The JI
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     
 

This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow Request Permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Hobby, P.
Right arrow Articles by Sutton, B. J.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Hobby, P.
Right arrow Articles by Sutton, B. J.
Right arrowPubmed/NCBI databases
*Substance via MeSH
Medline Plus Health Information
*Lupus
The Journal of Immunology, 1998, 161: 2944-2952.
Copyright © 1998 by The American Association of Immunologists

Molecular Modeling of an Anti-DNA Autoantibody (V-88) and Mapping of Its V Region Epitopes Recognized by Heterologous and Autoimmune Antibodies1

Paul Hobby*,{ddagger}, Francis J. Ward{dagger}, Andrew N. Denbury{dagger}, D. Gwyn Williams{ddagger}, Norman A. Staines{dagger} and Brian J. Sutton2,*

* The Randall Institute, Biomedical Sciences Division and {dagger} Infection and Immunity Research Group, Life Sciences Division, King’s College London, London, United Kingdom; and {ddagger} Renal Unit, Division of Medicine, United Medical and Dental Schools, Guy’s Hospital, London, United Kingdom


   Abstract
Top
 Abstract
Introduction
 Materials and Methods
 Results
 Discussion
 References
 
Anti-DNA autoantibodies are a characteristic feature of human systemic lupus erythematosus (SLE) and lupus diseases in the mouse. V-88 is an IgG1/{kappa} ssDNA-binding Ab, derived from a lupus mouse, that bears a cross-species, cross-reactive Id (CRI) that has been implicated in the pathogenesis of both human and murine disease. A linear epitope map of V-88 has been determined with anti-idiotypic antisera obtained from rabbits, and candidate sequences for the idiotopes of the CRI have been proposed. We now report the modeling of the three-dimensional structure of the V regions of Ab V-88, to map the location of these idiotopes. The V region framework structure was derived from those of crystallographically determined Ab structures, and the complementarity determining region (CDR) structures were based upon the set of canonical structures adopted by these loop regions in Abs of known structure. One of the idiotopes is an extensive, highly accessible epitope consisting of framework regions spatially adjacent to CDR2 in the heavy chain. Epitopes recognized by an anti-idiotypic rabbit antiserum were compared with those recognized by autoimmune sera from SLE-prone mice, and common features were identified. By analogy with the crystal structure of an anti-DNA Ab BV04-01 complexed with a trinucleotide, the modeled structure also suggests a mode of binding of ssDNA to V-88. The location of the candidate CRI, although within the framework region of VH, is such that it could influence Ag specificity.


   Introduction
Top
 Abstract
 Introduction
Materials and Methods
 Results
 Discussion
 References
 
Autoantibodies that react with DNA are a prime feature of human systemic lupus erythematosus (SLE) and the lupus diseases of MRL/Mp-lpr/lpr (MRL)3 and (NZB x NZW)F1 (BWF1) mice. They contribute to characteristic lesions of the kidney, skin, and brain by forming immune deposits, either by binding directly to the extracellular matrix or by combining with Ag. DNA-binding Abs of the IgM class with relatively low functional affinity are normal components of the Ab repertoire (1), but, in SLE, higher affinity IgG Abs are present in much greater amounts than in normal individuals (2). There is some evidence for functional Id control of anti-DNA Abs, but the relationship between physiologic and pathologic Abs, and their regulation, is not understood.

DNA-binding Abs are characterized by possession of public or cross-reactive Ids (CRI) defined by anti-Id Abs; many examples of such CRI have been recorded in studies of both human and murine Abs, and indeed the same CRI are expressed across species barriers. Thus each DNA-binding Ab possesses a distinctive set of idiotopes that are widely distributed not only on different anti-DNA Abs, but also on Abs that recognize other Ags. A CRI is not necessarily restricted to the Ab paratope, as may be the case with private Ids. Four crystal structures of idiotope-anti-idiotope complexes have been determined, and in all cases the private idiotopes consist predominantly of complementarity determining regions (CDR) and include the Ag binding site (3, 4, 5, 6). A comparison of VH gene sequences of Abs expressing Id 16/6 (a common CRI of DNA-binding Abs) led Young et al. (7) to conclude that a 5' sequence of the second CDR of the heavy chain (CDR-H2) encoded the Id. Such an observation can explain the heritability of some CRI and is in accord with findings that synthetic peptides corresponding to the VH sequences are immunogenic and can induce anti-Id Abs that react with the original native Id+ Ab (8, 9). Mutagenesis and mapping of substitutions onto the x-ray and modeled structures of anti-p-azophenylarsonate Abs have indicated that virtually all of the amino acid residues implicated as part of the idiotopes belong to the CDR (10, 11). In the cold agglutinin (anti-I/i) Ab system, the 9G4 Id (12) has been mapped to VH framework region 1 (FR-H1), but the implicated residues immediately precede CDR-H1 and could also influence Ag binding.

Linear peptide sequences may define continuous epitopes that are, in part, the component idiotopes of a CRI. A systematic approach to identify such epitopes has enabled us to construct linear Id maps of various DNA-binding autoantibodies (13, 14). The present study concerns the CRI of mAb (mAb) V-88, a class-switched (IgG1/{kappa}) and somatically mutated Ab, derived from a lupus BWF1 mouse, that binds to ssDNA (15) and expresses the 16/6 Id (16); the V-88 cross-species CRI (16) has been implicated in the pathogenesis of both human and murine disease (17, 18, 19). We have used anti-idiotypic Abs obtained from rabbits immunized with mAb V-88 (19) and from mice with lupus disease (20) in epitope scanning assays with synthetic peptides immobilized on pins, to create linear Id maps of V-88. These studies led to the identification of sequence similarities between the murine mAb V-88 and human Abs that also express Id 16/6 (16); these sequences may account in part for this common Id (19).

We now report the modeling of the three-dimensional structure of V-88 to investigate the spatial disposition of these epitopes. The crystal structures of a large number of Ab V domains have now been determined (21), revealing not only a conservation of structure within the framework regions, but also a pattern of canonical structures for five of the six CDR (22, 23, 24, 25, 26, 27). The model of V-88 draws on the crystal structure of the Fab fragment of an anti-DNA Ab, BV04-01, complexed with a trinucleotide (28), which permits prediction of the interaction between V-88 and DNA. The model also reveals the distribution of the epitopes recognized by rabbit and mouse anti-Id Abs on V-88 and identifies a major candidate structure for the CRI Id 88 that, although principally in a conserved framework region of the VH domain, could exert an influence upon Ag-binding specificity.


