T cell autoimmunity to histones and nucleosomes is a latent property of the normal immune system




Investigators in this study undertook to determine whether in vitro antigen-responsive immune (polyomavirus T antigen [T-ag]– specific) and autoimmune (histone-specific) T cells from normal individuals share structural and genetic characteristics with those from patients with systemic lupus erythematosus (SLE).


Histone-specific T cells were generated by stimulation of peripheral blood mononuclear cells (PBMCs) with nucleosome–T-ag complexes and were subsequently maintained by pure histones. T-ag–specific T cell clones were initiated and maintained by T-ag. The frequencies of circulating histone- and T-ag–specific T cells were determined in healthy individuals and in SLE patients by limiting dilution of PBMCs. T cell receptor (TCR) gene usage and variable-region structures were determined by complementary DNA sequencing. These sequences were compared between T-ag– and histone-specific T cells and between normal individuals and SLE patients for each specificity.


Individual in vitro–expanded histone- and T-ag–specific T cells from normal individuals displayed identical TCR Vα and/or Vβ chain third complementarity-determining region (CDR3) sequences, indicating that they were clonally expanded in vivo. The frequencies of in vitro antigen-responsive T-ag– or histone-specific T cells from normal individuals were similar to those from SLE patients. Although heterogeneous for variable-region structure and gene usage, histone-specific T cells from healthy individuals and SLE patients selected aspartic and/or glutamic acids at positions 99 and/or 100 of the Vβ CDR3 sequence.


Autoimmune T cells from healthy individuals can be activated by nucleosome– T-ag complexes and maintained by histones in vitro. Such T cells possessed TCR structures similar to those from SLE patients, demonstrating that T cell autoimmunity to nucleosomes may be a latent property of the normal immune system.

Experimental results consistently demonstrate that pathogenic autoimmune anti-DNA antibodies are secondary, antigen-driven, CD4+ T cell–dependent immune responses initiated by DNA itself (1–5). Since DNA most likely is not presented by antigen-presenting cells (APCs) in the context of HLA class II molecules, the contemporary paradigm favors DNA binding proteins as a sine qua non for generating T helper cell stimuli in the autoimmune anti-DNA antibody response (3, 5, 6). Such a hapten-carrier model, implying DNA-specific B cells and protein-specific T cells, enables the understanding of the molecular and the cellular basis for an immunoglobulin class switch and affinity maturation of anti–single-stranded DNA and anti– double-stranded DNA (anti-dsDNA) antibodies. Experimental and clinical results demonstrating that T cells specific for self or non-self DNA binding proteins may provide help for DNA-specific B cells support a model consistent with cognate B cell–T cell interaction (2, 3, 6–8). So far, however, it is unclear how self-specific T cells may be initially activated. A priori, one would expect, for example, nucleosome- or histone-specific T cells to be tolerant and the high-affinity ones to be deleted, since the thymus is a central locus for generation and presentation of apoptotic nucleosomes due to the high rate of cell death caused by T cell deletion by negative selection or by neglect.

Peripheral tolerance is important in silencing autoimmune T cells escaping central deletion. Such tolerized cells may be activated through a process implying linked presentation of non-self and self molecules by the same APC, provided that T cells responding to the non-self ligands are present (9). Precedent for this model relates to diversification of T cell responses to include responses to determinants within (10), but also between (11, 12), different polypeptides. In both situations, autoimmune T cells may be triggered. We recently obtained analogous results using polyomavirus T antigen (T-ag) complexed with nucleosomes (9). These results demonstrate that linked presentation of T-ag, for which virtually all normal individuals have responder T cells (3), and histones (both contained within the nucleosome–T-ag complex) by an APC may result in termination of T cell tolerance to histone peptides. This concept may be relevant to immunologically normal individuals and to patients with systemic lupus erythematosus (SLE) (9). Thus, the ability to activate autoimmune nucleosome-specific B cells and T cells in vivo may not be confined to syndromes such as SLE, but may instead represent a latent property of the immune system of all individuals.

