Autoantibodies against B23, a nucleolar phosphoprotein, occur in scleroderma and are associated with pulmonary hypertension




To determine whether the abundant nucleolar phosphoprotein B23 is a target of autoantibodies in scleroderma, and to examine the clinical phenotype associated with these antibodies.


Ninety-two randomly selected scleroderma sera were screened by enzyme-linked immunosorbent assay against recombinant human B23. Demographic, clinical, and serologic parameters associated with B23 autoantibody status were examined.


We demonstrated that autoantibodies against B23 occur in ∼11% of sera obtained from patients with scleroderma. B23 seropositivity was related to pulmonary hypertension, antifibrillarin antibody, anti-RNP antibody, and decreased lung capacity. In multivariate analysis, B23 autoantibodies remained strongly associated with moderate-to-severe pulmonary hypertension and antifibrillarin antibodies.


These data unite B23 with the group of nucleolar autoantigens targeted in scleroderma and thus focus attention on changes in the nucleolus that render its components immunogenic in this disease. The demonstration that antibodies to B23 are associated with an increased prevalence of pulmonary hypertension points to anti-B23 antibodies as a possible marker of a specific phenotype in scleroderma.


Autoantigens targeted in the systemic autoimmune diseases comprise a very diverse group of molecules, with no obvious unifying features (1). An exception to this generalization, however, is scleroderma, in which most of the major autoantigens are unified in their localization to the nucleolus (2). A subset of these molecules, including fibrillarin, RNA polymerase I, and NOR-90 (3–5) are predominantly nucleolar proteins, while others, including topoisomerase I and the centromere proteins, mainly reside in nuclear regions outside the nucleolus, although they can also be found in the nucleolus (6, 7). In addition to serving as disease-specific markers of scleroderma, autoantibodies against several of these nucleolar autoantigens further serve as phenotype-specific markers within scleroderma. For example, in diffuse scleroderma, antibodies to topoisomerase I and fibrillarin are associated with pulmonary fibrosis (8) and pulmonary hypertension (9, 10), respectively. Similarly, antibodies against the centromere proteins are markers of CREST syndrome (calcinosis, Raynaud's phenomenon, esophageal dysmotility, sclerodactyly, telangiectasias) or limited scleroderma (11, 12). Further identification and characterization of nucleolar autoantigens in scleroderma, including their unique phenotypic associations, may aid in understanding the etiology of the autoimmune response in this disorder and provide additional phenotype-specific markers.

Although the targeting of nucleolar autoantigens is most commonly associated with scleroderma, nucleolar autoantigens are also targeted in various malignancies, with the most extensive studies performed in hepatocellular carcinoma (12). Such molecules include fibrillarin, UBF/NOR-90, and the nucleolar phosphoprotein B23 (13). B23 is one of the most abundant proteins in the nucleolus (14) and has multiple functions, including a putative role in pre–ribosomal RNA processing and ribosome assembly (15–17). Levels of this nucleolar protein are significantly increased in tumor and proliferating cells compared with normal resting cells (18). Moreover, B23 has been implicated as playing a role in the regulation of cell proliferation (18, 19).

In contrast to the well-characterized targeting of B23 as an autoantigen in some cancers (13, 20), there are very limited data on the presence of B23 autoantibodies in autoimmune diseases. Three groups of investigators have previously identified autoimmune sera recognizing B23 by immunoblotting preselected sera exhibiting strong antinucleolar staining by immunofluorescence from patients with known systemic rheumatic diseases (21–23). In one study, B23 autoantibodies were detected in a patient with rheumatoid arthritis with sicca complex (21). In a second study, antibodies against B23 were observed in 2 patients with systemic lupus erythematosus (SLE) (22). Finally, in a third report, screening of 164 sera obtained from patients with autoimmune rheumatic disease yielded 7 sera (all from patients with lupus or a lupus variant) that were positive for B23 antibodies. Because antibodies reactive with nucleolar proteins are most frequently found in scleroderma (21, 24), and B23 is a major component of the nucleolus, we investigated whether B23 is also a target of the autoimmune response in scleroderma.

