Early targets of nuclear RNP humoral autoimmunity in human systemic lupus erythematosus

Authors

  • Brian D. Poole,

    1. Oklahoma Medical Research Foundation, and University of Oklahoma Health Sciences Center, Oklahoma City, and Brigham Young University, Provo, Utah
    Search for more papers by this author
  • Rebecca I. Schneider,

    1. Oklahoma Medical Research Foundation, and University of Oklahoma Health Sciences Center, Oklahoma City, and University of Missouri, Columbia
    Search for more papers by this author
  • Joel M. Guthridge,

    1. Oklahoma Medical Research Foundation, Oklahoma City
    Search for more papers by this author
  • Cathy A. Velte,

    1. Oklahoma Medical Research Foundation, Oklahoma City
    Search for more papers by this author
  • Morris Reichlin,

    1. Oklahoma Medical Research Foundation, Oklahoma City
    Search for more papers by this author
  • John B. Harley,

    1. Oklahoma Medical Research Foundation, University of Oklahoma Health Sciences Center, and Veterans Affairs Medical Center, Oklahoma City
    Search for more papers by this author
    • Dr. Harley has received consulting fees, speaking fees, and/or honoraria from Bio-Rad Laboratories, ImmunoVision, Inc., and IVAX Diagnostics (more than $10,000 each) and owns stock or stock options in IVAX Diagnostics.

  • Judith A. James

    Corresponding author
    1. Oklahoma Medical Research Foundation, and University of Oklahoma Health Sciences Center, Oklahoma City
    • Oklahoma Medical Research Foundation, 825 NE 13th Street, Oklahoma City, OK 73104
    Search for more papers by this author

Abstract

Objective

The U1 small nuclear RNPs are common targets of autoantibodies in lupus and other autoimmune diseases. However, the etiology and progression of autoimmune responses directed against these antigens are not well understood. The aim of this study was to use a unique collection of serial samples obtained from patients before and after the development of nuclear RNP (nRNP) antibodies to investigate early humoral events in the development of anti-nRNP autoimmunity.

Methods

Lupus patients with sera available from both before and after the development of nRNP antibody precipitin were identified from the Oklahoma Clinical Immunology Serum Repository. Antibodies in the serial samples were analyzed by enzyme-linked immunosorbent assay, Western blotting, solid-phase epitope mapping, and competition assays.

Results

The first-detected nRNP antibodies targeted 6 common initial epitopes in nRNP A, 2 in nRNP C, and 9 in nRNP 70K. The initial epitopes of nRNP A and nRNP C were significantly enriched for proline and shared up to 95% sequence homology. The initial nRNP 70K humoral epitopes differed from those of nRNP A and nRNP C. The initial antibodies to nRNP A and nRNP C were cross-reactive with the SmB′-derived peptide PPPGMRPP. Antibody binding against all 3 nRNP subunits diversified significantly over time.

Conclusion

Autoantibodies to nRNP A and nRNP C initially targeted restricted, proline-rich motifs. Antibody binding subsequently spread to other epitopes. The similarity and cross-reactivity between the initial targets of nRNP and Sm autoantibodies identifies a likely commonality in cause and a focal point for intermolecular epitope spreading.

Systemic lupus erythematosus (SLE) is a systemic autoimmune disease of complex and incompletely understood cause (for review, see refs. 1 and2). Antibodies against the nuclear RNP (nRNP) complex are found in the sera of 21–47% of SLE patients, and antibodies against the Sm antigen are found in the sera of ∼5–30% of SLE patients (3–5). The components of the nRNP complex, consisting of nRNP 70K, nRNP A, and nRNP C, may each be targeted by antibodies. In contrast to Sm autoantibodies, which are found almost exclusively in SLE patients, nRNP autoantibodies are often detected in patients with other autoimmune disorders, including mixed connective tissue disease (MCTD), Raynaud's phenomenon, and scleroderma. Antibodies against nRNP 70K are normally associated with MCTD, while antibodies against the other subunits are more common in SLE (6, 7).

The development of anti-nRNP antibodies in human SLE is inadequately understood despite considerable effort. Studies of murine models implicate both nRNP A and nRNP 70K in the initiation of nRNP antibodies (8, 9). A study investigating the sequence of nRNP antibody development in human rheumatic disease showed that nRNP 70K antibodies appear first (10). However, that study included both SLE patients and other subjects with nRNP antibodies, possibly obscuring the primary pathway in SLE. Epitopes of nRNP proteins targeted in an established autoimmune response have been mapped (11–22), but the key initial epitopes, which herald the onset of the loss of tolerance, have not been identified or investigated in detail. The understanding of the human SLE-specific pattern of nRNP antibody development is therefore still far from complete.