To substantiate this process, the following 4 sets of data were generated and compared between or within normal individuals and SLE patients: 1) frequencies of recirculating histone- and T-ag–specific T cells, 2) T cell receptor (TCR) α/β chain V–J and V–D–J gene usage and distribution among different clones within and between the two T cell specificities, 3) potential clonal antigen-selective expansion of autoimmune and immune T cells as determined by recurrent detection within individuals of identical TCR Vα and/or Vβ third complementarity-determining regions (CDR3), and 4) potential selection of charged amino acids at distinct positions in the TCR for autoimmune histone-specific and immune T-ag–specific T cells, which may contribute to antigenic specificity of T cells in analogy to variable-region structures that favor binding of antibodies to dsDNA (13, 14).

Such data may yield important information about whether the immune system is regulated differently in the two conditions. Our null hypothesis assumes there are no differences related to parameters that can be determined. If this is true, autoimmunity to DNA and nucleosomes may represent conventional immune responses exerted also by a normal immune system and may not be confined to SLE or related disorders, provided that self ligands are complexed with non-self molecules in analogy to a hapten-carrier model.


Patients and controls

Three female SLE patients (patients SLE1, SLE6, and SLE7) ages 37, 41, and 56 years, respectively, were selected for the present studies. All patients were Caucasians and fulfilled at least 4 of the 11 American College of Rheumatology 1982 revised criteria (15) for the diagnosis of SLE. All patients had anti-dsDNA antibodies as detected in enzyme-linked immunosorbent assay, but not in the Crithidia luciliae assay (data not shown). None of the patients were receiving systemic antiinflammatory drugs at the time of blood sampling. Five healthy individuals (4 women and 1 man), matched for age and ethnicity, were included as controls. Data on the T cells of these normal individuals have been described previously (3, 9). The Ethical Board at the University of Tromsø approved this study.

T cell lines. T-ag– or histone-specific T cell lines derived from SLE patients and healthy individuals were prepared and maintained as described (3, 9). The TCR α and β chain CDR3-region structures of T-ag– or histone-specific T cells were determined in monoclonal T cell lines. These were established by limiting-dilution cultures to clone antigen-specific T cells directly from peripheral blood mononuclear cells (PBMCs). Through this protocol, frequencies of T-ag– and histone-specific T cells were also determined.

Limiting-dilution cultures were established by adding varying numbers of PBMCs (104/well, 2 × 103/well, 4 × 102/well, 0.8 × 102/well) to individual culture wells (24 wells for each cell number) and stimulated with antigen (10 μ g/ml T-ag or 15 μg/ml nucleosome–T-ag complexes [ 3]) and irradiated (4,000 rads) APCs (5 × 105/well). The reason to select these PBMC concentrations derives from the observation that the frequency of in vitro–responsive antigen-specific T cells may constitute <1 × 10−4 of the entire T cell population (16). The Poisson distribution assumes that when ≥37% of the wells for each 24-well series were negative for proliferating cells, each positive well contained, on average, 1 antigen-specific T cell when initiating the cultures (17, 18). The number of T cells in the original PBMCs was calculated by flow cytometry as the percentage of CD3+ T cells of the entire CD45+ PBMC population.

T cell lines initiated with nucleosome–T-ag complexes were expanded by isolated histones (H1, H2A, H2B, H3, and H4 in a total concentration of 20 μ g/ml). Thus, these cells were primarily selected and activated in vitro by histone peptides derived from histones in the nucleosome–T-ag complex and were subsequently expanded by peptides processed from total histones added to the cultures. T cells activated and expanded by T-ag failed to proliferate when challenged with histones (3, 9) (Figures 1a and d). In some cases, T cells were finally expanded using anti-CD3–coated Dynabeads M-450 CD3 (Dynal, Oslo, Norway). For initial experiments, individual histone-class molecules (10 μg/ml) were used to define specificity of individual T cell lines, but these were later circumvented since the appearing T cell lines did not distinguish significantly between individual histone molecules (ref. 2 and present report).

Figure 1.

