Systemic Lupus Erythematosus Patients Nuclear RNA Autoantibodies ...

4 downloads 4919 Views 419KB Size Report
Receive free email-alerts when new articles cite this article. Sign up at: .... E-mail address: [email protected] or Dr. René M. A. .... DNA templates.
The Importance of the Light Chain for the Epitope Specificity of Human Anti-U1 Small Nuclear RNA Autoantibodies Present in Systemic Lupus Erythematosus Patients1 Rene´ M. A. Hoet,2*‡ Martijn Pieffers,* Maurice H. W. Stassen,* Jos Raats,* Ruud de Wildt,2* Ger J. M. Pruijn,* Frank van den Hoogen,‡ and Walther J. van Venrooij3* Abs to U1 RNA are frequently found in patients suffering from systemic lupus erythematosus overlap syndromes and Ab titers correlate with disease activity. We describe the isolation of the first human anti-U1 RNA autoantibodies from a combinatorial IgG library made from the bone marrow of a systemic lupus erythematosus patient. With the use of phage display technology, two anti-U1 RNA singlechain variable fragment (scFv) Abs were selected. Both high affinity anti-U1 RNA Ab fragments (Kd ; 1 nM) recognize stem II of U1 RNA and were derived from the same heavy chain gene (VH3–11) and the same l (3r) light chain gene although somatic mutations, predominantly present in the complementarity-determining regions, are different. Experiments, in which the heavy chain genes of both anti-U1 RNA scFvs were reshuffled with the original light chain repertoire of the patient resulted, after selection on stem loop II, in a large number of RNA-binding Ab fragments. All these stem loop II-specific RNA binding clones used a similar, but not identical, 3r l light chain. When scFvs were selected from the reshuffled libraries by stem loop IV, representing the other autoantigenic site of U1 RNA, most selected Ab clones did react with stem loop IV, but no longer with stem loop II. The stem loop IV-reactive Ab clones contained different, not 3r-related, light chains. These results point to a major role for the light chain in determining the sequence specificity of these disease-related anti-U1 RNA Abs. The possibility that secondary light chain rearrangements are involved in this autoimmune response is discussed. The Journal of Immunology, 1999, 163: 3304–3312.

P

atients with systemic lupus erythematosus (SLE)4 and SLE overlap syndromes often produce IgG autoantibodies directed to components of the U1 small nuclear ribonucleoprotein particle (snRNP) (reviewed in Ref. 1). Autoantibodies have been found recognizing the U1-specific proteins (U1A, U1C, and U1-70K) and the core or Sm proteins (B/B9, D1, D2, D3, and the EFG complex). Autoantibodies directed to the U1-specific proteins are often accompanied by Abs directed to the naked U1 RNA (2–5). The epitope regions on the U1 RNA recognized by these patient Abs are located within stem II and loop IV of U1 RNA (4, 6 – 8). Interestingly, the Ab titers to individual epitopes on U1 RNA appeared to correlate with disease activity (9), and Ab activity directed to U1 RNA was found in the more severely affected

*Department of Biochemistry, University of Nijmegen, Nijmegen, The Netherlands; † Department of Rheumatology, University Hospital Nijmegen, Nijmegen, The Netherlands; and ‡Department of Pathology, University Hospital Maastricht, Maastricht, The Netherlands Received for publication April 12, 1999. Accepted for publication July 8, 1999. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 1 This work was supported by the Netherlands Foundation for Chemical Research and the Technology Foundation with financial aid from the Netherlands Organization for Scientific Research (Grant 349-3206). 2 Current address: Medical Research Council, Centre for Protein Engineering, Hills Road, Cambridge CB2 2QH, U.K. 3 Address correspondence and reprint requests to Dr. Walther J. van Venrooij, Department of Biochemistry, University of Nijmegen, P.O.B. 9101, 6500 HB Nijmegen, The Netherlands. E-mail address: [email protected] or Dr. Rene´ M. A. Hoet, Department of Pathology, University Hospital Maastricht, P. O. Box 5800, 6202 AZ Maastricht, The Netherlands. E-mail address: [email protected] 4 Abbreviations used in this paper: SLE, systemic lupus erythematosus; snRNP, small nuclear ribonucleoprotein particle; scFv, single-chain variable fragment; CDR, complementarity-determining region; VSV, vesicular stomatitis virus.

Copyright © 1999 by The American Association of Immunologists

patient group (5), suggesting that this type of autoantibody has some intrinsic relationship with the disease. Both anti-stem II and anti-loop IV Abs have also been shown to interact with the native cellular U1 snRNP complex (10). Although it is generally accepted that most nucleic acids are poor immunogens, Abs to DNA and RNA are found in several autoimmune diseases. In patients with SLE, Abs have been described to single-stranded and double-stranded DNA, as well as to 28S ribosomal RNA (11) and U1 RNA (2– 4), whereas in myositis Abs have been described to tRNAAla (12), tRNAMet (13), and tRNAHis (14). Ab titers to dsDNA, present in a subpopulation of SLE patients, often correlate with disease activity and can be used as a predictive marker for this disease (15). Most of these Abs contain a net positive charge in the complementarity-determining region (CDR) loops of heavy and light chain. In this respect the CDR3 of the heavy chain, the most variable loop of the Ab, is thought to play the most important role. The reason why anti-nucleic acid Abs arise is not known. It has been proposed that the anti-RNA activity could be the result of accidental cross-reactivity with RNAs of other cellular or viral immunogens (4). Cross-reactivity between proteins and RNA is also a possible mechanism (7). To study the phenomenon of anti-nucleic acid autoantibodies in more detail, we prepared combinatorial Ab libraries representing the IgG repertoire present in the bone marrow of four patients with SLE overlap syndromes. The genes encoding the V domains of heavy (VH) and light (VL) chains were cloned in a phagemid vector and expressed as a fusion protein of the minor coat protein pIII. Two Ab fragments, specifically recognizing stem loop II of U1 RNA were isolated and analyzed in more detail. Both are derived from the same heavy chain gene (V3-11 5 DP-35) and the same l light chain gene (3r 5 DPL-23), but somatic mutations and CDR sequences were different. 0022-1767/99/$02.00

The Journal of Immunology To study to what extent the light chain was important in determining the specificity of these two related anti-U1 RNA Abs, we reshuffled both original anti-U1 RNA heavy chains with the light chain repertoire from the same patient. These reshuffled libraries were then analyzed for their U1 RNA-binding capacities. Stem loop II-specific binders were found only when similar, but not identical, light chains were used. Combinations of V3-11 with other light chains produced anti-RNA or anti-stem loop IV binders. These results underscore the important role of the light chain in the RNA-recognizing specificity of these Abs.

