SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES

Objective

To examine whether autoantibodies recognizing granzyme B (GB)-cleaved autoantigens are associated with ischemic digital loss (IDL) in limited systemic sclerosis (SSc).

Methods

Fifteen of 19 patients with limited SSc and IDL were matched by age, sex, race, and duration of disease to controls with limited SSc but without IDL. The sera were used to immunoblot HeLa cell lysates and chromosome preparations that had been incubated in vitro in the absence or presence of GB. Anticentromere antibodies (ACAs) were assayed by immunofluorescence and immunoprecipitation of in vitro-translated centromere proteins (CENP-B and CENP-C). Immunoprecipitation of GB-cleaved CENPs was also performed.

Results

GB-cleaved autoantigens were immunoblotted by 16 of 19 IDL sera (84.2%) compared with 6 of 15 non-IDL sera (40.0%) (odds ratio 8.0, 95% confidence interval 1.6–40.0). This association persisted after adjustment for ACA status. Furthermore, the presence of antibodies to centromere proteins as well as to GB-cleaved antigens was highly specific for IDL, occurring in 12 of 19 IDL patients (63.2%) and in none of 15 controls (P < 0.0001). An identical 60-kd GB-generated fragment was recognized by 5 of 16 IDL sera (31.3%) and was demonstrated to arise through GB-mediated cleavage of CENP-C. GB-cleaved CENP-C fragments were recognized preferentially over the intact CENP-C molecule by antibodies from patients with IDL.

Conclusion

The striking recognition of GB-generated autoantigen fragments by sera from patients with limited SSc and IDL constitutes the first in vivo evidence that antibodies against GB-generated centromeric peptide fragments identify a distinct clinical subset.

Systemic sclerosis (SSc; scleroderma) is a complex inflammatory, fibrogenic disorder in which vascular dysfunction underlies the major clinical manifestations (1). In common with other connective tissue disorders, the autoantibodies elaborated in this disease are predictive of characteristic disease phenotypes (2). Well-recognized associations include anticentromere antibodies (ACAs) with the CREST (calcinosis, Raynaud's phenomenon, esophageal dysmotility, sclerodactyly, telangiectasias) variant of SSc, and anti-topoisomerase I antibodies with the diffuse subset of SSc and pulmonary fibrosis (for review, see refs.3–5).

Recent studies on the connective tissue disorders have implicated modification of autoantigen structure during cell death in the selection of target molecules of the autoimmune response (6, 7). Relevant pathways include those occurring during both caspase-dependent and caspase-independent modes of cell death. One property increasingly recognized as unifying many autoantigens across the spectrum of autoimmunity (including SSc, systemic lupus erythematosus, Sjögren's syndrome, inflammatory muscle disease, and Rasmussen's encephalitis) is susceptibility to modification by proteases in the cytotoxic lymphocyte granule pathway. In this regard, numerous autoantigens are susceptible to proteolytic cleavage by granzyme B (GB) during cytotoxic lymphocyte granule-mediated cell death, generating unique fragments not observed during other forms of cell death (8–10). A smaller number of autoantigens targeted in systemic autoimmunity are also directly cleaved by granzyme A (11–13).

While several autoantigens are cleaved with similar efficiencies (but at different sites) by both GB and caspases during cell death, other autoantigens are exclusively or preferentially cleaved by GB (9). Prominent among the latter molecules are autoantigens targeted in scleroderma, including CENP-B and fibrillarin (which are cleaved by GB but not by caspases) as well as topoisomerase I (which is cleaved by GB ∼100-fold more efficiently than by caspases). Interestingly, in the one instance in which this has been studied, autoantibodies reactive against uniquely modified forms of autoantigens appear to be predictive of clinical phenotype. Specifically, recognition of the caspase-cleaved form of U1-70 kd by autoantibodies was associated with lupus skin disease (14). Since different modes of apoptotic cell death generate distinct modified forms of autoantigens, the demonstration of an association between reactivity against specifically modified autoantigens and unique clinical phenotypes may implicate a particular apoptotic cell death pathway in generating disease expression.

Of particular interest in this regard is the SSc phenotype of severe digital ischemia. Although vascular abnormalities are prominent and universal features in SSc patients (15, 16), only a distinct subset develops critical tissue ischemia resulting in ischemic digital loss (IDL) (17). There are 3 recognized predictors of this unique phenotype (17–23): limited cutaneous involvement, anti-endothelial cell antibodies (AECAs), and ACAs.