Primary responses of peripheral blood mononuclear cells (PBMCs) and subsequent responses of T cell lines. a, Proliferation of PBMCs from a normal individual (N1) or from a patient with systemic lupus erythematosus (SLE1) was tested in response to the indicated antigens and measured on day 5 of week 1. Subsequently, responses to medium (M), T antigen (T-ag), or histones (H) of T cell lines initiated by b, medium, c, histones, d, T-ag, e, nucleosomes, or f, nucleosome–T-ag complexes were determined during week 2, as indicated by arrows. Values are the mean and SD cpm of triplicate cultures. nd = not done in SLE.

After 4–5 weeks, proliferation of the T cell lines was measured by 3H-thymidine incorporation (3) in parallel cultures. On the basis of the number of wells having monoclonal T cells according to their Poisson distribution (see below) and the total number of T cells used in each experiment, the frequency of responder cells in peripheral blood was calculated. Of all T cell clones initially generated, a total of 54 clones from 2 of 3 SLE patients and from all 5 normal individuals were expanded and subjected to further analyses (see below). Of these, 24 T cell clones possessed specificity for histones (13 from SLE patients, 11 from normal individuals), while 30 T cell clones demonstrated specificity for T-ag (10 from SLE patients and 20 from normal individuals). Clones from patient SLE1 and most of the clones from normal individual 4 (individual N4) became extinct during the time needed for their expansion.

RNA extraction from T cells and synthesis of complementary DNA (cDNA). Isolation of total RNA and generation of cDNA have been described elsewhere (19, 20). The quality of cDNA was confirmed by polymerase chain reaction (PCR) amplification of the adenine–phosphoribosyltransferase transcript (20).

PCR amplification. The cDNA was amplified either with TCR Vα-region and Vβ-region consensus primers and nested C-region primers (21, 22) or with primers specific for individual Vα (23) or Vβ (24) genes. PCR using these latter primers was performed essentially as described elsewhere (25), except that a single-step PCR was run for 40 cycles with annealing temperatures of 60°C (Vα primers) or 61°C (Vβ primers).

Cloning and sequencing of PCR-amplified TCR α-chain and β -chain genes. PCR products were sequenced directly using Cα (21) and Cβ (22) primers or after being cloned into the pGEM-T Easy Vector System (Promega, Madison, WI). Sequencing of purified or cloned PCR products was performed with the BigDye Terminator Cycle Sequencing Ready Reaction kit (Applied Biosystems, Warrington, UK). Sequencing reactions were analyzed on an ABI377 Prism Sequencer (Perkin-Elmer, Boston, MA). The sequences were compared against the GenBank database using the program BLAST (National Center for Biotechnical Information, Bethesda, MD). Following are European Molecular Biology Laboratory (Heidelberg, Germany) accession numbers of sequences used in this study: AJ278904–AJ278906, AJ293221–AJ293234, AJ278630–AJ278639, and AJ403866–AJ403945.

Statistical analysis. The chi-square test or Fisher's exact test and t-test were used to test differences between groups. Two-sided tests were used. The Poisson distribution for rare events was tested as described (17, 18). Calculations were performed using SAS software, version 6.12 (SAS Institute, Cary, NC).


Generation of T cell lines specific for T-ag, nucleosomes, and histones. The principal protocol for generation of (immune) T-ag– and (autoimmune) histone-specific T cell lines using PBMCs from normal individuals is demonstrated in Figure 1 for individual N1. Primary T cell cultures responded to pure T-ag and to T-ag complexed with nucleosomes, but not to nucleosomes or histones (Figure 1a). T cells initially stimulated with medium, histones, or nucleosomes did not respond to subsequent restimulation with medium, T-ag, or histones (Figures 1b, c, and e, respectively). T cells initiated with T-ag responded as expected (3) to subsequent restimulation with T-ag (Figure 1d), but not with histones. T cells initially stimulated with nucleosome–T-ag complexes, however, responded subsequently to histones (Figure 1f), a finding consistent with previous results (9).

