately resulting in transplant rejection. The topologies of
the H13 pMHC complexes are remarkably similar, with the most notable
difference being the extension of the P4 side chain by a single methyl
group in SIL9 compared with SVL9. Modeling of the 2C TCR juxtaposed
onto H13 peptides suggests that while a large surface area engages the
TCR CDR loops, CDR3 is positioned for direct contact with the
terminal end of the P4 side chain (40). Allelic
discrimination was preserved by anti-H13 CTLs, even in the context
of modest reorganization of the peptide conformation imposed by the
substitution of a P5Asn anchor (Fig. 5, C and D), suggesting that CDR3 has sufficient
flexibility to bind the P4Val side chain
repositioned by the P5Asn substitution, and in
doing so, trigger the TCR. X-ray crystallographic studies of agonist
and superagonist pKb ligands for the 2C11 TCR
suggested that the ability to form H-bonds between CDR3 loops and a
critical peptide side chain is necessary to induce optimal TCR
triggering (40). The presence/absence of a single methyl
group may be an even more subtle trigger in which a small number of van
der Waals contacts between the TCR CDR3 and the P4 side chain forms
the basis of H13 allelic discrimination.
However, it is also possible that TCRs specific for
H13a recognize SVL9 in a way that alters the
peptide conformation, which can occur with valine and other side chains
(Fig. 4A), but not isoleucine. Structures of TCR complexed
to pMHC have shown that TCR binding can distort the peptide
conformation (37, 38, 39, 40). The crystal structures of H13
specific TCRs bound to respective ligands may assist in distinguishing
these possibilities.
Anchor imposition decreases antigenicity by causing conformational changes in a key TCR contact residue
Allele-specific MHC motifs have been a lynchpin of antigenic epitope definition and peptide-based vaccine design. Previous studies have all been consistent with obligate contacts between central anchor residues and MHC, either by direct interaction or by water-mediated H-bonds. Water molecules are frequently observed to fill cavities at molecular interfaces such as the TCR/pMHC interface (43) and the peptide/MHC interface (25). Because the loss of a single H-bond can decrease affinity by several orders of magnitude (44), direct or water-mediated H-bonds between anchor residues and MHC have been previously thought to be necessary for display of the characteristic peptide backbone configurations for each specific MHC allele (6, 7). However, some naturally occurring peptides, including the non-MHC encoded H Ags H13 (2) and H47 (45), are immunogenic despite the lack of a canonical central anchor residue. A clearer understanding of how such nonmotif peptides accomplish this task is a matter of considerable importance. In this report, we show two unprecedented features of the naturally occurring nonmotif H13 peptides (containing P5Gly instead of P5Asn): neither direct nor water-mediated H-bond interactions bridge anchor residues and MHC; and they assume the prototypic backbone structure for Db motif-bearing peptides. Thus, our studies provide evidence that interactions between central anchor residues and MHC side chains are not necessary for Ag presentation to T cells. In addition, the ability of water to stabilize the C pocket without the assistance of a peptide side chain has implications on the initial formation of pMHC complexes in the endoplasmic reticulum as well as intra- and extracellular peptide exchange.
Innovative insights into unresolved problems associated with
modification of T cell epitopes are provided by the structure and
function of an anchor-modified minor H Ag (Figs. 1D,
2B, 3, B and D, 4, and 5B).
The extent to which peptides can be improved as immunogens by
modification of anchor residues is a significant unresolved issue.