Materials and Methods Library construction Patient libraries were made from bone marrow cells obtained from 4 SLE overlap syndrome patients (D18, HO, O11, and Z5) (as synonym for SLE overlap, these patients are in the literature also classified as mixed connective tissue disease or anti-RNP-positive connective tissue disease patients). Serum samples from these patients were able to immunoprecipitate naked U1 RNA (data not shown). The patient libraries were constructed essentially as described (16, 17). All libraries contained .108 individual clones, and .75% of all clones contained full length inserts. From each library, 96 clones were analyzed by PCR-fingerprinting (BstN1 digestion) to confirm that the patterns were highly diverse.

Isolation of single IgG-positive B cells, amplification and sequencing of VH /VL regions PBL were obtained from a healthy donor and two SLE patients, Z5 and D101, using Ficoll-Paque (Pharmacia Biotech, Uppsala, Sweden). Single IgG-positive B cells were isolated using a Coulter Epics Elite flow cytometer (Coulter, Hialeah, FL) equipped with an automatic cell deposit unit as described previously (18, 19). These individual B cells were cultured for 10 days after which isotype-specific Ab production was tested. From IgGpositive cultures (76% or more), RNA was isolated, heavy and light chain V regions were amplified with family-specific primers, and the sequences were determined.

Selection and purification of Ab fragments The first round of panning was performed with in vitro T7 RNA polymerase-transcribed U1 RNA (4). Immunotubes (Maxisorp, Nunc, Roskilde, Denmark) were precoated with magic coating (2 h at 4°C) (20) and blocked with 0.5% BSA for 2 h at 4°C. After 5 washings with RNase-free milliQ water, U1 RNA was coated (20 mg/ml) in PBS overnight at 4°C. The remaining binding sites on the coated tubes were blocked by a 2-h incubation at 4°C with a mixture of PBS containing 2% nonfat dry milk powder (Marvel, Food Store, U.K.) 0.5% BSA, 50 U/ml RNasin (Promega, Madison, WI), and 20 mg/ml total yeast RNA (Mix-1). Phages were isolated from the libraries as described (18) and incubated in the Mix-1 mixture in an end-over-end rotator for 30 min at 4°C, after which the tubes were stored on ice for 90 min. Next the tubes were rinsed 15 times with RNase-free PBS containing 0.05% Tween 20 (PBS-T) and 15 times with RNase-free PBS. Elution of the phages (with 100 mM triethylamine), infection, and plating was performed as described (18). The second round of selection was performed with biotinylated U1 RNA. First streptavidin (20 mg/ml) in 0.1 M NaHCO3 (pH 8.6) was coated overnight at 4°C. After three washings with PBS, blocking was performed for 1 h at 4°C with biotin-free BSA in PBS. After another three washings with PBS, a mixture of Mix-1 containing biotinylated U1 RNA (10 mg/ml) was added and incubated in an-end-over-end rotator wheel for 1 h at 4°C. Phage incubation, washings, elution, infection, and plating were performed in the same way as described for selection round I. After the second round, 48 individual clones from each combinatorial library were grown, and scFv production was induced with IPTG as described (21). A 2-ml culture supernatant sample was screened for binding to 32P-labeled U1 RNA using a previously described nitrocellulose dot blot binding assay (9). ScFv sequences were isolated from pHENIX as NcoI-NotI DNA fragments and subcloned into PUC119(His)8 vesicular stomatitis virus (VSV) (17). The sequences (GenBank accession numbers AJ241377, AJ241378, AJ241420, and AJ241421) were compared with the germline sequences in the V-base sequence directory (37). Individual scFv-producing clones were grown in 500-ml cultures. Expression was induced with 1 mM isopropyl b-D-thiogalactoside. The cultures were grown for 3 h at 30°C, and Ab fragments were harvested and purified from the periplasmic fractions using nickel-agarose as described (21).

3305 Light-chain shuffling libraries The heavy chains of Z5scFv3 and Z5scFv7 were isolated by NcoI/SalI digestion of the full length-containing pHENIX clones. The DNAs were gel purified and used in ligation with NcoI/SalI-digested and gel-purified Z5Vl-pHENIX and Z5Vk-pHENIX DNA (patient Z5) in the same way as described for the library construction. The four reshuffled libraries all contained .107 individual clones with .90% full length inserts. The libraries (1 round) were selected with biotinylated stem loop II and stem loop IV RNA in the same way as described above for the second round of selection. ScFv sequences were isolated from pHENIX as NcoI-NotI DNA fragments and subcloned into PUC119(His)8VSV (17). The light chain sequences of the reshuffled scFv clones (GenBank accession numbers AJ241379 –AJ241419) were compared with the germline sequences in the V-base sequence directory (37).

Determination of the Kd The Kd of Z5scFv3 was determined by a nitrocellulose filter binding assay. A constant amount of purified scFv Ab was used with a series of stem loop II RNA concentrations (fixed amount of 32P-labeled stem loop II RNA (20,000 cpm) mixed with varying amounts of unlabeled stem loop II RNA). First, a saturation curve was made to determine the scFv concentration in which a linear fit was found between the concentration of stem loop II RNA and the scFv concentration. Secondly, the maximum amount of bound stem loop II RNA was determined. Using a constant amount of scFv, the amount of U1 RNA was increased, and the bound U1 RNA (B) was quantified with the nitrocellulose-binding assay counting the 32P activity (9). Assuming a 1:1 complex, the Kd was calculated (in triplicate) by Scatchard analysis.

Radiolabeled RNA and immunoprecipitation Radiolabeled RNA (32P-labeled) was prepared either by deproteinization of extracts of HeLa monolayer cells, cultured overnight in the presence of radiolabeled orthophosphate (3), or by in vitro transcription. For in vitro transcription the T7 RNA polymerase system was used as previously described (3). All immunoprecipitations (unless stated otherwise) were performed at 150 mM NaCl as described previously (21). DNA templates encoding wild-type U1 RNA wt, stem loop II, stem loop IVDDE (4), U3 RNA (22), and RNase MRP-RNA (25) have also been described previously.