Since T cells specific for T-ag did not respond to histones (Figure 1d) (3, 9), histone peptides processed from nucleosome–T-ag complexes selected and activated bona fide histone-specific T cells (see Discussion). Similarly, cultures initiated with a mixture of histones and T-ag proliferated when restimulated with pure histones (data not shown). For SLE patients, however, histones regularly had the potential both to initiate and to maintain T cell lines (3) (exemplified by PBMCs from patient SLE1) (Figures 1a and c). Nucleosome–T-ag complexes could also initiate histone-specific T cells when PBMCs from patient SLE1 were tested (Figure 1f). The experimental approach as outlined in Figure 1 enabled us to compare in vitro–activated autoimmune histone-specific T cells from normal individuals with those of the same molecular specificity spontaneously activated in SLE.

Frequencies of in vitro antigen-responsive monoclonal T-ag– or histone-specific T cells in normal individuals and SLE patients. In order to compare the frequencies of peripheral T cells able to respond to T-ag or histones in SLE with those in normal individuals, direct cloning of T cells by limiting dilution of PBMCs in primary cultures was first performed. Established clones were subsequently maintained with T-ag. After 4 weeks, wells containing responder T cells demonstrated a Poisson distribution consistent with the presence of monoclonal T cells in the wells (data not shown). The frequency of T-ag–specific T cells, as calculated on the basis of the number of positive wells and the amount of CD3+ cells added to the wells, was similar in normal individuals and in SLE patients (mean ± SD number of recirculating T cells × 10−6 94 ± 54 and 134 ± 38, respectively; P = 0.3) (Table 1).

Table 1. Frequency of monoclonal T-antigen (T-ag)– or histone-specific T cells in peripheral blood of normal individuals and patients with systemic lupus erythematosus (SLE)*
 T-ag–specific CD3+ T cellsHistone-specific CD3+ T cells
  • *

    Values are the mean ± SD number of recirculating T cells ×10− 6. Numbers are corrected for CD3+ cells among the purified peripheral blood mononuclear cells (range 55–70%).

  • Data were generated from 3 SLE patients and 5 normal individuals.

  • Data were generated from 3 SLE patients and 4 normal individuals.

SLE patients133.7 ± 37.634.3 ± 14
Normal individuals94.4 ± 53.666.7 ± 44

The same set of data was generated for recirculating histone-responsive T cells in 4 normal individuals and in 3 SLE patients. Three of the normal individuals lacked readily responsive histone-specific T cells, as determined by the absence of proliferation after primary in vitro stimulation with histones or with nucleosomes (3) (e.g., individual N1 in Figure 1a). Cells from one normal individual (subject N6 in ref. 3, denoted subject N8 in the present study) responded weakly (stimulation index of 3.3) to histones in the primary in vitro culture. We have recently presented individual data on responses to T-ag, nucleosomes, and histones of T cells from these normal individuals (3). After an initial stimulus of limiting-dilution cultures of PBMCs from these 7 individuals with nucleosome–T-ag complexes, resulting in activation of silent histone-specific T cells (Figure 1f), these clones were further expanded by weekly stimulation with histones. By testing their specificity for separate histone classes, individual T cell clones established by this protocol regularly responded to more than one histone class (e.g., clones N2H53, N2H59, N2H70, and N2H75 in Figures 2A and B), while few clones were specific for individual histone classes (e.g., clone N2H71 in Figure 2B).

Figure 2.

A, Representative examples of the fine molecular specificities of monoclonal T cells for individual histone classes. Established monoclonal T cell lines were tested against the individual histone classes H1, H2A, H2B, H3, or H4 by measuring proliferation as 3H-thymidine incorporation. B, Examples of rare monoclonal T cells specific for individual histone classes or for peptides common to most or all histones.

