Programmed cell death 1 gene ( PDCD 1 ) polymorphism and pemphigus foliaceus ( fogo selvagem ) disease susceptibility

Pemphigus foliaceus, also known as fogo selvagem, is an autoimmune disease of the epidermis characterized by superficial blisters and antibodies against desmoglein 1. It is a multifactorial disease and genetic susceptibility is oligogenic or polygenic. Considering the crucial function of the programmed cell death 1 molecule (PD-1) in the immune response, the aim of this study was to verify if variants of the PDCD1 gene influence susceptibility and resistance to pemphigus foliaceus, in a case control disease association study. We analyzed patients (n = 154) and unaffected control individuals (n = 325) of the Brazilian population, in respect to the PD1.3(G,A) PD1.5(C,T) and PD1.6(A,G) single nucleotide polymorphisms (SNPs) and also investigated, for the first time, the exon 5 PDCD1 microsatellite (CTG)n. The patient and control samples were divided into strata, according to the predominant ancestry of the individuals (African or European). The PD1.5 genotype distribution in the patients sample was almost indistinguishable from that in the control sample, in both population strata. A possible negative association between pemphigus foliaceus and allele PD1.3A was observed in the total African and European ancestry population sample (odds ratio (OR) = 0.55, p = 0.066) and should be investigated in forthcoming studies. The PD1.6A allele was over-represented among the patients of predominantly European ancestry due to an increase of both the G/A and the A/A genotypes (OR = 2.12 and 1.74, respectively; p = 0.035). We conclude that polymorphisms of the PDCD1 gene may influence susceptibility to pemphigus foliaceus, at least in Brazilians of predominantly European ancestry.


Introduction
Pemphigus foliaceus is an epidermal autoimmune disease, characterized by autoantibodies against a cellular adhesion molecule, desmoglein 1, expressed in keratinocytes (Anhalt et al., 1990).This complex multifactorial disease is endemic in Brazil, with inter-individual differences in susceptibility being due to multiple factors, which include variants of several genes.According to the polygenic model of disease susceptibility, the presence of any of the susceptibility alleles is neither sufficient nor required for the disease to occur.The strongest associations thus far described are those with HLA class II alleles, known since 1989 and well established (Petzl-Erler and Santamaria, 1989;Moraes et al., 1991;Pavoni et al., 2003;VMMS Roxo, personal communication).Other candidate genes have also been analyzed in pemphigus foliaceus association studies, including genes of several cytokines and cytokine receptors (IL1A, IL1B, IL1R, IL4, IL4R, IL6, IL10, TNF, LTA) (Pereira et al., 2004;Roxo et al., 2003), TP53 and BAX, whose products participate in apoptosis (Köhler and Petzl-Erler, 2006) and DSG1, which encodes desmoglein 1, the major autoantigen in pemphigus foliaceus (Petzl-Erler and Malheiros, 2005).
The programmed cell death 1 (PD-1) molecule belongs to the CD28 family of receptors and is expressed by activated T, B and myeloid cells.PD-1 is postulated to play an important role in induction and/or maintenance of peripheral tolerance.Upon interaction with its ligands (PD-L1 and PD-L2), PD-1 inhibits antigen stimulation of T and B cells (Sharpe and Freeman, 2002).The PD-Ls belong to the B7 family and may be expressed on both lymphoid and non-lymphoid organs.The balance of stimulatory and inhibitory signals provided by cell-surface interactions between lymphocytes and antigen presenting cells or target cells, provided by co-stimulatory and inhibitory molecules, is crucial for maintaining peripheral tolerance, while retaining a protective effect against foreign antigens.Disruption of this delicate balance could lead to breakdown of selftolerance and might be involved in the pathogenesis of autoimmune diseases (Okazaki and Honjo, 2006).
The function of PD-1 as a regulator of peripheral tolerance was further supported by the observation that PD-1 deficient mice developed autoimmune diseases, despite having distinct phenotypes on different genetic backgrounds (Nishimura et al.,1998;Nishimura et al., 2001).Indeed, PD-1 -/-mice on the C57BL/6 background show spontaneous development of a lupus-like syndrome, including glomerulonephritis and destructive arthritis (Nishimura et al., 1999).PD-1 -/-mice on a BALB/c background show development of dilated cardiomyopathy (Nishimura et al., 2001), which appears to be mediated by autoantibodies to a cardiac protein and deposition of IgG on the surface of cardiomyocytes (Okazaki et al., 2002).These findings support the notion that PD-1 might suppress immune activation and prevent autoimmune diseases (Okazaki and Honjo, 2006).
The human gene for PD-1, PDCD1, is localized on chromosome 2q37.Various single nucleotide polymorphisms (SNPs) in the PDCD1 gene have been identified (Prokunina et al., 2002) and some associations have been reported with autoimmune disorders such as systemic lupus erythematosus (SLE), rheumatoid arthritis and type I diabetes (Prokunina et al., 2002;Nielsen et al., 2003;Prokunina et al., 2004).An intronic SNP, PD1.3 (A,T) has been associated with SLE, with an analysis of familial and sporadic cases of SLE having found consistent over-representation of the A allele among American SLE patients of European ancestry and Mexican and Swedish SLE patients.The authors suggested that the associated A allele could interfere with the binding site of the RUNX1 transcription factor and thus affect expression of the PDCD1 gene (Prokunina et al., 2002).In addition, allele T of another SNP, named PD1.5 (C,T), was observed to be associated with the development of rheumatoid arthritis but not SLE in Chinese patients living in Taiwan (Lin et al., 2004).
To test the hypothesis that polymorphisms of the PDCD1 gene influence susceptibility to pemphigus foliaceus, we performed a case-control association study.We analyzed three SNPs of the PDCD1 gene: PD1.3(G,A) in nucleotide position 7146 of intron 4; PD1.5(C,T) in nucleotide 7785 of exon 5; and PD1.6(A,G) in nucleotide 8737 of the 3' untranslated region (UTR).We also investigated the exon 5 (CTG) n microsatellite polymorphism.