Results Selection of U1 RNA-specific Ab fragments Four combinatorial Ab libraries (all containing .108 individual clones) were made from bone marrow cells of four SLE overlap syndrome patients and screened for the presence of anti-U1 RNA activity. In the first round of selection U1 RNA was directly coated to the immunotubes, while in the second round biotinylated U1 RNA was used (see Materials and Methods). Both selections were conducted in the presence of an excess of yeast RNA to avoid selection of general nucleic acid-binding Abs and to prevent degradation of U1 RNA. The two different selection methods were used to avoid nonspecific binding of phages to biotin or streptavidin. After two rounds of selection, a 100-fold enrichment for binding phage compared with the first round was obtained for all 4 libraries. At this stage, 96 individual clones of each library were screened for binding to 32P-labeled U1 RNA with the nitrocellulose filter binding assay (4). In addition, scFv expression levels were determined with a dot blot assay, and fingerprint patterns were determined using BstNI digestion. Twelve U1 RNA-binding clones were subsequently selected and characterized (four from D18, one from HO, and seven from Z5). We also tried to isolate anti-U1 RNA Abs from semisynthetic Ab libraries (23, 24). Although we succeeded in isolating Abs directed to nucleic acids from these libraries, they were not specific for U1 RNA. Libraries derived from PBLs from patient Z5 and another SLE patient (D101) were analyzed as well, but possibly because of the low number of plasma cells in these preparations, no VH/VL combinations able to recognize U1 RNA were obtained from these libraries.

3306 Specificity of anti-U1 RNA Ab fragments To determine the specificity of the Ab fragments, competition experiments were performed in which radiolabeled stem loop II or IV of U1 RNA (the two major autoepitopes of U1 RNA) was incubated with the scFvs in combination with an excess of either unlabeled stem loop II or unlabeled stem loop IV as competitor and analyzed in a nitrocellulose filter binding assay. Most clones reacted with both stem loop II and IV of U1 RNA except Z5 scFv3, scFv5, and scFv7, which showed only reactivity and competition with stem loop II (Fig. 1A). To further analyze the specificity of these Ab fragments (scFv3, scFv5, and scFv7) competition experiments were performed with poly(A), poly(G-C), poly(G-U), poly(G), poly(C), HeLa cell ribosomal RNA, dsDNA, and U1 RNA. With scFv3 and scFv7, none of the competitors (except U1 RNA) showed inhibition of binding to 32P-U1 RNA, even when a 500-fold molar excess of competitor was used, whereas scFv5 binding was strongly competed by ribosomal RNA (data not shown). Next, the Ab fragments that seemed most specific for U1 RNA, scFv3 and scFv7, were used in immunoprecipitations in which the scFvs were indirectly bound via their VSV tag to protein A-agarose and incubated with a radiolabeled RNA mixture containing stem loop II RNA, stem loop IV RNA, U1 RNA, and RNase MRPRNA. The result shows that only stem loop II RNA is recognized by these two scFv Abs (Fig. 1B). Using various stem loop II mutants, we were able to show that both scFv3 and scFv7 recognized the stem of stem loop II of U1 RNA. A detailed analysis of the epitope has been published elsewhere (25). The Kd of the scFv3 for stem loop II was determined with a nitrocellulose filter binding assay (Scatchard analysis) to be 1.0 6 0.2 nM. Although the affinity of scFv7 could not be determined as accurately, because of its lower expression levels the estimated affinity of this scFv for stem loop II appeared to be of the same order of magnitude. Recognition of the native U1 snRNP particle Because of the low expression levels of scFv7, these experiments were performed only with scFv3. With 32P-labeled total RNA isolated from HeLa cells, scFv3 was shown to specifically immunoprecipitate U1 RNA (Fig. 1C). The Ab was also tested on Western blots containing HeLa nuclear extract and by ELISA using recombinant U1A and U1C autoantigens as Ag. No reactivity of scFv3 with proteins could be detected. Using a [35S]methionine-labeled cell extract in immunoprecipitation experiments, we have shown recently that scFv3 is able to immunoprecipitate the U1 snRNP complex from a HeLa S100 extract (25). The U1 snRNP proteins precipitated by scFv3 were similar to the pattern of proteins precipitated by an anti-Sm Ab which was used as a positive control and showed that the U1A protein, which binds to the loop of stem loop II of U1 RNA, is not interfering with the binding of scFv3 to the stem of stem loop II (25). V gene usage of anti-U1 RNA Ab fragments To determine the V gene usage of the two U1 stem loop II-specific scFvs, the sequences of both heavy chain (VH) and light chain (VL) were analyzed (Table I). Both VH genes align best with the VH3-11 (DP-35) gene, and both contain a relatively large number of somatic mutations (21 aa for scFv3 and 16 aa for scFv7), the majority being located in CDR2. The CDR3 of the heavy chain of both Ab fragments contains a number of positively charged amino acids at identical positions (for scFv3 lysines at positions 3, 6, and 9 and for scFv7 lysines at positions 3, 6, 8, and 9). Also in the CDR1 and CDR2 some replacement mutations introducing basic amino acids (arginine, lysine, and histidine) are found. No signif-

RECOMBINANT ANTI-U1 RNA Abs FROM SLE PATIENTS icant homology to any D segment could be detected. Both light chains of the two Abs align to the same l germline gene 3r (DPL23), and also here in both cases a relatively large number of somatic replacement mutations are found (for scFv3, 12 aa, and for scFv7, 21 aa). In this case, the CDR1 is the main target. In all light chain CDR sequences, except for the CDR1 of scFv7, basic amino acids are found, in part resulting from germline gene sequences and in part introduced by somatic mutations. In scFv3, as a result of somatic mutations, also two acidic amino acids are replaced by either a neutral (D3 N) or a basic (D3 R) amino acid. To compare the frequency of the selected VH/VL genes with the total VH/VL use in this patient, we isolated IgG-positive B-cells from the peripheral blood of this patient by FACS selection (18). Individual IgG-positive B cells were cultured for 10 days, and the IgG isotype of the B cells was confirmed by ELISA. Next we amplified and determined the sequences of 100 VH/VL pairs of patient Z5 (19). In the IgG-positive B cell population of patient Z5, 4% of the Abs contained VH gene VH3-11, which is not different from the frequency found in normal individuals (19). We also investigated the frequency by which the 3r (DPL-23) light chain gene is expressed in IgG positive B cells from a normal individual and from patient Z5. In only 1% of the normal IgG repertoire could 3r light chain-related clones be identified, whereas sequencing of the light chains of 100 IgG-positive B-cells from patient Z5 failed to detect any clone expressing this particular light chain. Light chain reshuffling It is generally accepted that the heavy chain determines in most cases the major part of the specificity of an Ab, although the light chain can also influence the affinity or specificity. Because both anti-U1 RNA Abs aligned to the same light chain germline sequence, we wondered whether, in this case, the light chain is important in determining the specificity of the anti-stem loop II activity. To investigate this, we reshuffled the heavy chains of both Abs with the original light chain repertoire of the same patient. The four sublibraries (each .107 in size, referred to as VHscFv3k, VHscFv3l, VHscFv7k, and VHscFv7l) were first analyzed without selection. Forty-eight clones were randomly picked from each sublibrary, and scFv production was induced using isopropyl b-Dthiogalactoside. Ab fragment expression levels were analyzed with a dot blot probed with an anti-VSV Ab. Consistent with the reproducibly low expression levels of scFv7 observed before, the scFv expression levels of most VHscFv7-derived clones were found to be very low compared with the expression by the scFv3derived clones. The culture supernatants were also tested in a nitrocellulose filter binding assay for reactivity to 32P-U1 RNA. One positive clone was found for VHscFv3 (V17l) and 3 for VHscFv7 (V18l, V19l, and V16k). These four clones were sequenced (Table II, V16 –V19). Only one of these four RNA-binding clones (V17) used a 3r (DPL-23)-related light chain. Also 15 nonbinders (7 l and 8 k clones) that showed scFv expression on dot blot were analyzed by DNA sequencing (Table II, V1–V15). None of these nonbinders used a 3r-related light chain. Then the four libraries were subjected to one round of selection on biotinylated stem loop II RNA. After selection, again 48 clones derived from each library were randomly chosen and tested for reactivity to U1 RNA using the filter binding assay. For scFv3 ;50% of the clones were positive, whereas for scFv7 95% of all selected clones were reactive with U1 RNA. Fingerprinting was performed for all positive clones, and 17 different clones were analyzed in more detail by DNA sequencing (named II-3.1–II-3.9 and II-7.1–II-7.8). DNA sequencing revealed that a large number of the stem loop II-selected clones (seven of nine derived from VHscFv3, and three of eight derived from VHscFv7) used a light