These observations are in accordance with previous results obtained with T cell clones derived from SLE patients (2). The observed cross-reaction may be based on amino acid or charge homologies between individual histone classes. This assumption originates from our group's comparison analyses of amino acid sequences of the 5 histone classes using the data bank of Protein Information Resource (PIR; National Biomedical Research Foundation, Washington, DC) (data not shown). (PIR identifiers for the 5 histone classes are as follows: H1, B40335; H2A, HSHUA1; H2B, HSHUB1; H3, HSHU33L; H4, HSHU4.) In addition, for this autoimmune T cell specificity, the frequencies (as determined by direct cloning of PBMCs) of T cells from normal individuals and SLE patients were comparable (mean ± SD number of recirculating T cells × 10−6 67 ± 44 and 34 ± 14, respectively; P = 0.26) (Table 1).

These estimates depended on whether the T cells were really monoclonal, which may have been true for two reasons. First, the wells containing proliferating T cells demonstrated a Poisson distribution, consistent with monoclonality. Second, as described in detail below, TCR cDNA generated from messenger RNA of T cells from positive wells could be PCR amplified using consensus V-gene primers and sequenced directly, consistent with homogenous cDNA. Thus, these data demonstrate that the normal individuals examined here had a repertoire of in vitro antigen-responsive autoimmune T cells with frequencies in peripheral blood comparable to those specific for polyomavirus T-ag.

Variable gene usage by T-ag– and histone-specific T cells. One purpose of establishing monoclonal T cell lines was to analyze the structure of TCR α and β chain CDR3 regions. TCR variable-region genes were analyzed to determine whether spontaneously expanded histone- or T-ag–specific T cells from SLE patients (1, 8, 9) and in vitro–activated and –expanded T cells of the same specificities from normal individuals shared structural and genetic characteristics. Amino acid sequences obtained from 24 histone-reactive T cell lines (11 from normal individuals and 13 from SLE patients) and 30 T-ag–specific T cell lines (20 from normal individuals and 10 from SLE patient 6) were compared. Diverse Vα- and Vβ-encoding genes were utilized by each group of antigen-specific T cells (Table 2) with a relative overrepresentation of Vβ14/14.1 and Jβ2.1. However, the selection of these genes between normal individuals and SLE patients, or between the different antigen-specific T cell populations, was not significantly different (P = 0.95 and P = 0.18, respectively). Thus, the TCR variable-region gene usage of immune or autoimmune T cells is not principally different in SLE patients compared with normal individuals.

Table 2. Primary structures of T cell receptor (TCR) Vβ and Vα third complementarity-determining regions (CDR3) of monoclonal T cell lines generated from systemic lupus erythematosus (SLE) patients and normal individuals*
Clinical conditionT cell specificity (n)T cell cloneRelated clones (n)Vβ geneV–D–Jβ CDR3 (positions 95–106)Jβ geneVα geneV–Jα CDR3 (positions 93–104)Jα gene
  • *

    T cell clones were either initiated and maintained by T antigen (T-ag–specific T cell clones) or initiated by the nucleosome–T-ag complex and thereafter maintained by histones (histone-specific T cell clones). The sequences of Vα and Vβ CDR3 regions are shown in single-letter amino acid code, and only negatively charged residues (D and E) located within the Vβ CDR3 regions are shown in bold. The CDR3 regions occupy positions 95–106 and 93–104 for TCR Vβ and Vα chains, respectively, according to Chothia et al (44). The transition between the junction region and the Jα/β genes is marked with an apostrophe. The resulting sequences were analyzed for identity with sequences accessible in the GenBank database using the program BLAST (see Patients and Methods). For the T cell clones SLE7H10 and N2T14, the Vα sequences were not determined. This explains why the number of charged amino acids presented in Figures 3C and D may be lower than indicated by this table. ND = not determined.

  • The Jα sequence of T cell clones N3H52 and N1T7 matches that of the L sequence reported in ref. 45, while clone N1T16 uses a Jα sequence similar to that of clone MTC5 (see ref. 46).

SLEHistone (13)
SLE7H2SLE7H3, H9, H10 (4)13.2SYLGP'EAF FGQ1.121RA'GKLI FGQG14.1
NormalHistone (11)
N3H54N3H56, H58 (3)12.1ASS'NEQF FGP2.121.1LP'TGTASKLT FGTG44
SLET-ag (10)
NormalT-ag (20)
N2T2N2T4, T7, T10 (4)14.1TPWGQGN'NEQF FGP2.123.1Y'SGNTPLV FGKG9.14

Clonal, antigen-selective expansion of histone- and T-ag–specific T cells. An important aspect of this study was to search for evidence that autoimmune T cells in normal individuals were clonally expanded. This would directly demonstrate that autoimmune T cells are not only present in healthy individuals, but may also be activated in vivo. Several of the T cell clones of individual normal subjects or SLE patients maintained by histones or T-ag expressed identical TCR Vβ- and/or Vα-region sequences (Table 2). For example, patient SLE7 and subject N3 respectively possessed 4 and 3 clonally related histone-specific T cells expressing identical TCR Vα/Vβ CDR3 regions, indicating that these autoimmune T cells may be clonally expanded. For T-ag–specific T cells, patient SLE6 had 2 groups of related clones (Table 2). Similarly, groups of clonally related T-ag–specific T cells were also detected among cells from subjects N2 and N8 (Table 2). Interestingly, clones N8T59 and N8T60 had identical Vα chains, but different Vβ chains. Collectively, these data demonstrate that immune and autoimmune T cells from normal individuals and SLE patients may be clonally expanded in vivo.

Selection of primary TCR Vαand VβCDR3-region structures, with special reference to charged amino acids, of T cells specific for T-ag or histones. Although heterogeneous, the TCR Vβ CDR3-region sequences of histone-specific T cells consistently contained the negatively charged aspartic (D) or glutamic (E) amino acids preferentially at positions 99 and 100, representing the center of this region (Table 2). D and E were predominantly encoded by Jβ genes and less frequently by N-nucleotides, as deduced from sequences in Table 2. Nine of 13 germline Jβ genes (69%) had codons for D and/or E at different positions (Table 3). Of 30 T-ag–specific T cell clones, 22 (73%) used Jβ genes expressing these codons, while such codons were present in 20 of 24 Jβ genes (83%) expressed by histone-specific T cells. Thus, there was no significant selection of Jβ genes with codons for D and/or E, or for Vβ CDR3 regions with such codons at any position for histone-specific T cells compared with those specific for T-ag (79% versus 67%, respectively; P = 0.31) (Table 3).

Table 3. Positions of acidic amino acids aspartic acid (D) and/or glutamic acid (E) in the TCR Vβ CDR3 regions of T-ag– and histone-specific T cells*
T cell clonesnCDR3No. of monoclonal T cell lines with amino acids D and/or E in Vβ
PositionsJ genes
99, 10099100TheoreticalIn clones
  • *

    Except where indicated otherwise, values are the number (%). See Table 2 for other definitions.

  • Positions are determined as described in Table 2.

  • Of 13 Jβ genes, 9 possess codons for D and/or E at any position.

T-ag–specific3020 (67)2 (7)2 (7)1 (3)9/13 (69)22 (73)
Histone-specific2419 (79)15 (63)7 (29)9 (38)20 (83)

The theoretical variability of positions for D and E in the reorganized Vβ CDR3 region is unpredictably higher compared with the distribution within individual Jβ genes due 1) to addition of varying numbers of N-nucleotides 3′ of the Vβ gene and 2) to processes related to V–D–J gene recombination. Selection of certain amino acids at distinct locations from such a chaotic repertoire of positions must indicate selection of TCR structures suitable to combine with a given antigen. This is evident for the histone-specific T cells described here, since 15 of 24 of their TCR Vβ regions (63%) possessed amino acids D and/or E at positions 99 and/or 100. Only 7% of those for T-ag–specific T cells contained D or E at these positions (Table 3). A remarkable selection of D and/or E at amino acid positions 99 (P = 0.027) or 100 (P = 0.0029) was therefore noted for histone-specific versus T-ag–specific T cells. For D and/or E in positions 99 and 100, the selection was even stronger (P = 0.000011) (Figures 3A and B and Table 3). Comparing frequencies and positions of D and/or E in TCR Vβ structures expressed by histone-specific T cells from normal individuals with those from SLE patients, no differences were noted (P = 0.14 and P = 0.68 for positions 99 and 100, respectively).