More is known about the initial humoral immune response to SmB′ in SLE patients. SmB′ antibodies initially target the amino acid sequence PPPGMRPP (23–25), and then diversify first to a repeated, proline-rich region, eventually targeting a variety of autoantigens in a process called epitope spreading (26–28). The close physical proximity and regions of similar amino acid sequences of Sm and nRNP proteins, as well as the temporal linkage of the appearance of antibodies to these proteins, suggest that autoantibodies to nRNP and SmB′ may have common originating events in select subsets of SLE patients.

The availability of serial samples of serum from SLE patients who had at least 1 sample that was negative for nRNP autoimmunity and a later sample that was positive for nRNP antibodies provided a unique opportunity to examine the initiation and development of nRNP antibodies. Our experiments evaluated the hypotheses that nRNP humoral autoimmunity in SLE begins with a limited number of epitopes, that this response diversifies over time, and that the development of nRNP and Sm humoral autoimmunity are intertwined in a subset of individuals.

PATIENTS AND METHODS

Patient sera.

This project was performed in accordance with the Declaration of Helsinki and was approved by the Institutional Review Boards of the Oklahoma Medical Research Foundation (OMRF) and the Oklahoma University Health Sciences Center (OUHSC). Uniquely coded, serial serum samples from 20 patients who were classified has having SLE according to the American College of Rheumatology (ACR) revised classification criteria (29, 30) and who had samples available from both before and after the development of nRNP autoantibodies were obtained from the Oklahoma Clinical Immunology Serum Repository (Oklahoma City, OK). Sera from more than 27,000 individuals with rheumatic diseases, of which 1,794 were positive for nRNP antibodies at some time point, have been collected in this repository.

The 20 patients whose sera we studied had a mean of 6.3 samples available (range 2–19), spanning an average of 8.0 years (range 2–20 years). All patients met the ACR criteria for SLE (29, 30). Patients with MCTD who did not meet at least 4 of the ACR classification criteria for SLE were excluded from the study. Sera from 11 unaffected volunteer donors were used as controls.

Autoantibody analysis.

All samples were tested for autoantibodies against nRNP by Ouchterlony immunodiffusion assays (31), as well as by standard enzyme-linked immunosorbent assay (ELISA) as described previously (14, 32). Briefly, 1 μg of nRNP antigen purified from calf thymus (ImmunoVision, Springdale, AR) was coated onto each well of 96-well polystyrene plates. Serum samples were diluted 1:100 and 1:1,000 and incubated in the antigen-coated wells for 2 hours at room temperature. Alkaline phosphatase–conjugated goat anti-human IgG (Jackson ImmunoResearch, West Grove, PA) was allowed to bind for 16 hours at 4°C, followed by the addition of paranitrophenyl phosphate disodium (PNPP; Sigma-Aldrich, St. Louis, MO) as the colorimetric substrate. Absorbance was measured at 410 nm with a Dynex MRX II ELISA reader (Dynex, Alexandria, VA), when the OD405 of the positive control was 1.0, which enabled standardization of results among multiple plates.

Protein expression and peptide construction.

Full-length nRNP A (gene ID 6626) cloned into the pRSET-B vector (Invitrogen, Carlsbad, CA) was received from Dr. Carol Lutz (University of Medicine and Dentistry of New Jersey, Newark, NJ) (33) and transformed into BL21 gold Escherichia coli (Stratagene, Carlsbad, CA). The coding sequence for nRNP 70K (gene ID 6625) was cloned from HeLa complementary DNA by polymerase chain reaction using the primer set 5′-ACG-CGT-CGA-CAT-GAC-CCA-GTT-CCT-GCC-GC-3′ (forward) and 5′-ATA-AGA-ATG-CGG-CCG-CTC-ACT-CCG-GCG-CAG-CCT-C-3′ (reverse), and then ligated into the expression vector pET28b (Novagen, Carlsbad, CA). The protein was expressed in BL21Star competent cells (Invitrogen). The plasmid containing the nRNP C (gene ID 6631) insert was obtained from American type Culture Collection as IMAGE clone no. 6391327 and was ligated into the expression vector pET28c at the Sal I and Not I sites and expressed in BL21Gold (DE3) pLysS competent cells (Stratagene). All proteins were expressed with 6 His tags. Expression of recombinant antigens was induced with 0.5 mM isopropyl β-D-thiogalactoside for 4 hours at 37°C. After sonication, the expressed recombinant proteins were purified from the cells by using nickel–nitrilotriacetic acid affinity columns (Qiagen, Valencia, CA) (33).