The Journal of Immunology

3307

FIGURE 1. Specificity test of patient-derived anti-U1 RNA-selected scFvs. A, Filter binding assay using radiolabeled domains of U1 RNA. 32P-stem loop II RNA (top, 20,000 cpm) or 32P-stem loop IV RNA (bottom, 20,000 cpm) were incubated with scFvs (2 ml culture supernatant) in the presence of 10 mg yeast RNA and different amounts of competitor (stem loop II or stem loop IV) RNA. After incubation (2 h, 4°C), samples were transferred to a dot blot manifold and washed three times with PBS as described previously (9). Bound RNA was visualized by autoradiography. Twelve scFvs were analyzed, seven derived from patient Z5, one from patient HO), and four from patient D18. (In the bottom panel of Z5, 6 (0 ng stem loop IV) no signal is detected; repeated experiments (not shown) resulted in a signal with the same intensity as that of Z5, 6 (0 ng stem loop II), indicating that this point is an artifact in this particular experiment). B, Immunoprecipitation of radiolabeled domains of U1 RNA by patient-derived scFvs. scFvs were indirectly coupled to protein A-agarose (using anti-VSV tag Abs) and incubated with a mixture of radiolabeled in vitro transcribed RNAs (stem loop II, stem loop IV, U1 RNA, RNase MRP-RNA). Immunoprecipitation was performed as described previously (21). Samples were analyzed on 10% polyacrylamide-8.3 M urea gels. The immunoprecipitations were performed with: Z5scFv3 (lane 2), Z5scFv7 (lane 3) and control anti-U1A protein scFv (lane 4). Lane 1, 10% of the input RNA mixture. C, Immunoprecipitation of U1 RNA from total HeLa cell RNA. Abs (serum or scFv) were coupled to protein A-agarose and incubated with 32 P-labeled HeLa cell RNA. Immunoprecipitation was performed as described previously (21). Samples were analyzed on 10% polyacrylamide-8.3 M urea gels. The immunoprecipitations were conducted with: patient serum Z5 (lane 3), normal human serum (lane 4), Z5scFv3 (lane 5), control scFv (lane 6). Lane 1, 5% input RNA; lane 2, 1% input RNA. The positions of the most abundant RNAs are indicated on the left.

3308

RECOMBINANT ANTI-U1 RNA Abs FROM SLE PATIENTS

Table I. Amino acid sequences of heavy and light chains of scFvs binding to stem loop II of U1 RNA (Z5scFv3 and Z5scFv7), aligned to their most homologous germline sequencea FR 1 Heavy chain VH 3-11 (DP-35) Z5scFv3b Z5scFv7b

VH 3-11 (DP-35) Z5scFv3 Z5scFv7

Light chain 3r (DPL-23) Z5scFv3b Z5scFv7b

3r (DPL-23) Z5scFv3 Z5scFv7 a b c

CDR 1

QVQLVESGGGLVKPGGSLRLSCAASGFTFS qvqlqqsg.A............G....S.. qvqlqesg.D.....A......T.......

DYYMS .S..N .H..T

FR 2

CDR 2

WIRQAPGKGLEWVS .............A ..............

YISSSGSTIYYADSVKG H.NGT.TSTK.....E. H..G..TSTN......D

FR 3

CDR 3

FR 4

Mutationsc (aa/nt)

RFTISRDNAKNSLYLQMNSLRAEDTAVYYCAR .........E...S...D...I......F..T .............S.......T....L....T

SFKAGKGQKYFDL SVKNGKGKKYFDL

WGRGTLVT. WGRGTLVT.

21/38 16/21

FR 1

CDR 1

FR 2

CDR 2

SYELTQPPSVSVSPGQTASITC qsvltqpp.............. qsvltqpp.......E..I.P.

SGDKLGDKYAC .RRT....FVS ...G..NS.TS

WYQQKPGQSPVLVIY ............... ....QA.........

QDSKRPS .NN.... .......

FR 3

CDR 3

FR 4

Mutationsc (aa/nt)

GIPERFSGSNSGNTATLTISGTQAMDEADYYC ..................V....TL....... ...D...A.S...S.....V.A.PT..G....

QAWDSSTA .....R.....RR.-

VLFGGGTKLT. VIFGGGTQLT.

12/19 21/35

Alignments using V-base homology search (37). Primer-encoded sequences are depicted in lower case letters. Basic amino acids in the CDRs are bold and underlined. The number of amino acid (nucleotide) mutations are given.

chain most homologous to l germline gene 3r (DPL-23). As stated above, from the four binders obtained before selection, which recognized U1 RNA in the nitrocellulose filter binding assay, one (V17) used the 3r light chain. In contrast, among the 15 randomly selected nonbinders (the V1–V15 series) no 3r-related clone was found, although the scFv expression levels were at least as high as those of the binders, suggesting that the presence of a 3r-related light chain favors the binding to U1 RNA. The specificity of all binders was also analyzed via RNA immunoprecipitation. For this purpose, periplasmic fractions of clones II-3.1–II-3.9, clones II-7.1–II-7.8, clones V16 – V19, and as a negative control an anti-U1A scFv (21) were analyzed for their capacity to immunoprecipitate RNA from a mixture of 32P-labeled U1 RNA, stem loop II RNA, stem loop IV RNA, and U3 RNA. A selected panel of such precipitation results is shown in Fig. 2A; the overall results are summarized in Table III. Although the dot blots showed binding to U1 RNA for all clones, in immunoprecipitation assays only a subset of these clones showed binding. This can be caused either by a lower scFv expression level (as for example in the case of II-3.3) or by a lower affinity for U1 RNA (as for example in the case of II-3.9). When the immunoprecipitation results and the DNA sequences were compared with each other, a striking observation was made. All VHscFv3 and VHscFv7 clones selected with stem loop II that showed specific binding to stem loop II in immunoprecipitation used a 3r related light chain. The clones with another light chain, which appeared to bind U1 RNA in the nitrocellulose filter binding assay, showed in immunoprecipitation assays no detectable binding to RNA or RNA binding with altered specificity (Table III, Fig. 2). The lack of RNA binding might be explained by a low scFv expression level or by a lower affinity of the Ab fragment. In addition, the results show that altering the light chain appears to influence the specificity of the RNA binding substantially (e.g., clone V19 which contains a DPL-11 light chain (Fig. 2A)). To investigate the influence of the light chain further, the same four sublibraries were used in a selection with biotinylated stem loop IV RNA. Stem loop IV contains the other major U1 RNA

epitope recognized by SLE patients’ sera including this particular patient serum (Z5). The selection was again conducted in the presence of a large excess of yeast RNA. After selection, 20% of the clones originating from VHscFv3 and 50% of the clones from VHscFv7 appeared to bind U1 RNA in the nitrocellulose dot blot assay. Five clones were then analyzed for binding specificity by immunoprecipitation, and their sequences were determined. Most of the immunoprecipitation data are shown in Fig. 2B, and the overall results are summarized in Table III (stem loop IV-selected VHscFv clones are designated IV-3.x and IV-7.x depending on the library from which they originated). As expected, most stem loop IV binders showed no or only very weak precipitation of stem loop II and much stronger precipitation of stem loop IV. However, the Abs were not specific for stem loop IV because U3 RNA, used as a control, was immunoprecipitated in all cases as efficiently as stem loop IV. In subsequent control experiments, it was found that some other RNAs (e.g., RNase MRPRNA) were often recognized by these scFvs, whereas other small RNAs (e.g., hY RNAs) were not recognized (data not shown). One stem loop IV-selected scFv Ab (scFv IV-7.3) immunoprecipitated both stem loop II and stem loop IV RNAs at 150 mM NaCl, but lost stem loop IV reactivity, and not stem loop II reactivity, at 350 mM NaCl (data obtained at 250 mM NaCl are shown in Fig. 2B). These results underline the idea that the selected scFvs are not stem loop IV specific. Nevertheless, the four stem loop IV binders (IV-3.1, IV-3.2, IV-7.1, and IV-7.2) all used different light chains, whereas IV-7.3, which also showed high affinity binding to stem loop II, again used a 3r-related light chain. Interesting, but as yet unexplained, was the observation that all 3r-related clones, in series II as well as series IV, contained a serine at position 34 instead of a cysteine (see Discussion). The latter results are thus in complete agreement with the results from the stem loop II selections and show that the stem loop II specificity of these anti-U1 RNA Abs is very much dependent on the identity of the light chain used. In Fig. 3, all sequenced clones are classified according to their most homologous germline gene. Fig, 4 illustrates that within the

The Journal of Immunology

3309

Table II. Amino acid sequences of the l (A) and k (B) light chains of scFvs obtained after light chain reshuffling of clones Z5scFv3 and Z5scFv7

a Sequences are given of randomly chosen scFvs with no affinity for U1 RNA (V1–V4: VHscFv3k; V5–V7: VHscFv7k; V8 –V12: VHscFv3l; V13–V15: VHscFv7l), nonselected scFvs with affinity for U1 RNA (V16 –V19), scFvs selected on stem loop II of U1 RNA (II-3.1–II-3.9 and II-7.1–II-7.8), and scFvs selected on stem loop IV of U1 RNA (IV-3.1, IV-3.2, and IV-7.1, IV-7.2, IV-7.3). U1 RNA binding clones are bold. Primer-encoded amino acid mutations are shown in lower case letters. Basic amino acid mutations (K, R, H) are underlined.

group of stem loop II-specific scFvs there is a dramatic overrepresentation of 3r-related light chains, whereas this particular light chain is not used by 15 randomly selected nonbinders. As noted above, the 3r light chain gene in IgG-positive B cells of the normal repertoire and in patient Z5 was used in 1% of the cases or less (19), indicating that there certainly is no general overrepresentation of this light chain product. Our results thus indicate strongly that in this SLE patient there is a clear light chain restriction for anti-stem loop II reactivity.

Discussion We have isolated a number of patient autoantibodies with specificity for the U1 small nuclear RNA molecule. In patient sera, the anti-U1 RNA population of autoantibodies mostly recognizes the

stem of stem loop II and/or the loop of stem loop IV of the RNA. The scFvs selected by us are specifically directed to the stem of stem loop II (see also Ref. 25). Both scFvs were derived from the same heavy (VH3-11) and light chain (l 3r) germline genes but differ because of a number of somatic mutations. Reshuffling experiments, using the VH3–11 gene of the Abs and the complete light chain repertoire of the patient, resulted in a second group of anti-U1 RNA Abs with different specificities. Abs recognizing stem loop II, however, always contained a 3r-related light chain, while Abs with different specificities did not. This indicates that, at least in this patient, the choice of the light chain strongly influences target RNA specificity. The two original stem loop II specific scFvs are characterized by a number of somatic mutations in both heavy and light chains

3310

RECOMBINANT ANTI-U1 RNA Abs FROM SLE PATIENTS

FIGURE 2. Immunoprecipitation patterns of light chain reshuffled scFvs. A mixture of radiolabeled RNAs (stem loop II, stem loop IV, U1 RNA, U3 RNA) was incubated with different scFvs coupled to protein A-agarose. Immunoprecipitations were performed (at 250 mM NaCl) as described (21). Precipitated RNAs were analyzed on 8% polyacrylamide-8.3 M urea gels. A, Immunoprecipitation patterns of light chain reshuffled scFvs selected with stem loop II. Precipitations were conducted with the original clones Z5scFv3 (K3) and Z5scFv7 (K7), control scFv (anti-U1A, C1), control anti-VSV Ab (C2), a randomly reshuffled, nonselected, but positive scFv (V25), and a selected panel of reshuffled scFvs, derived from Z5scFv3, selected with stem loop II (II-3.1, II-3.4, II-3.7, II-3.8, indicated by 31, 34, 37, and 38, respectively). The most homologous light chain germline gene of the various clones is indicated. The first two lanes show the mixture of input RNAs (1 and 0.1%, respectively). The positions of the RNAs are indicated on the left. B, Immunoprecipitation patterns of light chain reshuffled scFvs selected with stem loop IV. Precipitations were conducted with a selected panel of reshuffled scFvs, derived from Z5scFv7, selected with stem loop IV (IV-7.1, IV-7.2, IV-7.3, indicated by 71, 72, and 73, respectively) and with one of the original clones (Z5scFv3, indicated by K3). The most homologous light chain germline gene of the various clones is indicated. The positions of the RNAs are indicated on the left.