In this study, we identified B23 as a novel nucleolar autoantigen in scleroderma. Interestingly, scleroderma patients with antibodies against B23 (B23-positive) manifested a significantly increased prevalence of pulmonary arterial hypertension (PAH) and decreased lung function compared with those without B23 antibodies. Of note, the presence of B23 autoantibodies was enriched in patients with limited disease and associated with antibodies to U1 RNP, although none of the B23-positive sera possessed anticentromere antibodies. We also observed that 80% of the B23-positive scleroderma sera additionally contained antibodies against fibrillarin, a nucleolar autoantigen associated with pulmonary hypertension in the diffuse form of this disease (9, 10). The targeting of the structurally unrelated nucleolar autoantigens B23 and fibrillarin in scleroderma patients with a similar pulmonary phenotype suggests that a unique process affecting nucleoli of the pulmonary vasculature drives their co-selection and renders these molecules immunogenic.



Sera were randomly selected, without knowledge of clinical information, from a serum bank derived from patients evaluated at the Johns Hopkins and University of Maryland Scleroderma Center. Demographic data of these patients were compared with the demographic profile of the entire 1,011 patients followed up at the Scleroderma Center, to assess whether the randomly selected samples were representative of the overall Center population. Clinical information was prospectively collected at the time of clinic visits. All patients either met American College of Rheumatology (ACR) criteria for the diagnosis of scleroderma (25) or demonstrated 3 of the 5 features of CREST syndrome.

Standard clinical definitions of organ involvement were derived from published criteria (26). Patients were determined to have lung involvement if their forced vital capacity (FVC) or single breath diffusing capacity (DLCO) was <80% of predicted values or if there was echocardiographic, cardiac catheterization, or clinical evidence of PAH. PAH was defined by a right ventricular systolic pressure (RVSP) of ≥36 mm Hg as measured by echocardiography or by right heart catheterization. The severity of PAH was graded as mild (36–45 mm Hg), moderate (46–55 mm Hg), or severe (≥56 mm). Patients with no estimated RVSP but normal right ventricular size were considered normal. In the event that multiple pulmonary functions tests or echocardiograms were performed for an individual patient, the result of the most recent test performed before a defined cutoff date (December 1, 2000) was used in this study. The serologic profiles of patients were analyzed for antinuclear antibody (ANA) and anticentromere antibody (ACA) positivity by immunofluorescence and for anti Scl-70 and anti–U1 RNP antibodies by immunodiffusion, as previously described (27).

Enzyme-linked immunosorbent assays (ELISAs).

The complementary DNA for human B23 (a gift from Sophie Roy, Merck-Frosst, Pointe Claire-Dorval, Quebec, Canada) and fibrillarin (a gift from John Aris, University of Florida, Gainesville) were cloned into pET-28a+ vectors (Novagen, Madison, WI) containing an N-terminal histidine tag. Expression was induced with IPTG in BL21(DE3) cells (Novagen) for 2.5 hours, and cells were lysed in a buffer containing 20 mM imidazole, 0.5M NaCl, 20 mM Tris (pH 7.9), 0.05% Triton X-100, 20 mM β-mercaptoethanol, and protease inhibitors (leupeptin, chymostatin, pepstatin, antipain, phenylmethylsulfonyl fluoride). Purification was carried out over columns containing Ni-NTA resin (Qiagen, Valencia, CA). The purity of the proteins was assessed by Coomassie blue staining following sodium dodecyl sulphate (SDS)–polyacrylamide gel electrophoresis, and was 98% (Figure 1).

Figure 1.

Purification of His-B23 and His-fibrillarin. 1 μg of purified His-B23 and His-fibrillarin were electrophoresed on a 12% sodium dodecyl sulfate/polyacrylamide gel and visualized by Coomassie blue staining. A filled arrow and open arrow indicate the positions of full length His-tagged B23 and fibrillian, respectively. Molecular weight markers are shown to the left in kDa.

For ELISAs, 96-well high-binding plates (ImmunoPlate/MaxiSorp surface; Nalge Nunc International, Naperville, IL) were coated overnight with either 5 μg/ml purified His-B23 or 1 μg/ml His-fibrillarin in carbonate/bicarbonate binding buffer (pH 9.5) at 4°C. Plates were washed 3 times with distilled water, filled with blocking buffer (phosphate buffered saline [PBS] plus 5% bovine serum albumin [BSA] plus 0.05% Tween 20 plus 0.02% NaN3) and incubated for 2 hours at room temperature. Plates were washed again 3 times with distilled water, followed by the addition (in triplicate) of sera diluted 1:1,000 in PBS plus 1% BSA plus 0.05% Tween 20 (diluting buffer) for 2 hours at room temperature. Plates were washed extensively, and horseradish peroxidase (HRP)–conjugated goat anti-human IgG (Jackson ImmunoResearch, Avondale, PA) was added at 1:5,000 in diluting buffer and incubated for another 2 hours at room temperature. After extensive washing, detection was performed by adding an equal mixture of ABTS peroxidase substrate and peroxide H2O2 (Kirkegaard & Perry, Gaithersburg, MD) to each well. Absorbance was read at 405 nm.

Positive sera were defined as those with mean values, in at least 2 independent experiments, greater than the mean plus 2 standard deviations of 5 normal controls. The 5 healthy controls routinely used in each assay were representative (in terms of mean ± SD) of a pool of 20 sera tested from nonautoimmune individuals. Healthy controls had no history, symptoms, or signs of scleroderma or other autoimmune disease.


HeLa cells were lysed in RIPA buffer (150 mM NaCl, 50 mM Tris HCl [pH 8.0], 0.5% NP40, 5 mM EDTA, 0.5% sodium deoxycholate, and 0.1% SDS) containing the protease inhibitors pepstatin, antipain, chymostatin, leupeptin, and PMSF. The lysates were centrifuged at 16,000g for 20 minutes at 4°C. The supernatant was used in an immunoprecipitation assay to confirm the absence or presence of anti-B23 antibodies in the selected sera. Briefly, HeLa lysates were diluted in NP40 lysis buffer (150 mM NaCl, 20 mM Tris [pH 7.4], 0.1% NP40, and 1 mM EDTA) containing protease inhibitors and incubated, with mixing, at 4°C for 2 hours with sera from nondisease controls (n = 3) or from scleroderma patients who tested negative (n = 2) or positive (n = 10) for anti-B23 antibodies by ELISA. Antibody–antigen complexes were precipitated with protein A–agarose and electrophoresed on 12% SDS–polyacrylamide gels following extensive washing. Proteins were transferred to nitrocellulose, and the presence of B23 was assessed by immunoblotting with a rabbit polyclonal antibody raised against recombinant His-B23 followed by detection with a HRP-labeled secondary antibody (New England Biolabs, Beverly, MA) and chemiluminescence (Pierce, Rockford, IL).

Statistical analysis.

The distribution of demographic, clinical, and serologic variables between the B23-positive and B23-negative patients was compared using Student's t-test and chi-square testing for continuous and categorical variables, respectively. We performed univariate and multivariate logistic regression to estimate the association of these variables with B23 serologic status. The multivariate model consisted of age, disease subtype, and those variables found to be statistically significant in the univariate analyses. The strength and precision of these associations were assessed using odds ratios (ORs) and 95% confidence intervals. Statistical significance was defined as α = 0.05 using a 2-tailed test.


B23 autoantibodies occur in scleroderma: demographic data.

In order to determine whether B23 is an autoantigen in scleroderma, we screened 92 randomly selected sera obtained from scleroderma patients by ELISA, using recombinant human B23. Demographic information pertaining to these 92 patients was as follows: the mean (± SD) age of the patients was 53 ± 13.8 years, 80 (88%) were female, 63 (68%) were Caucasian, and 24 (26%) were African American. These data are consistent with the demographic profile of the overall patient population followed up at the Scleroderma Center, in which the mean age is 52 ± 13.7 years, 83% are female, 75% are Caucasian, and 18% are African American. Seventy (76%) of these 92 patients met ACR criteria for a diagnosis of scleroderma. The other 22 patients had at least 3 of the 5 features of CREST syndrome.

Antibodies against B23 were observed in 10 (10.9%) of the 92 scleroderma sera. The presence of anti-B23 antibodies in these 10 sera was confirmed by the ability of these sera to immunoprecipitate B23 from HeLa lysates (Figure 2, and data not shown). Sera from 3 nondisease controls as well as from 3 scleroderma patients who tested negative for anti-B23 antibodies by ELISA did not immunoprecipitate B23 (Figure 2 and data not shown). The sera depicted in Figure 2 are representative of those tested. The average age of the patients with anti-B23 antibodies was younger (45 ± 13 years) than that of patients without B23 antibodies (54 ± 14 years), although this difference was of borderline statistical significance. The sex and race distribution did not differ between the groups of patients with B23 antibodies and those without B23 antibodies (Table 1).

Figure 2.

Immunoprecipitation of B23 from HeLa lysates with scleroderma patient sera. 50 μg of HeLa lysate was used in an immunoprecipitation reaction with 1 μl of the indicated patient serum. The immunoprecipitated protein was visualized by Western blotting with a rabbit polyclonal antibody to B23. C1 and C2 are non-disease controls. S119 is a representative scleroderma serum that was B23-negative (−) by enzyme-linked immunosorbent assay (ELISA). The remaining six scleroderma sera are representative of the ten that were B23-positive (+) by ELISA.

Table 1. Demographic, clinical, and serologic features of scleroderma patients with and without B23 autoantibodies*
 B23-positive (n = 10)B23-negative (n = 82)P
  • *

    Unless other wise indicated, values are the number (%). RP = Raynaud's phenomenon; ACR = American College of Rheumatology; FVC = forced vital capacity; DLCO = single breath diffusing capacity; ANA = antinuclear antibody; ACA = anticentromere antibody.

  • One patient in this group did not have pulmonary function test data.

  • Moderate/severe pulmonary hypertension in the B23-positive group versus the B23-negative group.

Age, mean ± SD years45.3 ± 13.054.0 ± 13.70.06
No. females/males9/171/111
 White5 (50)58 (71)0.18
 Black4 (40)20 (24)0.28
 Other1 (10)4 (5)0.45
Duration of disease (non-RP), mean ± SD years7.6 ± 6.39.1 ± 9.80.71
Duration of disease (RP), mean ± SD years9.8 ± 5.712.0 ± 11.70.46
Duration of disease (diagnosis), mean ± SD years5.4 ± 6.17.3 ± 9.40.54
Scleroderma type   
 Limited7 (70)42 (51)0.33
 Diffuse3 (30)40 (49)0.33
ACR criteria7 (70)63 (77)0.7
Organ involvement   
 Gastrointestinal6 (60)52 (63)1
 Kidney1 (10)9 (11)1
 Muscle1 (10)7 (9)1
 Joint2 (20)15 (18)1
 Endocrine3 (30)11 (14)0.18
 Sicca3 (30)20 (68)0.71
 Lung9 (90)58 (71)0.28
Smoking status   
 Current smoker1 (10)17 (21)0.68
 Former smoker2 (20)22 (27)0.73
 Never smoked7 (70)43 (52)1
Digital loss0 (0)8 (10)0.59
Cancer0 (0)0 (0) 
Pulmonary function tests   
 % FVC, mean ± SD64.1 ± 15.081.8 ± 20.20.01
 % DLCO, mean ± SD56.1 ± 22.270.7 ± 27.00.10
Pulmonary hypertension6 (60)20 (24)0.03
 Mild1 (10)10 (12) 
 Moderate1 (10)5 (6)0.002
 Severe4 (40)5 (6) 
 ANA-positive10 (100)77 (94)1
 ACA-positive0 (0)13 (16)0.35
 Scl-70-positive2 (20)11 (13)0.63
 U1-RNP-positive4 (40)7 (9)0.02
 Fibrillarin-positive8 (80)15 (18)0.0002

Clinical data and specific features.

The duration of disease (as defined from the onset of Raynaud's phenomenon, from the first appearance of a non-Raynaud's phenomenon feature, or from the date of a physician diagnosis of scleroderma) did not differ between the patients with B23-positive and those with B23-negative sera (Table 1). Seven (70%) of the 10 patients in the B23-positive group were classified as having limited scleroderma, compared with 42 (51%) of 82 patients in the B23-negative group (P = 0.33).

No differences in several clinical parameters were found when comparing the B23-positive versus the B23-negative groups (Table 1). The frequency of organ involvement of the gastrointestinal tract, the kidney, the musculoskeletal or endocrine system, and of overall lung involvement was similar in the 2 groups. There was also no significant difference between these groups in the frequency of sicca symptoms, smoking status, history of cancer, or ischemic digital loss.

Pulmonary function testing.

Pulmonary function testing was available on all but one patient (who was in the B23-negative group). B23-positive patients had significantly reduced lung function compared with the B23-negative group, with a mean predicted FVC of 64.1% ± 15.0% for the B23-positive group versus 81.8% ± 20.2% for the B23-negative group (P = 0.01) (Table 1). Of note, 9 (90%) of the 10 patients in the B23-positive group had FVC values <80% of predicted, compared with 35 (43%) of the B23-negative group (P = 0.006). In 8 of these 9 B23-positive patients with FVC values <80%, the forced expiratory volume in 1 second/FVC ratio was >80% (9 of 9 patients had ratios >75%), consistent with a restrictive ventilatory defect. Similar to the FVC values, the mean predicted DLCO values were also reduced in the B23-positive group (56.1% ± 22.2%) compared with the B23-negative group (70.7% ± 27.0%); however, this difference did not reach statistical significance (P = 0.10).

Echocardiographic data.

All 10 patients in the B23-positive group and 70 of the 82 patients in the B23-negative group had 2-D echocardiographic studies performed. The RVSP was estimated in all 10 of the B23-positive and in 53 of the B23-negative patients who received echocardiograms. The 17 patients with no estimated RVSP on echocardiography (2 of these patients also had right heart catheterization) but normal right ventricular size were considered not to have PAH. The presence of PAH was more frequent in scleroderma patients with B23 antibodies than in those without antibodies against B23 (60% versus 24%; P = 0.03) (Table 1).

We additionally analyzed the B23-positive versus B23-negative populations with respect to the frequency of moderate-to-severe PAH. Interestingly, the association between B23 seropositivity and PAH was strengthened by this analysis, with 5 (50%) of 10 scleroderma patients with B23 antibodies demonstrating moderate-to-severe PAH, versus 10 (12%) of 82 patients without B23 antibodies (P = 0.002) (Table 1). Of note, all 5 of the B23-positive patients with moderate-to-severe PAH had limited scleroderma and FVC values <80% of predicted. The mean FVC value for these 5 patients was 57.0% ± 9.1% of predicted.

Serologic data.

The frequency of antibodies against U1 RNP was greater in scleroderma patients with anti-B23 antibodies than in those without anti-B23 antibodies (40% versus 8.5%; P = 0.02). All 4 of the B23-positive patients with anti-RNP autoantibodies had limited scleroderma yet demonstrated features of an overlap syndrome. The first of these 4 patients displayed photosensitivity and developed polyarthritis and histologic evidence of glomerulonephritis. The second manifested biopsy-proven inflammatory myositis. The third patient had a history of pericarditis, malar rash, and fever, and developed a polyarthritis. The fourth patient had episodic fever, oral ulcers, and polyarthritis. Of note, the association between antibodies against B23 and U1 RNP was maintained upon analysis of these autoantibody frequencies among the patients with limited scleroderma only. Four (57%) of the 7 patients with limited scleroderma and anti-B23 antibodies had antibodies to U1 RNP, compared with 4 (9.5%) of 42 patients without B23 antibodies (P = 0.009). In contrast to the situation with anti–U1 RNP antibodies, none of the 10 B23-positive patients were ACA-positive, compared with 16% (13 of 82) of the B23-negative patients (P = 0.35). The prevalence of ANA and antibodies to Scl-70 was also not significantly different between the B23-positive and B23-negative patients (Table 1).

Antibodies against another nucleolar autoantigen, fibrillarin, have previously been shown to occur in association with pulmonary fibrosis (28) and pulmonary hypertension (9, 10) in patients with diffuse scleroderma. We therefore investigated whether sera obtained from B23-positive scleroderma patients might contain antifibrillarin antibodies. Upon screening sera obtained from the same cohort of 92 scleroderma patients for antibodies against fibrillarin by ELISA, it was found that 80% (8 of 10) of the B23-positive scleroderma sera demonstrated antibodies against fibrillarin, compared with 18% (15 of 82) of the B23-negative sera (P = 0.0002) (Table 1).

Multivariate analysis.

After determining the demographic, clinical, and serologic associations with B23 seropositivity in the above analyses, we next used multivariate analysis to address the possibility of confounding. Such an approach may assist in teasing out the independent association of a given parameter with B23 autoantibodies status after simultaneously taking into account the other observed associations with B23 derived from the unadjusted analyses. Using multiple logistic regression, moderate-to-severe PAH remained associated with B23 autoantibody status (Table 2). Of note, there was a nearly 10-fold increased risk of pulmonary hypertension (OR 9.84) associated with B23 seropositivity. Fibrillarin antibody positivity remained associated with B23 seropositivity (OR 21.9) as well. In contrast, the association of B23 seropositivity with anti-RNP antibody and with reduced FVC greatly diminished (OR of 1.2 and 1.65, respectively).

Table 2. Association of demographic, clinical, and serologic variables with B23-positive status*
VariableOdds ratio95% confidence interval
  • *

    Odds ratios and confidence intervals calculated using logistic regression. Age is represented as a dichotomous variable, at or above the median age of 53 years in comparison to below median age. Pulmonary hypertension is represented as with compared to without moderate-severe pulmonary arterial hypertension. Forced vital capacity is represented as below, compared to at or above the 80% predicted value. Multivariate analysis adjusted for each of the other variables in the above model.

Limited versus diffuse disease  
Anti-fibrillarin antibody  
U1-RNP antibody  
Pulmonary hypertension  
Reduced forced vital capacity  

Longitudinal B23 data.

Because autoantibodies sometimes predate the development of disease symptoms, we determined whether antibodies against B23 might predict the subsequent development of pulmonary hypertension. A longitudinal analysis of B23 autoantibody status was possible in 3 B23-positive patients, each of whom had limited scleroderma. For these three individuals, multiple sera were available, taken both before (at a point in time when echocardiography was normal) and after detection of moderate-to-severe pulmonary hypertension. In all 3 patients, autoantibodies against B23 predated echocardiographic detection of moderate-to-severe pulmonary hypertension; this time lag was 16 months, 2 years, and 3 years. The occurrence of autoantibodies to B23 prior to development of moderate-to-severe pulmonary hypertension suggests that these antibodies may serve as useful predictors for the subsequent development of pulmonary arterial hypertension in limited scleroderma.


We used an ELISA with recombinant human B23 to establish whether this abundant nucleolar phosphoprotein is an autoantigen in scleroderma. Because scleroderma is characterized by the frequent reactivity of patient sera with nucleolar components, the potential targeting of B23 in this disease was of particular interest. We found that ∼11% (10 of 92) of scleroderma sera contained antibodies to B23. Previously, in a screen of 164 sera containing ANA, obtained from patients with autoimmune rheumatic disease, Li et al demonstrated that none of the 44 sera from patients known to have scleroderma were reactive against B23 by immunoblotting (23). Due to the fact that conformational epitopes are frequently the target of patient autoantibodies (e.g., fibrillarin) (29), the denaturing method of immunoblotting may not always identify autoantibodies in patient sera. Interestingly, although none of the B23-positive sera in our study recognized B23 (either recombinant or in cell lysates) in a Western blotting assay, all of these sera recognized native B23 in an immunoprecipitation assay (Figure 2). Together with immunoblotting data from previous studies (21), these data suggest that the epitopes targeted by anti-B23 antibodies in other rheumatic diseases may differ from those targeted in scleroderma, where the antibodies are likely directed against a conformational epitope. Several previous studies have demonstrated that the phosphorylation status of autoantigens may influence their recognition by autoantibodies (30, 31). It is not presently known whether the phosphorylation status of B23 affects its recognition by patient autoantibodies, though such studies would clearly be of significant interest.

The clinical and serologic profiles associated with B23 autoantibodies in scleroderma were quite distinct. Antibodies against B23 in scleroderma were associated with PAH and reduced FVC. Of particular note, in 3 patients with limited scleroderma and restrictive lung function, the presence of antibodies to B23 predated the detection by echocardiography of moderate-to-severe pulmonary hypertension. Antibodies to B23 may therefore provide a useful biomarker to predict the development of moderate-to-severe pulmonary hypertension in patients with scleroderma. It is not yet known whether anti-B23 antibodies might also be present in patients with PAH or pulmonary hypertension associated with other rheumatic diseases (e.g., SLE, mixed connective tissue disorder), or may precede the pulmonary hypertension phenotype in these diseases.

Studies suggest that more severe interstitial lung disease is a feature of diffuse scleroderma that is associated with antibodies against topoisomerase I (8, 32). In contrast, limited scleroderma is associated with ACA and a higher prevalence of isolated pulmonary hypertension (8). Our study suggests that patients with anti-B23 antibodies appear to have a unique clinical phenotype that is different from previously described associations. For example, the B23-positive patients who had moderate-to-severe pulmonary hypertension had limited skin disease and a reduced FVC typical of restrictive lung function secondary to interstitial lung disease. They were also negative for ACA.

Another notable feature of the B23-positive population of scleroderma patients is the strong association observed with antibodies to U1 RNP. Of note, all 4 of the B23-positive patients with anti-RNP antibodies had limited scleroderma with overlap features, including polyarthritis and histology evidence of myositis and biopsy-proven glomerulonephritis. It will be of interest to determine whether antibodies against B23 may also define a subpopulation of patients with a scleroderma overlap syndrome characterized by an increased prevalence of pulmonary hypertension and restrictive lung disease, and hence increased morbidity. An analysis of B23 autoantibody status in a cohort enriched with patients with an overlap syndrome will be informative.

Although antibodies against B23 do not appear to be specific for any one disease, it is possible that these antibodies serve as markers for a pathogenic process leading to a distinct phenotype within a given disease. For example, antibodies against B23 and other nucleolar autoantigens (e.g., fibrillarin and NOR-90) have previously been documented in hepatocellular carcinoma (HCC) and other cancers (13, 33), as well as in occasional patients with various autoimmune rheumatic diseases, including lupus (23) and, in this study, scleroderma. In one report, the presence of antinucleolar antibodies was shown to coincide with the progression of chronic liver disease to HCC (13), again indicating that these autoantibodies can potentially serve as predictors of an imminent disease phenotype. Of note, none of the patients included in our study had cancer.

It is remarkable that in the majority (80%) of the B23-positive scleroderma sera used in our study, antibodies against the nucleolar autoantigen fibrillarin were also present. Because B23 and fibrillarin are not similar at the level of amino acid sequence, and because there were numerous fibrillarin-positive sera that did not contain antibodies to B23, it is unlikely that this association is attributable to a cross-reactivity between these antigens. Similar to the B23 autoantibodies, antifibrillarin autoantibodies have been shown to be associated with restrictive lung disease and in a subset of patients enriched with pulmonary hypertension (9, 10, 28). The similar clinical associations between antibodies to B23 and fibrillarin suggest that there may be unique features of these nucleolar autoantigens that determine these shared phenotypic associations. Interestingly, both B23 and fibrillarin have been shown to co-immunoprecipitate in ribonucleoprotein complexes from nuclear extracts (34). These complexes were also present in extracts made from mitotic cells, in which the nucleolus is disassembled, implying that pools of B23 and fibrillarin may remain in close association during all phases of the cell cycle. Identifying potential structural modifications that B23 and fibrillarin may undergo during pulmonary damage may provide insights into the pathogenic events driving such pulmonary damage and the associated autoantibody response in scleroderma.