Figure 3.

Distribution of acidic (E, D) and basic (R, K, H) amino acids in T cell receptor Vβ and Vα CDR3 regions. A and C, Vβ and Vα CDR3 regions of monoclonal histone-specific T cell lines. B and D, Vβ and Vα CDR3 regions of T antigen (T-ag)–specific T cell lines.

All 24 TCR Vα CDR3 regions of histone-specific T cells and 25 of 30 Vα CDR3 regions of T-ag–specific T cells were characterized by the presence of one or several of the positively charged amino acids arginine (R), lysine (K), and histidine (H) (Table 2 and Figures 3C and D), while few contained acidic amino acids. These were all randomly distributed along the entire CDR3 regions for both T cell specificities (Figures 3C and D).


This study is part of a research program aimed at describing the molecular and cellular processes responsible for the generation of B cell and T cell autoimmunity to DNA and nucleosomes (3, 6, 9, 26, 27). Based on previous results, polyomaviruses and their transcription factor T-ag seem to be recurrently expressed in SLE patients and rarely in normal individuals (27–29). Clonal expansion of T-ag–specific T cells is therefore expected to take place in SLE. Consistent with this view, one should expect higher frequencies of T-ag–specific T cells in SLE patients compared with normal individuals. However, this study demonstrates that normal individuals and SLE patients had comparable numbers of recirculating T-ag–specific T cells as determined by limiting-dilution cultures.

An explanation for the relatively high frequency of T-ag–specific T cells in normal individuals may relate to the nature of polyomavirus infection among such individuals. After a transient primary infection, polyomaviruses normally establish a latent infection throughout a person's lifetime (28, 29). In normal cells, the virus genome remains largely episomal (30). To persist in the host, however, the virus or its genome must replicate to ensure reinfection or transmission to new cells. Maintenance of latent polyomavirus DNA therefore requires constitutively low levels of T-ag expression, since this protein is indispensable for viral replication (30). As a consequence of this T-ag–dependent low-grade virus replication, T-ag–specific T cells may constitutively be activated and clonally expanded. On the one hand, T-ag expression related to latent infection may be sufficient to stimulate T cells, intrinsically explaining the unexpectedly high number of recirculating T-ag– specific T cells in normal individuals, reaching levels comparable to those in SLE patients having active polyomavirus infection (27, 31). On the other hand, this assumed low-grade T-ag expression is obviously not sufficient to evoke humoral immune responses, since normal individuals rarely produce anti–T-ag antibodies (9, 31, 32).

T-ag forms stable complexes with nucleosomes in vitro and in cells expressing T-ag (3, 9), allowing linked presentation of T-ag– and histone-derived peptides by an APC processing this complex. Responder T-ag–specific T cells may then, through secretion of interleukin-2, activate autoimmune, histone-specific T cells. Subsequently, these T cells may clonally expand, provided that histone peptides are presented in the context of HLA class II and provided that sufficient costimulatory signals are available (33, 34). Such a process may have the consequence that T-ag expression may result directly in clonal expansion of T-ag–specific T cells and indirectly in expansion of bona fide histone- or nucleosome-specific T cells with the potential to help DNA-specific B cells (2, 8). These autoimmune, histone-specific T cells have previously been shown to respond to histone peptides processed from total histones and from pure nucleosomes (3, 9) such as those derived from SLE patients. T-ag– specific T cells do not respond to histones (Figure 1d) or pure nucleosomes (3, 9).