(Table I) indicating that they have arisen via an Ag-driven process. Because the U1 RNA is contained in the human U1 snRNP particle, and because the Abs are directed to a free, accessible part of this complex (25), it is reasonable to assume that the patient’s U1 snRNP complex somehow became the target of the autoimmune response. Although the underlying cause for this is still unknown, there are indications that the group of patients with anti-U1 RNA autoantibodies is immunogenetically and clinically distinct from the anti-U1 RNA-negative patient group (5). Anti-U1 RNA Abs have been reported to be present in the more severely affected patient group (5) and, as is the case with anti-dsDNA Ab, the anti-U1 RNA Ab titer appears to correlate with disease activity (9). Also HLA-DR2/DR4, Raynaud’s phenomenon, and synovitis are significantly increased in the anti-U1 RNA-positive patient group (5). It is clear that further studies are needed to explain why these Abs develop. VH gene restriction Both anti-U1 stem loop II specific Abs use the same germline VH gene (DP-35, VH3-11). Although several previous reports suggested an overexpression of particular V genes coding for autoantibodies (26, 27), more recent studies seem to support the idea that the frequency of VH gene usage in autoantibodies is comparable with that of the normal repertoire (19). For example, in the case of anti-DNA autoanti-

bodies no particular overexpression of VH genes could be found (28). Nevertheless, when one focuses on one particular autoantigen, preferential VH gene usage sometimes seems to occur (as in the case of the U1A and U1C autoantigens (17, 21, 29)), which could be explained by the fact that these VH genes probably have a natural fit for the immunodominant epitopes present on these autoantigens, resulting in their selection by the immune system for affinity maturation.

VL gene restriction Both anti-stem loop II Abs were derived from the same l light chain (3r). This light chain gene is used in ,1% of the cases in the normal IgG repertoire, and after sequencing the light chains of 100 IgG-positive B cells from patient Z5 this particular light chain was not found at all. This means that there is no general overexpression of this light chain V gene product, neither in normal individuals (see also Ref. 19) nor in this patient. It is generally accepted that the heavy chain plays a major role in Ag recognition, although there are several reports that indicate that the light chain is important for Ag binding as well. Collet et al. (30) reported that when a number of anti-HIV p120-positive VH genes were reshuffled with different VLs, 43 to 100% of the VLs supported binding to p120, and this strongly depended on the par-

The Journal of Immunology

3311

Table III. Reactivity of the U1 RNA-positive clones containing reshuffled light chains

Clone

Z5, scFv3 (original) Z5, scFv7 (original) Unselectedb V16 V17 V18 V19 Stem loop II selectedc II-3.1 II-3.2/3.4/3.5 II-3.3 II-3.6 II-3.7 II-3.8 II-3.9 II-7.1 II-7.2 II-7.3 II-7.4 II-7.5 II-7.6 II-7.7 II-7.8 Stem loop IV selectedc IV-3.1 IV-3.2 IV-7.1 IV-7.2 IV-7.3

Germline Gene

aa Mutationsa

U1 RNA Dot Blot

U1 RNA IP

Stem Loop II IP

Stem Loop IV IP

Anti-VSV

U3 RNA IP

3r (l) 3r (l)

12 21

111 11

11 11

11 11

2 2

11 1/2

2 2

L16 (k) 3r (l) 6a (l) 2a2 (l)

13 13 4 8

11 111 1 111

2 11 2 11

2 11 2 1/2

2 2 2 111

1/2 11 1/2 111

2 2 2 111

3r (l) 3r (l) 6a (l) 3r (l) 3r (l) 3r (l) A19 (k) L6 (k) 3r (l) 3l (l) 3r (l) 6a (l) 2A2 (l) 4b (l) 3r (l)

20 24 7 27 19 23 8 8 12 5 6 8 16 29 17

111 1 1 1 111 11 1 11 1 1 1 1 1 1 1

11 1 2 1/2 11 11 2 1 1 1/2 2 2 2 2 1

11 1 2 1/2 11 11 2 2 1 1/2 2 2 2 2 11

2 2 2 2 2 2 2 1 2 2 2 2 2 2 2

1 1/2 2 2 11 1 11 1 1/2 1/2 2 1/2 2 2 1/2

2 2 2 2 1/2 2 2 1 2 2 2 2 2 2 2

L6 (k) A27 (k) 3l (l) 5c (l) 3r (l)

2 7 9 6 11

1 1 11 111 111

1 1 11 11 11

2 2 2 2 111

1 1 11 11 1/2d

1 1 11 11 11

1 1 11 11 1/2d

a aa mutations, number of amino acid mutations in V gene compared with most homologous germline gene; U1 RNA dot blot, reactivity to U1 RNA was measured using a filter binding (dot blot) assay (9); IP, immunoprecipitation; Anti-VSV, relative expression levels of scFv was determined via anti-VSV reactivity; b V16 –V19 are positive clones from the unselected library reactive with U1 RNA in U1 RNA filter binding (dot blot) analyses. c Clones II-3.1–II-3.9 are U1 RNA-positive clones derived from Z5scFv3, and II-7.1–II-7.8 are positive clones derived from Z5scFv7 after selection on stem loop II, whereas IV-3.1, IV-3.2, IV-7.1, IV-7.2, and IV-7.3 are positive clones obtained via selection on stem loop IV. d Clone IV-7.3 was positive on both stem loop II, stem loop IV as well as U3 RNA when 150 mM NaCl was used in immunoprecipitation, but only reactive with stem loop II RNA at 350 mM salt.

ticular heavy chain sequence. Ohlin et al. (31) reported that the light chain could be of influence for the fine specificity of an Ab directed to cytomegalovirus Gb. Murine anti-DNA Abs have been extensively studied in MRL/lpr mice, and although the heavy chain plays in most cases a major role the light chain is important as well (see, e.g., Refs. 32–36). To study the influence of the light chain in