   Materials and Methods
Top
 Abstract
 Introduction
 Materials and Methods
Results
 Discussion
 References
 
Monoclonal Ab V-88 and its Fv sequence

The source, preparation and properties of this Ab have been described (15, 16, 19). It is an IgG1/{kappa} Ab, derived from an adult female BWF1 mouse, that reacts with ssDNA and, to a lesser extent, with RNA. The VH-88 gene is a member of the VH 7183 family and is thus a member of the third clan of VH families (27). The VL-88 gene is closely related to the K5.1 germline gene, a member of the V{kappa}1A subgroup.

Heterologous anti-Id sera

A rabbit anti-mAb V-88 antiserum was raised as described (19). A serum pool was also obtained from young normal rabbits. Each was absorbed against normal mouse serum IgG on a CNBr/Sepharose 4B column, and stored at -20°C in aliquots.

Autologous anti-Id sera

MRL and BALB/c (normal) mice were bled at 25 wk of age, and sera were stored at -20°C in aliquots.

Epitope mapping

The procedures used for the synthesis of immobilized peptides and epitope mapping by ELISA (pepscans) have been described (14, 19). Hexapeptides were constructed, each overlapping by five residues to span the VH and VL sequences of V-88 (with the exception of the first 24 residues of VH, which were not known at the time of epitope scanning, but which have been included in the modeling). The rabbit and mouse sera were applied in the assay and detected by anti-Ig species-specific reagents. The profiles display a small number of strong reactions and a larger number of weaker reactivities; a cut-off value of 3x the mean of the lowest 25% of values was taken to ensure that only the most significant peptides were mapped.

Modeling of V-88 Fv

V-88 heavy and light chain V region sequences were compared with those of Abs with structures deposited in the Brookhaven protein databank (PDB) using the NORMPAIRS and MULT-TREE programs in the multiple sequence comparison package MULTALIGN (29). NORMPAIRS provided sequence similarity scores, and MULT-TREE was used to align the sequences, delineate regions of structural conservation, and identify differences in FR and CDR length, as defined by Chothia et al. (23). Molecular modeling was performed on Silicon Graphics 4D30 IRIS and Indigo workstations running INSIGHT II and HOMOLOGY (Biosym Technologies/Molecular Simulations, San Diego, CA). All hypervariable loops except CDR-H2 and CDR-H3 were assigned using structures from the subsets of CDR canonical forms. For these two noncanonical loops, a conformational search algorithm (GENLOOPS, in HOMOLOGY) was employed (30). Conformers were considered if splice regions were acceptable, if C{alpha}-Cß bonds of the terminal residues aligned with those of the flanking template residues, and if backbone {phi} and {Psi} angles were acceptable. Minimal steric interaction with adjacent regions of the structure was the final criterion for selection. Energy minimization and molecular dynamics simulations were performed using DISCOVER (Biosym Technologies/Molecular Simulations). The program PROCHECK (31) was run between successive DISCOVER calculations to monitor main-chain and side-chain stereochemistry. Manual building and regularization of stereochemistry was conducted with FRODO (32) on an Evans & Sutherland PS390 graphics system, and the stereochemistry of the final structure was assessed by PROCHECK.

Epitope topology and accessibility

Solvent accessibilities were calculated with the program DSSP (33), using a probe radius of 1.4 Å. Epitopes identified by pepscan were displayed on the model of V-88 to assess their distribution and topology. The C{kappa} and C{gamma}1 domains of Ab 4-4-20 were included in the model to assess the accessibility of V region epitopes adjacent to the C domains.


   Results
Top
 Abstract
 Introduction
 Materials and Methods
 Results
Discussion
 References
 
Modeling of V-88 framework regions

The light chain of the anti-fluorescein Ab 4-4-20 (PDB code 4FAB) displays a remarkably high degree of sequence identity with V-88 (93%) throughout both FR and CDR and was selected as the template (Fig. 1Go). The heavy chain of 4-4-20 is also one of the best matches available for V-88 heavy chain (67% identity), and, to ensure compatibility in the mode of association between the heavy and light chain FR, the 4-4-20 structure was also taken as the principal template for the heavy chain. The anti-progesterone Ab DB3 (1DBB), which has an almost identical light chain to both 4-4-20 and V-88 but a different heavy chain, has an identical mode of VH:VL association to that of 4-4-20 (34), and the same is true for both the anti-ssDNA Ab BV04-01 (28) and the anti-ssRNA Ab Jel 103 (35). These structures support the choice of 4-4-20 as template for V-88 in both VH and VL.



View larger version (29K):
[in this window]
[in a new window]
  		FIGURE 1. Alignment of the heavy and light chain sequences of V-88 with those of the Abs used as modeling templates, 4-4-20 and BV04-01. The boxed regions indicate which of these two structures was used, and the boundaries of the FR and CDR are shown, together with the extent of the structurally defined CDR loops. Asterisks above residues in CDR-H2 and -H3 denote segments for which conformational searches were performed.

 
The light chain FR of V-88 were modeled entirely upon Ab 4-4-20 (Fig. 1Go). In the heavy chain, however, a contribution to the V-88 FR model, in the final few residues of FR-H3 and all of FR-H4, was taken from Ab BV04-01 (1CBV), since the sequences are virtually identical in much of this region (Fig. 1Go). Coordinates for this segment of V-88 were assigned from the structure of BV04-01, after local alignment with 4-4-20, and the splice point was chosen to lie in a region of very close structural similarity, adjacent to the cysteine residue (Cys92, Kabat numbering) of the conserved intradomain disulfide bridge.

Modeling of V-88 complementarity determining regions

Analysis of Ab V region crystal structures has revealed that for the three light chain CDR, and the first two heavy chain CDR, there is a limited repertoire of main chain conformations or canonical forms (22, 23, 24, 25, 26). Key residues both in the CDR loop structure and in the adjacent FR determine which canonical structure will be formed. A multiple sequence alignment of the heavy and light chains of V-88 with those of all Ab crystal structures deposited in the PDB allowed the CDR to be compared in both length and sequence and allowed their canonical forms to be identified. For canonical structures represented by more than one Ab in the structure database, the selected template was taken from the structure with greatest sequence similarity, subject to the proviso that the pre- and postloop geometry of the FR template structure (4-4-20 or BV04-01) and that of the structure from which the CDR was to be selected were compatible for splicing.