The linear peptides PPPGMMPV (nRNP C109–116), KIPPTPFS (nRNP C61–68), GPAPGMRPP (nRNP C117–125), MPPPGMIPP (nRNP A164–172), PPPGLAPG (nRNP A171–178), PRDPIPYLPP (nRNP 70K15–24), EFEDRPDAPPTRA (nRNP 70K47–59), and ERPGPSPLPH (nRNP 70K221–230) were used for inhibition analysis of the nRNP autoantibodies. The Sm-derived linear peptide PPPGMRPP (SmB′191–198) was used to test for cross-reactivity, and PPPPPPPP (poly-proline) was used as a negative control. Peptides were synthesized by the OUHSC Molecular Biology and Proteomics Facility.

Detection of nRNP A, C, or 70K by Western blot analysis.

HeLa cell extracts or recombinant nRNP A, C, or 70K proteins were subjected to electrophoresis in 12.5% polyacrylamide gels and transferred to a nitrocellulose membrane. Consistent protein loading and transfer were confirmed by staining with fast green (Fisher Scientific, Pittsburgh, PA). Blots were cut into strips and blocked with 3% (weight/volume) dry milk in Tris buffered saline. Sera were incubated at a 1:200 dilution with individual membrane strips for 2 hours at room temperature. After washing, alkaline phosphatase–conjugated goat anti-human IgG was added for 3 hours. Serum antibody binding to specific proteins was visualized by the addition of the nitroblue tetrazolium/BCIP substrate (Fisher Scientific, Waltham, MA).

Western blot analysis was used to determine peptide inhibition of antibody binding. Recombinant nRNP A, nRNP C, or nRNP 70K was subjected to electrophoresis and transferred to nitrocellulose membranes as described above. Serum samples were diluted from 1:50 to 1:20,000 and then incubated with 40 μg/ml of peptide for 2 hours at room temperature. A concentration of 40 μg/ml was experimentally determined to be the penultimate dilution in a dilution series of 10, 20, and 40 μg/ml of peptide (data not shown). In additional experiments, the nRNP 70K–positive samples were also incubated with poly-lysine (Fisher Scientific) at concentrations of 100, 500, and 1,000 μg/ml. The peptide-preadsorbed sera were incubated with the nitrocellulose-bound antigens, and antibody binding was detected as described above.

Blots were scanned, and band intensity was quantified by using Scion Image software for Windows (Scion, Frederick, MD). Western blot experiments were performed in triplicate, with average values used for comparisons. Inhibition is likely to be underrepresented due to the compressed dynamic range of the Western blot readout. Indeed, incubation with full-length recombinant nRNP at 40 μg/ml only showed a mean inhibition of 41%.

Solid-phase peptide synthesis and autoantibody assays.

Decapeptides overlapping by 8 amino acids, representing the protein sequences of nRNP 70K (locus NP_003080) (34) and nRNP C (locus PO9234) (35), as well as maximally overlapping octapeptides encompassing the sequence of nRNP A (locus PO9012) (36), were constructed on the ends of polyethylene pins by using FMOC solid-phase peptide chemistry, as previously described (14, 23). Serum antibody binding to each peptide in the solid-phase assay was detected with a modified ELISA technique. Briefly, individual solid-phase peptides were incubated with patient sera for 2 hours at room temperature. Peptides were washed and then incubated overnight at 4°C with an anti-human IgG alkaline phosphatase conjugate (Jackson ImmunoResearch). After washing, peptides were incubated at 37°C with PNPP substrate until the positive control wells had an OD405 of 1.0. A well-characterized positive control serum was used to normalize the results among multiple plates.

Statistical analysis.

Antibody binding to peptides in the solid-phase assays was considered significant if the mean absorbance of the positive samples was at least 3 SD above the mean absorbance in the control samples. Epitopes were defined as one or more positively bound, overlapping peptides. Chi-square analysis was used to determine significant differences in amino acid prevalence and the number of peptides recognized by initial and late serum samples. The Mann-Whitney U test was used to compare inhibition of antibody binding. Comparisons that resulted in a P value of less than 0.05 were considered significant.

RESULTS

Identification of sera containing early nRNP antibodies.

We identified 20 SLE patients for whom serum samples were available from both before and after the appearance of nRNP antibodies, as determined by precipitin reaction. These samples were from 11 African American, 6 European American, 2 American Indian, and 1 Asian SLE patient; 18 of them were women, and 2 were men. Their mean ± SD age was 43.5 ± 12.6 years. The patients met an average of 5.7 (range 4–10) ACR criteria for SLE (29, 30).

The presence of anti-nRNP antibodies was confirmed by nRNP antigen–specific ELISA and Western blot analysis. For every case, at least 1 sample was observed to be negative for nRNP antibodies by precipitin reaction, ELISA, and Western blotting, and nRNP antibodies were detected in at least 1 subsequent sample by precipitin reaction and ELISA. The mean ± SD time between the last negative sample and the first positive sample was 3.5 ± 2.7 years (range 2 months to 9.8 years). Patients were observed for 1–14 years after the development of anti-nRNP. No significant differences in the times of anti-nRNP appearance were noted when the methods of nRNP antibody detection were compared.