FIGURE 3. Light chain V gene usage of reshuffled scFvs. The light chain V gene usage is shown for nonselected scFvs (random reshuffled light chains; V1–V15 in Table II), for U1 RNA binding scFvs (II-3.1–II3.9, II-7.1–II-7.8, V16 –V19), for scFvs specifically binding to stem loop II of U1 RNA and of scFvs binding to stem loop IV but not to stem loop II (designated stem loop IV binders).

the anti-U1 RNA response, we performed reshuffling experiments with the total light chain repertoire of the patient. When selecting the reshuffled libraries with stem loop II RNA as Ag, we obtained only stem loop II specific binders when the VH chain had been combined with a 3r-derived light chain. A large number of different 3r-derived light chains were found, all having different somatic mutations (Table II), and most of them also containing several positively charged amino acids in their CDRs. All of the clones also contained a serine at position 34 instead of a cysteine. This may be an indication for a genetic polymorphism. Alternatively, this might be a somatic mutation important for stem loop II RNAspecific binding. This possibility could not be verified because this light chain was not found to be used in the 100 IgG-positive B cell clones from this patient that were analyzed (19). The striking finding that a combination of VH3-11 with another light chain than l 3r resulted in an Ab with another RNA recognition specificity was confirmed when selections were performed with the other autoepitope of U1 RNA, i.e., stem loop IV. Most clones that were obtained reacted with stem loop IV, although other RNAs like U3 RNA were recognized as well. In contrast to the situation for stem loop II (4, 25), the stem loop IV epitope has not been characterized well. Correct folding of the stem loop structure is necessary for autoantibody recognition, and the upper part of the stem and the whole loop are the main regions targeted by patient autoantibodies (4). The stem loop IV loop contains only four nucleotides (UUCG), which are stacked and form an extremely stable structure, which is also present in several unrelated RNAs. However, U3 RNA does not contain this

3312 sequence, suggesting that this element is not important for the recognition by the single-chain Abs selected by stem loop IV. Because the Abs selected from the reshuffled libraries are derived from scFvs (scFv3 and scFv7) recognizing an epitope composed of mainly double-stranded RNA (25), it is more likely that they target the doublestranded region of stem loop IV of U1 RNA and similar structures in other RNAs, e.g., U3 RNA. In conclusion, the light chain seems to play a very important role in the recognition of RNA by the Ab, and the specificity of anti-U1 RNA Abs can be altered dramatically by replacement of the light chain. Recently, we reported that secondary light chain rearrangements are likely to occur relatively frequently in peripheral organs during or after the process of hypermutation (19). It might be possible that autoreactive B cells, once they have been triggered to proliferate, attempt to prevent autoreactivity by changing their light chain. This could, however, also lead to reactivity to other closely related proteins or nucleic acids. Therefore, it seems possible that via such secondary light chain rearrangements U1 RNA reactive B cells can change not only the affinity but also the specificity for its anti-RNA Ab. Our in vitro light chain reshuffling experiments show that combining a highly mutated VH gene with another light chain indeed can abolish binding to its original RNA target (e.g., stem II of U1 RNA) and enhance the affinity for other RNAs (e.g., loop IV of U1 RNA). It is tempting to speculate that secondary light chain rearrangements might be a way for the immune system to accomplish the so-called “epitope spreading” phenomenon. Autoantibodies to stem loop II of U1 RNA could via light chain rearrangements evolve into anti-stem loop IV or nonspecific anti-RNA autoantibodies or vice versa. The fact that many SLE patients produce autoantibodies to both stem loops II and IV of U1 RNA supports this hypothesis. It is even possible that such a mechanism is responsible for the development of Abs that could play a role in the pathogenesis of the disease.

Acknowledgments We thank Dr. T. van Kuppeveld (Department of Biochemistry, University of Nijmegen, The Netherlands) for providing us with “RNA-magic coating”; patients Z5, D18, HO, and O11 for donating bone marrow samples that were essential for this study; and Dr. W. Ouwehand (Blood Transfusion Centre, Cambridge, U.K.) for providing us with the pHENIX vector.

References 1. Klein Gunnewiek, J. M., L. B. van de Putte, and W. J. van Venrooij. 1997. The U1 snRNP complex: an autoantigen in connective tissue diseases: an update. Clin. Exp. Rheumatol. 15:549. 2. Deutscher, S. L., and J. D. Keene. 1988. A sequence-specific conformational epitope on U1 RNA is recognized by a unique autoantibody. Proc. Natl. Acad. Sci USA 85:3299. 3. Van Venrooij, W. J., R. Hoet, J. Castrop, B. Hageman, I. W. Mattaj, and L. B. van de Putte. 1990. Anti-(U1) small nuclear RNA antibodies in anti-small nuclear ribonucleoprotein sera from patients with connective tissue diseases. J. Clin. Invest. 86:2154. 4. Hoet, R. M., P. De Weerd, J. K. Gunnewiek, I. Koornneef, and W. J. Van Venrooij. 1992. Epitope regions on U1 small nuclear RNA recognized by anti-U1RNA-specific autoantibodies. J. Clin. Invest. 90:1753. 5. Hoffman, R. W., G. C. Sharp, and S. L. Deutscher. 1995. Analysis of anti-U1 RNA antibodies in patients with connective tissue disease: association with HLA and clinical manifestations of disease. Arthritis Rheum. 38:1837. 6. Tsai, D. E., and J. D. Keene. 1993. In vitro selection of RNA epitopes using autoimmune patient serum. J. Immunol. 150:1137. 7. Keene, J. D. 1996. RNA surfaces as functional mimetics of proteins. Chem. Biol. 3:505. 8. St Clair, E. W., and J. A. Burch, Jr. 1996. In vitro RNA selection of an autoimmune epitope on stem-loop II of U1 RNA. Clin. Immunol. Immunopathol. 79:60. 9. Hoet, R. M., I. Koornneef, D. J. de Rooij, L. B. van de Putte, and W. J. van Venrooij. 1992. Changes in anti-U1 RNA antibody levels correlate with disease activity in patients with systemic lupus erythematosus overlap syndrome. Arthritis Rheum. 35:1202. 10. Hoet, R. M., B. Kastner, R. Luhrmann, and W. J. van Venrooij. 1993. Purification and characterization of human autoantibodies directed to specific regions on U1RNA; recognition of native U1RNP complexes. Nucleic Acids Res. 21:5130.