CDR-L1. Four canonical forms have been identified for this loop. The fourth of these, found in Ab 4-4-20, is the most homologous to V-88. All of the FR residues critical to this canonical structure are present, namely Val2, Ser25, Leu33 and Phe71.

CDR-L2. Only one canonical form is known for this very short loop. Ab 4-4-20 is identical in sequence to V-88 and was therefore taken as the template for this loop. The key residues Ile48 and Gly64 are present in V-88.

CDR-L3. There are three principal canonical structures for this loop, but the first accounts for over 80% of all the CDR-L3 sequences in the database, and it is to this group that V-88 belongs. It is identical to the sequence of 4-4-20 at six of the seven residues, and the key residue, Gln90, is also present.

CDR-H1. Three canonical structures have been identified for this loop, but the length, and presence of residues Met34 and Arg94, places V-88 in group 1. Because of the high structural homology that exists between the members of this group, it would be acceptable to assign coordinates from any of the Abs in group 1 (except for D1.3 and HyHEL-5, which do not have the same critical residues). Since a major consideration was to assign a loop with the best possible fit to the framework, CDR-H1 of 4-4-20 was taken.

CDR-H2. Four canonical structures have been observed for this loop, but V-88 does not match any of these in length. It is two residues longer than canonical form 1, one residue longer than forms 2 and 3, and one residue shorter than form 4. Indeed, at positions 52a, -b, and -c (Fig. 1Go), all Abs whose structures have been determined to date have either 0, 1, or 3 residues and conform to one of the four defined canonical structures. A conformational search was therefore performed for the sequence GDGGS (Fig. 1Go), as described in Materials and Methods. The possibility was considered that CDR-H2 conforms to either canonical form 3 or 4, since V-88 possesses the critical residue Arg71. This would require accommodating the insertion or deletion of one residue elsewhere, and the only feasible location would be in the following loop within FR-H3. However, there is no precedent for this among any of the Abs of known structure, and the virtual identity between the sequences of V-88, 4-4-20, and BV04-01 in the regions flanking CDR-H2 argues strongly against any alignment other than that shown in Figure 1Go.

CDR-H3. No canonical forms have been identified for CDR-H3, which displays considerable variation in both length, sequence, and structure. In V-88, CDR-H3 is relatively long (Fig. 1Go), and McPC603 (1 MCP) is the closest in length, at only one residue shorter. However, there is a striking sequence identity between V-88 and BV04-01 in the flanking part of FR3 and in FR4 (Fig. 1Go), and, since CDR-H3 in BV04-01 is only two residues shorter than in V-88, this was selected as the model for the loop. This choice is also supported by the fact that CDR-H3 in McPC603 adopts a conformation that is incompatible with the adjacent CDR-L1 of V-88, which was modeled on the structure of 4-4-20 (in turn identical to that of BV04-01). Selection of BV04-01 as the template for CDR-H3 ensures compatibility with CDR-L1. Coordinates were therefore assigned from BV04-01 to V-88 for the sequences CARL and AWFAY (Fig. 1Go), and a conformational search was performed for the intervening sequence PYYSNY.

V-88 model refinement

Steric overlaps that occurred between FR residues at the VH:VL interface were relieved by manual selection of alternative side chain conformers from the rotamer library in HOMOLOGY, with reference to the conformations observed for these residues in other structures. The side chains of only five core residues required adjustment in this way. The polypeptide backbone stereochemistry at the splice points was first screened by PROCHECK, and, where necessary, manual adjustment and regularization of the stereochemistry in FRODO was followed by the splice repair protocol (INSIGHT), in which the structure is tethered except for the splice region during energy minimization (100 cycles of steepest descent in DISCOVER). This was necessary only for certain of the splice points in loop regions and was not required for the single framework splice point in FR-H3. The energies of the loop regions were then minimized ("Relax" option in DISCOVER; 100 cycles of steepest descent minimization). The two loop regions for which conformational searches were performed (CDR-H2 and H3, as described above) were subjected to a dynamics/minimization protocol ("Explore" option in DISCOVER; temperature 300K). The final model was assessed by PROCHECK (data not shown), and the results of the tests were at least as satisfactory as those for crystallographically derived Fab structures, including BV04-01. The trace of the polypeptide chain in the three-dimensional structure may be seen later in Figure 5Go.



View larger version (31K):
[in this window]
[in a new window]
  		FIGURE 5. The V-88 model structure (C{alpha} backbone trace) showing residues (His27d and Tyr32 in CDR-L1; Trp100c in CDR-H3) that, in the structurally homologous Ab BV04-01, are involved in contacts with bound trinucleotide. Two of the thymine bases are shown in bold lines, superimposed upon the model of V-88 in the same orientation as found in the crystal structure of the complex with BV04-01. Two other residues (Tyr97 and Tyr98 in CDR-H3) are indicated. The stereo image is viewed down the pseudo twofold axis of the VL:VH pair.

 
Mapping pepscan epitopes onto the V-88 model

V-88 heavy chain. The linear epitopes defined on the V-88 heavy chain by rabbit anti-V-88 antiserum and MRL sera are compared with a solvent accessibility profile of V-88 in Figure 2Go. Rabbit anti-V-88 serum Abs reacted principally with peptides that mapped to CDR-H1, the FR-H2/CDR-H2 junction, and a cluster of sites in FR-H3 (Fig. 2Goa). The strong reaction with CDR-H1 was limited to the sequence SSYVMS. Only the first few of these residues are exposed, and the final residue (Ser35) is completely buried in the model at the VL:VH interface. A similar intensity and specificity of reaction was seen with the peptides EWVATI and VATISG at the FR-H2/CDR-H2 junction. The residues TISG represent the start of CDR-H2 as defined by Kabat, but the hypervariable loop of CDR-H2 begins at Gly52a (Fig. 1Go), and the preceding residues are predicted to be only poorly accessible (Fig. 2Gob). However, these two accessible regions of CDR-H1 and H2 lie adjacent to each other in the three-dimensional structure and form the patch designated a in Figure 3GoA. The third region of reactivity with rabbit Abs included the sequence SRDNAK, which is totally accessible; it corresponds to the FR-H3 loop, which lies adjacent to both CDR-H1 and CDR-H2 and includes the key residue Arg71, a determinant of the conformation of CDR-H2. This region, indicated as b in Figure 3Go, A and D, forms an extended epitope with patch a. Other weakly reactive peptides in this region, SSLRS/E/DTALY, map to an exposed ß-strand indicated as c in Figure 3Go, A and D.



