Demethylation of Promoter Regulatory Elements Contributes to Perforin Overexpression in CD4⁺ Lupus T Cells¹

Abnormal hypomethylation of CD4⁺ T cell DNA can contribute to the pathogenesis of lupus-like diseases. Human and murine CD4⁺ T cells treated with the DNA methylation inhibitor 5-azacytidine (5-azaC)⁴become autoreactive in vitro, and CD4⁺ murine T cells made autoreactive with 5-azaC induce a lupus-like disease in vivo (1, 2, 3, 4). The autoreactivity is due in part to LFA-1 (CD11a/CD18) overexpression, because LFA-1-transfected CD4⁺ T cells have an identical autoreactivity in vitro and cause a similar lupus-like disease in vivo (5). Further, some lupus-inducing drugs, including procainamide and hydralazine, are DNA methylation inhibitors (2), and CD4⁺ T cells treated with these drugs overexpress LFA-1 and become autoreactive in vitro, and induce autoimmunity in vivo, similar to the effects of 5-azaC (6). In addition, CD4⁺ T cells from patients with active lupus have hypomethylated DNA (7) and overexpress LFA-1 on an autoreactive subset (8). Finally, the same LFA-1 regulatory sequences are demethylated in T cells treated with 5-azaC and in T cells from patients with active lupus (9). However, the DNA hypomethylation caused by 5-azaC and the lupus disease process is genome-wide and probably affects multiple genes. Although LFA-1 overexpression is important in the hypomethylation model and possibly in human lupus, other genes may also be overexpressed in hypomethylated T cells and contribute to disease pathogenesis.

Our group has sought additional methylation-sensitive genes. PHA-stimulated T cells were treated with 5-azaC, and mRNA expression patterns were compared in treated and untreated cells using oligonucleotide arrays. Perforin was identified as a transcript that reproducibly increased >2-fold in the hypomethylated cells. Overexpression was observed in both CD4⁺ and CD8⁺ T cells and was due to demethylation of a conserved region located between the promoter and upstream enhancer (10). Perforin is primarily expressed in CTLs and contributes to target cell killing (11). Importantly, T cells from lupus patients promiscuously kill autologous monocytes/macrophages, which may contribute to disease pathogenesis by both increasing the amount of potentially antigenic apoptotic material as well as decreasing its clearance (8, 12). The killing is completely inhibited by Abs to class II MHC determinants and partially inhibited by Abs to class I, suggesting that CD4⁺ T cells are primarily responsible for the killing, but are assisted by CD8⁺ cells (8). This is supported by the observation that optimal monocyte/macrophage killing requires both subsets (12). Fas ligand, TRAIL, and TNF-like weak inducer of apoptosis play a role in this response, but inhibition of these molecules does not completely inhibit the killing (12). Thus, additional cytotoxic molecules, such as perforin, may contribute.

In this report we compared perforin expression in primary CD4⁺ and CD8⁺ T cells from healthy subjects and patients with active and inactive lupus. We also compared the methylation status of the perforin promoter and enhancer region in these groups, and determined the role of perforin in monocyte/macrophage killing mediated by lupus T cells. The results further support a role for DNA hypomethylation in the pathogenesis of human lupus.

Purified PHA was obtained from Murex (Norcross, GA). The following reagents were obtained from BD PharMingen (San Diego, CA) and unless specified otherwise, the Abs were all mouse anti-human mAbs: PE-anti-human perforin kit, unconjugated anti-human perforin, CyChrome (CYC)-anti-CD3, allophycocyanin- and CYC-anti-CD4, allophycocyanin-CD11a, FITC-anti-CD25, CYC- and FITC-anti-CD8, allophycocyanin-mouse IgG2a, PE-mouse IgG1, FITC-mouse IgG1, and CYC mouse IgG1. A pan-T cell isolation kit, CD4 T cell isolation kit, monocyte isolation kit, and CD8 microbeads were obtained from Miltenyi Biotec (Auburn, CA). Chloroquine, hydrocortisone, indomethacin, 6-mercaptopurine, and concanamycin A were obtained from Sigma-Aldrich (St. Louis, MO). ⁵¹Cr was obtained from PerkinElmer Life Sciences (Boston, MA).