RECOMBINANT ANTI-U1 RNA Abs FROM SLE PATIENTS 11. Uchiumi, T., R. R. Traut, K. Elkon, and R. Kominami. 1991. A human autoantibody specific for a unique conserved region of 28 S ribosomal RNA inhibits the interaction of elongation factors 1 a and 2 with ribosomes. J. Biol. Chem. 266:2054. 12. Bunn, C. C., and M. B. Mathews. 1987. Autoreactive epitope defined as the anticodon region of alanine transfer RNA. Science 238:1116. 13. Wilusz, J., and J. D. Keene. 1986. Autoantibodies specific for U1 RNA and initiator methionine tRNA. J. Biol. Chem. 261:5467. 14. Brouwer, R., W. Vree Egberts, P. H. Jongen, B. G. van Engelen, and W. J. van Venrooij. 1998. Frequent occurrence of anti-tRNA(His) autoantibodies that recognize a conformational epitope in sera of patients with myositis. Arthritis Rheum. 41:1428. 15. Swaak, A. J., J. Groenwold, and W. Bronsveld. 1986. Predictive value of complement profiles and anti-dsDNA in systemic lupus erythematosus. Ann. Rheum. Dis. 45:359. 16. Finnern, R., E. Pedrollo, I. Fisch, J. Wieslander, J. D. Marks, C. M. Lockwood, and W. H. Ouwehand. 1997. Human autoimmune anti-proteinase 3 scFv from a phage display library. Clin. Exp. Immunol. 107:269. 17. Hoet, R. M. A., J. Raats, R. De Wildt, H. Dumortier, S. Muller, F. Van den Hoogen, and W. J. Van Venrooij. 1999. Human monoclonal autoantibody fragments from combinatorial antibody libraries directed to the U1snRNP associated U1C protein; epitope mapping, immunolocalization and V-gene usage. Mol. Immunol. 35:1045. 18. De Wildt, R. M., P. G. Steenbakkers, A. H. Pennings, F. H. van den Hoogen, W. J. van Venrooij, and R. M. Hoet. 1997. A new method for the analysis and production of monoclonal antibody fragments originating from single human B cells. J. Immunol. Methods 207:61. 19. De Wildt, R. M. T., R. M. A. Hoet, W. J. Van Venrooij, I. M. Tomlinson, and G. Winter. 1999. Analysis of heavy and light chain pairings indicates that receptor editing shapes the human antibody repertoire. J. Mol. Biol. 285:895. 20. Van Kuppevelt, T. H., M. A. Dennissen, W. J. van Venrooij, R. M. Hoet, and J. H. Veerkamp. 1998. Generation and application of type-specific anti-heparan sulfate antibodies using phage display technology: further evidence for heparan sulfate heterogeneity in the kidney. J. Biol. Chem. 273:12960. 21. De Wildt, R. M., R. Finnern, W. H. Ouwehand, A. D. Griffiths, W. J. van Venrooij, and R. M. Hoet. 1996. Characterization of human variable domain antibody fragments against the U1 RNA-associated A protein, selected from a synthetic and patient-derived combinatorial V gene library. Eur. J. Immunol. 26:629. 22. Pluk, H., J. Soffner, R. Luhrmann, and W. J. van Venrooij. 1998. cDNA cloning and characterization of the human U3 small nucleolar ribonucleoprotein complex-associated 55-kilodalton protein. Mol. Cell. Biol. 18:488. 23. Griffiths, A. D., S. C. Williams, O. Hartley, I. M. Tomlinson, P. Waterhouse, W. L. Crosby, R. E. Kontermann, P. T. Jones, N. M. Low, T. J. Allison, et al. 1994. Isolation of high affinity human antibodies directly from large synthetic repertoires. EMBO J. 13:3245. 24. Nissim, A., H. R. Hoogenboom, I. M. Tomlinson, G. Flynn, C. Midgley, D. Lane, and G. Winter. 1994. Antibody fragments from a “single pot” phage display library as immunochemical reagents. EMBO J. 13:692. 25. Teunissen, S. W., M. H. Stassen, G. J. M. Pruijn, W. J. van Venrooij, and R. M. Hoet. 1998. Characterization of an anti-RNA recombinant autoantibody fragment (scFv) isolated from a phage display library and detailed analysis of its binding site on U1 snRNA. RNA 4:1124. 26. Bensimon, C., P. Chastagner, and M. Zouali. 1994. Human lupus anti-DNA autoantibodies undergo essentially primary V k gene rearrangements. EMBO J. 13:2951. 27. Logtenberg, T. 1994. How unique are pathogenic anti-DNA autaonatibody V regions? Curr. Opin. Immunol. 6:921. 28. Isenberg, D., M. A. Rahman, C. T. Ravirajan, and J. K. Kalsi. 1997. Anti-DNA antibodies: from gene usage to crystal structures. Immunol. Today 18:149. 29. De Wildt, R. M. T., R. Ruytenbeek, W. J. Van Venrooij, and R. M. A. Hoet. 1997. Heavy chain CDR3 optimization of a germline encoded recombinant antibody fragment predisposed to bind the U1A protein. Protein Eng. 10:835. 30. Collet, T. A., P. Roben, C. F. Barbas, D. R. Burton, and R. A. Lerner. 1992. A binary plasmid system for shuffling combinatorial antibody libraries. Proc. Natl. Acad. Sci. USA 89:10026. 31. Ohlin, M., H. Owman, M. Mach, and C. A. Borrebaeck. 1996. Light chain shuffling of a high affinity antibody results in a drift in epitope recognition. Mol. Immunol. 33:47. 32. Retter, M. W., R. A. Eisenberg, P. L. Cohen, and S. H. Clarke. 1995. Sm and DNA binding by dual reactive B cells requires distinct VH, V k, and VH CDR3 structures. J. Immunol. 155:2248. 33. Ibrahim, S. M., M. Weigert, C. Basu, J. Erikson, and M. Z. Radic. 1995. Light chain contribution to specificity in anti-DNA antibodies. J. Immunol. 155:3223. 34. Jang, Y. J., D. Sanford, H. Y. Chung, S. Y. Baek, and B. D. Stollar. 1998. The structural basis for DNA binding by an anti-DNA autoantibody. Mol. Immunol. 35:1207. 35. Friedmann, D., N. Yachimovich, G. Mostoslavsky, Y. Pewzner-Jung, A. Ben-Yehuda, K. Rajewsky, and D. Eilat. 1999. Prodution of high affinity autoantibodies in autoimmune New Zealand Black/New Zealand White F1 mice targeted with an anti-DNA heavy chain. J. Immunol. 162: 4406. 36. Chen, Y., and B. D. Stollar. 1999. DNA binding by the VH domain of anti-ZDNA antibody and its modulation by association of the VL domain. J. Immunol. 162: 4663. 37. Tomlinson, I. M., S. C. Williams, O. Ignatovich, S. J. Corbett, and G. Winter. 1998. V Base Sequence Directory. MRC Centre for Protein Engineering, Cambridge, U.K. http://www.mrc-cpe.cam.ac.uk/imt-doc/index.html