Since CENP-B (a major component of the centromere antigen [24]) is cleaved by GB but not by caspases, we addressed whether autoantibodies from patients with limited SSc and digital loss preferentially immunoblot GB-cleaved forms of autoantigens. We demonstrated a striking association between recognition of GB-cleaved autoantigens and IDL in this population. Interestingly, the presence of both ACAs and antifragment antibodies was highly specific for IDL. In several cases, autoantibodies preferentially recognized the cleaved fragment over the intact form of the antigen. Moreover, in 1 subject with serum available prior to the onset of IDL, these autoantibodies preceded digital loss, suggesting that antibodies recognizing GB-generated fragments may predict subsequent phenotype. Taken together, these studies strongly implicate the cytotoxic lymphocyte granule pathway in the generation of the immune response associated with this specific clinical phenotype.

PATIENTS AND METHODS

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES

Study design and patient selection.

Nineteen patients with limited SSc were selected based on a history of IDL and availability of serum, from among patients attending The Johns Hopkins and University of Maryland Scleroderma Center between 1991 and 2000. We attempted to match each patient individually with a control subject with limited SSc but without IDL. Matching variables included age (by decade of life), sex, race, and duration of disease (within 5 years). A single control subject was randomly selected when more than 1 matched control was available. For 4 patients, we were unable to find matched controls by our age and disease duration criteria. Overall, we therefore evaluated sera from 19 patients and 15 controls. In addition, serum samples from a single patient with IDL collected sequentially over the course of 9 years were available for analysis. The initial sample for this patient was obtained 3 years prior to the first episode of digital loss.

All study participants satisfied the American College of Rheumatology (formerly, the American Rheumatism Association) criteria for SSc (25) and had skin thickening limited to the hands, forearms, feet, legs, and/or face (26). The extent of cutaneous involvement was also dichotomized as follows: subjects with type I disease had sclerodactyly with or without face or neck involvement; those with type II disease had type I disease plus skin thickening over the forearm or leg. IDL was defined as the loss of all or part of a digit secondary to an ischemic event. In all but 1 patient, the first episode of digital loss occurred prior to collection of serum samples.

Clinical and laboratory evaluation.

Clinical data were obtained in a uniform manner at the initial evaluation in the Scleroderma Center. The presence of pulmonary hypertension, interstitial lung disease, and renal insufficiency was determined using standard criteria (27). Risk factors for vascular disease were recorded as dichotomous variables. These included hypertension (defined as systolic blood pressure >140 mm Hg and/or diastolic blood pressure >90 mm Hg on at least 2 separate occasions) or the use of antihypertensive medication. Since calcium channel blockers were almost universally prescribed for the management of Raynaud's phenomenon, only non-calcium channel blocker antihypertensives were considered within this definition. Diabetes mellitus was defined as at least 2 random blood glucose measurements yielding levels >200 mg/dl. The smoking status of subjects was dichotomized as ever smoked (current or former smokers) and never smoked. Duration of disease was calculated from the onset of the first scleroderma feature (other than Raynaud's phenomenon) to the date of serum collection. In addition, disease duration was determined from the onset of Raynaud's phenomenon and from the date of physician diagnosis of SSc to the date of serum collection.

Following initial consultation in the Scleroderma Center, serum was obtained for routine hematologic and biochemical measurements and conventional serologic assays and stored at –80°C for future analyses. Conventional serologic assays included determination of antinuclear antibody (ANA) titer and immunofluorescence pattern using HEp-2 cells as substrate. ANA titers ≥1:160 were considered positive. The presence of anti-topoisomerase I antibodies was determined by standard immunodiffusion techniques.

Immunoblotting of HeLa lysates to detect cleavage of endogenous antigens.

HeLa cells were cultured using standard conditions. Lysates for in vitro reactions were prepared in buffer A, containing 10 mM HEPES/KOH pH 7.4, 2 mM EDTA, 1% Nonidet P40, and the protease inhibitors phenylmethylsulfonyl fluoride, antipain, pepstatin, and leupeptin (28). In vitro incubations were performed by incubating these lysates at 37°C for 1 hour in the absence or presence of granule contents (an amount equivalent to 42 nM purified GB), 42 nM purified GB, or 50 nM caspase 3. Identical fragments were detected by immunoblotting when cleavage was performed using granule contents or purified GB. Granule contents and GB were prepared from YT cells as previously described (8), and caspase 3 was a kind gift from Dr. N. Thornberry (Merck, Rahway, NJ). GB (and granule content) cleavage reactions were performed in the presence of 10 mM iodoacetamide (IAA) to inhibit activation of endogenous caspases by GB. Dithiothreitol (DTT; 5 mM) was added to reactions performed with caspase 3 and the IAA was omitted, to enable the added caspase 3 as well as endogenous caspases to be active.

All reactions were terminated with sodium dodecyl sulfate (SDS)-sample buffer, and 50-μg aliquots were electrophoresed on 10% SDS-polyacrylamide gels. Proteins were transferred to nitrocellulose, and intact antigens and cleaved fragments were immunoblotted (29) using patient sera at 1:2,000 dilution. Blotted proteins were detected with horseradish peroxidase-labeled secondary antibody diluted at 1:10,000 (Jackson ImmunoResearch, West Grove, PA) and chemiluminescence (Pierce, Rockford, IL).

