SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. AUTHOR CONTRIBUTIONS
  7. Acknowledgments
  8. REFERENCES
  9. Supporting Information

Objective

In HLA–B27–transgenic rats, the development of a disorder that mimics spondyloarthritis (SpA) is highly correlated with dendritic cell (DC) dysfunction. The present study was undertaken to analyze the underlying mechanisms of this via transcriptome analysis.

Methods

Transcriptome analysis of ex vivo–purified splenic CD103+CD4+ DCs from B27-transgenic rats and control rats was performed. Transcriptional changes in selected genes were confirmed by quantitative reverse transcriptase–polymerase chain reaction. A meta-analysis of our rat data and published data on gene expression in macrophages from ankylosing spondylitis (AS) patients was further performed.

Results

Interferon (IFN) signaling was the most significantly affected pathway in DCs from B27-transgenic rats; the majority of genes connected to IFN were underexpressed in B27-transgenic rats as compared to controls. This pattern was already present at disease onset, persisted over time, and was conserved in 2 disease-prone B27-transgenic rat lines. In DCs from B27-transgenic rats, we further found an up-regulation of suppressor of cytokine signaling 3 (which may account for reverse IFN signaling) and a down-regulation of interleukin-27 (a cytokine that opposes Th17 differentiation and promotes Treg cells). The meta-analysis of data on conventional DCs from rats and data on monocyte-derived macrophages from humans revealed 7 IFN-regulated genes that were negatively regulated in both human and rat SpA (i.e., IRF1, STAT1, CXCL9, CXCL10, IFIT3, DDX60, and EPSTI1).

Conclusion

Our results suggest that expression of HLA–B27 leads to a defect in IFNγ signaling in antigen-presenting cells in both B27-transgenic rats and SpA patients, which may result in Th17 expansion and Treg cell alteration (as shown in B27-transgenic rats) and contribute to disease pathogenesis.

Spondyloarthritis (SpA) comprises a spectrum of inflammatory disorders that predominantly affect spinal and sacroiliac joints, the prototypical form of which is ankylosing spondylitis (AS). Other characteristic manifestations that may variably combine with each other include peripheral joint arthritis and enthesitis, as well as several extraarticular features, including psoriasis, uveitis, and inflammatory bowel disease ([1]). The precise molecular and cellular basis underlying the strong association of SpA with the class I major histocompatibility complex (MHC) antigen HLA–B27 remains uncertain today, 40 years after the initial description of the association, despite the ongoing efforts of many groups to explain it ([2, 3]).

Studies of HLA–B27/human β2-microglobulin (hβ2m)–transgenic rats (B27-transgenic rats) have advanced our understanding of this relationship. Several lines of B27-transgenic rats have proven to be suitable models of SpA, since B27-transgenic rats develop a spontaneous multisystem inflammatory disease that closely resembles human SpA (with arthritis, colitis, and psoriasiform skin lesions developing in rats with an intact gut microbiome) ([4, 5]).

In this model, cell transfer experiments have shown that bone marrow–derived cells (thought to be antigen-presenting cells [APCs]) that express high levels of HLA–B27 are necessary and sufficient to induce SpA in healthy rats ([6]) and that disease expression requires the intervention of CD4+ T cells ([7]). Interestingly, professional APCs, such as dendritic cells (DCs) from disease-prone B27-transgenic rats, exhibit several strikingly abnormal functions, including impaired capacity to stimulate T cell responses, altered cytoskeletal dynamics, reduced expression of class II MHC molecules, and enhanced apoptotic death ([4, 8-10]). A strict correlation has been established between altered DC function, high expression of the HLA–B27/hβ2m transgene, and disease susceptibility across several transgenic lines ([11]). Finally, aberrant function has been observed in DCs from premorbid B27-transgenic rats and from rats expressing the disease-associated HLA–B27/hβ2m–transgenic locus on a genetic background that is protective against disease development, thereby demonstrating that this defect was not a consequence of disease and could be a principal factor in the spontaneous development of SpA, e.g., by altering the interaction of DCs with CD4+ T cells ([4]).

Consistent with the foregoing hypothesis, CD4+ T cells expressing a proinflammatory Th17 profile accumulate in B27-transgenic rats in parallel with disease development ([12]), and DCs from B27-transgenic rats induce biased expansion of Th17 cells that could play a pathogenic role in SpA ([13]). Moreover, we recently observed that Treg cell function was altered in B27-transgenic rats, as shown by a decreased ratio of interleukin-10 (IL-10) to IL-17 production, and that this imbalance favoring proinflammatory over antiinflammatory cytokine production was driven by the interaction of CD4+ T cells with DCs ([14]).

