High Basal Activity of the PTPN22 Gain-of-Function Variant Blunts Leukocyte Responsiveness Negatively Affecting IL-10 Production in ANCA Vasculitis

Consequences of expression of the protein tyrosine phosphatase nonreceptor 22 (PTPN22) gain-of-function variant were evaluated in leukocytes from patients with anti-neutrophil cytoplasmic autoantibody (ANCA) disease. The frequency of the gain-of-function allele within the Caucasian patient cohort was 22% (OR 1.45), compared to general American Caucasian population (16.5%, p = 0.03). Examination of the basal phosphatase activity of PTPN22 gain-of-function protein indicated persistently elevated activity in un-stimulated peripheral leukocytes, while basal activity was undetectable in leukocytes from patients without the gain-of-function variant. To examine consequences of persistently high PTPN22 activity, the activation status of ERK and p38 MAPK were analyzed. While moderate levels of activated ERK were observed in controls, it was undetectable in leukocytes expressing PTPN22 gain-of-function protein and instead p38MAPK was up-regulated. IL-10 transcription, reliant on the ERK pathway, was negatively affected. Over the course of disease, patients expressing variant PTPN22 did not show a spike in IL-10 transcription as they entered remission in contrast to controls, implying that environmentally triggered signals were blunted. Sustained activity of PTPN22, due to the gain-of-function mutation, acts as a dominant negative regulator of ERK activity leading to blunted cellular responsiveness to environmental stimuli and expression of protective cytokines.

The aim of the studies presented here is to investigate effects of the gain-of-function variant on signaling responses in leukocytes from patients with ANCA disease and how these influence immunological events. ANCA have two primary targets, PR3 and myeloperoxidase (MPO), which are expressed solely on the surface of neutrophils and monocytes. Because PTPN22 is uniquely expressed in hematopoietic cell types, studies of the gain-of-function polymorphism are fundamentally important in this disease [25]. Binding of ANCA to its antigens stimulates cellular signal transduction pathways causing changes in gene transcription, cell activation status, and ultimately, neutrophil degranulation [45][46][47][48][49]. It is the aberrant release of neutrophils' noxious constituents that causes inflammation of vessel walls and injury of highly vascularized organs such as the kidney and lung [50][51][52][53].
What makes a gain-of-function variant of PTPN22 particularly interesting, especially in the framework of multifactorial systemic autoimmune diseases like ANCA disease, is that protein tyrosine phosphatases (PTPs) serve as ''sensors'' and ''transmitters'' for environmental signals [54]. Alterations in genes encoding protein tyrosine phosphatases broadly affect kinase-phosphatase systems with deleterious effects on cellular equilibrium. PTPN22 is known to modulate the activity of the RAS and SRC-family signaling pathways, both of which are major pathways involved in immune modulation [55,56]. Due to the proximal position of the RAS and SRC-family in numerous signal transduction cascades, including extracellular signal-regulated kinase (ERK), JNK, and p38 MAPK [57], inappropriate regulation would impact immune cell functions, including those emanating from integrins, Fc receptors, growth factor receptors, and cytokine receptors [58][59][60][61].
Intuitively, a function-altering, genetic polymorphism in PTPN22 coupled with environmental exposures would place an individual at a higher risk for developing autoimmune disease. Environmental factors known to impact ANCA disease, at both disease onset and relapse, include bacterial and viral infections [62][63][64], aging [2,65], seasonal changes [66] and silica exposure [3]. We have evidence that one manifestation of these factors is perturbation of epigenetic regulation of gene transcription. We found that gene silencing marks were altered in leukocytes of patients with ANCA disease resulting in aberrant transcription at the gene locus for PR3 and MPO [4,67]. The gain-of-function variant could also be deviant in transmission of environmentallyinduced epigenetic signals. For example, the JmjC-domain containing histone demethylase, JMJD3, is ''induced'' when the cell ''senses'' bacterial products and inflammatory cytokines within the microenvironment [68,69]. We draw particular attention to JMJD3 because mRNA levels were abnormally high in ANCA disease patients concurrent with loss of epigenetic methylation marks at PRTN3 and MPO loci [4].
Here we describe a study demonstrating how a genetic polymorphism can disrupt ''sensors'' of the signaling milieu. The data indicate that the PTPN22 gain-of-function variant confers abnormally high basal phosphatase activity perturbing proper responses to external stimuli in circulating neutrophils and lymphocytes of patients with ANCA disease.

Patients and clinical analysis
Patients with biopsy-proven ANCA disease enrolled in this study were diagnosed between 1985 and 2009, and followed in a life-long registry by physicians in the Glomerular Disease Collaborative Network (GDCN). Methods of identifying and enrolling patients in the GDCN have been described [65,70,71]. All study materials were given Institutional Review Board approval for human subjects' research (IRB study #97-0523) by the UNC-CH Office of Human Research Ethics. Study subjects gave informed, written consent and participated according to UNC Institutional Review Board guidelines. A total of 230 Caucasian patients with ANCA disease participated in the PTPN22 genotyping study. Patients were categorized by diagnosis: granulomatosis with polyangiitis (GPA) [72][73][74], microscopic polyangiitis (MPA), Churg-Strauss syndrome (CSS), and renallimited disease (Lim) [75,76]. ANCA serotypes were determined by indirect immunofluorescence and/or antigen-specific PR3 and MPO enzyme-linked immune-absorbent assays (ELISA) (Invitrogen, Carlsbad, CA, USA) [77,78]. Of 230 Caucasian ANCA-

PTPN22 genotyping
Genomic DNA was extracted from leukocytes in EDTA-treated blood using the Puregene DNA Purification System (Puregene, Minneapolis, MN, USA). DNA quality was spectrophotometrically determined by OD 260/280 nm ratios and by agarose gel visualization. Genotyping for SNP C1858T (rs2476601) and G788A (rs33996649) was performed using TaqMan-SNP-Genotyping Assay (Applied Biosystems, Foster City, CA). The primer sequences for G788A were: forward 59 TTTGAACTAAT-GAAGGCCTCTGTGT 39 and reverse 59 ATTCCTGA-GAACTTCAGTGTTTTCAGT 39. The specific minor groove binder probe sequences were 59 TTGATCCGGGAAATG 39 (FAM) and 59 TTGATCCAGGAAATG 39 (VIC). The primer and the specific minor groove binder probe sequences for C1858T were commercially available and pre-designed by Applied Biosystems. TaqMan was performed by ABI PRISM 7900HT sequence detection system (Applied Biosystems).
Microtiter plate wells were coated in duplicate with mouse antihuman PTPN22 antibody (Abnova, Taipei, Taiwan) at a 1:100 dilution and incubated overnight at 4uC. Normal mouse IgG served as a mock control. Leukocytes were lysed in lysis buffer (20 mM Tris-HCl, 150 mM NaCl, and 1 mM EDTA, pH 7.4, with 1 mM of phenylmethanesulphonylfluoride, 10 mg/ml of aprotinin, 10 mg/ml of leupeptin, 10 mg/ml of soybean trypsin inhibitor) at a concentration of 5610 6 cells/ml. Lysate was added into each pre-coated well (100 ml) and incubated for 3 hrs at room temperature (RT). After washing with lysis buffer without protein inhibitor, 100 ml of phosphatase substrate (p-NPP, Bio-Rad, Hercules, CA, USA) (in 100 mM Bis-Tris, pH 6.0, 5 mM DTT buffer) was added to each well [80]. Phosphatase activity was detected by a VersaMax Microplate Reader (Molecular Devices, Sunnyvale, CA, USA) at 405 nm.
PTPN22 protein was quantitated by capture enzyme-linked immunosorbant assay. Wells were coated with mouse anti-human PTPN22 antibody, or normal mouse IgG as a mock control. An aliquot of fresh cell lysate was added and PTPN22 protein detected with rabbit anti-PTPN22 (1:200, Lifespan, Providence, RI, USA) and secondary antibody AKP-conjugated goat anti rabbit IgG (H+L) (1:5000, Pierce, Rockford, IL, USA). For PTPN22-responsiveness studies, total leukocytes were pre-treated with PMA (100 ng/ml) as described (Sigma) for 10 mins at 37uC.
Densitometric scanning analysis of the ratio of intensity pERK/ ERK and pp38/p38 was performed by ImageMaster VDS software. Analysis of microarray data RNA was isolated from circulating leukocytes of gain-offunction (n = 4) and non-variant (n = 12) patients with ANCA disease [67,79,81]. The Affymetrix microarray gene chip was used for identification of gene expression levels, as previously described [67,79,81]. The data were then imported into the Partek Genomics Suite 6.4 program (Partek, St Louis, CA, USA) for an ANOVA statistical analysis and differentially expressed genes ($2.0-fold, p-value ,0.05) within the gain-of-function group were compared to non-variant group. The molecular network analysis was performed using Ingenuity Pathways Analysis (IPA) (Ingenuity Systerms, Redwood City, CA, USA) and the expression profile of genes from gain-of-function and non-variant groups was visualized using a principal components analysis (PCA) mapped scatter plot in Partek program.
Taqman PCR analyses for IL-10 gene expression IL-10 primers and probes were purchased from Applied Biosystems. Fluorescence emission was monitored using the ABI PRISM 7900 HT sequence detection system. Relative level of total leukocyte RNA was determined by standard 2 (2DDCt) calculations and expressed as fold change of reference control samples. Cytochrome c oxidase subunit 5B (COX5B) was used as a RNA loading standard [67,81,82].

Statistical analysis
Differences in genotyping tests between ANCA patients and controls were analyzed by chi-square test. The direction and strength of these differences were assessed by calculating odds ratios. All of the alleles detected in our study were tested for the Hardy-Weinberg equilibrium. Clinical comparisons between patients with and without C1858T SNP for categorical measures were performed using chi-square tests. Continuous measures were compared using Wilcoxon rank sum test. A corrected p-value of ,0.05 was considered significant. Wilcoxon Two-Sample test were used for comparisons of continuous measures and paired data were analyzed by the Signed Rank Test. A corrected p-value of ,0.05 was considered significant. All statistical analyses were performed using SAS statistical program (SAS Institute, Inc., Cary, NC, USA).

Identification of patients with the risk-associated allele of PTPN22
The gain-of-function allelic variant has a SNP changing a cytosine to a thymine (C1858T) which converts the codon from one coding for arginine (R) to one for tryptophan (W) (R620W), and this amino acid change confers a gain-of-function phenotype. A total of 230 Caucasian patients were genotyped for the riskassociated allele of PTPN22 (C1858T) using a TaqMan-SNP- Genotyping Assay. Different from cohorts studied in previous reports, the patient cohort studied here included both PR3-and MPO-ANCA groups. There was a significant association of the C1858T SNP allele in patients of Caucasian descent (22.2%) compared to the general American Caucasian population frequency (16.5%, p = 0.03) [32], with an odds ratio (OR) of 1.45 (95% confidence interval 1.02-2.04) ( Table 1). The frequency was significantly higher in patients with a PR3-ANCA serotype (24.3%, p = 0.03, OR 1.63, 95% confidence interval 1.02-2.60) compared to American Caucasian population, but not in those with a MPO-ANCA serotype (20.2%, p = 0.32) ( Table 1). Carriage of the variant allele had no influence on a diagnosis of GPA, (23.6%, p = 0.11) or with MPA (21.6%, p = 0.16) compared to the general American Caucasian population.
For completeness, genotypic analysis was performed to determine the frequency of the loss-of-function variant in PTPN22 (G788A, rs33996649) in the patient cohort. This polymorphism results in an amino acid change in residue 263 from arginine (R) to glutamine (Q) (R263Q) conferring a loss-of-function phenotype, and has been proposed to have a protective effect in SLE [80]. The frequency of this allele in ANCA patients was similar to the general American Caucasian population (4.8% vs 3.0%, p = 0.21) ( Table 2) [80].
Combining the reported frequencies of the risk allele, C1858T, in ANCA disease [24,27] with the frequency observed in this USA cohort, a meta-analysis was performed. Even with the population differences, the combined odds ratio was 1.49 (95% confidence interval 1.28-1.73) (p,0.0001) ( Table 3).

Assessment of functional changes attributed to the gainof-function variant of PTPN22
Based on the assertion that the PTPN22 variant (R620W) confers a gain-of-function phenotype, we hypothesized that unstimulated peripheral leukocytes from patients carrying this allele would have higher basal activity. Evaluations included both patients in remission and with active disease (Table 4). To determine amount of activity/protein concentration, PTPN22 protein was captured from total leukocyte lysates using an anti-PTPN22 antibody on two separate micro-titer plates. One was analyzed for total protein captured and the other for activity status of the captured protein. All patients with the gain-of-function variant (R620W) (n = 12) expressed high basal PTPN22 phosphatase activity in un-stimulated leukocytes, in stark contrast to controls with undetectable activity, including both leukocytes expressing the loss-of-function alleles (n = 3) and non-variant alleles (n = 12) (1.2260.14 versus 0.4160.12, p,0.0001) ( Figure 1A). High basal phosphatase activity was present in R620W neutrophils (n = 6) (5.8761.50 versus 1.0460.38, p = 0.0004) as well as lymphocytes&monocytes (n = 6), but not in non-variant controls (n = 6) (p = 0.0003) ( Figure 1B).
We asked if we could modulate this high basal phosphatase activity by treating the leukocytes with the powerful stimulant PMA (n = 10 with sufficient sample). PTPN22 phosphatase activity was signifi-  Figure 1C). The data imply that constitutive phosphatase activity of the variant remains susceptible to pharmacological agents.

Downstream effects of PTPN22 variant high basal activity
To examine whether the high basal activity of the gain-offunction PTPN22 variant, observed in un-stimulated leukocytes, was affecting primary signaling pathways, we mined our existing Affymetrix array database. Four of the patients enrolled in that study were carriers of the gain-of-function PTPN22 variant (Table 5). Comparisons between groups indicated that the high basal activity of the PTPN22 variant caused global changes in gene transcription. Analysis identified that 151 genes (98 up and 53 down) were differentially regulated ($2.0-fold and p,0.05). Bioinformatic analysis using principal component analysis (PCA) showed remarkably different gene expression profiles comparing leukocytes with the gain-of-function genotype compared with non-variants ( Figure 2). Genes with correlated expression profiles tend to cluster tightly into a small-size elliposoid by the wire mesh, while genes with less similar expression profiles form a looser cluster with a larger size of ellipsoid (Figure 2). The analysis indicates that there are dramatic intrinsic differences in signaling pathways associated with the gainof-function polymorphism. Analysis using the Ingenuity Pathway Tools (IPA) software indicated that the primary networks affected were those involving ERK, p38MAPK and NFkB ( Figure 2).
These data are consistent with reports that PTPN22 function regulates signaling molecules leading to activation of ERK1,2 [83,84]. We examined the phosphorylation status of ERK1,2 in four patients with PTPN22 (R620W), two patients with PTPN22 loss-of-function allele (R263Q) and three patients with the normal allele (Table 4). Phosphorylated/active ERK was undetectable with PTPN22 gain-of-function activity, in contrast to controls. Instead, the phosphorylated/active p38 mitogen-activated protein kinase (p38 MAPK) form was elevated (Figure 3).

IL-10 gene expression is down regulated in leukocytes with the PTPN22 gain-of-function (R620W) variant
Maximal IL-10 production requires signaling through activated ERK and the downstream phosphorylation of the Sp1 transcrip- tion factor [85,86].    It was reported that lower IL-10 levels in remission are associated with a higher relapse rate in long-term follow-up [87]. Analysis of our cohort indicated that higher IL-10 transcript levels were associated with non-relapsing disease (n = 14, 6.664.4, p,0.0001), while lower levels were associated with a relapsingdisease history (n = 39, 1.861.15) ( Figure 4D). There were 14 patients in this study that had the gain-of-function SNP. Three (21%) were in the non-relapsing group while 11(79%) were in the relapsing-disease group.

Discussion
This is the first report of studies on the basal activity of the PTPN22-gain-of-function protein in non-stimulated leukocytes, immediately following blood draw. High basal PTPN22 phosphatase activity was detected in leukocytes of every patient tested who had the PTPN22 SNP (C1858T) genotype, while their non-variant counterparts had undetectable activity. We were intrigued that high basal PTPN22 phosphatase activity was present in neutrophils expressing the gain-of-function variant. Interestingly, one consequence was activation of the p38 MAPK pathway. P38 MAPK regulates macrophage and neutrophil functional responses, including respiratory burst activity, and chemotaxis. High activity was also present in the lymphocyte&monocyte pool which would support the findings of altered T cell function in Type-1 diabetes and Jurkat T leukemia cells overexpressing the transfected gain-of-function variant [80,[88][89][90].
Constitutive activity of PTPN22 gain-of-function variant was associated with global changes in the transcriptome, inasmuch as the assessment of the small cohort studied here. Activity of normal PTPN22 is under regulatory constraints, one of which is inhibition by phosphorylation on Ser-35 by protein kinase C (PKC) [91]. The mechanistic pathway of PTPN22 is under study by many groups and to date, functional partners of normal PTPN22 include growth factor receptor-bound protein 2 (GRB-2), and C-SRC kinase (CSK) [89]. The schematic in Figure 5A illustrates some of the signaling pathways reportedly linked with PTPN22 function, and in Figure 5B a prediction -based on the data presented hereof how they may be perturbed by constitutively high PTPN22 activity.
It is reported that decreased IL-10 production during remission is a predictor of relapse in ANCA disease [87]. Results in our study are in agreement. Moreover, we found that IL-10 message was lower in patients expressing the PTPN22 gain-of-function variant and that a high proportion of these fell into the relapsing-disease category. Reduced IL-10 responses in patients with the gain-offunction variant can be considered as a deviation in transmission of signals within the microenvironment of the body. We examined gene expression of the anti-inflammatory cytokine IL-10, because it is known to be responsive to environmental triggers [86,92]; for example, neutrophils secrete high amounts of IL-10 in response to  [83,84]. It can affect the activity of SRC-family kinases through regulation of CSK (cSRC Kinase) [96,97]. PTPN22 can affect RAS activity through binding to GRB-2 (Growth factor receptor-bound protein 2). ERK1,2 phosphorylates and activates many transcription factors, including the transcription factor Sp1 depicted here, which regulates the transcription of IL-10 [86]. (B) Changes in PTPN22 function due to the gain-of-function phenotype. PTPN22 (R620W) amino acid change lies within a domain that binds CSK, resulting in a greatly reduced binding [84]. Thus CSK is available for binding and inhibiting SRC. Also, a gain-of function alteration could act as a super-antagonist of epigenetic nucleosome remodeling, based on reports that PTPs directly dephosphorylate histone tails [98]. The gain-of-function phosphatase activity also affects the function of GRB2 restraint of RAS signaling and a loss of ERK1/2 phosphorylation/activation and loss of Sp1 transcriptional activity. doi:10.1371/journal.pone.0042783.g005 bacterial products [93]. Relevant to this discussion, expression of IL-10 is regulated through epigenetic mechanisms at the IL-10 locus through signals transmitted by the ERK signaling pathway, which is responsive to many cytokines, growth factors, and importantly environmental stress [85,86,94].
Although the allelic variant of PTPN22 has been reported as a predisposing factor in ANCA disease [95], this is the first report of its frequency in a USA patient cohort. In this patient cohort, association of the disease-associated allele was skewed toward PR3-ANCA disease. In concordance, the geographic distribution of this variant allele is highest in countries where PR3-ANCA predominates with the highest in Finland (15.5%), Sweden (12%) and then UK (8%), decreasing southward to Spain (6%) and Italy (2%). Combined, these studies provide statistical power to support an association between the PTPN22 variant and ANCA disease (OR 1.49, p = 4.15610 26 ). The allele is nearly absent in African American and Asian populations [90].
PTPN22 integrates and transmits environmental changes through dynamic signalling molecules. Studies of the gain-offunction variant in multifactorial autoimmune diseases, such as ANCA disease, provides the opportunity to understand how disease outcome can be influenced by a complex interplay of genetic regulation and environmental influences.