Patients with active systemic lupus erythematosus (SLE), inactive SLE, and rheumatoid arthritis (RA) were recruited from the out-patient rheumatology clinic and in-patient services at the University of Michigan. Healthy controls were obtained by advertising. This protocol was approved by the University of Michigan internal review board, and written informed consent was obtained from all subjects. All SLE and RA patients fulfilled the American College of Rheumatology criteria for these diseases (13, 14). SLE activity was assessed by the SLE disease activity index (SLEDAI) (15), and those with SLEDAI above 5 were considered to have active disease. Patient and control T cells were paired and studied in parallel. The demographics and medications received by the subjects are shown in Table I.

PBMC were isolated by Ficoll-Hypaque (Amersham Pharmacia Biotech, Uppsala, Sweden) density gradient centrifugation as previously described (1). T cells were isolated by negative selection using magnetic beads and instructions provided by the manufacturer (pan T cell isolation kit; Miltenyi Biotec) or by rosetting with sheep erythrocytes (Colorado Serum, Denver, CO) as previously described (7). When indicated, CD4⁺ or CD8⁺ T cells were isolated using negative selection with magnetic beads after T cell purification.

Unstimulated and PHA-stimulated human T cells were cultured in 24-well plates in RPMI 1640/5% FBS (100,000 cells/ml) in the presence or the absence of graded concentrations (0.1–100 μM) of indomethacin, chloroquine, hydrocortisone, or 6-mercaptopurine (6-MP). A stock solution of 6-MP was prepared in dimethylformamide (DMF) at a concentration of 10 mg/ml. The stock solution was diluted in assay diluent (80% culture medium/20% ethanol) to yield a 6-MP working solution of 80 μg/ml or less as indicated (12). The working solution of 6-MP as well as the other materials prepared in assay diluent were then further diluted 1/40 into the cell cultures for the tests. The final concentrations of ethanol (0.5%) or DMF (<0.02%) do not yield significant effects on the cell cultures (12). Indomethacin was prepared in a concentration of 500 mM in absolute ethanol, then diluted to final concentrations in the cell culture medium immediately before use. Control cells were treated with an equal volume of the solvent; the highest concentrations of ethanol in the cell culture medium was 0.1%. After different treatment times, cells were washed, resuspended in standard buffer (SB; PBS, 2% horse serum, and 0.1% sodium azide), and stained with fluorescent Abs as below.

Perforin transcripts were quantitated by real-time RT-PCR using a LightCycler (Roche, Indianapolis, IN) and previously published primers and protocols (10). Perforin mRNA levels are presented relative to β-actin transcripts as described previously (10).

Perforin protein was quantitated by immunoblotting using previously described protocols (16) and anti-perforin Abs (BD PharMingen). Bisulfite sequencing of the perforin promoter was performed as described (10).

PBMCs were isolated from SLE patients, healthy controls, and RA patients. PBMCs or purified T cells (isolated with magnetic beads) were washed with SB and incubated for 30 min at 4°C with 0.06–0.15 μg/ml fluorochrome-conjugated mAb following the manufacturer’s directions, washed three times again with SB, and fixed in 10% formaldehyde. For perforin expression, cells were then permeabilized with 0.1% saponin, stained with mouse anti-human perforin-PE or control, washed and resuspended in 10% formaldehyde, and analyzed in a FACScan flow cytometer (BD Biosciences, Mountain View, CA).

⁵¹Cr release assays were performed as previously described by our group (12). Briefly, 250,000 PBMC were adhered to round-bottom microtiter wells in 100 μl of RPMI 1640/10% FBS for 1–2 h, after which the nonadherent cells were removed. The adherent cells were labeled with 2 μCi ⁵¹Cr/well for 3 h and then washed. T cells were obtained by rosetting or negative selection using magnetic beads. Where indicated, fractionated CD4⁺ or CD8⁺ T cells were used after isolation with magnetic beads. T cells were preincubated with 100 nM concanamycin A or DMSO for 1 h and added to the labeled monocytes at a ratio of 25:1. Cells were then incubated in the presence or the absence of 100 nM concanamycin A or DMSO. ⁵¹Cr release was determined 18 h later as previously described (12).

The difference between means was analyzed using Student’s t test or ANOVA with post-hoc analysis and Bonferroni correction, using Systat 10 software (Systat, Evanston IL) and Stata 6.0 (Stata Corp., College Station, TX). Differences in the degree of methylation of various DNA segments between groups was tested using the method of generalized estimating equations. This method fits a logistic regression model for correlated data (17). Robust SEs were used to assess the statistical significance. To determine whether current treatment with immunosuppressive agents was associated with perforin expression, univariate linear regression was performed. Control subjects were excluded from this portion of the analysis because a high degree of colinearity existed between case/control status and treatments due to the lack of immunosuppressive use in controls.

Initial experiments used flow cytometry to compare perforin expression in CD4⁺ and CD8⁺ cells from lupus patients and controls. Fig. 1,A shows perforin staining in purified CD4⁺ T cells from a healthy control, a patient with lupus, and a patient with RA. Fig. 1bsimilarly shows perforin staining in purified CD8⁺ T cells from the same three subjects. More CD4⁺ T cells from the lupus patient express perforin than those from controls, whereas perforin expression was similar in the CD8⁺ populations. Fig. 1,C compares the percentage of perforin-expressing T cells in the CD4⁺ and CD8⁺ subsets from 16 healthy controls, 25–26 lupus patients, and nine RA patients. More CD4⁺ T cells from lupus patients express perforin than those from normal controls (p = 0.03), whereas perforin expression in the CD8⁺ subset was not significantly different between lupus patients and controls. As the extent of T cell DNA hypomethylation increases with lupus disease activity, the patients shown in Fig. 1,C were classified as inactive (SLEDAI = ≤5; n = 12) or active (SLEDAI = >5; n = 13). Fig. 2,A shows that perforin is preferentially expressed in CD4⁺ T cells from patients with active lupus, whereas expression in T cells from patients with inactive lupus is not significantly different from that in controls (p < 0.001 overall, by ANOVA; p = 0.001, active vs control and active vs inactive, by post-hoc analysis). In Fig. 2 B the percentage of CD4⁺ T cells expressing perforin is plotted against the SLEDAI. In the CD4⁺ subset, perforin expression increased with disease activity (p < 0.001, by regression analysis).

These results were confirmed at the protein level by immunoblotting and at the mRNA level by real-time RT-PCR. Fig. 3,A shows a representative immunoblot of perforin and β-actin using CD4⁺ T cells, isolated by negative selection, from a healthy control and a patient with active (SLEDAI = 6) lupus. There was a relative increase in perforin protein in the lupus T cells. Fig. 3,B shows perforin protein expressed relative to β-actin in CD4⁺ T cells from eight controls, five patients with inactive lupus, and six patients with active lupus. Overall, CD4⁺ T cells from patients with active lupus expressed greater amounts of perforin than those from healthy controls (p = 0.03, by t test, control vs active lupus), and the relationship between disease activity and perforin expression detected by flow cytometry was confirmed (p < 0.001, by ANOVA). The difference between active and inactive lupus patients was not significant (p > 0.05). Fig. 3,C shows immunoblot analyses of perforin expression in CD8⁺ T cells from four of the same controls and seven of the lupus patients. Although there was an apparent increase in perforin in T cells from patients with active lupus, this was due to a large increase in one patient, whereas the rest were lower (60%) relative to the controls, and the average increase was not statistically significant. Fig. 3,D shows a representative real-time RT-PCR amplification curve of perforin transcripts from CD4⁺ T cells of a lupus patient and a healthy control. Fewer cycles were required to amplify the transcripts from the lupus patient relative to the control. Fig. 3,E shows perforin mRNA levels in CD4⁺ T cells from 14 of the same subjects shown in Fig. 3,B, again expressed relative to β-actin. CD4⁺ T cells from patients with active lupus also had greater amounts of perforin mRNA relative to CD4⁺ T cells from normal controls (p = 0.006) or patients with inactive lupus (p = 0.019). Fig. 3 F compares perforin mRNA in CD8⁺ T cells from nine controls and 10 lupus patients. In contrast to CD4⁺ T cells, there was no difference in perforin mRNA levels in T cells from patients with active or inactive disease, and perforin mRNA levels were significantly higher in CD8⁺ T cells from the controls relative to the combined lupus patients (p = 0.013). The greater decrease in perforin mRNA expression relative to protein suggests post-transcriptional control in the CD8⁺ subset.

As DNA hypomethylation contributes to LFA-1 overexpression on lupus T cells (18), we tested whether those cells overexpressing LFA-1 also overexpress perforin. Fig. 4, A and B, shows representative histograms of purified T cells from a control and a lupus patient, stained for LFA-1 and perforin expression. Relatively few control T cells expressed perforin. In contrast, greater numbers of T cells from a patient with active lupus (SLEDAI = 12) expressed perforin, and perforin was preferentially expressed in cells with high LFA-1 levels. Fig. 4 C compares the mean channel fluorescence (MCF) of CD11a expression on CD4⁺perforin⁺ and CD4⁺ perforin⁻ cells from 10 lupus patients and eight controls. In lupus patients, the CD11a MCF was significantly greater in perforin-expressing CD4⁺ T cells compared with perforin⁻ cells (p = 0.024). In contrast, there was no significant difference in CD11a MCF in perforin⁺ and perforin⁻ CD4⁺ T cells from the controls. This suggests that the perforin-expressing lupus T cells also overexpress LFA-1, which was previously shown to be due to DNA hypomethylation (9). To exclude the possibility that T cell activation contributed to LFA-1 and perforin overexpression, perforin was correlated with IL-2R (CD25) expression on T cells from 17 lupus patients. No correlation was observed (p = 0.89). Similarly, there was no relationship with CD4⁺ T cell class II MHC expression (p = 0.55). However, perforin was preferentially expressed in CD4⁺CD69⁻ cells (8.4 ± 3.2% perforin⁺CD69⁻ vs 1.4 ± 0.4% perforin⁺CD69⁺; n = 12; p = 0.015, by paired t test) and in CD4⁺ CD40L⁻ cells (5.4 ± 2.2% perforin⁺CD40L⁻ (CD40L⁻) vs 1.9 ± 1.3% perforin⁺CD40L⁺; n = 12; p = 0.07), arguing against this interpretation.

Alternatively, the increase in perforin expression may be related to the medications used to treat the lupus patients. However, patients with rheumatoid arthritis were receiving similar medications, and no effect on perforin was noted (Table I). To further exclude this possibility, unstimulated and stimulated human T cells were cultured in the presence or the absence of graded concentrations (0.1–100 μM) of medications that are commonly used in lupus treatment: indomethacin (for nonsteroidal anti-inflammatory drugs), chloroquine (for antimalarial agents), hydrocortisone (for steroids), 6-MP (for azathioprine), and their combinations, as previously reported (12). After culture for 6 and 24 h, perforin expression was determined by flow cytometry and compared with that in untreated cells. None of these medications caused significant changes in perforin expression, suggesting that medications do not account for the differences in expression (data not shown). To determine whether treatment with immunosuppressive agents was associated with abnormal perforin expression, univariate linear regression was performed. Perforin was modeled as a dependent variable, with medications modeled as dichotomous independent predictors. There was no correlation of specific medications with perforin expression.

To test whether changes in DNA methylation might contribute to perforin overexpression in CD4⁺ lupus T cells, the methylation status of a 1.3-kb region encoding the perforin promoter, enhancer, and 5′-flanking region was analyzed using bisulfite sequencing. Fig. 5,A shows the key elements of the PRF1 promoter and 5′-flanking region. The first 1300 bp 5′ to the transcription start site contains 29 potentially methylatable CG pairs. The first 55 bp 5′ to the start site contains a core promoter with a GC box. A series of repetitive elements is located between −396 and −83, and a region containing enhancer elements is located between −1089 and −835 (19). Earlier studies demonstrated that hypomethylation of the seven CG pairs located between −929 and −577, indicated by thedouble arrow in Fig. 5 A, is associated with perforin expression, and that methylation of the region suppresses promoter function in reporter constructs (10). Therefore, particular attention was focused on the methylation status of this methylation sensitive region.

Fig. 5 B shows the promoter/enhancer methylation pattern in normal CD4⁺ T cells determined using five cloned and sequenced fragments from seven (−1326 to −599 bp) or six (−457 to −4 bp) healthy individuals for a total of 35 or 30 fragments, respectively. The region from −1326 through the CG pair at −650 is relatively heavily methylated, whereas the more proximal CG pairs become progressively less methylated the closer they are to the transcription start site, with most of the methylated residues in the proximal sequences located in the region containing repetitive elements. Overall, the average methylation is 47%.

We then examined the methylation pattern of the perforin promoter on T cells from lupus patients with a range of SLEDAI scores. Fig. 5 C shows the methylation status of the same region in CD4⁺ T cells from four lupus patients with SLEDAI scores of 0–2. Again, five fragments from each donor were cloned and sequenced, so each point represents the mean of 20 determinations. The overall methylcytosine content was 51%, slightly higher than that in controls, and the difference was not statistically significant (p = 0.72). The region from approximately −900 to −600, responsible for methylation-induced suppression (10), was also not significantly different from that in controls (overall methylation, 0.80 vs 0.79; control vs lupus, p = 0.92).

Fig. 5 D shows the methylation patterns of the same region in CD4⁺ T cells from three or four lupus patients with SLEDAI scores of 4–6, determined from five sequenced fragments from each region. The average methylation of the region is less than that in controls (41%), but this was not statistically significant (p = 0.63). The enhancer 3′-flanking region was somewhat, but not significantly, hypomethylated (average methylation, 0.80 vs 0.63; control vs lupus, p = 0.31). The region closest to the promoter (−500 to the start site) was slightly less methylated than controls (total methylation, 0.21 vs 0.19; control vs lupus), but the overall difference was not statistically significant (p = 0.69), and methylation of this region had a lesser effect on gene expression relative to the enhancer (10).

Fig. 5 E shows the methylation pattern of the same region in CD4⁺ T cells from four lupus patients with SLEDAI scores of 8–12. The average methylation was 0.28, less than that in controls or patients with less active lupus (p = 0.006, control vs active lupus). The methylation-sensitive enhancer flanking region was also significantly hypomethylated relative to controls (0.80 vs 0.50; control vs lupus, p = 0.05).

Fig. 5,F shows the average methylation status of the same region in CD8⁺ T cells isolated from five (−1326 to −599 bp) or four (−457 to −4 bp) healthy individuals, for a total of 25 or 20 fragments, respectively. Overall, the entire region tended to be less methylated in normal CD8⁺ T cells compared with CD4⁺ cells (total methylation, 0.23 vs 0.47; p = 0.002). The region from −1326 to −1000 was less methylated than the corresponding region in the CD4⁺ T cells (mean fraction methylated, 0.89 vs 0.52; CD4 vs CD8; p = 0.002). The region from −900 to −599, containing the methylation-sensitive region, was also significantly (p = 0.004) less methylated in CD8⁺ T cells relative to CD4⁺ T cells (mean fraction methylated, 0.41 vs 0.71), although there was one exception at −720. The region from −457 to the transcription start site was also less methylated in the CD8⁺ T cells (CD4 vs CD8, 0.21 vs 0.07; p = 0.002). Interestingly, the overall pattern resembled the CD4⁺ T cells from the patients with the most active lupus (Fig. 5 E).

Fig. 5,G shows the methylation pattern in CD8⁺ T cells from three lupus patients with SLEDAI scores of 0–2. The overall methylation was greater than that in controls (0.46 vs 0.23, lupus vs control), and the enhancer flanking region appeared more methylated (0.67 vs 0.41), although neither was statistically significant. The perforin promoter/enhancer in CD8⁺ T cells from two or three patients with SLEDAI scores of 4–6 (Fig. 5,H) was only slightly less methylated than those from patients with scores of 0–2 (overall, 0.46 vs 0.37) and was still somewhat greater than that in controls (0.23; p = 0.07), but there were no significant differences in the methylation of the enhancer flanking region (0.67 vs 0.59; p = 0.3, SLEDAI of 0–2 vs 4–6). CD8⁺ T cells from four lupus subjects with SLEDAI scores of 8–12 (Fig. 5 I) were again only slightly less methylated than those from patients with scores of 4–6 (overall methylation, 0.31 vs 0.37) and was still more methylated than controls (0.23). The enhancer flanking region was also only slightly less methylated than that in patients with scores of 4–6 (0.49 vs 0.59). However, it was still more heavily methylated than that in controls (0.49 vs 0.41), although this was also not statistically significant.

The data presented in Fig. 5support a relationship among perforin expression, perforin promoter methylation, and disease activity in CD4⁺ T cells. To further examine the relationship between disease activity and promoter methylation, perforin quantitation and bisulfite sequencing were performed on CD4⁺ T cells from a patient during times of active (SLEDAI = 6; 21.1% CD4⁺perforin⁺ by flow cytometry) and inactive (SLEDAI = 2; 3.4% CD4⁺perforin⁺) disease. Fig. 6shows that the region from −900 to −600 is more heavily methylated in CD4⁺ T cells when her disease was inactive and fewer CD4⁺ T cells expressed perforin.

Overall, these results indicate progressive demethylation of the methylation-sensitive perforin enhancer flanking region in CD4⁺, but not CD8⁺, T cells from patients with active lupus. The progressive hypomethylation correlates well with perforin mRNA and protein levels in CD4⁺ lupus T cells.

As lupus T cells spontaneously kill autologous monocytes/macrophages (8, 12), we determined whether perforin plays a role in the monocyte killing. T cells from patients with active lupus were cultured with autologous monocytes with and without the addition of concanamycin A, a selective perforin inhibitor (20). Fig. 7,A shows that concanamycin A inhibits monocyte killing in a dose-dependent fashion. Fig. 7,B shows the effect of 100 nM concanamycin A on monocyte killing using T cells from 11 patients with active lupus. Concanamycin A causes significant (p = 0.001, by paired t test) inhibition of killing, suggesting a role for perforin in this response. These studies were repeated using purified CD4⁺ and CD8⁺ T cells from five patients with active lupus (Fig. 7 C). Concanamycin A (100 nM) caused nearly complete inhibition of killing by CD4⁺ T cells, but only partial inhibition by CD8⁺ T cells (p = 0.036, percent killing in the presence of concanamycin A, CD4 vs CD8). This suggests that nonperforin-mediated mechanisms may contribute to macrophage killing to a greater extent in CD8⁺ T cells than in CD4⁺ T cells.

Several novel observations are described in this report. First, CD4⁺ T cells from patients with active lupus abnormally overexpress perforin. Second, methylation-sensitive perforin regulatory elements are hypomethylated in CD4⁺ T cells from patients with active lupus. Third, perforin may participate in the spontaneous monocyte/macrophage killing that characterizes lupus T cells. Finally, as DNA-hypomethylating agents cause LFA-1 and perforin overexpression in CD4⁺ T cells, and CD4⁺ lupus T cells have hypomethylated DNA and overexpress LFA-1 and perforin, the results suggest that genes overexpressed in experimentally hypomethylated CD4⁺ T cells may predict genes overexpressed in lupus T cells.

Perforin expression was identified in CD4⁺ lupus T cells using flow cytometry, immunoblotting, and real-time RT-PCR and was directly proportional to disease activity. Perforin expression has been reported previously in CD4⁺ T cells, but in most instances has been in cultured T cell lines (19, 21). As lupus T cells are frequently activated (22), it is possible that T cell activation contributes to the increase in perforin expression. However, no correlation was seen between IL-2R and perforin expression, and perforin was preferentially expressed in CD69⁻ T cells, arguing against this interpretation. Further, we have not observed perforin expression in stimulated primary CD4⁺ T cells unless they were first treated with DNA methylation inhibitors (10). It thus seems reasonable to propose that the abnormal expression is due to mechanisms other than T cell stimulation alone. As the DNA methylation inhibitor 5-azaC induces perforin expression in CD4⁺ T cells by demethylating the enhancer 3′-flanking sequence (10), and the same sequences are demethylated in CD4⁺ lupus T cells, DNA hypomethylation may be contributing to the abnormal expression in CD4⁺ lupus T cells.

Our observation that the same sequences are not significantly hypomethylated in the CD8⁺ population suggests that perforin gene methylation is regulated differently in CD4⁺ and CD8⁺ T cells. Elucidating the mechanisms that establish and maintain methylation patterns in normal CD4⁺ and CD8⁺ T cells may help clarify the differential regulation in lupus. However, as we have reported that inhibiting DNA methylation in CD4⁺ T cells is sufficient to cause a lupus-like disease (3, 4), the contribution of the CD8 subset is not necessary for disease induction. Interestingly, we have been unable to induce autoreactivity in the CD8 subset using 5-azaC (23), suggesting differences in the regulation of some genes by DNA methylation in CD4⁺ and CD8⁺ T cells.

To the best of our knowledge there is only one other report of perforin expression in human lupus T cells. In this report, low expression was detected in CD3-stimulated lupus PBMC cultured for 13 days (24). However, in this report freshly isolated T cells did express perforin, although comparisons with controls were not made, and CD4⁺ and CD8⁺ subsets were not studied. Further, the decrease after 13 days of culture was also seen in the controls. Thus, perforin expression in primary human lupus T cells has not previously been studied.

Perforin expression in CD4⁺ T cells may contribute to the pathogenesis of lupus. Others have reported that injecting apoptotic thymocytes into mice causes anti-DNA Abs (25), suggesting that increasing apoptosis could contribute to an autoimmune response. Other groups have also reported that mice genetically deficient in mechanisms involved in clearing apoptotic material will develop a lupus-like autoimmune disease (26), suggesting that decreased clearance of apoptotic material can also induce autoimmunity. As macrophages are responsible for the clearance of apoptotic material, their death would both increase the amount of apoptotic debris as well as decrease its clearance. We have reported that lupus T cells use FasL, TNF-like weak inducer of apoptosis, and TRAIL to induce monocyte apoptosis (12), and perforin now appears to contribute as well. Interestingly, perforin appears to play a greater role in macrophage killing mediated by CD4⁺ lupus T cells than CD8⁺, at least as assessed by concanamycin A inhibition. It should be noted that MRL/lpr mice deficient in perforin have a more severe disease, suggesting that in some instances perforin can be protective (27). However, perforin expression is usually confined to CD8⁺ T cells and NK cells, so the decreased expression in this animal model may influence a regulatory CD8⁺ population. An abnormal overexpression in CD4⁺ T cells, on the other hand, could participate in killing of the APC with which CD4⁺ T cells normally interact, with disease inducing results.

Finally, these studies are the second example in which a gene found to increase expression in experimentally hypomethylated T cells is also abnormally hypomethylated and expressed in lupus T cells, the first being CD11a (9). We have identified >100 additional methylation-sensitive genes using oligonucleotide arrays. It is possible that other genes contributing to the pathogenesis of human lupus may be identified using a similar strategy.

We thank Theresa Vidalon and Cindy Bourke for their expert secretarial assistance, and Donna Ray for her outstanding technical help.