Although vaccination of melanoma patients with an anchor-imposed
peptide derived from the gp100 melanoma-associated Ag,
gp100209–2 M, was reported to be significantly
more efficient in generating clinical responses to melanoma in clinical
trials (46), a study assessing the effect of modified
anchors on melanoma-reactive CTL reported that only 2 of 47 modified
peptides actually increased binding, immunogenicity, and recognition by
established CTL lines (47). X-ray crystallographic
analysis of the H13 SVNL9 pDb complex indicates
that the P5Asn side chain is engaged in
prototypic H-bond contacts with the Db C pocket,
which can account for the increased stability of the SNVL9
pDb complex compared with that of the SVL9
pDb (Table I). However, despite
substantially increased binding, structural changes induced by the
P5Asn anchor not only failed to increase agonist
activity, but for SVL9 pDb, the
P5Asn substitution went so far as to reduce
agonist activity of the SVL9 pDb complex (Fig. 4C). Our results provide the first direct structural
evidence that anchor imposition decreases antigenicity by causing
subtle conformational changes in critical TCR contact residues of the
peptide. Careful consideration of the consequences of anchor residues
and water in peptide-MHC interactions should assist in epitope
identification and vaccine design.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
2 Address correspondence and reprint requests to Dr. Derry Roopenian, The Jackson Laboratory, 600 Main Street, Bar Harbor, ME 04609. E-mail address: dcr@aretha.jax.org; or Dr. Steven C. Almo or Dr. Stanley G. Nathenson, Albert Einstein College of Medicine, 1600 Morris Boulevard, Bronx, NY 10461. E-mail addresses: almo{at}aecom.yu.edu or nathenso{at}aecom.yu.edu
3 Abbreviations used in this paper: H, histocompatibility; H-bonds, hydrogen bonds; SVL9, SSVVGVWYL; SIL9, SSVIGVWYL; SVNL9, SSVVNVWYL.
Received for publication August 21, 2001. Accepted for publication October 26, 2001.
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
T cell receptor structure at 2.5 A and its orientation in the TCR-MHC complex. Science 274:209.This article has been cited by other articles:
|
|
 |
|
  T. Kanaseki and N. Shastri Endoplasmic Reticulum Aminopeptidase Associated with Antigen Processing Regulates Quality of Processed Peptides Presented by MHC Class I Molecules J. Immunol., November 1, 2008; 181(9): 6275 - 6282. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 V. Mitaksov and D. H. Fremont Structural Definition of the H-2Kd Peptide-binding Motif J. Biol. Chem., April 14, 2006; 281(15): 10618 - 10625. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
T. Takaki, S. M. Lieberman, T. M. Holl, B. Han, P. Santamaria, D. V. Serreze, and T. P. DiLorenzo Requirement for Both H-2Db and H-2Kd for the Induction of Diabetes by the Promiscuous CD8+ T Cell Clonotype AI4 J. Immunol., August 15, 2004; 173(4): 2530 - 2541. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
F. R. Santori, K. Holmberg, D. Ostrov, N. R. J. Gascoigne, and S. Vukmanovic Distinct Footprints of TCR Engagement with Highly Homologous Ligands J. Immunol., June 15, 2004; 172(12): 7466 - 7475. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
H. W. H. G. Kessels, K. E. de Visser, F. H. Tirion, M. Coccoris, A. M. Kruisbeek, and T. N. M. Schumacher The Impact of Self-Tolerance on the Polyclonal CD8+ T Cell Repertoire J. Immunol., February 15, 2004; 172(4): 2324 - 2331. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
R. Yadav, Y. Yoshimura, A. Boesteanu, G. J. Christianson, W. U. Ajayi, R. Shashidharamurthy, A. K. Stanic, D. C. Roopenian, and S. Joyce The H4b Minor Histocompatibility Antigen Is Caused by a Combination of Genetically Determined and Posttranslational Modifications J. Immunol., May 15, 2003; 170(10): 5133 - 5142. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
B. Mommaas, J. Kamp, J.-W. Drijfhout, N. Beekman, F. Ossendorp, P. van Veelen, J. den Haan, E. Goulmy, and T. Mutis Identification of a Novel HLA-B60-Restricted T Cell Epitope of the Minor Histocompatibility Antigen HA-1 Locus J. Immunol., September 15, 2002; 169(6): 3131 - 3136. [Abstract] [Full Text] [PDF]
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
1 helix behind the plane of the page, the 