Taken together, these data suggest a plausible link between aberrant DC function and the development of SpA in B27-transgenic rats. However, the molecular events that result from HLA–B27 expression in DCs and lead to functional impairment remain incompletely understood. In an attempt to more thoroughly investigate this issue, we analyzed the transcriptome, comparing ex vivo–purified CD103+CD4+ splenic DCs from B27-transgenic rats with those from control rats (transgenic for HLA–B7/hβ2m [B7-transgenic] or nontransgenic). Next, we compared our results with the results of a study of monocyte-derived macrophages from AS patients ([15]). Strikingly, we observed that several genes induced by interferon (IFN) were similarly down-regulated in both disease-prone B27-transgenic rats and SpA patients, thereby highlighting a coordinate dysregulation of the IFN pathway in APCs that might be critical for disease development.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. AUTHOR CONTRIBUTIONS
  7. Acknowledgments
  8. REFERENCES
  9. Supporting Information

Animals

The HLA–B/hβ2m–transgenic rat lines used in this study were originally produced at the University of Texas Southwestern Medical Center. The disease-prone HLA–B27/hβ2m–transgenic rat line 33-3, bearing 55 copies of HLA–B*2705 and 28 copies of hβ2m, and the disease-free HLA–B7/hβ2m–transgenic rat line 120-4, bearing 52 copies of HLA–B*0702 and 26 copies of hβ2m, were both on a F344 background (in this study, the number of hβ2m transgene copies in the 33-3 and 120-4 rat lines was reevaluated using a more sensitive quantitative polymerase chain reaction [PCR] method than was used for the previously reported dot-blot estimation [16]). HLA–B27/hβ2m–transgenic 21-3 rats and hβ2m-transgenic 283-2 rats were crossed to obtain disease-prone (21-3 × 283-2)F1 rats, bearing 20 copies of HLA–B*2705 and 50 copies of hβ2m, and disease-free 283-2 rats, bearing 35 copies of hβ2m, which were both on a Lewis background. Nontransgenic F344 rats and their Lewis littermates were used as controls. All rats were bred and housed under conventional conditions. Age- and sex-matched rats (2–12 months of age) were used in each experiment. All animal procedures were approved by the Institutional Animal Experimentation Ethical Committee (CEB-26-2012).

CD103+CD4+ splenic DC isolation

DCs were purified ex vivo according to methods that have been previously described by Josien et al ([17]). Spleens were digested with 2 mg/ml collagenase D (Roche Diagnostics) for 20 minutes at 37°C and in the presence of EDTA at 10 mM during the last 5 minutes. Cell suspension was washed and resuspended in phosphate buffered saline, 0.5 mM EDTA, and 1% bovine serum albumin, and low-density cells, containing most of the conventional CD103+ DCs, were obtained after centrifugation on a 14.5% Nycodenz gradient (Nycomed).

For magnetic sorting, T cells and B cells were depleted by incubating low-density cells with anti-rat CD45RC (OX22; Santa Cruz Biotechnology), anti-rat Ig κ chain (OX12; Santa Cruz Biotechnology), anti-rat α/β T cell receptor (R73; BD PharMingen), and anti-rat CD25 (OX39; BD PharMingen) monoclonal antibodies at 4°C for 20 minutes. Negative selection of CD103+ cells was performed with goat anti-mouse IgG MicroBeads on CS columns according to the instructions of the manufacturer (Miltenyi Biotec). Cells were then incubated with anti-CD4 MicroBeads (Miltenyi Biotec) at 4°C for 20 minutes. Positive selection of CD103+CD4+ DCs was performed on MS columns (Miltenyi Biotec). Purity of the selection was routinely in the range of 70–80%.

For fluorescence-activated cell sorting (FACS), low-density cells were incubated with anti-CD103+ (OX62) MicroBeads (Miltenyi Biotec) at 4°C for 20 minutes. Positive selection was performed using automated cell sorting (autoMACS Pro; Miltenyi Biotec). Cells were then stained with fluorescein isothiocyanate–conjugated anti-rat CD103 (OX62; Santa Cruz Biotechnology), allophycocyanin-conjugated anti-rat CD4 (OX35; BD PharMingen), and phycoerythrin-conjugated anti-rat CD45RC (OX22; BD PharMingen) monoclonal antibodies, and CD103+CD4high DCs were sorted on a BD FACSAria III cell sorter (BD Biosciences) after excluding OX22+ cells (including plasmacytoid DCs). Purity of the selection was routinely ≥98%.