Western blot analysis with HeLa cell extracts as the antigen source was used to detect antibodies against the nRNP subunits nRNP A, nRNP C, and nRNP 70K. Antibodies against either a single nRNP protein (8 of 15 samples) or multiple proteins simultaneously (7 of 15 samples) were common in the initial nRNP-positive serum samples. Antibodies against single nRNP proteins were most commonly targeted to nRNP A (4 of 8 samples), with 2 patients each developing initial antibodies against only nRNP C or nRNP 70K. Five samples that were positive for nRNP by precipitin reaction and ELISA did not have detectable antibodies against nRNP subunits in denaturing Western blots. Antibody binding to nRNP subunits diversified in subsequent samples, with 11 of 15 patients positive for antibodies to multiple subunits in the last available samples. Of the patients with antibodies against a single subunit, 1 had antibodies to nRNP C, and 3 had antibodies to nRNP A.

Early nRNP A humoral fine specificity is directed against proline-rich epitopes.

Sufficient quantities of serum were available to map the initial binding of the first nRNP A–positive serum samples from 6 SLE patients by solid-phase peptide analysis. The autoantibodies detected in these patients, along with the sequence of autoantibody appearance are provided in Table 1.

Table 1. Antibody profiles of patients whose sera were used in solid-phase epitope-mapping studies*
Patient/race/sexAntibody subunit, by solid-phase assaySequence of antibody appearance, by precipitin reactionSequence of nRNP subunit appearanceTime between negative and positive sera, years
  • *

    nRNP = nuclear RNP; AA = African American; EA = European American; A = Asian; AI = American Indian.

1/AA/FA, C, 70KRo then nRNPA, C, and 70K together1.17
2/EA/FAnRNP onlyA and 70K together2.75
3/AA/FA, C, 70KSm/nRNP togetherA, C, and 70K together9.8
4/AA/FAnRNP onlyA, C, and 70K together6.25
5/AA/FA, C, 70KSm/nRNP together70K then A and C together2.0
6/AA/MAnRNP onlyA only3.0
7/A/FCRo then nRNPC only1.17
8/AI/FCnRNP onlyA, C, and 70K together1.75
9/AA/FCRo then nRNPC then 70K0.5

Antibody binding was analyzed for each unique octapeptide in the nRNP A sequence (Figure 1). The mean antibody-binding pattern in the SLE patients was compared with that in 4 normal volunteer donors and in 3 anti-nRNP–negative individuals who later developed anti-nRNP.

Figure 1.

Sequence of development of autoantibodies to nuclear RNP (nRNP) proteins. Serial serum samples from 6 patients with nRNP A–specific, 6 patients with nRNP C–specific, and 3 patients with nRNP 70K–specific antibodies collected concurrent with and after the first observation of antibodies were analyzed by solid-phase peptide mapping to identify initial humoral epitopes. Serum antibody binding to each overlapping octapeptide in the sequence of nRNP A and each overlapping decapeptide in the sequences of nRNP C and nRNP 70K was measured. The development of autoantibodies to nRNP proteins began by targeting a limited number of epitopes and then expanded. Representative antibody binding profiles are shown.

Twelve peptides comprising 6 epitopes were identified from the first nRNP A–positive samples (Figure 2A). The initial nRNP A epitopes were significantly enriched for proline (χ2 = 12.5, P = 0.0004) as compared with the peptides that were not targeted by autoantibodies. No other amino acids were significantly enriched in the early epitopes. The most commonly bound epitope GQPPYMPPPGMIPPPGLAPG (amino acids 159–178) was recognized by 5 of 6 (83.3%) of the nRNP-positive samples. The only sample that did not bind to this peptide did not bind significantly to any sequential peptide sequence tested. This sequence, especially PPPGMIPP, bears a striking resemblance to the prominent initial epitope PPPGMRP(G)P found in SmB′.

Figure 2.

Identification of sequential humoral epitopes in early and late nuclear RNP A (nRNP A) humoral autoimmunity. A, Early pattern of antibody binding to the sequential overlapping peptides of nRNP A in the first nRNP A antibody–positive sera. Solid-phase peptide octamers were used to map the pattern of antibody binding to nRNP A. Six epitopes (defined as overlapping, significantly bound peptide sequences) recognized by the first nRNP A–positive serum antibodies (n = 6) are shaded and numbered. Values are the number of SDs above the mean absorbance in controls (n = 7); only epitopes exhibiting an increased absorbance of at least 3 SD are shown. B, Amino acid positions, numbers of samples showing nRNP A antibody binding, and sequences of the 6 epitopes identified in A. C, Inhibition of antibody binding to nRNP A accounted for by the dominant first epitope. Antibody binding was quantified by blocking the binding of epitope 4 with 2 peptides representing epitope 4 and then detecting residual binding in Western blots. Significant inhibition as compared with the poly-proline negative control was detected when peptide nRNP A164–172 peptide (A 164; P = 0.026) or a mixture of both nRNP A164–172 and nRNP A171–178 (A 171) peptides (P = 0.011) was incubated with the patient sera (n = 9). Horizontal bars show the mean. D, Late pattern of antibody binding in the last available serum sample after the development of nRNP A antibodies (n = 6). Shaded areas and numbers show the antibody binding pattern in the initial positive samples for comparison. Values are the number of SDs above the mean absorbance in controls (n = 7).