View larger version (34K):
[in this window]
[in a new window]
  		FIGURE 2. a) The recognition of overlapping hexapeptides corresponding to the heavy chain sequence of V-88 by rabbit Abs. Each element of the histogram lies above the first residue of the hexapeptide. (Data are taken from Ref. 19 with permission). b, Calculated solvent accessibility of each heavy chain residue in the model of V-88. c, The recognition of overlapping hexapeptides corresponding to the heavy chain sequence of V-88 by MRL mouse Abs. Each element of the histogram lies above the first residue of the hexapeptide. The labeling of the reactive peptides corresponds to the epitopes shown in Figure 3Go; the peptide labeled "x" is occluded by the C domains in the Fab structure.

 


View larger version (51K):
[in this window]
[in a new window]
  		FIGURE 3. Mapping of epitopes, defined by pepscan studies, onto the three-dimensional model structure of the VH and VL domains of V-88. Heavy chain atoms are shown in blue, light chain atoms in gray, unless part of an epitope that is colored red or green (for heavy and light chain, respectively) in A, C, D, and F. In B and E, the six CDR are colored according to the code shown and encompass both the structural and hypervariability definitions of these regions (Fig. 1Go). A through C are views down the pseudo twofold axis of the VL:VH pair; Panels D through F are orthogonal views to those in A through C. A and D show the epitopes recognized by rabbit Abs, and C and F show the epitopes recognized by MRL mouse Abs. The letters refer to epitopes discussed in the text and indicated in Figures 2Go and 4Go.

 
The principal reactivity of the MRL Abs (Fig. 2Goc) corresponded to parts of FR-H3 that are similar to those recognized by the rabbit serum, including the SRDNAK segment, and are again labeled b in Figure 3Go, C and F. The considerably more extensive region labeled c in Figure 3GoF (cf Fig. 3GoD) reflects the additional reactivity with the peptides YPDSV/K/GRFTI of CDR-H2/FR-H3 (Fig. 2Goc). In fact the region defined by epitopes b and c may be even more extensive, since the N-terminal region of FR-H1, which was not included in the pepscan analysis, lies adjacent to these epitopes. One further reactive peptide, RQTPEK, following CDR-H1, can be seen in Figure 2Goc. The final three residues are solvent accessible (Fig. 2Gob), but they lie adjacent to the C{gamma}1 domain and would not be accessible to Ab; they are not visible in the orientation of Figure 3Go, either C or F.

V-88 light chain. Rabbit Abs reacted with peptides mapping to FR-L1, CDR-L1, FR-L3, and CDR-L3 (Fig. 4Goa). The last three segments, EHSNGY (CDR-L1), F/SGSGS/G (FR-L3), and THVPYT (CDR-L3), were mapped to accessible regions in the structure that form a contiguous patch labeled d and d' in Figure 3GoA (cf Fig. 3GoB). The remaining peptide, ASISCR in FR-L1, maps to a ß-strand that is visible in Figure 3GoD as patch e.



View larger version (42K):
[in this window]
[in a new window]
  		FIGURE 4. a, The recognition of overlapping hexapeptides corresponding to the light chain sequence of V-88 by rabbit Abs. Each element of the histogram lies above the first residue of the hexapeptide. (Data are taken from Ref. 19 with permission). b, Calculated solvent accessibility of each light chain residue in the model of V-88. c, The recognition of overlapping hexapeptides corresponding to the light chain sequence of V-88 by MRL mouse Abs. Each element of the histogram lies above the first residue of the hexapeptide. The labeling of the reactive peptides corresponds to the epitopes shown in Figure 3Go; the peptide labeled "x" is occluded by the C domains in the Fab structure.

 
The MRL Abs also reacted strongly with peptides that corresponded to FR-L3, again including FSGSGS (Fig. 4Goc); this can be seen as patch d' in Figure 3GoC. The weaker reactivity with the CDR-L1 peptide HSNGYT (Fig. 4Goc) corresponds to the patch d in Figure 3GoC. This is smaller than the corresponding epitope recognized by the rabbit serum (Fig. 3GoA) because there was no significant reactivity from peptides corresponding to CDR-L3. The reactivities of other Abs in the MRL serum, against peptides ISRVEA (in FR-L3) and KPGQSP (in FR-L2), map to regions far from the paratope, close to the C domains; indeed the latter peptide is predicted to be occluded by the CL domain. The former region is shown as patch f in Figure 3GoC but is almost entirely hidden at the "back" of the domain in Figure 3GoF.

Ag combining site of V-88

The Ag combining site region of the model was compared with that of the Ab BV04-01, the structure of which has been solved both free and as a complex with the trinucleotide dT3 (28). In this complex, the principal interaction is with the central thymine base, which is intercalated between Trp100a of CDR-H3, and Tyr32 of CDR-L1. Both of these residues are present, and in the same relative positions, in V-88 and can be seen in Figure 5Go. (In V-88, the structurally homologous Trp residue is numbered 100c). The structures of BV04-01, both free and complexed, and V-88 were superimposed to assess whether the trinucleotide could bind to V-88 in the same way. While the central and third thymine bases could be accommodated in the site of V-88, the first clashed with residues of CDR-H3. However, CDR-H3 in the uncomplexed BV04-01 also adopts a conformation that is incompatible with dT3 binding but then undergoes a substantial conformational change upon ligand binding (28). The predicted structure of V-88 CDR-H3 is very similar to that of the uncomplexed BV04-01, and, if it too underwent a conformational change similar to that of BV04-01, the trinucleotide could be accommodated completely within the binding site. In Figure 5Go, only the central and third thymine bases of the trinucleotide are shown, superimposed upon the modeled structure of V-88. His27d in CDR-L1 is also shown, another contact residue in the BV04-01 complex that is present in V-88 in the same position. Other residues nearby may be involved in binding, such as tyrosine residues 97 and 98 of the sequence YYSNY in CDR-H3 (Fig. 5Go). No positively charged residues are located close enough to be putative contacts for the phosphate groups. Only a single arginine residue (position 52) was found to make contact with the ligand in the BV04-01 complex, but this is absent in the CDR-H2 of V-88, which is shorter and of a different conformation.