A process as outlined above may explain why the frequencies of both T-ag– and histone-specific T cells, as determined by direct cloning of T cells from PBMCs (18), may be relatively high in normal individuals. Direct cloning of antigen-specific T cells avoids the potential artifact given by the fact that one or few clones may dominate when oligoclonal T cell lines are established prior to limiting-dilution cultures. This would eventually result in a skewed picture of potential heterogeneity of the TCR structure, and also of the frequency of clonally unrelated, antigen-specific T cells in the original sample (18). This problem is omitted if T cells are cloned directly by limiting dilution of PBMCs, as was done in the present study. However, direct cloning by culturing T cells in vitro may give a falsely low estimate of the number of recirculating autoimmune T cells in active SLE, since in vivo–activated, Fas-sensitive, nucleosome-reactive T cells may undergo activation-induced apoptosis upon antigenic stimulation in vitro (35). Thus, although characterization of individual T cells by direct cloning may have inherent advantages and problems, this study indicates that a significant proportion of T cells in normal individuals possesses specificity for nucleosomal autoantigens.

All available data generated so far (3, 6, 9, 27, 31) are compatible with this model as one for activation of autoimmune T cells, implying linked presentation of peptides from non-self polyomavirus T-ag and self histones by individual APCs, and this model may be designated “spreading of T cell responses.” Consistent with previous results (9), initiation of such a process requires dominant responder (here T-ag–specific) T cells. Provided with linked presentation of T-ag and an autoantigen like a histone, these T-ag– specific T cells may terminate an unresponsive state of autoimmune histone-specific T cells. Precedent for an in vivo correlate to this model derives from previously reported data (e.g., see refs. 10–12).

Why then do not all individuals have autoimmune disorders? Since nucleosomes (as opposed to T-ag, for example) are ubiquitously distributed in the body, such complexes may be presented in the periphery by cells expressing HLA class II, but lacking costimulatory molecules like B7 (CD80/86) and CD40, or having a low density of such molecules on their surface. For example, if such cells present histones to activated histone-specific T cells, these may enter a state of anergy (36, 37), but may later be reactivated (9) upon newly encountering APCs copresenting peptides from T-ag and histones. Through such a process, autoimmune T cells may fluctuate between an active state, induced by linked presentation of self and non-self ligands, and a state of nonresponsiveness, when expression of non-self ligands is terminated and the autoantigens are presented to the autoimmune T cells by nonprofessional APCs. Such a cyclic process may be relevant to explain why healthy individuals occasionally harbor T cells that readily proliferate when stimulated in vitro by histones or nucleosomes (2, 3).

Our protocol for activating silent autoimmune T cells allowed us to analyze the TCR of T cells generated both in normal individuals and in SLE patients. The CDR3 regions of TCRs are thought to contact the antigenic peptide (38). An interaction between oppositely charged residues at the center of the TCR Vβ CDR3 loops and the corresponding antigenic peptides has been found in several antigenic systems (7, 8, 38). This is consistent with the view that this region is more important for binding to peptide–HLA class II complexes than are the Vα CDR3 regions contributing less to the energy of binding or specificity (39). This study revealed negatively charged residues (D and/or E) in positions 99 and 100 of the TCR Vβ CDR3 region of histone-specific T cells, and these were contributed by Jβ genes, D genes, or N-nucleotide addition. This particular distribution was not observed for T-ag–specific T cells or, for example, for tetanus toxin–specific T cells (40). The presence of D and/or E at the center of the Vβ CDR3 regions may be explained by the fact that histones have long runs or clusters of positively charged regions (41). T-ag, however, displays short runs of positively charged residues alternating with neutral or negatively charged ones (42), which is consistent with a more random pattern of charged amino acids in the TCR Vβ CDR3 loops of T cells with this specificity.

Taken together, the findings of this study indicate that autoimmune nucleosome- (or histone-) specific T cells can be clonally expanded in vivo in normal individuals as in SLE patients. Furthermore, this study did not reveal any structural or genetic characteristics of autoimmune TCRs that are unique to SLE. From these data, one may also speculate whether antigen-selective T cell autoimmunity may depend (at least in part) on the in vivo generation of complexes of self and non-self ligands. In that sense, these and previous data (3, 6, 9, 26, 27), combined with the fact that autoimmunity in SLE seems to be antigen restricted (43), are not compatible with a general lack of deletion or silencing of autoimmune T cells in SLE.