Population sample
The sample consisted of 154 patients (76 women and 78 men, age range 6 to 88 years) with pemphigus foliaceus and 325 controls (176 women and 149 men, age range 12 to 86 years) without any history of pemphigus foliaceus.Blood samples were collected from patients at Hospital Adventista do Pênfigo, Campo Grande municipality, Mato Grosso do Sul (MS) state, Brazil.Diagnosis of pemphigus foliaceus was based on clinical criteria, histopathology and indirect immunofluorescence.Blood samples were also collected from unaffected control individuals in Campo Grande-MS and Curitiba (Paraná state, Brazil).The control sample was similar to the patient sample in respect to age, birthplace, residence, socio-economic status, sex ratio and occupation.The total population sample (patients plus controls) was of predominantly European and African ancestry and extensively admixed, but not panmictic.Therefore, we considered that the different population strata might differ regarding allelic frequencies.In view of the interest of association analysis in the sub-populations and the need of matching the patient and the control samples for ancestry, individuals were classified according to morphological characteristics, as a proxy to an estimate of the relative contribution of European and African ancestry to each individual's genome.On the basis of facial morphological features, hair type and skin color, individuals were grouped as of mixed African and European ancestry (AF: patients = 46.5%,controls = 53.5%)or of predominantly European ancestry (EU: patients = 26.1%,controls = 73.9%).
This approach to subdividing the population into strata according to ancestry and degree of admixture is validated by population genetics studies (Braun-Prado et al., 2000;Probst et al., 2000).They have shown that the distribution of alleles known to be restricted to populations from one continent follows a gradient among the strata of the Southern Brazilian population classified as in the present study, as expected if the classification discriminated among the population strata according to the relative contributions of the ancestral populations.
To verify the population frequencies of the PDCD1 (CTG) n microsatellite, we analyzed a sample of 95 unaffected individuals of four population strata according to ancestry: 40 individuals of predominantly European ancestry (EU), 20 of Eastern Asian ancestry (EA), 20 of Amerindian ancestry (AM) and 15 of predominantly African ancestry (AF).The Amerindians are Guarani and the samples were collected in Rio das Cobras, Laranjeiras do Sul (Paraná state, Brazil) (Petzl-Erler et al.,.1993); the samples of the other groups were obtained in Curitiba (Paraná state, Brazil).
Individuals were made aware of the study aims and signed an informed consent document prior to participating in the project.The study has been approved by the Human Subjects Research Ethics Committee of the Federal University of Paraná, according to Brazilian Federal laws.

Genotyping
The three PDCD1 SNPs have been previously described by Prokunina et al. (2002) and we have used the nomenclature suggested by these authors.The intron 4 nucleotide 7146 PD1.3(G,A) SNP was genotyped as described by Prokunina et al. (2002).To genotype the exon 5 nucleotide 7785 PD1.5(C,T) SNP and the 3' UTR nucleotide 8737 PD1.6(A,G) SNP we performed the polymerase chain reaction followed by hybridization of sequencespecific oligonucleotide probes (PCR-SSOP) with the denatured PCR products immobilized on a nylon membrane (dot blot).Primers and probes were designed based on the sequence of the human PDCD1 gene (GenBank accession AF363458) (Table 1).For PCR, a 20 μL reaction mixture was set up to contain 0.06 μg of genomic DNA, 1X PCR buffer, 6 pmoles of each primer, 0.2 mM of each dNTP, 2.5 mM MgCl 2 , and 0.5 units of Taq DNA polymerase (Invitrogen), and the DNA segment was amplified for 35 cycles, after initial denaturation at 95 °C for 5 min.Each cycle consisted of three steps: denaturation for 15 s at 95 °C, annealing for 30 s at 67 °C for PD1.5 and 68 °C for PD1.6, and primer extension for 15 s at 72 °C.The PCR products (568 bp for PD1.5 and 727 bp for PD1.6) were dot blotted, hybridized with digoxigenin-labeled probes, washed and the hybridized probes were detected using chemiluminescence according to previously described procedures (Bignon and Fernandez-Viña, 1997).Genotyping results for PD1.5 were confirmed by the PCR-restriction fragment length polymorphism (PCR-RFLP) method using the PvuII restriction enzyme (Fermentas) and the genotype of each individual was detected by fractionating the digest by size in a 4% agarose gel.The size of the resulting fragments were 491 bp and 77 bp for allele PD1.5T, with only the larger fragment being visualized in the gel.Microsatellite typing was performed for 95 individuals (15 AF, 20 AM, 40 EU and 20 EA).The PCR was carried out with 6 pmol of each the forward fluorescent (FAM-6) primer 5'ctacagggag ggccagatgca 3' and the reverse primer 5' tgctccaaggccatctc caac 3' in a total volume of 20 μL, containing 30 ng of genomic DNA, 1X PCR buffer, 0.2 mM of each dNTP, 2.5 mM MgCl 2 , and 0.5 units of Taq DNA polymerase (Invitrogen).The PCR conditions were initial denaturation at 95 °C for 5 min, followed by 35 cycles of denaturation at 95 °C for 15 s, annealing at 60 °C for 30 s, and extension at 72 °C for 15 s.The PCR products were denatured for 3 min at 94 °C and separated by size on a MegaBACE 1000 capil-lary array electrophoresis system (Amersham/GE Healthcare).Electropherograms were generated using the Fragment Profiler software 1.2 (Amersham/GE Healthcare).

Statistical analysis
Genotypic frequencies, allelic frequencies, and allele carrier frequencies (i.e., the frequency of individuals having the allele in either homozygosity or heterozygosity) were estimated by direct counting.Maximum likelihood estimates of haplotype frequencies were obtained using an expectation-maximization (EM) algorithm included in the program package ARLEQUIN 3.0 (Excoffier, Laval and Schneider, 2005).Lewontin's D' (relative linkage disequilibrium coefficient) between the SNPs was calculated using the ARLEQUIN 3.0 package.Hardy-Weinberg equilibrium was assessed using a modified version of the Markov-chain random walk algorithm, described by Guo and Thompson (1992) and included in the ARLEQUIN 3.0 program package.
For disease association analysis, genotypic frequencies, allelic frequencies and allele carrier frequencies were compared between the patient and control groups in the AF and EU population strata, by analysis of contingency tables using an exact test to estimate the probability (p) value via the metropolis algorithm and the RxC computer program (Miller, 1997).Using the same test, allelic frequencies were compared between the AF and EU population strata.If the two strata did not differ significantly, association analysis was also performed for the total population sample.The p value of 0.05 was adopted as the significance limit in all comparisons.Whenever the p value was lower than 0.1, odds ratios (OR) with 95% confidence intervals (CI) were calculated by Woolf's method.

Results
This is the first work investigating the PDCD1 (CTG) n microsatellite.To describe the allele frequencies and to evaluate the utility of this microsatellite as a marker 316 Pemphigus foliaceus and PDCD1 genetic variants for the disease association study, we analyzed a sample of the Brazilian population, stratified according to ancestry.The most frequent allele in all four ethnic groups was 197 (Table 2).The degree of polymorphism was low in the Amerindian, African and European strata but in the East Asian group only the 197 allele occurred.Because of the low degree of polymorphism, this marker was not used for the disease association analysis.
Comparison of the allele frequencies between the AF and EU population strata revealed significant differences (p < 0.05) for the PD1.6(G,A)SNP and homogeneity (p > 0.05) for the PD1.3(G,A) and PD1.5(C,T) SNPs.Therefore, association analyses were performed also for the Braun-Prado and Petzl-Erler 317 total population sample (AF plus EU) for PD1.3 and PD1.5 (Table 3) but not for PD1.6 (Table 4).
Comparisons of the genotypic, allelic, and allele carrier frequencies between the patient and the control groups did not reveal any statistically significant difference for the PD1.3 and PD1.5 SNPs (Table 3).In fact, for both the AF and the EU population strata, the PD1.5 frequencies in the patient group were very similar to those observed in the control group.The decreased frequency of allele PD1.3A observed among patients of both population strata approached the significance limit in the total population sample (OR = 0.55, p = 0.066) (Table 3).
Statistical analysis of the PD1.6 SNP revealed significant differences in the genotypic and allelic frequencies between the patient and control groups in the EU but not the AF population stratum (Table 4).The PD1.6A allele was over-represented among the EU patients, due to an increase of both the G/A and the A/A genotypes (OR = 2.12 and 1.74 respectively; p = 0.035).These results indicate that the PD1.6 SNP is associated with susceptibility to pemphigus foliaceus and the disease risk is increased by the presence of the PD1.6A allele in the population stratum of predominantly European ancestry.
Association analyses considering the haplotypes of the three polymorphic positions (PD1.3,PD1.5 and PD1.6) showed a positive association with the GCA haplotype in the EU stratum (OR = 3.32, p = 0.004) but in the AF stratum a negative association with the GTA haplotype with the p value close to the significance limit was observed (OR = 0.39, p = 0.048) (Table 5).The three SNPs were in linkage disequilibrium (Table 6).

Discussion
Co-inhibitory molecules of the CD28 family are increasingly being recognized as important players in the control of T-cell immunity (Chambers, 2001).Inhibitory signals produced by the PD-1/PD-L pathway are of crucial relevance to the mechanisms maintaining tolerance and to terminate immune response in peripheral tissues, thus limiting organ damage (Freeman et al., 2000;Latchman et al., 2001;Nishimura and Honjo, 2001).The causes of initiation and perpetuation of the auto-aggressive responses in pemphigus foliaceus patients are largely unknown, although there is consensus about the influence of both genetic and environmental factors on susceptibility to this condition (Eaton et al., 1998;Petzl-Erler and Santamaria, 1989).The pathogenic concepts include defects in the maintenance of peripheral immune tolerance towards potential autoantigens (Coyle and Gutierrez-Ramos, 2001).
Our study is the first to consider PDCD1 a candidate gene for differential susceptibility to pemphigus foliaceus and also the first to analyze the exon 5 microsatellite in a human population.The low degree of polymorphism of this microsatellite may reflect functional constraints resulting from involvement of this region of the PDCD1 gene in the control of gene expression.
The mechanisms that account for the observed positive association between pemphigus foliaceus and the PD1.6A allele remain to be clarified.Although the relative contribution of the PD1.6A allele to genetic predisposition to pemphigus foliaceus appears to be small, it is possible that this allele could lead some individuals to cross the threshold of disease manifestation, if these individuals bear susceptibility alleles (polygenes) at other loci and are exposed to the putative non-genetic susceptibility factors.The PD1.6(G,A) SNP is located in the 3' untranslated region of the PDCD1 transcript and could potentially alter the expression of the PDCD1 gene.A lowered expression of the PD-1 receptor at the cell membrane due to the PD1.6A allele might result in faulty control of peripheral tolerance.A functional effect and a direct involvement of this SNP in susceptibility to pemphigus foliaceus should be addressed in future studies.An alternative interpretation is supported by the absence of association in the population stratum of predominantly African ancestry.The association might be secondary, resulting from linkage disequilibrium of the PD1.6 SNP with other, as yet unanalyzed, polymorphisms of PDCD1 or other, closely linked, genes.
In the proximity of the PDCD1 gene is a cluster of homologous genes (CTLA4, CD28 and ICOS) related to immune activation and considered promising candidate genes for susceptibility to autoimmune disease (Holopainen et al., 2004).A more comprehensive analysis of the 2q33-2q37 genomic region might reveal whether variants of one or more of these genes are involved in susceptibility and/or resistance to pemphigus foliaceus.
According to our results, the polymorphic variants of PD1.5 do not seem to be involved in the pathogenesis of pemphigus foliaceus and do not alter the risk of developing the disease in the population studied.The lack of association leads us to suggest that this single nucleotide polymorphism does not influence susceptibility to pemphigus foliaceus.
We observed a possible negative association between allele PD1.3A and pemphigus foliaceus.It has been suggested that this SNP is functional, altering a binding site for the transcription factor RUNX1, and positive associations of allele PD1.3A with systemic lupus erythematosus and other autoimmune diseases have been reported.However, results of the autoimmune disease association studies are contradictory (Nielsen et al., 2003;Prokunina et al., 2002Prokunina et al., , 2004;;Ferreiros-Vidal et al., 2004, among others).It should be noticed that the PD1.3A allele is not common.Highest frequencies (5%-16%) have been reported for European populations; in other populations the PD1.3A allele is uncommon (Mori et al., 2005;Kong et al., 2005, Prokunina et al., 2002;Sanghera et al., 2004).The frequency of the PD1.3A allele was lower in our AF than in our EU population sample, in agreement with low frequency or absence of the PD1.3A allele in African populations and the known gene flow from Brazilians of European ancestry in the Brazilian population of African ancestry (Braun-Prado et al., 2000;Probst et al., 2000).The difference of allele frequencies between populations might partly account for the conflicting results of disease association studies.
Braun-Prado and Petzl-Erler 319 In conclusion, we have shown for the first time that the presence of the PD1.6A allele at nucleotide 8737 of the PDCD1 gene is associated with inter-individual differences in susceptibility to pemphigus foliaceus in the Brazilian population.

Table 1 -
PDCD1 single nucleotide polymorphisms (SNPs) and typing methods used.The variants of the SNPs detected by the probes are highlighted in boldfaced type in the last column.Probes SND3CT and SND3GA anneal to monomorphic positions and were used as controls for the quality of the PCR and the dot blots.

Table 2 -
Allelic frequencies (%) of the PDCD1 (CTG) n microsatellite in four Brazilian population strata of different ancestry.

Table 3 -
Genotype, allele and allele carrier frequencies (%) and association analysis between pemphigus foliaceus and the PD1.3 7146(G,A) and PD1.5 7785(C,T) single nucleotide polymorphisms (SNPs) in samples of patients and unaffected control subjects of the Brazilian population.

Table 4 -
Genotype, allele and allele carrier frequencies (%) and association analysis between pemphigus foliaceus and the PD1.6 8737(G,A) single nucleotide polymorphism (SNP) in samples of patients and unaffected control subjects of the Brazilian population.

Table 5 -
Haplotype frequencies (%) and association analysis between pemphigus foliaceus and PDCD1 haplotypes, as defined by the PD1.3, PD1.5, and PD1.6 single nucleotide polymorphisms (SNPs), in samples of patients and unaffected control subjects of the Brazilian population.

Table 6 -
Lewontin's D' relative linkage disequilibrium (LD) coefficient estimation for pairs of the three single nucleotide polymorphisms of the PDCD1 gene in samples of patients with pemphigus foliaceus and unaffected control subjects of the Brazilian population.