Preparation of chromosomes.

Crude chromosomes were prepared from mitotic K562 cells as described (30), with the following modifications. K562 cells were blocked in mitosis by adding nocodazole at 0.4 μg/ml to the cultures for 16 hours. The cells were swollen, lysed, and homogenized, and the nuclei and debris were removed by centrifugation (250g for 5 minutes at 4°C). Chromosomes were centrifuged (1,300g for 20 minutes at 4°C), resuspended in buffer A, and incubated in the absence or presence of GB for 60 minutes at 37°C. CaCl2 (2 mM) and micrococcal nuclease (50 μg/ml) were added to the reactions (for 20 minutes at 4°C) before addition of gel buffer and boiling. The samples were immunoblotted with a rabbit polyclonal antibody raised against CENP-C (a kind gift from Dr. Ann Pluta) (31, 32) (1:2,000 dilution) or patient sera, as described above.

Cleavage and immunoprecipitation of 35S-methionine-labeled CENP-B and CENP-C.

35S-methionine-labeled CENP-B and CENP-C were generated by coupled in vitro transcription/translation (IVTT) using the appropriate full-length complementary DNA (kind gifts from Dr. Ann Pluta). For in vitro cleavage reactions, the radiolabeled CENPs were diluted in buffer A and incubated for 60 minutes at 37°C in the presence of purified caspases 3, 6, 7, or 8 (each at 50 nM) or purified GB (at concentrations indicated in the figures). All purified proteases were gifts from Dr. N. Thornberry (Merck, Rahway, NJ). DTT (5 mM final concentration) was added to the reactions containing caspases. In some cases, cleavage of 35S-methionine-labeled CENP-C was performed in the presence of added unlabeled HeLa lysate (45 μg/reaction) and 10 mM IAA. In vitro cleavage reactions were either terminated by adding gel buffer and boiling or were used for immunoprecipitations. In the latter case, 70 mM chymostatin was added to stop GB activity, and all further processing was performed at 4°C. Immunoprecipitations to test which of the sera recognized intact 35S-methionine-labeled CENP-B and CENP-C, or the fragments generated by GB cleavage, were performed as described (33). Gel samples were electrophoresed on 10% SDS-polyacrylamide gels, and intact proteins and their cleaved fragments were detected by fluorography.

Statistical analysis.

Demographic, clinical, and serologic variables for the patients with IDL and for the control patients without IDL were summarized as the mean ± SD and as proportions. Differences in these variables between patients and controls were analyzed using Student's t-test for continuous variables and chi-square test or Fisher's exact test (if n ≤ 5) for categorical variables. We performed simple logistic regression to estimate the odds ratio (OR) and 95% confidence interval (95% CI) for recognition of GB-cleaved HeLa lysates in patients compared with controls. Using multiple logistic regression to evaluate for possible confounding, we adjusted for ACA status. The relationship between recognition of GB-cleaved HeLa lysates and IDL was also examined for the 15 matched patients and controls using McNemar's matched-pairs analysis. All statistical analyses were performed using Stata Statistical Software (Release 7.0 [2001]; Stata Corporation, College Station, TX), and reported P values are 2-tailed with α = 0.05.

RESULTS

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES

Patient characteristics.

The demographic, clinical, and serologic characteristics of study participants at the time of serum collection are summarized in Table 1. Thirty-two of the 34 study participants were women (94.1%) and 30 were white (88.2%). Their mean ± SD age was 57.1 ± 12.0 years and their mean ± SD duration of disease from SSc onset to the date of serum collection was 13.4 ± 11.3 years. The duration of disease from the onset of Raynaud's phenomenon was 17.7 ± 13.5 years and from the date of physician diagnosis of SSc, 10.1 ± 9.6 years. A higher proportion of patients with IDL had type I skin involvement and a history of smoking, although the difference was not statistically significant. None of the study participants had diabetes mellitus, and the frequencies of systemic hypertension, pulmonary hypertension, interstitial lung disease, and renal insufficiency did not differ between patients and controls. All subjects were ANA positive; a centromere immunofluorescence pattern was seen in 15 of 19 patients with IDL (78.9%) compared with 5 of 15 controls (33.3%) (P = 0.01).

Table 1. Demographic, clinical, and serologic characteristics among 19 SSc patients with and 15 control patients without ischemic digital loss*
 All study subjects (n = 34)Patients with digital loss (n = 19)Controls without digital loss (n = 15)P
  • *

    SSc = systemic sclerosis. ANA = antinuclear antibody.

  • Estimated by right ventricular systolic pressure >35 mm Hg on Doppler echocardiography.

  • Forced vital capacity <80% of predicted, or diffuse interstitial changes on chest radiography.

  • §

    Serum creatinine concentration >1.3 mg/dl.

Age, mean ± SD years57.1 ± 12.059.1 ± 13.954.5 ± 9.00.28
Female, no. (%)32 (94.1)17 (89.5)15 (100)0.49
Caucasian, no. (%)30 (88.2)17 (89.5)13 (86.7)1.0
Duration of SSc, mean ± SD years13.4 ± 11.314.2 ± 11.812.3 ± 10.90.63
Skin involvement, no. (%)    
 Type I21 (61.8)14 (73.7)7 (46.7)0.16
 Type II13 (38.2)5 (26.3)8 (53.3) 
Ever smoked, no. (%)22 (64.7)15 (78.9)7 (46.7)0.08
Hypertension, no. (%)8 (23.5)4 (21.1)4 (26.7)1.0
Major organ involvement, no. (%)    
 Pulmonary hypertension6 (17.6)2 (10.5)4 (26.7)0.37
 Interstitial lung disease5 (14.7)3 (15.8)2 (13.3)1.0
 Renal insufficiency§2 (5.9)0 (0)2 (13.3)0.19
Autoantibodies, no. (%)    
 ANA titer ≥1:16034 (100)19 (100)15 (100)1.0
 Centromere ANA pattern20 (58.8)15 (78.9)5 (33.3)0.01
 Antitopoisomerase4 (11.8)1 (5.3)3 (20.0)0.30

Novel fragments generated in GB-cleaved HeLa lysates are recognized by sera from patients with digital loss.

Sera from the 19 patients with IDL were immunoblotted against control HeLa cell lysates. Of these sera, 18 detected distinct bands (for representative examples, see left lanes of panels in Figure 1A). A robust signal was detected using serum dilutions of 1:2,000 and enhanced chemiluminescence exposure times of 10–30 seconds. When HeLa cell lysates were incubated in vitro with GB-containing granule contents, 16 of the 19 sera immunoblotted novel fragments (for representative examples, see right lanes of panels in Figure 1A; open arrows denote novel fragments). Identical results were obtained when immunoblots were performed on lysates made from human umbilical vein endothelial cells incubated in the absence or presence of GB (data not shown) and when purified GB was used instead of GB-containing granule contents (data not shown). Since in vitro incubations were performed in the presence of 10 mM IAA to prevent activation of endogenous caspases by GB, the new bands that were generated represent novel fragments resulting specifically from GB cleavage of intact antigens. Of note, the generation of identical fragments after cleavage with GB or granule contents demonstrates that the cleavage products observed with the latter are specifically produced by GB rather than by another granule component.

thumbnail image

Figure 1. Immunoblotting of novel fragments in granzyme B (GB)-cleaved HeLa cell lysates by sera from systemic sclerosis patients with ischemic digital loss. HeLa lysates containing 10 mM iodoacetamide were incubated for 60 minutes in the absence or presence of GB-containing granule contents. The samples were subsequently immunoblotted with A, sera from patients with digital loss, B, sera from patients without digital loss, and C, sera, obtained longitudinally on the indicated dates from a patient with digital loss (patient S-43). The migration of molecular weight marker standards (in kd) is indicated on each panel, and open arrows indicate new fragments detected after GB cleavage. In B, the identities of the bands denoted “topo 1” and “U1-70 kDa” that were immunoblotted by serum FW-100 were confirmed by comigration with bands blotted by reference sera and by generation of the signature 62-kd and 40-kd fragments produced by GB and caspase cleavage, respectively (in the case of U1-70 kd), and the 95-kd fragment generated by GB cleavage (topo I) (data not shown). Topo I = topoisomerase I.

Download figure to PowerPoint

When sera from the 15 matched control patients without IDL were assayed using the same approach as that shown in Figure 1, the results were strikingly different. Three sera did not immunoblot any bands (for a representative example, see FW-143 serum in Figure 1B). Four sera detected topoisomerase I and/or U1-70 kd (for a representative example, see serum FW-100 in Figure 1B). Both of these antigens are known to be specifically cleaved by GB to generate novel fragments (9). Six sera immunoblotted antigens in HeLa cell lysates, but these antigens were not cleaved by GB after in vitro incubation with the protease (for representative examples, see sera RFW-158, RFW-148, and FW-83 in Figure 1B). The final 2 sera immunoblotted antigens which were cleaved by GB, generating novel fragments.

We therefore determined that 16 of 19 sera from patients with IDL (84.2%) immunoblotted novel fragments generated in HeLa lysates by GB treatment compared with 6 of 15 control sera (40.0%) (OR 8.0 [95% CI 1.6–40.0], P = 0.012). This association persisted after adjusting for ACA status (OR 21.5 [95% CI 2.0–229.9], P = 0.011). In a McNemar matched-pairs analysis involving only the 15 matched patients and controls, GB-cleaved autoantigens were again associated with an increased risk of IDL (OR 8.0 [95% CI 1.1–355.0], P = 0.039).

Development of phenotype preceded by recognition of cleaved antigens in longitudinal study of sera from patient S-43.

Sera had been collected from patient S-43 for 3 years prior to the first IDL episode and for 6 years thereafter. To address the temporal sequence of reactivity to GB-cleaved HeLa cells and digital loss, these sera were used to immunoblot HeLa cell lysates incubated in the absence or presence of GB (Figure 1C). The sera were all used at a 1:2,000 dilution, and the lysates were electrophoresed on a single gel and immunoblotted simultaneously to facilitate direct comparison of the profiles. Although the first digit loss event in this patient occurred in 1994, antibodies detecting intact antigens and GB-cleaved fragments were present in 1991 (the first serum sample available) (Figure 1C). The sera obtained from 1991 to 1996 immunoblotted intact antigens of ∼250 kd (unidentified) and topoisomerase I, both of which are cleaved by GB. In addition to antibodies against these two antigens, the sera obtained in 1999 and 2000 contained antibodies against a 44-kd antigen, which was not cleaved by GB. Therefore, in this patient, recognition of cleaved antigens preceded the development of the unique associated phenotype.

Identical 60-kd fragment immunoblotted by sera from several patients with digital loss.

Interestingly, 5 of the 16 sera that immunoblotted novel fragments in GB-cleaved HeLa lysates recognized the identical 60-kd fragment (solid arrow, lanes 2, 4, 6, 8, and 10 in Figure 2). In contrast to previously described GB-cleaved autoantigens in which the intact molecule and fragment are recognized similarly by autoantibody (8, 9), this 60-kd fragment was much more strongly recognized than any of the high molecular weight proteins from which it may have potentially arisen. This was particularly evident for sera S-04, FW-14, and FW-121. This preferential immunoblotting of a GB-generated fragment is of great interest because it was enriched in sera from patients with IDL. When HeLa lysates were treated with caspase 3, the 60-kd fragment was not observed (compare lanes 12 and 13 in Figure 2). Caspase 3 activity in these lysates was verified by immunoblotting with a patient serum containing antibodies to U1-70 kd; the 40-kd fragment arising from caspase 3 cleavage was detected (data not shown). It is noteworthy that of the 2 sera from patients without IDL which immunoblotted antigens cleaved by GB, neither detected 60-kd fragments in GB-cleaved HeLa lysates.

thumbnail image

Figure 2. Recognition, after granzyme B (GB) cleavage, of a new 60-kd fragment by 5 of 19 sera from patients with digital loss. Lanes 1–11 and 13, HeLa lysates containing 10 mM iodoacetamide were incubated for 60 minutes at 37°C in the absence or presence of GB-containing granule contents. Lane 12, HeLa lysate containing 50 nM purified caspase 3 (C3) and 5 mM dithiothreitol was incubated for 60 minutes at 37°C. The samples were immunoblotted with the indicated sera from patients with digital loss. The migration of molecular weight marker standards (in kd) is indicated; solid arrows denote the 60-kd fragment generated by cleavage with GB.

Download figure to PowerPoint

Recognition of CENP-B and/or CENP-C, in addition to GB-cleaved antigens, is highly specific for digital loss in limited SSc.

Since an ACA immunofluorescence pattern was also associated with IDL, the presence or absence of antibodies to CENP-B and CENP-C in patient and control sera was confirmed by immunoprecipitation using 35S-methionine-labeled antigens generated by IVTT. As expected, the results were similar to ACA determination by immunofluorescence. Thus, 14 of 19 patient sera (73.7%) recognized CENP-B, compared with 4 of 15 control sera (26.7%) (P = 0.006), while 13 of 19 patient sera (68.4%) recognized CENP-C, compared with 4 of 15 control sera (26.7%) (P = 0.016). Four patient sera recognized non-60-kd fragments generated by GB cleavage of HeLa cells and did not immunoblot CENP-B or CENP-C. Importantly, sera from 12 of 19 patients with IDL (63.2%) had reactivity to GB-cleaved fragments as well as to one or both CENP antigens, compared with none of 15 control subjects (OR 51.7 [95% CI 2.6–1,002.2], P < 0.0001). Therefore, detection of antibodies reactive both against GB-cleaved antigens and against a CENP antigen was a very specific finding for subjects with IDL in this limited SSc population.

CENP-B and CENP-C are cleaved by GB, generating novel fragments not observed during other forms of apoptotic death.

The susceptibility of CENP-B and CENP-C to cleavage by GB and by a panel of caspases was tested in vitro. 35S-methionine-labeled CENP-B was not cleaved by caspases 3, 6, 7, and 8, but was cleaved by purified GB, generating fragments of 55, 40, 28, and 25 kd (kcat/Km = 1.1 × 104M−1 · s−1) (lanes 1–6 in Figure 3). In contrast, CENP-C was cleaved by each of the caspases tested (lanes 8–11 in Figure 3) and was very efficiently cleaved by GB (kcat/Km = 7.2 × 104M−1 · s−1) (lanes 7, 12, and 13 in Figure 3). It is noteworthy that cleavage of the radiolabeled CENP-C by purified GB resulted in loss of intact antigen, without good visualization of the fragments generated, despite careful titration of the amount of protease used (compare lanes 7, 12, and 13 in Figure 3).

thumbnail image

Figure 3. Granzyme B (GB) cleavage in vitro of 35S-methionine-labeled CENP-B and CENP-C. 35S-methionine-labeled CENP-B and CENP-C (generated by in vitro transcription/translation) are cleaved by GB in vitro, generating unique fragments. Lanes 1–15, CENP-B (lanes 1–6) or CENP-C (lanes 7–15) was incubated without added proteases (Minus) or with purified caspases (Casp) 3, 6, 7, or 8 (each at 50 nM) or GB (at the indicated concentrations). Dithiothreitol (5 mM) was added to the caspase reactions. Lanes 16–18, Unlabeled HeLa lysate (45 μg) was added to reactions containing 35S-methionine-labeled CENP-C, 10 mM iodoacetamide (to inhibit activation of endogenous lysate caspases by GB), and the indicated amounts of purified GB. Note that higher concentrations of GB were used in the latter reactions because the HeLa lysate itself provided a pool of alternate (unlabeled) GB substrates (compare lanes 13 and 18). All lanes, After incubating reactions for 60 minutes at 37°C, the samples were electrophoresed on 10% sodium dodecyl sulfate-polyacrylamide gels, and radiolabeled CENP-B, CENP-C, and their cleaved fragments were visualized by fluorography. The radiolabeled CENP-C fragments generated in the presence of HeLa lysate (lanes 17 and 18) were identical to, but detected much better than, those seen in the absence of added lysate (lanes 12 and 13).

Download figure to PowerPoint

Detection of these fragments was optimized by adding unlabeled HeLa lysate to the in vitro cleavage reactions in the presence of 10 mM IAA (which inactivates the endogenous caspases in HeLa lysate). Under these conditions, distinct CENP-C fragments of ∼90 kd (doublet), 60 kd (doublet), 40 kd, and 21 kd were observed (lane 18 in Figure 3). Radiolabeled CENP-C migration on SDS-polyacrylamide gel electrophoresis (PAGE) was much more discrete when HeLa lysate was added (compare lanes 14 and 16 in Figure 3), suggesting that interactions with components in the lysate alter CENP-C conformation. It is also possible that components in the lysate may be influencing the activity and/or specificity of the cationic granzyme. Future studies will further define the mechanisms of this effect, since they may provide important insights into the molecular interactions in vivo that ultimately result in generation of the cleavage fragments preferentially recognized by the patient sera.

Immunoblotted 60-kd fragment detected in GB-cleaved cell lysates is derived from CENP-C.

Several previous studies have shown that autoantibody recognition of centromere proteins by immunoblotting is enhanced in mitotic cells, possibly due to mitosis-specific posttranslational modifications (34, 35). To facilitate detection of CENP-C by immunoblotting, crude chromosomes, enriched in centromere proteins, were prepared from mitotic K562 cells. These were incubated in the absence or presence of GB and immunoblotted with serum S-396 or a polyclonal antibody raised against CENP-C.

After cleavage by GB, a novel 60-kd fragment was detected by serum S-396; this comigrated with the band detected by the polyclonal anti-CENP-C antibody (Figure 4). (Note that identical sets of samples were electrophoresed in adjacent gel lanes for this comparison.) Interestingly, both sera detected a slight alteration in SDS-PAGE migration of CENP-C (a loss of ∼1–2 kd) after incubation for 60 minutes at 37°C in the absence of GB, as compared with unincubated samples, probably due to phosphorylation of the CENP-C in mitotic cells which is removed by phosphatases during in vitro incubation of lysates. As noted in Figure 2, sera from patients with IDL recognized intact CENP-C poorly by immunoblotting, yet recognized the 60-kd fragment very well. Interestingly, 4 sera from the control group recognized CENP-C by immunoprecipitation, but none of these sera immunoblotted the 60-kd CENP-C fragment.

thumbnail image

Figure 4. Cleavage by granzyme B (GB) of CENP-C in crude chromosomal preparations. CENP-C in crude chromosomal preparations is cleaved by GB, generating a 60-kd fragment which comigrates with that immunoblotted by S-396 serum. Crude chromosomes, prepared from mitotic K562 cells, were incubated for 1 hour at 37°C in the absence or presence of 56 nM purified GB. A third sample in each set was not incubated; gel buffer was added directly to the freshly prepared chromosomes. All reactions contained identical amounts of crude chromosomes (equivalent to an amount prepared from 2.5 × 106 mitotic K562 cells), and the reactions were performed in a volume of 25 μl. Duplicate sample sets were electrophoresed adjacent to each other on 10% sodium dodecyl sulfate-polyacrylamide gels and immunoblotted with a polyclonal antibody raised against CENP-C or patient serum S-396.

Download figure to PowerPoint

GB-cleaved form of CENP-C protein is preferentially detected by autoantibodies recognizing CENP-C.

The ability of sera from patients with IDL to detect intact or GB-cleaved CENP-B or CENP-C was tested directly by immunoprecipitation. IVTT-generated CENPs were incubated in the absence or presence of GB, prior to immunoprecipitation with the sera. Patient sera containing antibodies to CENP-B recognized both the intact antigen and the GB-cleaved fragments, with a preference for the intact antigen (compare the profile of the cleaved CENP-B in lane 2 of Figure 5 with that of the cleaved, precipitated CENP-B in lane 4 of Figure 5). In contrast, patient sera containing antibodies to CENP-C recognized the GB-cleaved CENP-C fragments well but detected the intact protein poorly (compare the profile of the cleaved CENP-C in lane 6 of Figure 5 with that of the cleaved, precipitated CENP-C in lane 8 of Figure 5), indicating that these sera are fragment-specific/preferential. This is consistent with the results observed with immunoblotting (Figure 2), and is the first description of an autoantibody that preferentially recognizes the cleaved form of its cognate antigen. Of note, the patient sera recognized the prominently labeled 21-kd fragment generated by GB cleavage. It is likely that this fragment partially overlaps the 60-kd fragment observed by immunoblotting. The exact location will require determination of the GB cleavage sites in CENP-C by mutation analysis.

thumbnail image

Figure 5. Preferential detection of granzyme B (GB)-cleaved form of CENP-C. Patient sera recognizing CENP-C preferentially detect the GB-cleaved form of this molecule. 35S-methionine-labeled CENP-B (generated by in vitro transcription/translation [IVTT]) was incubated in the absence (lanes 1 and 3) or presence (lanes 2 and 4) of 40 nM purified GB for 60 minutes at 37°C. 35S-methionine-labeled CENP-C (generated by IVTT) was added to 45 μg unlabeled HeLa lysate containing 1 mM iodoacetamide and incubated in the absence (lanes 5, 7, and 9) or presence (lanes 6, 8, and 10) of 50 nM purified GB for 1 hour at 37°C. An aliquot of each of the cleavage reaction mixtures was used to prepare a gel sample to assess the extent of cleavage (lanes 1, 2, 5, and 6). The remainder of each cleavage reaction mixture was used for immunoprecipitation (Ipptn) with patient serum. The following sera were used: S-04 (lanes 3 and 4), S-396 (lanes 7 and 8), and FW-121 (lanes 9 and 10). Antibodies to CENP-B recognized the intact antigen better than the cleaved fragments. In contrast, antibodies to CENP-C preferentially recognized the CENP-C fragments generated by GB cleavage. Note that the additional bands (generated in IVTT reactions) that are smaller than the intact antigen arise from less efficient, alternate initiation sites and/or premature termination of mRNA translation.

Download figure to PowerPoint

DISCUSSION

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES

Modification of autoantigen structure during different forms of cell death has been implicated in the selection of molecules as targets of the autoimmune response. By demonstrating a striking association between recognition of GB-cleaved autoantigens and a specific clinical phenotype (IDL in limited scleroderma), this study strongly implicates the cytotoxic lymphocyte granule pathway in the generation of this phenotype and its associated autoantibody response. Furthermore, this is the first direct demonstration that a subgroup of phenotype-associated autoantibodies preferentially recognizes the cleaved form of autoantigens, indicating that GB-mediated cleavage plays an important role in unmasking cryptic B cell epitopes. Finally, since there are very few predictors of the development of digital loss in patients with limited scleroderma, these findings have potential clinical relevance for identifying patients who are at particularly high risk for developing IDL.

GB is a serine protease expressed in cytoplasmic granules of cytotoxic T lymphocytes (CTLs) and natural killer (NK) cells. It induces apoptosis in target cells during granule exocytosis-induced cytotoxicity by catalyzing the cleavage and activation of several caspases, as well as through caspase-independent pathways. Unlike most forms of apoptotic death, which occur in noninflammatory contexts and actively suppress the initiation of primary immune responses (36), the GB pathway has a major function at the proinflammatory host-pathogen interface, where it plays a central role in clearance of intracellular infections (37–39).

The following findings of several recent studies have strongly implicated the GB pathway in the pathogenesis of systemic autoimmune diseases. First, CTLs are present and express an activated phenotype with markedly up-regulated granzyme expression in several diseases, including scleroderma (40), Sjögren's syndrome (41, 42), and autoimmune myositis (43). Second, GB is expressed at high levels by several cell types (including T cells and NK cells) within the synovium in rheumatoid arthritis (RA) (44, 45). Third, levels of GB in joint fluid and serum appear to be predictive of the subsequent development of erosive joint disease in RA (46, 47). Fourth, GB specifically cleaves most autoantigens at sites not recognized by caspases (9), potentially revealing cryptic T cell epitopes. While this last property is strongly associated with a molecule's autoantigen status, previous studies have not demonstrated that recognition of GB-cleaved autoantigens associates with particular phenotypes, nor has there been any demonstration of preferential recognition by autoantibodies of GB-cleaved forms of antigens over the intact molecules (9).

This study demonstrates that autoantibody recognition of GB-cleaved antigens is strongly associated with IDL in limited SSc. The mechanism underlying this association is not yet known, but may involve a direct role for GB in altering the immunogenicity of these molecules (see below) or may identify another structural feature that influences immunogenicity.

Although they frequently coexist with ACAs, the antibodies that recognize cleaved antigens contribute independently to the prediction of digital loss. Approximately one-third of sera from patients with IDL recognized a 60-kd fragment arising from GB-mediated cleavage of CENP-C. Of note, antibodies to CENP-C preferentially recognized the fragment rather than the intact form of the antigen. Although both CENP-B and CENP-C are susceptible to cleavage by GB, proteolysis of these two proteins differed in 3 distinct ways. First, CENP-C was susceptible to efficient cleavage by several caspases as well as by GB, with generation of distinct fragments, while CENP-B was exclusively cleavable by GB. Second, GB cleaved CENP-C significantly more efficiently than CENP-B. Third, GB-generated fragments of CENP-C were preferentially recognized over the intact form by sera from patients with IDL, while fragments of CENP-B were not preferentially bound. Thus, among the numerous autoantigens that are cleaved by GB, CENP-C represents the first autoantigen in which fragment-preferential autoantibodies have been demonstrated (9).

Fragment-specific autoantibodies may arise when cleavage generates or reveals novel structure that is cryptic in the native molecule. This might include exposure of the new termini generated by cleavage, or other conformational modifications distant from the cleavage site. The presence of such fragment-specific antibodies argues strongly that the cleaved form of the antigen is responsible for driving the immune response in patients with this phenotype. The fact that the majority of autoantibodies targeted across the spectrum of autoimmune diseases do not preferentially recognize cleaved forms of their target molecules strongly suggests that the B cell epitopes in these molecules are not hidden in the intact molecule, while the relevant T cell epitopes are (and hence might require cleavage in order to be revealed). By demonstrating preferential recognition of GB-induced fragments in some individuals manifesting a unique clinical phenotype, this study shows the first in vivo evidence that the cytotoxic lymphocyte granule pathway plays a role in the selection of targets for an autoantibody response. Demonstration that a similar principle applies to the generation of T cell epitopes for many other autoantigens is a high priority.

The fact that most patients with limited scleroderma and without digital loss did not recognize GB-cleaved antigens, but did manifest other clinical features of limited SSc, suggests that the GB pathway plays a role in generating some aspects of this phenotype, but that it is not a universal component of pathogenesis in this disease. Previous studies have demonstrated that vascular cell apoptosis may be an early event in SSc (21), but the inducers of this cell death in vivo are not yet known. Potential candidates include cytotoxic lymphocytes (including NK cells and CTLs), antibody-mediated cytotoxicity (e.g., AECAs), and recurrent ischemia-reperfusion. The potential for additional autoantigen modifications (e.g., metal-catalyzed oxidation, glutathiolation) during the last two of these processes has been emphasized (48, 49). It will be important to determine whether the autoantigens that are shared between patients with and without IDL and that also are not cleaved by GB are otherwise modified during apoptotic cell death. Defining such a unifying process will focus attention on additional pathogenic pathways of universal importance in this disease spectrum. In this regard, it is noteworthy that pyruvate dehydrogenase E2, the major autoantigen in primary biliary cirrhosis, which is sometimes associated with limited SSc, is oxidatively modified during apoptotic death in a cell type-specific manner (49).

In summary, we have demonstrated a striking and (in some cases) preferential recognition of GB-generated autoantigen fragments in sera from patients with limited SSc and the unique phenotype of IDL. These findings constitute the first in vivo evidence that antibodies against GB-generated centromeric peptide fragments identify a distinct clinical subset. Furthermore, the identification of serologic risk factors for IDL in limited scleroderma may allow the opportunity for early targeted intervention in at-risk patients.

REFERENCES

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES