Gene Expression by PBMC in Primary Sclerosing Cholangitis: Evidence for Dysregulation of Immune Mediated Genes

Primary sclerosing cholangitis (PSC) is a chronic disease of the bile ducts characterized by an inflammatory infiltrate and obliterative fibrosis. The precise role of the immune system in the pathogenesis of PSC remains unknown. We used RNA microarray analysis to identify immune-related genes and pathways that are differentially expressed in PSC. Messenger RNA (mRNA) from peripheral blood mononuclear cells (PBMC) was isolated from both patients with PSC and age and sex matched healthy controls. Samples from 5 PSC patients and 5 controls were analyzed by microarray and based upon rigorous statistical analysis of the data, relevant genes were chosen for confirmation by RT-PCR in 10 PSC patients and 10 controls. Using unsupervised hierarchical clustering, gene expression in PSC was statistically different from our control population. Interestingly, genes within the IL-2 receptor beta, IL-6 and MAP Kinase pathways were found to be differently expressed in patients with PSC compared to controls. Further, individual genes, TNF-α induced protein 6 (TNFaip6) and membrane-spanning 4-domains, subfamily A (ms4a) were found to be upregulated in PSC while similar to Mothers against decapentaplegic homolog 5 (SMAD 5) was downregulated. In conclusion, several immune-related pathways and genes were differentially expressed in PSC compared to control patients, giving further evidence that this disease is systemic and immune-mediated.


Introduction
Primary sclerosing cholangitis (PSC) is a rare liver disease of chronic inflammation affecting primarily medium and large sized bile ducts leading to fibrosis and eventually cirrhosis. Histologically, PSC is characterized by an inflammatory infiltrate and obliterative fibrosis, which leads to the pathognomic biliary strictures observed on cholangiographic imaging. Whether this immune response is autoimmune in nature or simply a response to bacterial or viral pathogens remains unresolved.
Several lines of evidence have established that PSC is an immune-mediated process. First, the inflammatory infiltrate of PSC consists predominantly of T cells, which have been shown to have a Th-1 type phenotype. Second, patients with PSC and their family members have increased frequencies of autoimmune diseases (Saarinen et al. 2000;Bergquist et al. 2005), for example, approximately 80% of patients with PSC have coexistent inflammatory bowel disease (IBD), most often ulcerative colitis Bambha et al. 2003). Third, the human leukocyte antigen (HLA) complex, particularly the HLA B8-DR3 haplotype, has been positively associated with PSC (Schrumpf et al. 1982;Chapman et al. 1983;Spurkland et al. 1999;Donaldson and Norris 2001;Wiencke et al. 2001;Bittencourt et al. 2002;Neri et al. 2003). Finally, autoantibodies are frequent in PSC, the most common is the atypical perinuclear anti-neutrophil antibody (p-ANCA), present in 80% of patients (Terjung and Worman 2001;Aoki et al. 2005b). The purpose of this study was to compare the gene expression profile of peripheral blood mononuclear cells (PBMC) from PSC patients with age and sex matched controls using microarray technology.

Subjects
The PSC patients were clinically well characterized and diagnosed in accordance with accepted criteria with typical findings at cholangiography (Chapman et al. 1980;Wiesner and LaRusso 1980), the diagnosis and classification of IBD were based on endoscopic and histological criteria. Demographic and clinical characteristics of the patients are noted in Table I. Unrelated, healthy, age and sex matched controls were used throughout this study. The study protocol conformed to the ethical guidelines of the 1975 Declaration of Helsinki as reflected in a priori approval by the appropriate institutional review committee. Informed consent in writing was obtained from each subject.

RNA isolation
Heparinized venous blood samples were obtained and within 1 h of collection, PBMC isolated by Ficoll-Histopaque gradient centrifugation techniques using Histopaque-1077 (Sigma Chemical Co, St Louis, MO, USA). Trizol Reagent (Invitrogen, Carlsbad, CA, USA) was used to isolate total RNA, the RNA quality and integrity was determined using an Agilent 2100 Bioanalyzer (Agilent Technologies, Palo Alto, CA, USA) and absorbance at A260/A280.

Microarray analysis
In the first part of this study, five PSC subjects (PSC 1, 3, 4, 5 and 7) and matched controls were used for microarray analysis. Sample labeling, hybridization to chips and image scanning were performed as described (Affymetrix Expression Analysis Technical Manual). For each sample, 10 mg total RNA was labeled using One-Cycle Target Labeling and hybridized to the Affymetrix Human U133 Plus2 arrays that contain more that 54,000 probe sets that sample over 39,000 genes.

Data analysis
The ten CEL files were first imported into the R environment and analysed with the Affymetrix w package of Bioconductor. Raw data were normalized with quantile normalization; background corrected and gene expression levels computed with Robust Multichip Average (RMA) (http://www.bioconductor. org/). Presence/absence calls were computed using the Wilcoxon signed rank-based gene expression presence/absence detection algorithm. Genes present with a p , 0.005 significance level were included for further analysis. Significant differences between PSC and control groups were computed using Student t-test for genes that differed at least 2-fold between the medians of the cases and controls. A p-value of p , 0.005 was considered significant. Genes that were expressed in at least 80% of the samples and had a 2-fold or greater difference compared to the median of the samples were clustered using unsupervised hierarchical clustering with Gene Cluster 3.0. The leaves of the clustering trees were then folded with Self-Organizing Map (SOM) algorithm. The clustering result was presented using Java TreeView 1.0.1.

Semi-quantitative RT-PCR
For the second part of our study, mRNA from 10 PSC patients and 10 aged and sex matched controls were used for RT-PCR. Gene microarray results were validated for a select number of transcripts using semi-quantitative real-time polymerase chain reaction (Table II) performed on 1 ml of a 1:10 dilution of the cDNA reaction was used for template with primers in a 25-ml SYBR Green Assay GAPDH RNA was also measured using primers GAPDH.F (5 0 -GAAGGTGAAGGTCGGAGTC-3 0 ) and GAPDH.R (5 0 -GAAGATGGTGATGGGA-TTTC-3 0 ). All samples were assayed in duplicate. Quantification was performed using Sequence  Detector version 1.7 software. A standard curve was established with pooled cDNA derived from the age and sex matched controls and used as a calibrator in each assay. mRNA was recorded relative to the standard and normalized to GAPDH. Statistical analysis was performed using Graphpad Prism v 4.0 (Graphpad Inc, San Diego, CA, USA). Because the distribution was non-Gaussian, comparisons were made with non-parametric methods. A P value of less than 0.05 was considered significant.

Results
Unsupervised hierarchical clustering separated PSC cases from controls ( Figure 1). 1887 genes were clustered that were expressed in at least 80% of the samples with at least one of the samples having a 2-fold difference in expression compared to the median of the samples. Using a t-test comparison we identified 942 genes differentially expressed in patients with PSC compared to controls when significance was set at p , 0.005. Using the Database for Annotation and Visual Integration Discovery (DAVID) on these 942 genes we found evidence of several pathways potentially altered in PSC (Table III). These pathways included IL-2 receptor beta activation of T cells, IL-6 signaling, macrophage differentiation by M-CSF and MAP kinase-signaling pathways. We discovered in PSC PBMC three genes expressed greater than 2-fold and 18 genes expressed less than 0.5-fold the median expression index of the controls (Tables III, IV and V).
To validate the microarray data and to investigate potential pathways suggested by the array data, we chose nine genes to quantify by RT-PCR in10 PSC cases and 10 controls (Figure 2). Of the genes with greater than 2-fold expression in PSC by microarray, membrane-spanning 4-domains subfamily A member 4 and tumor necrosis factor alpha-induced protein six were found to be expressed significantly greater in PSC by RT-PCR. SMAD 5, which was 0.6-fold less than controls by microarray at a P value of 0.00003, was found to be significantly lower in PSC than controls. However, SMAD 4 and TGF-b receptor I which signals through SMAD 4 and SMAD 5 were not found to be significantly different between PSC and controls. No other genes tested were found to have statistically significant differences between PSC patients and controls by RT-PCR (data not shown).

Discussion
Using microarray technology, we screened PBMC from PSC patients in search of genes that might lead to insights in the pathophysiology of PSC. We chose to analyse this data with maximum sensitivity using a simple t-test and fold change comparison to the median of the control sample. From this data we then sought to determine immunological pathways that may be relevant to the disease. Analysis of the results suggested the involvement of the IL-2 receptor beta activation of T cells and IL-6 signaling. The IL-2 receptor is composed of the alpha, beta and gamma subunit (Leonard et al. 1984). In addition to resting and activated T cells, the IL-2 receptor beta has been found on B cells (Zola et al. 1991), natural killer cells (Nishikawa et al. 1990), monocytes (Espinoza-Delgado et al. 1990), dendritic epidermal cells (Tanaka et al. 1992) and neutrophils (Djeu et al. 1993). IL-2 receptor beta knock out mice develop an increase in activated CD4 þ T cells, an increase in total IgG, high-titers of anti-nuclear and anti-DNA autoantibodies, and an autoimmune hemolytic anemia. Interestingly, these mice also have an autoimmune mediated liver disease with immature myeloid cell infiltration of the liver sinusoids and portal tracts (Suzuki et al. 1995). The differentially expressed genes in this pathway found in patients with PSC may reflect a downregulation of the IL-2 receptor beta and potentially plays a role in the development of inflammation of the medium and large bile ducts observed in PSC.
The reduced expression of SMAD 5 in PSC may be relevant as a number of autoimmune diseases  including IBD have been reproduced in TGFb knock out mice (Hahm et al. 2001;Aoki et al. 2005a). TGFb has immunosuppressive properties that are mediated by SMAD signaling.
Other genes confirmed to be differentially expressed in PSC compared to control include TNF-a induced protein 6 (TNFaip6) and membrane-spanning 4-domains, subfamily A (ms4a). TNFaip6 is a 35 kilodalton (kda) protein that is elevated in a number of autoimmune diseases including, systemic lupus erythematosus, rheumatoid arthritis (Wisniewski and Vilcek 1997) and mucosal smooth muscle cells in IBD (Wisniewski and Vilcek 1997). TNFaip6 is a secreted protein (Lee et al. 1992) found also on PBMC after stimulation with LPS, TNFa, PHA and Con A, but is not present on PBMC from healthy human subjects Wisniewski et al. 1993). TNFaip6 is believed to contribute to chronic inflammation in a Figure 2. Confirmatory RT-PCR using cDNA derived from PBMC from patients with PSC compared to age-and sex-matched healthy control patients. Each gene is normalized to GAPDH and relative to pooled cDNA from the controls. The horizontal line represents the mean expression. number of ways. First, secreted TNFaip6 complexes with a serine protease inhibitor, inter-alpha-inhibitor (Wisniewski et al. 1994), these complexes have been found in the synovial fluid of rheumatoid arthritis patients and tightly bind hyaluronic acid (HA), a major component of extra cellular matrix tissue. The formation of the TNFaip6:inter-alpha-inhibitor may play a role in disease by binding to HA and cause a dysregulation of extracellular matrix remodeling or assembly (Milner and Day 2003). This may be occurring in the extracellular matrix of PSC patients where there is a chronic fibrosing process surrounding biliary epithelial cells, resulting in an "onion skin" like appearance on histopathology. Alternatively, the upregulation of TNFaip6 on PBMC in patients with PSC may be a marker of intense TNFa or IL-1 stimulation and is acting to inhibit the inflammatory process as has been shown in air pouch mice, a mouse model of arthritis (Wisniewski et al. 1996).
MSa4a is a member of a superfamily of which includes CD20, a marker of precursor and mature B cells, high affinity IgE receptor (Fc1R1b) found on mast cells and basophils, and a hematopoietic-cell specific protein found on a diverse group of myeloid and lymphoid cells (HTm4) of unknown function (Ishibashi et al. 2001). These proteins share 16 -32% homology and their genes are all found on chromosome 11q12 -13.1 (Tedder et al. 1988;Hupp et al. 1989;Adra et al. 1994). Interestingly, this gene has been mapped to a susceptibility locus in patients with asthma/atopy (Adra et al. 1999). The MSa4a is more closely related to Htm4 and Fc1R1b than CD 20 and found on a number of human lymphoblastoid cell lines, but more highly expressed in pre B cell (NALM-6) and B cell lines (BJAB) than T cell lines (Liang and Tedder 2001). This may confirm prior reports, which have shown B cells to be the predominant cell type in PBMC in patients with PSC (Lindor et al. 1987). The function of the Msa4a protein is unknown but given its amino acid homology to the other members of the superfamily it is suggested it is a component of an oligomeric cell surface complex that plays a role in signal transduction (Liang and Tedder 2001).
A number of other candidate genes remain to be examined further using RT-PCR. We have initially chosen only those thought to be immunologically relevant; for many of the genes differentially expressed, the function remains to be identified. In addition, further work needs to be performed, to define the cell type and protein expression of the identified genes. There has been one other study looking at genome wide expression in autoimmune liver disease (Shackel et al. 2001). RNA was extracted from liver biopsies of patients with PBC (n ¼ 6), PSC (n ¼ 4) and normal liver and the cDNA array used consisted of 588 genes and nine housekeeping genes. None of the genes differentially expressed in that study were identified in this present study. This lack of correlation may be related to the different tissue and gene chip used in that prior study. Our demonstration of several immune-related pathways and genes differentially expressed in PSC PBMC compared to control patients, suggests that the disease is more than a local inflammatory process in the liver and reflects a systemic immune response.