   Discussion
Top
 Abstract
 Introduction
 Materials and Methods
 Results
 Discussion
References
 
In this study we describe the construction of a three-dimensional model of the VH:VL dimer of V-88, an IgG1/{kappa} anti-ssDNA Ab. The model has been used to visualize the distribution and topology of linear hexapeptide sequences of V-88, reactive with anti-Id 88 Abs raised in rabbits by immunization and with anti-Id Abs arising naturally in autoimmune MRL mice. This has led us to the identification of a candidate structure for the CRI of Ab V-88.

The framework regions of the model were based upon the crystal structure of the anti-fluorescein Ab 4-4-20 (4FAB), with four of the hypervariable loops modeled on known canonical structures. All three CDR in the light chain conformed to known canonical structures and, like the framework, were modeled on Ab 4-4-20. In the heavy chain, only CDR-H1 could be matched to a known canonical structure. CDR-H2 differed in length by at least one residue from the known canonical forms (22, 23, 24, 25, 26), and no Ab structure has yet been determined with this length of CDR-H2. The variability in CDR-H3 is so great that no canonical forms have been defined for this region. The structures of CDR-H2 and CDR-H3 in the V-88 model were therefore generated by conformational search.

In most of the Ag-binding loops, the number of residues that comprise the loop structure is not the same as in the CDR as defined by Kabat (see Fig. 1Go). The model, together with the results of solvent accessibility calculations, shows that some of the Kabat-defined CDR residues are too deeply buried to be accessible to Ag. Others have stressed that the role of some CDR residues is not only to participate directly in binding of Ag, but also to influence the structure of the rest of the loop (36, 37). A relevant example of this is the S107 anti-phosphorylcholine Ab, which attains specificity for DNA upon mutation of residue 35 in CDR-H1 from Glu to Ala (38). Although this residue is buried and cannot interact directly with Ag, it has a marked influence on specificity, and, in the crystal structure of the anti-phosphorylcholine Ab McP603, it appears to maintain the conformation of the binding site (39). We may expect therefore that the Id of V-88, like its Ag-binding specificity, is also influenced by CDR residues that do not contribute to the paratope surface.

V-88 shares several features with the anti-ssDNA Ab BV04-01 (40), for which crystal structures have been obtained for the unliganded and complexed form with dT3 (28). Both BV04-01 and V-88 possess the highly conserved V{kappa}1A light chain, which is also found in the anti-fluorescein Ab 4-4-20 and the anti-progesterone Ab DB3, with the same canonical loop structures and only minor differences in sequence. Virtually the same light chain as in BV04-01 (with one substitution) is also used by the anti-ssRNA Ab Jel 103, the structure of which has been determined as complexes with three mononucleotides (35). In BV04-01, two key residues Tyr32L and Trp100aH engage a thymine moiety in aromatic stacking interactions, and, in Jel 103, Tyr32L similarly interacts with the mononucleotide. Both of these residues are also present in V-88 in the same relative orientations and could therefore be involved in DNA binding (Fig. 5Go), especially if a conformational change occurs upon ligation, as seen in BV04-01. A tryptophan residue at this position (100aH in BV04-01, 100cH in V-88) is a common feature of murine anti-DNA Abs and is found in 16 of 53 sequences in which the V{kappa}1A light chain is used (Kabatman database; 41 . Another important contact residue in the BV04-01 complex, His27d in CDR-L1, is also present in V-88 (Fig. 5Go). The paratope of V-88 is rich in aromatic residues however, and Tyr97H and Tyr98H in CDR-H3 are particularly accessible (Fig. 5Go); they may also be involved in base stacking since V-88 binds predominantly to single-stranded polynucleotides.

Many other anti-DNA Abs are characterized by abundant cationic residues in the paratope, often clustered in CDR-H3 (42, 43). V-88 does not fit this pattern, perhaps reflecting the fact that recognition of ssDNA, in contrast to dsDNA, need not involve interaction with the phosphate groups if these are exposed, as in the model (Fig. 5Go). However, a histidine residue is present in each of CDR-L1 and CDR-L3, a lysine and an arginine in CDR-L2, and an arginine in CDR-H3. In the model, these lie on the paratope surface, and one or more of them could be involved in interactions with polynucleotides larger than dT3.

Epitope mapping revealed that rabbits make IgG Abs reactive with continuous sequence epitopes in the V regions of V-88. Solvent accessibility calculations show that only some of these epitopes are surface accessible. The epitope at the FR-H2/CDR-H2 junction (Fig. 2Goa) is an example of an Id that is apparently not surface accessible in the native molecule, suggesting that this region is not itself part of the 16/6 Id as had been suggested through gene sequence analysis (7). On the other hand, the model reveals that the epitopes recognized by the mouse Abs are largely surface accessible. There were quantitative differences between the anti-peptide Abs in MRL and BWF1 mice, with lower overall levels in the latter and without the larger amounts of IgG Abs that characterize MRL mice.

BALB/c mice made very much lower amounts of anti-peptide Abs and only of the IgM class. Thus it appears that the Abs in the lupus mice are made as a result of an active immune response, because of the class switching and the much greater signal to noise ratio in the epitope-scanning assays. The fact that IgG Abs such as V-88 can be made implies the involvement of T cells, so, in this regard, it is pertinent that T cells reactive against V region peptides are found in lupus mice, at least during a defined time window of disease development (44, 45, 46). They give recall proliferative responses to peptides in vitro, but, in normal mice (such as BALB/c), such responses are evoked only after intentional immunization. We conclude from this that mice are not tolerant to self V region epitopes and that B and T cells are spontaneously activated as part of the lupus disease pathogenesis. A role for B cells in presenting their own idiotopes as processed self Ig peptides can be imagined as can their ability to present processed peptides from the anti-Id Abs complementary to their own surface Igs. This could account for T cell activation, but whether it represents the primary lesion in lupus is not known.

Abs in the sera from immunized rabbits and lupus mice recognized a number of common structures, as can be seen by comparing Figure 3Go, A and D (rabbit sera) with Figure 3Go, C and F (mouse sera). The principal common structures lie in CDR-L1 (epitope d), FR-L3 (epitope d'), and FR-H3 (epitopes b and c). The latter may be even more extensive than that shown in Figure 3Go, since the first 24 residues of FR-H1, which lie immediately adjacent to regions b and c, could not be analyzed in this study. Other than these, the epitopes defined by Abs in these sera differed from each other. This demonstrates that the Id of an Ab is not an absolute structure but is a set of idiotopes described by the Abs that react with it. Consequently, heterologous anti-Id Abs may, or may not, define idiotopes that participate in endogenous regulatory Id interactions. There must, however, be many opportunities for idiotypic interactions between Abs, because the CRI clearly extends well beyond the paratope.

Most of the epitopes mapped onto the model of V-88 consist of polypeptide segments that are separated in the sequence. However, the ability of the peptide ELISA to detect Abs at relatively low functional affinity probably allows linear segments of such discontinuous epitopes to be identified. The location of some peptide idiotopes of V-88 correspond to the Id-determining regions (IDR) predicted by Kieber-Emmons and Kohler (47) on the basis of surface accessibility and relative lack of sequence conservation. The two epitopes in FR-H2/CDR-H2 (VATISG; see Fig. 2Goa and epitope a in Fig. 3GoA) and in FR-H3 (RDNAKS; see Fig. 2Goa and epitope b in Fig. 3Go, A-F) correspond to IDR B and D respectively, but IDR A predicted in FR-H1/CDR-H1 of V-88 is partially buried, although anti-peptide Abs against the sequence SSYVMS in CDR-H1 (Fig. 2Goa) were found. In the light chain, of the six predicted IDR, five coincided with peptide epitopes described in the present study. Some epitopes here did not coincide with predicted IDR, and thus surface accessibility and sequence heterogeneity are not the only criteria that determine the immunogenicity of V region structures. It is notable that some of the idiotopes defined by the synthetic peptides are located in framework regions. The Abs are therefore directed against structures that are widely expressed in different Abs, and this reinforces the notion that these are part of the CRI of mAb V-88.

From analysis of germline VH and VL gene sequences we conclude that a number of these epitopes are encoded by unmutated germline sequences in V88. This indicates that B cells reactive with self Ig V region epitopes are not deleted in ontogeny and that animals are not tolerant of their own idiotopes. This lack of tolerance extends beyond B cells because lupus mice have been shown to have T cells spontaneously activated against V region epitopes of this and other DNA-binding Abs (44, 45, 46, 48). Some of these epitopes, such as those in FR-L1, L3, and FR-H3, are recognized by both the mouse autoantibodies and the rabbit xeno-Abs. Comparison of mouse and rabbit germline sequences indicates that, although there is strong homology throughout the FRs, there are sequence differences in all of the epitopes that we have identified. Until we know the precise structures of the idiotopes and which amino acid residues are involved in Ab binding, it is impossible to know whether the rabbit Abs are recognizing foreign or self structures. However, the topographical coincidence of these auto- and xeno-reactivities implies a conservation of the idiotopes at the functional level.

Thus Abs from both rabbits and mice react with peptides mapping to the third framework regions of both light and heavy chains of V-88. The peptides of FR-H3 are of particular interest, especially those adjacent to the paratope (epitope b in Fig. 3Go, A-F). This prominent, solvent accessible region lies adjacent to the CDR1 and CDR2 loops of the heavy chain and may therefore contact Ag. Sequences in FR-H1 and FR-H3 characterize three clans of VH families (27). FR-H1 is highly conserved both in sequence and structure within each clan and, in this study, is not antigenic. FR-H3, however, does contain a major epitope for rabbit Abs (epitope b in Fig. 3Go, A and D) and several epitopes for MRL Abs (epitopes b and c in Fig. 3Go, C and F). Thus family-restricted motifs can contribute to CRI, and these are, in genetic terms, probably isotypic or allotypic structures (49). The FR-H3 region, which is a protrusive structure in the V-88 model, thus has some of the properties expected of a CRI: proximity to the paratope, influence on paratope conformation, and common distribution with minor variations in primary sequence.


   Acknowledgments
 
We thank Allen Edmundson for providing the coordinates of the BV04-01 complex structure before deposition in the Brookhaven databank, and Kate Kirwen for preparing the color figures.


   Footnotes
 
1 Supported by the Arthritis Research Campaign (U.K.) and the Special Trustees of Guy’s Hospital, London (to P.H.). Back

2 Address correspondence and reprint requests to Dr. Brian J. Sutton, The Randall Institute, King’s College London, 26–29 Drury Lane, London WC2B 5RL, U.K. E-mail address: Back

3 Abbreviations used in this paper: MRL, MRL/Mp-lpr/lpr; PDB, protein databank; FR, framework regions; CDR, complementarity determining regions; CRI, cross-reactive Id; PDB, protein databank; IDR, Id-determining regions. Back

Received for publication February 10, 1997. Accepted for publication May 14, 1998.


   References
Top
 Abstract
 Introduction
 Materials and Methods
 Results
 Discussion
 References
 

  1. Shoenfeld, Y., D. A. Isenberg. 1993. Natural Autoantibodies: Their Physiological Role and Regulatory Significance CRC Press, Boca Raton.
  2. Buskila, D., Y. Shoenfeld. 1992. Anti-DNA antibodies. R. G. Lahita, ed. Systemic Lupus Erythematosus 2nd Ed.205. Churchill Livingstone, New York.
  3. Bentley, G. A., G. Boulot, M. M. Riottot, R. J. Poljak. 1990. Three-dimensional structure of an idiotype-anti-idiotype complex. Nature 348:254.[Medline]
  4. Ban, N., C. Escobar, R. Garcia, K. Hasel, J. Day, A. Greenwood, A. McPherson. 1994. Crystal structure of an idiotype-anti-idiotype Fab complex. Proc. Natl. Acad. Sci. USA 91:1604.[Abstract/Free Full Text]
  5. Evans, S. V., D. R. Rose, R. To, N. M. Young, D. R. Bundle. 1994. Exploring the mimicry of polysaccharide antigens by anti-idiotypic antibodies. J. Mol. Biol. 241:691.[Medline]
  6. Fields, B. A., F. A. Goldbaum, X. Ysern, R. J. Poljak, R. A. Mariuzza. 1995. Molecular basis of antigen mimicry by an anti-idiotope. Nature 374:739.[Medline]
  7. Young, F., L. Tucker, D. Rubinstein, T. Guillaume, J. André-Schwartz, K. J. Barrett, R. S. Schwartz, T. Logtenberg. 1990. Molecular analysis of a germline-encoded idiotypic marker of pathogenic human lupus autoantibodies. J. Immunol. 145:2545.[Abstract]
  8. Mazza, G., V. Guigou, D. Moinier, S. Corbet, O. Ollier, M. Fougerau. 1987. Molecular interactions in the "GAT" idiotypic network: an approach using synthetic peptides. Ann. Inst. Pasteur Immunol. (Paris) 136D:259.
  9. Meek, K., M. Takei, H. Dang, I. Sanz, M. J. Dauphinee, J. D. Capra, N. Talal. 1990. Anti-peptide antibodies detect a lupus-related idiotype that maps to the H chain CDR2. J. Immunol. 144:1375.[Abstract]
  10. Hasemann, C. A., J. D. Capra. 1991. Mutational analysis of the cross-reactive idiotype of the A strain mouse. J. Immunol. 147:3170.[Abstract]
  11. Strong, R. K., G. A. Petsko, J. Sharon, M. N. Margolies. 1991. Three-dimensional structure of murine anti-p-azophenylarsonate Fab 36–71. II. Structural basis of hapten binding and idiotypy. Biochemistry 30:3749.[Medline]
  12. Potter, K. M., Y. Li, V. Pascual, R. C. Williams, L. C. Byers, M. Spellerberg, F. Stevenson, D. Capra. 1993. Molecular characterization of a cross-reactive idiotope on human immunoglobulins utilizing the VH 4-21 gene segment. J. Exp. Med. 179:1419.
  13. Staines, N. A., C. T. Ravirajan, A. Morgan, A. J. Belcher, A. J. Henry, R. A. Lake, D. A. Smith, A. S. Hamblin, M. Hara, D. Adu, C. Morland, D. A. Isenberg. 1993. Expression and relationships of seven public idiotypes of DNA-binding autoantibodies on monoclonal antibodies and serum immunoglobulins. Lupus 2:25.[Abstract/Free Full Text]
  14. Staines, N. A., F. J. Ward, P. Hobby, A. N. Denbury, S. Bansal, B. J. Sutton, D. G. Williams, D. A. Isenberg, J. Kalsi, S. Oleinick. 1994. Idiotypes of DNA-binding autoantibodies: mapping by epitope scanning and molecular modeling. M. Zouali, ed. Experimental and Fundamental Aspects of Autoimmunity 173. Springer-Verlag, Berlin.
  15. Morgan, A., R. R. C. Buchanan, A. M. Lew, I. Olsen, N. A. Staines. 1985. Five groups of antigenic determinants on DNA identified by monoclonal antibodies from (NZB x NZW)F1 and MRL/Mp-lpr/lpr mice. Immunology 55:75.[Medline]
  16. Morgan, A., D. A. Isenberg, Y. Naparstek, J. Rauch, D. Duggan, R. Khiroya, N. A. Staines, A. Schattner. 1985. Shared idiotypes are expressed on mouse and human anti-DNA autoantibodies. Immunology 56:393.[Medline]
  17. Isenberg, D. A., W. Williams, J. Axford, R. Bakimer, D. Bell, T. Casasceca-Grayson, B. Diamond, F. Ebling, B. Hahn, G. Harkiss, C. Mackworth-Young, S. Le Page, H. Massicottee, J. Rauch, C. T. Ravirajan, R. S. Schwartz, Y. Shoenfeld, N. A. Staines, A. Todd-Pokrpoek, L. Tucker, R. A. Watts, M. Zouali. 1990. Comparison of DNA antibody idiotypes in human sera: an international collaborative study of 19 idiotypes from 11 different laboratories. J. Autoimmun. 3:393.[Medline]
  18. Ravirajan, C. T., N. A. Staines. 1991. Involvement in lupus disease of idiotypes Id. F-423 and Id.IV-228 defined, respectively, upon fetal and adult MRL/Mp/lpr/lpr DNA-binding monoclonal autoantibodies. Immunology 74:342.[Medline]
  19. Staines, N. A., F. J. Ward, A. N. Denbury, J. Mitchiner, O. Hartley, D. Eilat, D. A. Isenberg, S. Bansal. 1993. Primary sequence and location of the idiotopes of V-88, a DNA-binding monoclonal autoantibody, determined by idiotope scanning with synthetic peptides on pins. Immunology 78:371.[Medline]
  20. Ward, F. J., A. N. Denbury, C. Cunningham, W. Harris, N.A. Staines. 1997. Natural antibodies that react with V region peptide epitopes of DNA-binding antibodies are made by mice with systemic lupus erythematosus as disease develops. Immunology 92:354.[Medline]
  21. Padlan, E. A.. 1994. Anatomy of the antibody molecule. Mol. Immunol. 31:169.[Medline]
  22. Chothia, C., A. M. Lesk. 1987. Canonical structures for the hypervariable regions of immunoglobulins. J. Mol. Biol. 196:901.[Medline]
  23. Chothia, C., A. M. Lesk, A. Tramontano, M. Levitt, S. J. Smith-Gill, G. Air, S. Sheriff, E. A. Padlan, D. R. Davies, W. R. Tulip, P. M. Colman, S. Spinelli, P. M. Alzari, R. J. Poljak. 1989. Conformations of immunoglobulin hypervariable regions. Nature 342:877.[Medline]
  24. Chothia, C., A. M. Lesk, E. Gherardi, I. M. Tomlinson, G. Walter, J. D. Marks, M. B. Llewelyn, G. Winter. 1992. Structural repertoire of the human VH segments. J. Mol. Biol. 227:799.[Medline]
  25. Tomlinson, I. M., J. P. L. Cox, E. Gherardi, A. M. Lesk, C. Chothia. 1995. The structural repertoire of the human V{kappa} domain. EMBO J. 14:4628.[Medline]
  26. Martin, A. C. R., J. M. Thornton. 1996. Structural families in loops of homologous proteins: automatic classification, modeling and application to antibodies. J. Mol. Biol. 263:800.[Medline]
  27. Kirkham, P. M., F. Mortari, J. A. Newton, H. W. Schroeder. 1992. Immunoglobulin VH clan and family identity predicts variable domain structure and may influence antigen binding. EMBO J. 11:603.[Medline]
  28. Herron, J. N., X. M. He, D. W. Ballard, P. R. Blier, P. E. Pace, A. L. M. Bothwell, E. W. Voss, A. B. Edmundson. 1991. An autoantibody to single-stranded DNA: comparison of the three dimensional structures of the unliganded Fab and a deoxynucleotide-Fab complex. Proteins Struct. Funct. Genet. 11:159.[Medline]
  29. Barton, G. J., M. J. E. Sternberg. 1987. A strategy for the rapid multiple alignment of protein sequences: confidence levels from tertiary structure comparisons. J. Mol. Biol. 198:327.[Medline]
  30. Shenkin, P. S., D. L. Yarmush, R. M. Fine, H. Wang, C. Levinthal. 1987. Predicting antibody hypervariable conformations. I. Ensembles of random conformations for ring-like structures. Biopolymers 26:2053.[Medline]
  31. Laskowski, R. A., M. W. MacArthur, D. S. Moss, J. M. Thornton. 1993. PROCHECK: a program to check the stereochemical quality of protein structures. J. Appl. Cryst. 26:283.
  32. Jones, T. A.. 1978. A graphics model building and refinement system for macromolecules. J. Appl. Cryst. 11:268.
  33. Kabsch, W., C. Sander. 1983. Dictionary of protein secondary structure: pattern recognition of hydrogen-bonded and geometrical features. Biopolymers 22:2577.[Medline]
  34. Arevalo, J. H., E. A. Stura, M. J. Taussig, I. A. Wilson. 1993. Three-dimensional structure of an anti-steroid Fab' and progesterone-Fab' complex. J. Mol. Biol. 231:103.[Medline]
  35. Pokkuluri, P. R., F. Bouthillier, Y. Li, A. Kuderova, J. Lee, M. Cygler. 1994. Preparation, characterization and crystallization of an antibody Fab fragment that recognizes RNA: crystal structures of native Fab and three Fab-mononucleotide complexes. J. Mol. Biol. 243:283.[Medline]
  36. Foote, J., G. Winter. 1992. Antibody framework residues affecting the conformations of hypervariable loops. J. Mol. Biol. 224:487.[Medline]
  37. Schildbach, J. F., R. I. Near, R. E. Bruccoleri, E. Haber, P. D. Jeffrey, J. Novotny, S. Sheriff, M. N. Margolies. 1993. Modulation of antibody affinity by a non-contact residue. Protein Sci. 2:206.[Medline]
  38. Giusti, A. M., N. C. Chien, D. J. Zack, S. Shin, D. Scharff. 1987. Somatic diversification of S107 from an anti-phosphorylcholine to an anti-DNA autoantibody is due to a single base change in its heavy chain variable region. Proc. Natl. Acad. Sci. USA 84:2926.[Abstract/Free Full Text]
  39. Padlan, E. A., G. H. Cohen, D. R. Davies. 1985. On the specificity of antibody-antigen interactions: phosphocholine binding to MCPC603 and the correlation of 3-dimensional structure and sequence data. Ann. Inst. Pasteur Immunol. (Paris) 136C:271.
  40. Smith, R. G., E. W. Voss. 1990. Variable region primary structures of monoclonal anti-DNA autoantibodies from NZB/NZW F1 mice. Mol. Immunol. 27:463.[Medline]
  41. Martin, A. C. R.. 1996. Accessing the Kabat antibody sequence database by computer. Proteins Struct. Funct. Genet. 25:130.[Medline]
  42. Radic, M. Z., J. Mackle, J. Erikson, C. Mol, W. F. Anderson, M. Weigert. 1993. Residues that mediate DNA binding of autoimmune antibodies. J. Immunol. 150:4966.[Abstract]
  43. Radic, M. Z., M. Weigert. 1994. Genetic and structural evidence for antigen selection of anti-DNA antibodies. Annu. Rev. Immunol. 12:487.[Medline]
  44. Ebling, F. M., B. P. Tsao, R. R. Singh, E. E. Sercarz, B. H. Hahn. 1993. A peptide derived from an autoantibody can stimulate T cells in the (NZB x NZW)F1 mouse model of systemic lupus erythematosus. Arthritis Rheum. 36:355.[Medline]
  45. Ward, F. J., N. Harper, A. N. Denbury, D. Gveric, N. A. Staines. 1994. Mapping T cell epitopes on DNA-binding autoantibodies associated with SLE. Arthritis Rheum. 37:S387.
  46. Singh, R. R., F. M. Ebling, E. E. Sercarz, B. H. Hahn. 1995. Immune tolerance to autoantibody-derived peptides delays development of autoimmunity in murine lupus. J. Clin. Invest. 96:2990.
  47. Kieber-Emmons, T., H. Kohler. 1986. Toward a unified theory of immunoglobulin structure-function relations. Immunol. Rev. 90:29.[Medline]
  48. Williams, W. M., N. A. Staines, S. Muller, D. A. Isenberg. 1995. Human T cell responses to autoantibody variable region peptides. Lupus 4:464.[Abstract/Free Full Text]
  49. Jefferis, R.. 1993. What is an idiotype?. Immunol. Today 14:119.[Medline]



This article has been cited by other articles:


Home page
 LupusHome page
J K Kalsi and B H Hahn
The role of VH determinants in systemic lupus erythematosus
Lupus, December 1, 2002; 11(12): 878 - 884.
[Abstract] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow Request Permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Hobby, P.
Right arrow Articles by Sutton, B. J.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Hobby, P.
Right arrow Articles by Sutton, B. J.
Right arrowPubmed/NCBI databases
*Substance via MeSH
Medline Plus Health Information
*Lupus

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS