|Home | About | Journals | Submit | Contact Us | Français|
Fogo selvagem (FS), the endemic form of pemphigus foliaceus (PF), is characterized by pathogenic anti-desmoglein 1 (Dsg1) autoantibodies. To study the etiology of FS, hybridomas that secrete either IgM or IgG (predominantly IgG1 subclass) autoantibodies were generated from the B cells of eight FS patients and of one individual four years prior to FS onset, and the H and L chain V genes of anti-Dsg1 autoantibodies analyzed. Multiple lines of evidence suggest that these anti-Dsg1 autoantibodies in FS are antigen selected. First, clonally related sets of anti-Dsg1 hybridomas characterize the response in individual FS patients. Second, H and L chain V gene use appears to be biased, particularly among IgG hybridomas, and third, most hybridomas are mutants and exhibit a bias in favor of CDR amino acid replacement (R) mutations. Strikingly, pre-FS hybridomas also exhibit evidence of antigen selection, including an overlap in VH gene use and shared multiple R mutations with anti-Dsg1 FS hybridomas, suggesting selection by the same or a similar antigen. We conclude that the anti-Dsg1 response in FS is antigen driven, and that selection for mutant anti-Dsg1 B cells begins well before the onset of disease.
Pemphigus encompasses a group of autoimmune blistering diseases exhibiting pathogenic autoantibodies against desmogleins (Dsg), a family of desmosomal cell adhesion glycoproteins (Beutner and Jordon, 1964; Ding et al., 1997; Lever, 1953; Udey and Stanley, 1999). The hallmark of these diseases is the presence of intraepidermal vesicles (Lever, 1953) and anti-epidermal autoantibodies (Beutner and Jordon, 1964; Ding et al., 1997; Udey and Stanley, 1999). Pemphigus foliaceus (PF) and pemphigus vulgaris (PV) are the two major phenotypes of pemphigus (Lever, 1953). Immunologically, the sera of PF patients show anti-Dsg1 antibodies, while the sera of PV patients contain antibodies to Dsg3 (mPV) or both Dsg1 and Dsg3 (mcPV) (Ding et al., 1997; Udey and Stanley, 1999). PV and PF in North America are sporadic (Lever, 1953; Udey and Stanley, 1999), but endemic PF is described in certain states of Brazil, where it is known as Fogo Selvagem (FS) (Diaz et al., 1989b). FS shows similar clinical, histological and immunological features to those observed in non-endemic PF (Diaz et al., 1989a; Stanley et al., 1986). The published epidemiological studies of FS strongly suggest that this disease is precipitated by an environmental agent(s) harbored in certain regions of Brazil. One of these sites, under investigation by our group for the last 15 years, is the Amerindian Reservation of Limao Verde (Hans-Filho et al., 1996). We have reported the serological transition from preclinical to clinical stage of FS in several cases from Limao Verde (Li et al., 2003; Qaqish et al., 2009; Warren et al., 2003; Warren et al., 2000).
FS is mediated by pathogenic autoantibodies against Dsg1 (Roscoe et al., 1985; Stanley et al., 1986). These pathogenic autoantibodies are IgG4 restricted (Rock et al., 1989) and their appearance in the serum heralds the onset of clinical disease (Warren et al., 2003). In fact, a recent study by our group has identified IgG4 anti-Dsg1 autoantibodies as the serological marker of disease in FS (Qaqish et al., 2009). Non-pathogenic IgG anti-Dsg1 autoantibodies (Li et al., 2003; Warren et al., 2000) as well as IgM anti-Dsg1 autoantibodies (Diaz et al., 2008) have been detected in healthy individuals living in endemic areas of FS. The underlying mechanism of autoantibody formation in FS however, specifically anti-Dsg1 autoantibodies, is still poorly understood. Whether these autoantibodies developed through a polyclonal activation or an antigen-driven mechanism is still a mystery, but this information is needed to identify the cause of FS. To provide a definite answer to this question requires the genetic analysis of anti-Dsg1 autoantibody gene repertoire.
Analyses of the V gene sequences encoding the autoantibodies in PV by our group and by Payne et al. (Payne et al., 2005; Qian et al., 2007) have yielded important clues to the development of anti-Dsg1 and anti-Dsg3 antibodies. The potentially pathogenic IgG anti-Dsg response in PV is has been shown to be antigen selected (Qian et al., 2007). The results of these studies are similar to findings reported in other autoimmune diseases, such as systemic lupus erythematosus (SLE) and rheumatoid arthritis (RA). Recently, two single-chain variable fragments of pathogenic antoantibodies from a pemphigus foliaceus patient were isolated and the H chain V regions of the autoantibodies from this patient were shown to be encoded by restricted number of genes (Ishii et al., 2008).
Although there has been no genetic study on the autoantibodies in individuals before the onset of autoimmune diseases, the high prevalence of FS in Limao Verde provides us with a unique opportunity to address this question since we have preserved peripheral blood mononuclear cells (PBMCs) of selected FS and healthy individuals from this human settlement. In this study, we have examined anti-Dsg B cells from multiple FS patients and from an individual who was healthy at the time of blood draw, but developed FS four years later. We conclude that a) anti-Dsg response in FS is composed of mutant B cells that have been subject to extensive antigen selection, and b) the pre-clinical anti-Dsg response is composed of mutant B cells that have been undergone the same or similar selection pressures. These findings indicate the presence of inciting antigen(s) in FS endemic areas and the inciting antigen(s) play an important role in the etiology of FS.
We fused EBV-transformed PBMCs from eight FS patients with myeloma cells and screened for the resulting hybridomas for anti-Dsg1 production by ELISA. We generated seventy-seven anti-Dsg1 hybridomas, 40 IgM producers and 38 IgG producers (Table 1). Most IgG anti-Dsg1 hybridomas were IgG1. Despite the fact that the IgG4 autoantibodies are the main pathogenic antibodies in the sera of FS patients (Rock et al., 1989), only two IgG4 hybridomas were identified from two FS patients. There are possible explanations for the disparity between the high levels of serum IgG4 from FS patients and the low frequency of IgG4 hybridomas generated in this study. Firstly, serum IgG4 antibodies were produced by plasma cells. Plasma cells mainly reside in the tissues and bone marrow, and are rarely in peripheral blood (Benner et al., 1981; Slifka and Ahmed, 1996). Thus, they were not abundantly present in the FS patients blood samples collected. Secondly, EBV transformation predominantly immortalizes B cells (Middleton et al., 1991), and the IgG4 secreting plasma cells might not be immortalized and thus might be absent from the subsequent hybridoma generation. Thirdly, this may simply reflect the relative amount of IgG1 and IgG4 expressing B cells in the peripheral blood of FS patients. The Dsg1 specificity of the antibodies produced by these hybridomas was confirmed by immunoprecipitation (Fig 1a) and indirect immunoflorescence (Fig 1b). Most of these hybridomas (35 of 38) also recognized the ectodomain of Dsg3 by ELISA (data not shown).
The mRNA of the expressed VH and VL genes of these 77 anti-Dsg1 hybridomas was PCR amplified and sequenced to determine VH and VL gene use and to identify somatic mutations. Table 1 summarizes this analysis. Clonally related sets of hybridomas from an individual fusion are a hallmark of secondary responses to foreign antigen and are also a characteristic of autoimmune responses, including in human PV, as we have shown previously (Qian et al., 2007). This oligoclonality is due to antigen-driven clonal expansion of a limited number of B cell clones. Clonally related hybridomas will express the identical V region genes and have identical VH CDR3 sequences. By these criteria, we identified twelve sets of clonally related hybridomas from seven of the eight FS patients. In patient FS8, all seven of the anti-Dsg1 hybridomas identified belonged to a single clonal set, while those from patients FS12 and FGS belonged to only two clonal sets each, indicating dominance of a small number of clones in some patients. Most clonally related hybridomas were identical in sequence suggesting that a limited number of clones in each patient had undergone extensive clonal growth in the absence of somatic mutation. However, sequence identity raises the possibility that the presence of clonal sets is an artifact of in vitro growth. To discriminate between these possibilities, we compared the extent of somatic mutation among the clonal sets before divergence with that of singlet hybridomas. Based on sequence comparison to the most similar germline VH gene most (9 of 11) clonally related sets had at least 10 differences from germline, indicating the occurrence of extensive somatic mutation before clonal divergence. This was significantly different from that of singlet hybridomas, in which only half had 10 or more mutations (16 of 35) (χ2, p=0.036). This argues against an in vitro artifact to explain the presence of clonal sets in FS patients, since any in vitro expansion will be independent of the extent of somatic mutation. This is further suggested by intraclonal sequence difference in the clonal set from patient FS12 (FS12-1F10 and FS12-3A7). Thus, the presence of clonal sets of hybridomas in these FS patients likely reflects in vivo clonal expansion, presumably because of the selective advantage in antigen binding conferred by the somatic mutations they acquired before clonal divergence.
We identified eighteen VH genes used by the 48 clonally independent anti-Dsg1 hybridomas (Table 1 and Fig. 2A upper panels) and found that IgM and IgG anti-Dsg1 hybridomas differed significantly in their expressed VH repertoires. VH3 gene family use increased from 43.5% among IgM hybridomas to 68.2% (χ2, p=0.095) among IgG hybridomas, and VH1 gene family use decreased from 34.8% to 9.1% (χ2, p=0.038). No VH gene dominated the IgM repertoire, but IGHV3-23 may be favored in the IgG repertoire, as it was used by 5 of 22 clonally independent IgG hybridomas, although this does not reach the level of significance in comparison to IgM hybridomas (2 of 23 clonally independent hybridomas:χ2, p=0.1942) (Fig. 2a upper panels), or in comparison to healthy controls (12/71 B cells; χ2, p=0.5367) (Brezinschek et al., 1995). JH use by IgM and IgG anti-Dsg1 included JH3, 4, 5, and 6, and was not significantly different between IgM and IgG hybridomas and not different from healthy control B cells (Brezinschek et al., 1995). In addition, we observed no restriction in VH CDR3 length, as it ranged from 6 to 21 amino acids (Table 1).
We identified eleven κ and six λ VL genes used by 29 sequenced clonally independent anti-Dsg1 hybridomas (12 IgM and 17 IgG) (Table 2 and Fig. 2b upper panels). As with VH gene use, VL gene use appears to be more restricted among IgG hybridomas than among IgM hybridomas. With one exception, each IgM hybridoma expressed a unique VL gene, whereas four VL genes were used by two or more clonally independent IgG hybridomas. IGKV1D-39 was expressed by four of 17 independent hybridomas (23.5%), suggesting that this VL gene provides a selective advantage in binding Dsg. This biased distribution of VL is similar to that observed in responses to foreign antigens and therefore suggests that the anti-Dsg1 B cells in FS patients undergo antigen selection.
The anti-Dsg1 hybridomas from FS patients were extensively mutated (Table 1). The number of VHDJH mutations in IgM hybridomas (166 mutations in 23 sequences; 7.2 mutations/gene) was half that in IgG hybridomas (324 mutations in 22 sequences; 14.7 mutations/gene) (χ2, p=0.021). Moreover, all 5 unmutated sequences were from IgM hybridomas (Table 1). We used a multinomial distribution model (Lossos et al., 2000) to determine bias in the frequency of amino acid replacement (R) and silent (S) mutations in FWRs and CDRs. We found that five of 23 IgM hybridomas showed a biased distribution of mutations in either FWR or CDR encoding regions and one (FS33-3H3) showed a bias distribution in both FWR and CDR encoding regions. In contrast, 15 of 22 independent IgG hybridomas showed a biased distribution in both FWR and CDR encoding regions, and most of the remaining hybridomas showed a biased distribution in either FWR or CDR encoding regions (Table 1). This difference between IgM and IgG hybridomas is statistically significant (χ2, p<0.001). We conclude that anti-Dsg B cells are antigen selected, and that mutant IgG B cells in particular have a selective advantage in contributing to this response in FS patients.
The expressed VL genes from these hybridomas were also extensively mutated, although as is typical, VL mutations were less frequent than VH mutations (195 mutations in 29 genes (6.7 mutations/gene) vs. 490 mutations in 45 genes (10.9 mutations/gene) (χ2, p=0.055) (Table 2). In common with the VH mutations in these hybridomas, the frequency of VL mutation in IgM hybridomas (5.2 mutations/gene) was lower than that in IgG (7.7 mutations/gene) (χ2, p=0.4434). However, few IgG VL genes exhibited a bias in distribution of CDR mutations (2 of 17 VL genes) (Table 2), suggesting that VL mutations are less likely to improve antigen binding than VH mutations.
Bias in somatic mutation was also evident by the large number of shared mutations among these hybridomas Fig 3. For example, four of seven IGHV3-23 expressing hybridomas acquired a mutation at position 31 in the CDR1 encoding region of this VH (Fig 3) with S31N occurring twice. In addition, CDR1 mutation S35N occurred twice and S35T occurred twice, and in CDR2, both A50S, and A50G occurred twice. There were also shared mutations located in FWRs, such as N77K, Y80F, and M83I. Parallel mutations were not limited to the more frequently used IGHV3-23 and IGHV3-30 genes, as they occurred in other VH genes (Fig. 3). Interestingly, S31N occurred in three other members of the VH1 family, and a member of the VH3 and VH5 families, further suggesting its importance to antigen binding.
We have demonstrated that healthy individuals from Limao Verde possess IgM and IgG anti-Dsg1 autoantibodies (Diaz et al., 2008; Warren et al., 2000), strongly suggesting the existence of an environmental factor in endemic areas that sensitizes and triggers anti-Dsg1 autoantibody formation in these individuals. To understand the development of anti-Dsg1 antibodies in FS, we investigated the anti-Dsg1 response before the onset of disease. We have collected serum samples and PBMC from selected individuals with or without FS living in endemic regions for the past twenty years, and one non-FS individual developed FS four years later (patient FS45). The PBMCs of this individual, kept in liquid nitrogen, were EBV transformed and fused with MSP-2S myeloma cells.
We identified 28 anti-Dsg1 hybridomas (17 IgM, 11 IgG1) by ELISA (Table 3) and confirmed their specificity by immunoprecipitation (data not shown). Two clonal sets of hybridomas were identified, the largest of which consisted of six IgM hybridomas. Fourteen VH genes encode anti-Dsg1 antibodies, but only IGHV3-23 and IGHV3-30 were used by both IgM and IgG hybridomas (Fig 2a lower panels). IGHV3-30, used by three independent FS IgM hybridomas, was the most frequently used VH gene by pre-FS IgG hybridomas (3 of 10 clonally independent pre-FS IgG hybridomas).
VL gene use by pre-FS hybridomas (Table 4) was also diverse, similar to that of FS hybridomas. We identified only two VL genes used by both pre-FS and FS hybridomas (Fig 2a lower panels and Fig 2b lower panels). The IgG hybridomas were very restricted in VL use since 5 of the 7 clonally independent hybridomas sequenced used IGKV2D-28 and IGKV4-1 (Fig. 2a lower panel). Interestingly, we did not observe V gene use among pre-FS hybridomas (0 of 17 vs. 10 of 29 among FS hybridomas; χ2, p=0.006) (Fig. 2b lower panel). Overall, although the pre-FS and FS anti-Dsg1 responses overlap in VH gene use, they exhibit notable differences in VL gene use.
As with the FS hybridomas, the overall VH and VL mutation frequencies differ (266 mutations in 20 VH genes; 59 mutations in 17 VL genes; χ2, p<0.001). All VH genes of pre-FS hybridomas were somatically mutated (Table 3), and IgM and IgG anti-Dsg1 hybridomas exhibited similar mutation rates (123 mutations in 10 genes for IgM; 143 mutations in 10 genes for IgG; χ2, p=0.7451). These rates were also similar to that of FS IgG hybridomas (χ2, p=0.6125 and p=0.9404, respectively). Multiple pre-FS hybridomas exhibited evidence of antigen selection based on biases in VH R and S mutations. However, only 3 of 12 clonally independent IgM hybridomas showed a significant bias, and then only in FWR encoding regions, whereas 6 of 10 clonally independent IgG hybridomas exhibited a significant bias in mutation in the regions encoding FWRs or CDRs or both. Interestingly, comparison of IGHV3-23 and IGHV3-30 sequences from pre-FS and FS hybridomas reveals multiple shared amino acid R mutations (Fig. 3). The IGHV3-23 mutations A23T, S31N, S35N, A50S, A50V, S57R, and N74S, as well as IGHV3-30 mutation S31N, occurred in both groups. In addition, the IGHV3-30 mutations S30R and A88P each occurred twice among pre-FS hybridomas. Thus, like FS hybridomas, the pre-FS anti-Dsg1 hybridomas exhibit evidence of selection for mutant B cells, particularly among anti-Dsg1 IgG B cells. Moreover, the shared VH mutations suggest that the same or a similar antigen is responsible for the selective pressure in the pre-clinical and clinical stages of the disease.
In this study, we report the genetics of anti-Dsg1 autoantibodies from eight FS patients and one individual four years before the clinical onset of FS. Our results show that the anti-Dsg1 response in FS patients living in endemic regions of the disease in Brazil is antigen selected and that selection begins well in advance of the onset of clinical disease.
The hypothesis of antigen selection of anti-Dsg1 B cells in FS patients is based on several lines of evidence. First, multiple groups of clonally related hybridomas were identified among hybridomas from each patient indicating that certain clones have a selective advantage in growth. Clonally related hybridoma sets are a characteristic of secondary responses to foreign antigen (Blier and Bothwell, 1987; Clarke et al., 1985; Scott et al., 1989) and have been observed among hybridomas derived from autoimmune patients (Qian et al., 2007). Second, a limited VH and VL gene repertoire may be used to encode these anti-Dsg1 antibodies in pre-FS and FS patients. Most VH genes belong to VH families 1, 3, and 4, similar to those in normal and SLE individuals (Brezinschek et al., 1997; de Wildt et al., 2000; Dorner et al., 1999). However, IGHV3-23 is common among IgG anti-Dsg1 hybridomas, increasing from 9% among IgM hybridomas to 24% among IgG hybridomas, suggesting selective pressure in favor of its use in this response. Selective VL gene use also occurs, since we find that IGKV1D-39 increases from 8.3% among IgM hybridomas to 23.5% among IgG hybridomas. Thus, V gene use by Dsg1 specific B cells may become increasingly restricted during the course of the response. Third, FS anti-Dsg1 hybridomas exhibit a bias favoring the accumulation of amino acid R mutations in CDRs and S mutations in FWRs in either VH or VL or both. This pattern is consistent with the selection for R mutations that provide an advantage in antigen binding (in CDRs) and against R mutations that could harm antigen binding or the structural integrity of the antibody molecule (FWRs). Consistent with this are the numerous parallel VH mutations among these hybridomas, many of which occurred in CDRs. Altogether, the anti-Dsg1 response of FS patients resembles antigen-selected responses to foreign antigen, arguing that the anti-Dsg response in FS is antigen selected. This parallels our findings with the anti-Dsg1 response in PV (Qian et al., 2007).
The endemic nature of FS, which allowed the freezing of PBMCs from individuals with a relatively high probability of developing FS, makes possible for the first time to examine the autoreactive B cell repertoire in humans before development of an autoimmune disease. Our analysis of one pre-FS individual indicates a remarkable similarity in pre-clinical and clinical anti-Dsg1 responses. Clonal sets of hybridomas were evident among pre-FS hybridoma panels indicating uneven clonal expansion. In addition, VH and VL use by pre-FS and FS hybridomas overlap, particularly with the use of IGHV3-23 and IGHV3-30 (Fig. 2a). There is also overlap in Vκ use by pre-FS and FS hybridomas (6 of 14 Vκ genes, Fig 2b). Although IGKV1D-39 is expressed, it is not dominant, as it is in the FS response (Fig. 2b). The most notable difference between pre-FS and FS hybridomas is the expression of Vλ. One-third (10 of 29) of FS hybridomas, but none of 17 pre-FS hybridomas expressed Vλ genes (Fig. 2b). Unfortunately, a PBMC sample from this patient after FS diagnosis is not available. However, analysis of anti-Dsg1 antibodies in sera from this patient indicates very low levels of λ anti-Dsg1 before compared to after FS diagnosis (data not shown). In contrast, there was no change in the levels of κ anti-Dsg1 before and after active FS. This change in λ anti-Dsg1 antibodies was observed in 4 of 11 individuals for which sera before and after FS diagnosis were available (data not shown). This raises the possibility that λ anti-Dsg antibodies are a clinical marker for FS that may predict the development of disease. We are currently testing this possibility.
The CDR replacement mutations are the most striking similarity between the pre-FS and FS hybridoma panels. Eighteen amino acid replacement mutations in CDR1 and CDR2 occurred in IGHV3-23 and IGHV3-30 of pre-FS and FS hybridomas. In some cases these mutations occurred multiple times within a panel. We interpret this to mean that many of the same mutations provide a selective advantage in the clonal expansion of anti-Dsg1 B cells, providing strong evidence that the same or a similar antigen is responsible for clonal selection before and after active FS. The differences between these responses, such as Vλ gene use, may be due to targeting of different epitopes before and after active FS. Nevertheless, these data indicate that selection of anti-Dsg1 B cells begins well before the onset of clinical disease. What distinguishes between those clones that produce pathogenic autoantibodies and those that are benign or even protective has yet to be elucidated, but the current data suggests that antigen selection of mutant IgG1 anti-Dsg1 B cells is not sufficient for pathogenicity, since the pre-FS individual analyzed here did not develop clinical disease for another four years.
The two IgG4 hybridomas overlap in VH and VL gene use with IgG1 hybridomas (IGHV3-48, IGHV2-5, and IGLV6-57), and like IgG1 hybridomas exhibit a bias in the distribution of mutations (Table 1). Moreover, the IgG4 hybridoma shares the CDR mutations Y32F and S53T with IgG1 hybridomas (Fig. 3). Further analysis of more IgG4 autoantibodies is required to conclusively determine whether potential pathogenic IgG4 B cells are subject to the same or similar selective pressures in vivo as those of IgG1 B cells. Nevertheless, this study provides a view of the anti-Dsg1 repertoire in FS patients indicating that the response to this self-antigen is antigen selected and begins well in advance of clinical disease. The source of the driving antigen could be environmental or self. We previously demonstrated that patients with parasitic diseases where insect bites are involved, such as onchocerciasis, leishmaniasis, and Chagas, often possess serum anti-Dsg1 autoantibodies (Diaz et al., 2004). Epidemiological studies of FS suggest that insect bites are a risk factor in FS (Aoki et al., 2004). It has been proposed that arthropod salivary antigen(s) induces the production of cross-reactive anti-Dsg1 antibodies. Alternatively, the inflammatory reaction to insect bites may expose Dsg1, and allowing an anti-Dsg1 response.
In conclusion, we have demonstrated that the development of autoantibodies in FS is antigen-driven, similar to other autoimmune diseases, such as SLE and PV, and that antigen selection of anti-Dsg1 B cells can begin years before the onset of active FS.
Heparinized peripheral blood (PB) samples were collected from 8 FS patients and an individual 4 years before the onset of clinical FS. All FS patients were living in Limao Verde, Brazil, except 3 that were hospitalized in the Penfigo Hospital, in Campo Grande, Brazil. Clinical and serological features of these patients have been reported previously, i.e. FS6 (patient #6), FS7 (patient #7), FS8 (patient #8) and FS 12 (patient #12) (Hans-Filho et al., 1996). FS33 is a patient from Limao Verde that developed FS at age 44 and was not included in the previous report. The patient donating blood 4 years before the onset of disease was FS45 that developed FS at age 22 and is the son of FS12 and brother of FS46. Patient JLDO was a 23 year old female with 5 months history of a generalized FS. Patient GCDS was a 13 year old male with a 1 year history of generalized FS. Finally, FGS was a 33 year old female with a 5 months history of generalized FS. Collection of clinic information and patients’ samples were approved by the Institutional Review Boards of the University of North Carolina and the University of Sao Paulo, Brazil.
Hybridomas were generated by fusion of Epstein-Barr virus (EBV) transformed PBMC with mouse myeloma cells (P3X63Ag8.653) (Kumpel, 2000) or MFP-2S myeloma cells(Kalantarov et al., 2002) as described previously (Qian et al., 2007). MFP-2S myeloma cells were kindly provided by Drs. Kalantarov and Trakht from Columbia University. Hybridomas secreting anti-Dsg1 antibodies were screened using a Dsg1-specific ELISA as described previously (Diaz et al., 2004; Li et al., 2003; Warren et al., 2003; Warren et al., 2000). The specificity of these autoantibodies was confirmed by immunoprecipitation (Li et al., 2003; Warren et al., 2000) and indirect immunofluorescence using monkey esophagus (IF) as described (Anhalt et al., 1982; Ding et al., 1997).
Messenger RNA isolation, PCR amplification and sequence analysis were conducted as described previously (Qian et al., 2007). The distribution of somatic mutations was analyzed according to the multinomial distribution model established by Lossos et al (Lossos et al., 2000) and the p-values were calculated using the JAVA applet at http://www-stat.stanford.edu/immunoglobin. A p-value of less than 0.05 is taken as evidence for antigen selection at framework region (FWR) and complementary determining region (CDR) (Lossos et al., 2000).
This work was supported in part by U.S. Public Health Service Grants R01-AR30281, RO1-AR32599, and T32 AR07369 awarded to L.A.D., R01-AI43587 to S.H.C, and a Dermatology Foundation Research Grant and an American Skin Association Alice P. Melly Research Grant to Y.Q..
Conflict of Interest:
The authors state no conflict financial interests.