To determine the proportion of the anti-nRNP A initial antibody response accounted for by binding to the epitope at amino acids 159–178, the peptides nRNP A164–172 and nRNP A171–178 were tested for their ability to inhibit binding to recombinant nRNP A. Binding was significantly inhibited by both nRNP A164–172 and by a combination of both peptides; mean inhibition was 17.0% for nRNP A164–172 (P = 0.026) and 33.2% for the combination of both peptides (P = 0.011). Up to 75% inhibition of binding in individual samples was observed (Figure 2C).

Evolution of the humoral fine specificity for several lupus autoantigens has been described previously (25, 26, 37–39). The extent of epitope spreading in antibodies to nRNP A was examined by comparing the initial nRNP antibody-positive sample with the antibody binding profile in sera collected a mean ± SD of 2.3 ± 1.4 years after the initial sample (Figure 1). The later response was significantly more diverse than the initial response; 40 of the 276 octapeptides bound in the later response and 12 octapeptides were identified from the first nRNP A–positive samples (χ2 = 6.068, P = 0.014) (Figure 2D). The pattern of antibody binding, with prominent binding to the N-terminus, that was found in the late sera was substantially different from that of the initial sera.

Initial nRNP C humoral epitopes are limited and enriched for proline.

Antibodies to nRNP C were most commonly found in tandem with nRNP A antibodies. The sequence of nRNP C resembles that of nRNP A in terms of the prevalence of large proline-rich regions. Initial autoantibody binding to nRNP C was examined by using solid-phase mapping of sera with the first detectable nRNP C antibodies, which were collected a mean ± SD of 3.6 ± 3.4 years after the last negative sample. The mean antibody binding pattern in the initial nRNP C–positive sera from 6 SLE patients was compared with the mean antibody pattern in 5 normal controls and in 6 SLE patients before onset of nRNP C antibodies (Figure 1). The autoantibody development pattern found in these patients is provided in Table 1.

Epitope mapping revealed significant antibody binding to only 2 initial epitopes of nRNP C (Figure 3). These epitopes are located at amino acids 57–66 (FQQGKIPPTP) and amino acids 109–128 (PPPGMMPVGPAPGMRPPMGG). Both epitopes were recognized by antibodies from 3 of the 6 samples. One sample bound only to 3 contiguous decapeptides whose mean was not significantly increased overall (amino acids 63–76 [PPTPFSAPPPAGAM]). One sample did not bind to any of the sequential epitopes. The initial epitopes targeted by nRNP C autoantibodies were significantly enriched for proline as compared with the epitopes in the unbound regions (χ2 = 3.90, P < 0.48). No other amino acids were significantly enriched in the initial epitopes.

Figure 3.

Identification of key epitopes in the early and late nuclear RNP C (nRNP C) humoral immune response. A, Early pattern of antibody binding to the sequential overlapping peptides in the first nRNP C antibody–positive sera. Solid-phase peptide decamers were used to map the pattern of antibody binding to nRNP C. Two epitopes recognized by the first nRNP C–positive serum antibodies (n = 6) are shaded and numbered. Values are the number of SDs above the mean absorbance in controls (n = 11); only epitopes exhibiting an increased absorbance of at least 3 SD are shown. B, Amino acid positions, numbers of samples showing nRNP C antibody binding, and sequences of the 2 epitopes identified in A. C, Inhibition of antibody binding to nRNP C by incubation with peptides corresponding to the dominant epitopes. Significant inhibition as compared with the poly-proline negative control was seen after incubation with peptide nRNP C117–126 (C 117; mean 46.1% [P = 0.035]) or with a mixture of nRNP C117–125, nRNP C61–68 (C 61), and nRNP C109–116 (C 109) peptides (mean 40.3% [P = 0.035]). Horizontal bars show the mean. D, Late pattern of antibody binding in sera collected a mean of 2.5 years after the initial development of nRNP C antibodies (n = 6 pairs of first and last nRNP C–positive sera). Shaded areas and numbers show the antibody binding pattern in the initial positive samples for comparison. Values are the number of SDs above the mean absorbance in controls (n = 6).

Antibody binding to recombinant nRNP C was significantly inhibited by incubation with peptides corresponding to the initial epitopes. Incubation with a mixture of the peptides nRNP C61–68, nRNP C109–116, and nRNP C117–125, encompassing both of the commonly bound initial epitopes, resulted in a mean of 40.3% inhibition and up to 74.5% inhibition in individual samples (P = 0.035) (Figure 3C). Peptide nRNP C61–68 inhibited antibody binding in only 2 serum samples, by 56% and 90%, respectively. Interestingly, the time interval between the last negative sample and the first positive sample for nRNP C was only 1.2 years and 1.75 years, respectively, in these 2 patients, while the mean interval for the remaining patients was 5.2 years. This short interval for the samples that were highly inhibited by incubation with the peptides supports the idea that these epitopes are more important early in the autoantibody response, which diversifies over time. Incubation with nRNP C109–116 inhibited the binding of 3 of the 6 samples by >20%. Peptide nRNP C117–125 exhibited the strongest individual inhibitory effect of the 3 peptides; 5 of the 6 samples were inhibited by >25% (P = 0.034) (Figure 3C).

Autoantibody binding to nRNP C diversified considerably with time. Serum samples from 5 patients drawn a mean ± SD of 4.1 ± 5.2 years after the initially reactive sample significantly bound to 23 individual decapeptides, while only 4 decapeptides were bound by the first positive samples (χ2 = 14.6, P = 0.0001) (Figure 3D). Individual late nRNP C–positive samples recognized between 18 and 25 decapeptides, with a mean ± SD of 20.2 ± 2.8 samples.

Early epitopes in nRNP 70K humoral immunity are diverse and enriched for basic amino acids.

The fine antibody specificities of the negative and first-reactive sera from 3 patients were mapped by using decapeptides that overlapped by 8 amino acids and encompassed the entire nRNP 70K sequence (Figure 1). The antibody development pattern in these 3 patients showed diverse first epitopes, some of which were enriched for proline and some for basic amino acids (Table 1). A mean ± SD of 1.67 ± 1.15 years elapsed between the negative samples and the first samples showing nRNP 70K binding. The initial sera bound to 21 common decapeptides, comprising 9 epitopes (Figure 4B). Individual samples bound to 5, 41, and 66 decapeptides. The epitope at amino acids 63–84 (REERMERKRREKIERRQQEVET) was bound by all 3 samples. The highest average absorbance was found in epitope 4 (amino acids 217–234 [SRYDERPGPSPLPHRDRD]), which was bound by 2 of the 3 samples.

Figure 4.

Epitopes in nuclear RNP 70K (nRNP 70K) humoral immunity. A, Early pattern of antibody binding to the sequential overlapping peptides in the first nRNP 70K antibody–positive sera. Solid-phase peptide decamers were used to map the pattern of antibody binding to nRNP 70K. Nine epitopes recognized by the first nRNP 70K–positive serum antibodies (n = 3) are shaded and numbered. Values are the number of SDs above the mean absorbance in controls (n = 3); only epitopes exhibiting an increased absorbance of at least 3 SD are shown. B, Amino acid positions, numbers of samples showing nRNP 70K antibody binding, and sequences of the 9 epitopes identified in A. C, Late pattern of antibody binding in sera collected a mean of 2 years after the initial development of nRNP 70K antibodies (n = 3 pairs of first and last nRNP 70K–positive sera). Shaded areas and numbers show the antibody binding pattern in the initial positive samples for comparison. Values are the number of SDs above the mean absorbance in controls (n = 6).

Unlike the initial epitopes in nRNP A and nRNP C humoral immunity, in which proline-rich sequences were predominant, the initial epitopes in nRNP 70K were diverse. However, 5 of 9 initial epitopes were highly basic, having theoretical isoelectric points (pI) of 10.67 (epitope 2), 12.48 (epitope 3), 11.83 (epitope 5), 11.42 (epitope 6), and 10.67 (epitope 8). The theoretical pI of the nRNP 70K protein is 9.94. The initial epitopes as a whole were not significantly enriched for any amino acid.

To determine whether the antibody binding to the basic epitopes was a result of charge–charge interactions rather than sequence specificity, the 3 first sera reactive with 70K antibody were incubated with poly-lysine, and inhibition of binding was tested by Western blotting. No decrease in antibody binding was observed after incubation with poly-lysine at concentrations up to 1 mg/ml (data not shown).

Serial serum samples from 2 years after the development of nRNP autoimmunity were available for all 3 patients that developed anti-nRNP 70K antibodies. Epitope mapping was performed on these samples and revealed epitope spreading (Figure 1). The total number of decapeptides that were significantly bound, as determined by comparing the mean of the late samples with the mean of the negative samples, increased to 66 the 21 decapeptides that were identified by comparing the mean of the first positive sample with that of the negative samples (χ2 = 27.9, P < 0.0001) (Figure 4C).

Cross-reactivity of initial epitopes with an Sm-derived peptide.

The proline enrichment that was observed in the initial epitopes of nRNP A and nRNP C was especially interesting, since a proline-rich epitope, PPPGMRP(G)P, has been described as a crucial, early humoral epitope in the development of autoimmunity to SmB′ (23, 24). Portions of the most commonly bound initial epitope of nRNP A, GQPPYMPPPGMIPPPGLAPG (nRNP A159–178), are nearly identical to the repeated motif PPPGMRPP from SmB′, as are the 2 initial epitopes from nRNP C, FQQGKIPPTP (nRNP C57–66) and PPPGMMPVGPAPGMRPPMGG (nRNP C109–128). Given the tight temporal association of antibodies to nRNP and Sm, antibodies against these complexes may develop at the same time and under the same initiating conditions. If this is the case, the earliest antibodies would likely cross-react between the 2 protein subsets.

We examined the extent of cross-reactivity between the antibodies that bound to the initial epitopes in nRNP A or nRNP C with the SmB′-derived peptide. The initial anti-nRNP A–positive or nRNP C–positive patient sera were incubated with the SmB′-derived peptide PPPGMRPP or with poly-proline as a negative control. Incubation with PPPGMRPP significantly inhibited binding to nRNP A (P = 0.0029) (Figure 5A). Incubation with PPPGMRPP significantly eliminated the total binding to nRNP C (P = 0.0087) (Figure 5B). No inhibition of binding to recombinant nRNP 70K was observed after incubation of the first positive sera with PPPGMRPP (data not shown).

Figure 5.

Inhibition of binding to nuclear RNP (nRNP) A and C proteins by an SmB′-derived peptide. Sera with the first observable anti–nRNP A (A) or anti–nRNP C (B) autoantibodies were incubated with the SmB′-derived peptide PPPGMRPP or the negative control peptide poly-proline (top). Significant inhibition of serum antibody binding to recombinant nRNP A (n = 9; P = 0.0029) and recombinant nRNP C (n = 6; P = 0.0087) was seen after incubation with peptide PPPGMRPP, as compared with the poly-proline negative control. Horizontal bars show the mean. Results are representative of 3 independent experiments for each sample. Inhibition of binding to recombinant proteins was quantified by Western blotting (bottom). Three representative Western blots are shown for each of the nRNP proteins. Molecular weight markers are shown at the left. Consistency of protein loading and transfer was confirmed by staining with fast green.

DISCUSSION

By using serial serum samples from our Clinical Immunology Serum Repository, which were collected from more than 27,000 individuals, we identified SLE patients who progressed from nRNP antibody–negative to nRNP antibody–positive status. This resource allowed the investigation of the initial targets of the humoral nRNP autoimmune response. We identified 6 common, early epitopes of nRNP A in these sera. One of these epitopes, GQPPYMPPPGMIPPPGLAPG (amino acids 159–178), was recognized by every initial sera that had detectable binding to sequential epitopes and was bound more strongly than the other initial epitopes. We found that the early response to nRNP C commonly bound only 2 epitopes, both of which were proline-rich. The early antibody binding to nRNP 70K showed a predominance of basic amino acids and was mainly dissimilar to the binding to nRNP C and nRNP A.

Proline-rich sequences were commonly bound by the first autoantibodies against all 3 nRNP proteins. The proline-rich sequences that were bound by initial antibodies from nRNP A and nRNP C were 44–74% similar to each other and to the sequence PPPGMRPP, which is the initial humoral epitope targeted in SmB′ humoral autoimmunity and the major target of anti-SmB′ antibodies (24, 25). Enrichment for proline-rich motifs in the initial antibody binding may be due to a number of factors, including potentially high surface expression, protein conformation, or cross-reactivity with similar sequences in other antigens.

The inhibition of binding to nRNP proteins after incubation with PPPGMRPP indicates that cross-reactive antibodies make up a major portion of the initial autoantibody response to nRNP A and C proteins in this patient collection. Cross-reactivity of antibodies is suggested as a mechanism for the development or spreading of autoimmunity through molecular mimicry (39, 40). Antibodies against proline-rich regions from nRNP A, nRNP C, SmB′, and nRNP 70K have been shown to cross-react (16, 17, 41), and immunization of mice with the SmB′-derived sequence was shown to lead to a lupus-like autoimmunity (42). It is plausible that antibodies against these similar proline-rich regions could facilitate epitope spreading between nRNP and Sm proteins.

Comparison of findings from the present study and from studies of the mature nRNP A response (11–13) reveals that the pattern of antibody binding changes from the initial response to the mature response, in which antibody binding is focused on the N-terminal sequences. This shift of pattern as the antibody response matures suggests that separate selective forces may drive the antibody specificity as time passes. For example, the initial response may be due to cross-reactivity with environmental triggers, while the later antibody specificity may be determined by other factors, such as access to the binding sites on the antigen, selective antigen processing/presentation, or preferential V–D–J recombination. These findings indicate that care should be taken in attempting to gain insight into the cause of autoimmune processes by studying samples collected late in the disease process.

Antibody binding to nRNP C was initially directed against just 2 regions. These initial epitopes were also found in other studies that mapped the mature nRNP C epitopes (14–17). Unlike epitope spreading in nRNP A, the epitope spreading found in nRNP C does not diminish the role of the initial epitopes in the mature humoral response. In contrast, the epitopes from amino acids 109–128 remained predominant in the mature antibody-binding pattern.

The initial response to nRNP 70K was not similar to the responses to nRNP A and nRNP C. Interestingly, nRNP 70K immunity has been shown to be more strongly associated with MCTD than with SLE (43, 44), and the difference in initial antibody binding patterns may reflect different pathogenic processes. The initial epitopes that we discovered in nRNP 70K humoral autoimmunity were consistent with those found in other studies of nRNP 70K humoral epitopes (12, 18). By using fine-resolution peptide mapping, this study identified the specific amino acids that were bound in larger, previously described nRNP 70K epitopes (19, 20, 45). An epitope previously found in the mature nRNP 70K humoral immune responses is amino acids 135–194 (21, 22). Although the initial nRNP-positive sera did not bind this region, the later samples did, indicating the importance of epitope spreading in the development of autoimmunity.

The targeting of proline-rich similar epitopes in early humoral autoimmunity may be partly due to environmental factors. Epstein-Barr virus (EBV) infection is associated with SLE with a high degree of significance (46, 47), and the Epstein-Barr nuclear antigen 1 (EBNA-1) protein has a sequence (PPPGRRP) that is very similar to the initial epitopes of nRNP A and nRNP C identified in the current study (48). Immunization with the EBNA-1–derived peptide on a branching poly-lysine backbone causes lupus-like autoimmunity (48, 49), as does DNA vaccination of mice with EBNA-1 (50). The similarity of epitopes from 4 major autoantigens targeted in lupus to the EBNA-1 sequence suggests that the role of EBNA-1 in lupus may be more extensive than was previously thought and could help to explain the association of EBV infection with the development of lupus, even in patients who do not develop antibodies against Sm.

These studies used a powerful resource, the OMRF Clinical Immunology Serum Repository. While a prospective longitudinal study may yield more data than this retrospective study, obtaining sufficient samples and participation for such a study is impractical. Using the serum collection allowed for a cost-effective means to approach these questions. An additional limitation of the study is that solid-phase peptide assays may miss conformational epitopes, while providing high resolution of the antibody-binding pattern. Although this is a concern, the level of agreement between the current and previous studies in which the peptide assay was used with studies of protein fragments and larger peptides is high (11–22, 45). Antibodies against epitopes identified through this method make up a large proportion of the total antibody population (24, 25). Indeed, in this study, incubation with the most predominant peptides detected by solid-phase epitope mapping led to a mean decrease in antibody binding of >30% for both nRNP A and nRNP C.

This study identified a consistent proline-rich motif in early humoral autoimmunity to nRNP proteins, a motif that is shared with early autoimmunity to SmB′. The data support a role of this motif in epitope spreading during the initiation and development of the autoantibody response to nRNP proteins. The motif is absent in early nRNP 70K antibody targets, suggesting that the development of nRNP 70K autoantibodies follows a different path than antibodies against the other nRNP components. These findings are consistent with the previously observed close temporal association of nRNP and Sm autoantibodies (51), as well as with the association of nRNP 70K and nRNP A antibodies with different clinical outcomes (6, 7). These results identify a key pattern in the development of a number of lupus-associated autoimmune specificities in human disease and support the possibility that molecular mimicry is involved in the initiation of autoimmunity against these components in naturally occurring SLE in humans.

AUTHOR CONTRIBUTIONS

Dr. James had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study design. Poole, Schneider, Harley, James.

Acquisition of data. Poole, Schneider, Guthridge, Velte, Reichlin, James.

Analysis and interpretation of data. Poole, Schneider, Guthridge, Harley, James.

Manuscript preparation. Poole, Schneider, Guthridge, Velte, Reichlin, Harley, James.

Statistical analysis. Poole, James.

Acknowledgements

The authors thank Terri McHugh and Jody Gross for their technical assistance. We are grateful for the kind gift of the expression plasmid for nRNP A (U1-A) from Carol Lutz, PhD (University of Medicine and Dentistry of New Jersey, Newark, NJ). We thank Kristina Wasson-Blader, PhD, for review and revision of the manuscript. We also thank the Oklahoma University Health Sciences Center Molecular Biology-Proteomics facility for performing the peptide synthesis.

Ancillary