RNA isolation

For gene expression studies, total RNA was isolated from DC pellets using an RNeasy kit (Qiagen), genomic DNA was eliminated by deoxyribonuclease treatment (RNase-Free DNase set; Qiagen), and samples were immediately stored at −80°C. The RNA quality was assessed using a Bioanalyzer 2100 (Agilent), and samples were quantified using a NanoDrop Spectrophotometer (Thermo Scientific); the RNA integrity number was 7.5–10.

Hybridization and microarray analysis

Microarray experiments were performed using an Affymetrix GeneChip RatGenome 230.2.0 genome-wide array containing 31,100 probes. Total RNA from magnetically sorted CD103+CD4+ DCs was reverse transcribed using an Affymetrix GeneChip 3′ in vitro transcription express kit. Briefly, the resulting double-stranded complementary DNA was used to synthesize biotin-labeled complementary RNA. After purification, complementary RNA was fragmented and hybridized to chips. After overnight hybridization, chips were washed in a Fluidic Station FS450 according to protocol and scanned using an GCS3000 7G (Affymetrix). The image was then analyzed with Expression Console software (Affymetrix) to obtain raw data (CEL files) and metrics for quality control. The microarray data accession number is E-MEXP-3808.

Gene expression levels were normalized using the GeneChip robust multiarray averaging algorithm, and flags were computed using MAS5. Quality assessment of the chips was performed using an affyQCReport R package. Gene expression was compared in B27-transgenic rats and control rats. For statistical analyses, a t-test was used to calculate P values (significance threshold set at P ≤ 0.05). Hierarchical clustering was applied to mean centered data to perform sample and gene classification (Spearman's correlation similarity measure and an average/Ward's linkage algorithm were used). R software was used to create heatmap source files, and pictures were produced with Java TreeView software. For probe sets and annotation and examination of the gene networks, including canonical pathways, molecular functions, and genetic networks, data were analyzed using IPA tools (http://www.ingenuity.com).

Meta-analysis of human and rat microarray data

A meta-analysis of the rat data from this study and the data from a published gene expression study of macrophages from 8 AS patients and 9 healthy control subjects ([15]) was performed as previously described ([18]). Human data extraction, normalization, and analysis were performed using the same method and cutoff values that were used in this rat study.

To identify possible corresponding features between samples from the 2 species, we performed an unsupervised clustering analysis of merged human and rat expression data. For this, we identified differentially expressed genes that are common to human and rat chips (251 orthologous genes). The corresponding probes were then selected and grouped by gene, using their mean to obtain a single intensity measure by gene and by sample. We merged the human and rat coded data sets and submitted this to hierarchical clustering, which was performed using Spearman's correlation similarity measure and an average linkage clustering algorithm.

Quantitative reverse transcriptase–PCR (qRT-PCR).

The qRT-PCR experiments were conducted according to the recommendations of the Minimum Information for Publication of Quantitative Real-Time PCR Experiments guidelines ([19]). Due to the large amount of data generated by our study, detailed data on each target gene are available upon request from the corresponding author.

RNA replicates of the same sample were reverse-transcribed using SuperScript II reverse transcriptase with random primers (both from Invitrogen) and then pooled to homogenize expression profiles. Quantitative PCR was performed using a LightCycler 480 SYBR Green I Master Kit (Roche) and the real-time PCR system LC480 (Roche). Primers were designed using a Primer3 tool and tested for specificity and efficiency (see Supplementary Table 1, available on the Arthritis & Rheumatology web site at http://onlinelibrary.wiley.com/doi/10.1002/art.38318/abstract). Thermocycling conditions were as follows: 95°C for 5 minutes, and then 45 cycles of 95°C for 10 seconds, 60°C for 10 seconds, and 72°C for 15 seconds. This was followed by the standard denaturation curve. Duplicates were run for each sample in a 96-well plate; Gapdh and Pgk1 were used as endogenous reference genes as previously described ([20]). Each qPCR analysis included negative controls without RNA.

Data were analyzed using the ΔCq method. For each sample, messenger RNA (mRNA) abundance (Cq value) of the target genes was normalized to the reference genes and compared to a calibrator value, corresponding to the average expression of all samples (mean of all sample Cq values). The results were depicted as relative expression corresponding to the normalized relative quantity.

Statistical analysis

Quantitative RT-PCR data are expressed as mean ± SEM, and the significance of differences between series of results was assessed using Student's unpaired 2-tailed t-test. P values less than 0.05 were considered significant.

RESULTS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. AUTHOR CONTRIBUTIONS
  7. Acknowledgments
  8. REFERENCES
  9. Supporting Information

Whole-transcriptome study of B27-transgenic rat CD103+CD4+ splenic DCs

We compared the transcriptome of CD103+CD4+ splenic DCs from B27-transgenic rats with those from control rats (B7-transgenic rats and nontransgenic rats). We identified 422 probe sets that were differentially expressed between B27-transgenic rat DCs and B7-transgenic rat DCs (164 up-regulated and 258 down-regulated in B27-transgenic rat DCs) as well as 623 probe sets that were differentially expressed between B27-transgenic rat DCs and nontransgenic rat DCs (252 up-regulated and 371 down-regulated in B27-transgenic rat DCs). Altogether, 255 probe sets corresponding to 172 annotated genes were differentially expressed in B27-transgenic rat DCs as compared to both control groups, with consistent direction of variation (69 up-regulated and 103 down-regulated genes in B27-transgenic rats) (Figure 1A). Using these data, B27-transgenic rats were readily separated from both control groups by clustering analysis (Figure 1B).

image

Figure 1. Whole-transcriptome study of HLA–B27–transgenic (Tg) rat CD103+CD4+ splenic dendritic cells (DCs). Gene array analysis of 8 pools (2 rats per pool) of 6–10-month-old male B27-transgenic rats and 2 control groups of age- and sex-matched rats (HLA–B7–transgenic rats [6 pools] and nontransgenic rats [NTG] [7 pools]) was performed. A, The Venn diagram represents the extent to which probe sets that were differentially expressed between B27-transgenic rats and nontransgenic rats and probe sets that were differentially expressed between B27-transgenic rats and B7-transgenic rats overlap. The direction of variation in differentially expressed probe sets (up-regulated or down-regulated) is shown. In only 2 of 257 overlapping probe sets did the directions of variation differ according to the control group used. B, Hierarchical clustering was performed using 255 overlapping probe sets corresponding to 172 annotated genes, and consistent variations between B27-transgenic rats and both control groups were exhibited. The blue-coded bar corresponds to the transcripts that were down-regulated, and the magenta bar corresponds to the transcripts that were up-regulated, in B27-transgenic rats.

Download figure to PowerPoint

Global functional analysis conducted with IPA software indicated that IFN signaling was the most significantly affected pathway (see Supplementary Figure 1, available on the Arthritis & Rheumatology web site at http://onlinelibrary.wiley.com/doi/10.1002/art.38318/abstract). Indeed, several of the most profoundly down-regulated genes in B27-transgenic rat DCs were connected to the IFN pathway, indicating a “reverse” (i.e., down-regulated) IFN signature. For example, Irf1, Irf7, Ifit2, Ifit3, Mx1, Ly6e, Cxcl9, Cxcl10, and Cxcl11, which are known to be up-regulated by IFN, were down-regulated in DCs from B27-transgenic rats (see Supplementary Table 2, available on the Arthritis & Rheumatology web site at http://onlinelibrary.wiley.com/doi/10.1002/art.38318/abstract). Altogether, 36 genes that were down-regulated in B27-transgenic rat DCs (35%) were directly connected to the IFNγ pathway (including Ifng itself), and a large fraction of those genes was connected to IFNα as well (see Supplementary Table 3). Surprisingly, in B27-transgenic rats the most highly up-regulated genes were involved in antimicrobial peptide/innate immunity pathways, including genes in the defensin family associated with neutrophil function (Defa, Np4, and RatNP-3b), mastocyte proteases (Mcpt1, Mcpt8, and Cpa3), macrophage receptors (Cd163, Camp, and Pglyrp1), or the inflammatory pathway (S100a8 and S100a9) (see Supplementary Table 2), which indicates that DC samples were possibly contaminated by other cells (see below).

In the same samples, we further measured the expression of several differentially expressed genes by qRT-PCR, including one group of genes down-regulated in B27-transgenic rat DCs (which included, principally, genes linked to the IFN pathway and IL-10), and 2 groups of up-regulated genes that corresponded to genes associated with innate/antimicrobial immunity or with endoplasmic reticulum (ER) stress (see Supplementary Table 4). For most of the genes, we confirmed significant differences between B27-transgenic DCs and control samples (see Supplementary Figure 2).

Defective IFN signaling is consistent across age groups in B27-transgenic rat DCs

To replicate and monitor variations among the selected genes across age groups, we measured gene expression by qRT-PCR in CD103+CD4+ splenic DCs from B27-transgenic and control rats at age 2 months, 6 months, and 1 year. Reduced expression of several cytokines/chemokines and transcription factors associated with the IFN pathway (Figure 2) and increased expression of ER stress–associated genes (see Supplementary Figure 3A, available on the Arthritis & Rheumatology web site at http://onlinelibrary.wiley.com/doi/10.1002/art.38318/abstract) and innate/antimicrobial immunity–associated genes (see Supplementary Figure 3B) were detected in DCs from B27-transgenic rats at age 2 months (i.e., at disease onset) and persisted until 1 year of age. Since no major variations in gene expression patterns were observed at different ages, we grouped together all time points for statistical analysis. Between B27-transgenic rats and control rats, statistically significant differential expression was apparent in most of the genes studied (Figure 2 and Supplementary Figure 3). Because such changes in DCs from B27-transgenic rats were already present at the time of disease onset and persisted throughout the disease course, it is conceivable that at least some of those changes could contribute to the chronic inflammatory process in those animals. Some of the genes that were down-regulated in B27-transgenic rat DCs (i.e., Cxcl10, Il10, Irf7, and Stat1) (Figure 2) were also significantly decreased in B7-transgenic rats, as compared to the nontransgenic controls, albeit to a lesser extent than in the B27-transgenic rats.

image

Figure 2. Monitoring of interferon (IFN) signaling and Il10 expression in B27-transgenic rat DCs across age groups, using quantitative reverse transcriptase–polymerase chain reaction (PCR). IFN pathway–related gene expression of cytokines/chemokines (A) and transcription factors (B) was quantified using ex vivo–magnetically sorted CD103+CD4+ splenic DCs from B27-transgenic rats and control rats (B7-transgenic rats and nontransgenic rats) at the indicated ages. PCR data were normalized to results obtained with Gapdh and Pgk1. Bars show the mean ± SEM of 5–6 pools (2 rats per pool) per group. Statistical analysis was performed using Student's unpaired 2-tailed t-test. See Figure 1 for other definitions.

Download figure to PowerPoint

Defective IFN signaling signature is shared with SpA patients

We have previously reported differences in the transcriptome of monocyte-derived macrophages from AS patients as compared to healthy controls (i.e., we found a “reverse” IFN signature) ([15]). Given the parallels between these data sets, we performed a meta-analysis of the human and rat transcriptome data.

Hierarchical clustering analysis of 251 orthologous rat and human genes resulted in samples that were separated into 2 groups: 16 patients and B27-transgenic rats in one group and 19 (of 22) human and rat controls in the other (Figure 3A). Meta-analysis of rat and human data revealed 9 genes that were differentially expressed in both AS patients and B27-transgenic rats, as compared to all controls (Figure 3B). In 7 of the genes, the direction of variation was similar between patients and B27-transgenic rats (indicative of a “reverse” IFN signature), although it was divergent for the 2 remaining genes (Table 1).

image

Figure 3. Combined gene expression profiling and meta-analysis of human and rat transcriptome data. A, Hierarchical clustering analysis of 251 combined rat and human orthologous genes was performed, resulting in a striking separation, with the 16 ankylosing spondylitis (AS) patients and B27-transgenic rats in one group and 19 (of 22) human healthy controls (hHC) and rat controls in another group. B, The Venn diagram represents the extent to which genes that were differentially expressed in DCs from B27-transgenic rats as compared to nontransgenic rats, genes that were differentially expressed in DCs from B27-transgenic rats as compared to B7-transgenic rats, and genes that were differentially expressed in macrophages from AS patients as compared to human healthy controls overlap. See Figure 1 for other definitions.

Download figure to PowerPoint

Table 1. Dysregulated genes common to dendritic cells from HLA–B27–transgenic rats and macrophages from ankylosing spondylitis patients
GenePatients vs. controlsHLA–B27–transgenic rats vs. nontransgenic ratsHLA–B27–transgenic rats vs. HLA–B7–transgenic ratsGene product
PRatioPRatioPRatio
Cxcl98.39 × 10−30.289.58 × 10−70.223.08 × 10−60.24Chemokine (CXC motif) ligand 9
Cxcl107.61 × 10−30.201.42 × 10−60.165.65 × 10−70.22Chemokine (CXC motif) ligand 10
Ifit36.34 × 10−30.644.20 × 10−50.108.59 × 10−30.21Interferon-induced protein with tetratricopeptide repeats 3
Irf16.76 × 10−30.322.62 × 10−60.592.23 × 10−50.62Interferon regulatory factor 1
Stat15.41 × 10−30.461.57 × 10−40.535.39 × 10−50.67Signal transducer and activator of transcription 1
Ddx604.20 × 10−20.562.37 × 10−50.433.93 × 10−50.35DEAD box polypeptide 60
Epsti11.95 × 10−30.304.43 × 10−30.617.37 × 10−30.62Epithelial stromal interaction 1
Tpd52l12.94 × 10−21.891.17 × 10−50.101.14 × 10−20.27Tumor protein D52-like 1
Sell3.88 × 10−20.502.15 × 10−21.801.22 × 10−21.97L-selectin

Defective IFN signaling is confirmed in FACS-sorted B27-transgenic rat DCs

Since possible contamination of magnetically sorted rat DC samples by other cell populations was a concern, as mentioned above, we repeated qRT-PCR experiments on FACS-sorted CD103+CD4+ splenic DCs with higher purity (>98%). Data confirmed that expression of most of the genes connected to the IFN pathway in DCs from B27-transgenic rats was significantly decreased, as compared to expression in DCs from nontransgenic rats (Figures 4A and B), including among those genes that were differentially expressed in both humans and rats (Cxcl9, Cxcl10, Ifit3, Irf1, Stat1, and Il10). Up-regulation of genes linked to ER stress in B27-transgenic rat DCs was also confirmed (Figure 4C).

image

Figure 4. Expression of selected genes, determined by quantitative reverse transcriptase–polymerase chain reaction (PCR), in rat DCs after fluorescence-activated cell sorting. Expression of interferon (IFN) pathway–related and interleukin-10 (IL-10)–related cytokines/chemokines (A), IFN pathway–related and IL-10–related transcription factors (B), genes associated with endoplasmic reticulum stress response (C), and genes for suppressors of cytokine signaling (D) was quantified in CD103+CD4+ splenic DCs from B27-transgenic rats and nontransgenic control rats. PCR data were normalized to results obtained with Gapdh and Pgk1. Each data point represents a pool of 2 rats (n = 5–6 pools per group); horizontal lines show the mean. Statistical analysis was performed using Student's unpaired 2-tailed t-test. See Figure 1 for other definitions.

Download figure to PowerPoint

Among genes related to innate/antimicrobial immunity, low-level expression was observed both in B27-transgenic rats and nontransgenic rats, confirming that overexpression of these genes, as detected in magnetically sorted B27-transgenic rat DCs, most likely reflected contaminant cell population(s) (see Supplementary Figure 4, available on the Arthritis & Rheumatology web site at http://onlinelibrary.wiley.com/doi/10.1002/art.38318/abstract).

It has recently been shown that engagement of the IFNγ pathway in DCs induces IL-27 while inhibiting osteopontin expression, leading to suppression of IL-17 production and induction of IL-10 from T cells ([21]). Interestingly, we found down-regulated Il27 gene expression in highly purified CD103+CD4+ B27-transgenic rat DCs (Figure 4A). Additionally, we investigated the expression of genes for suppressor of cytokine signaling (SOCS) 1 and 3, which are known for their crucial role in limiting cytokine-mediated inflammatory responses (including IFNγ-related responses), in those cells. No difference in Socs1 expression was observed between B27-transgenic rats and nontransgenic rats. In contrast, Socs3 was overexpressed in DCs from B27-transgenic rats, providing a possible underlying explanation for the down-regulated IFN pathway (Figure 4D).

Defective IFN signaling is common in SpA-prone B27-transgenic rat DCs

To investigate the putative role of unfolded protein response and ER stress in SpA, rats of the 21-3 transgenic line expressing moderate levels of HLA–B27/hβ2m and free of arthritis were crossed with 283-2 rats transgenic for hβ2m (both strains of mice were on a Lewis background). This resulted in F1 offspring with a high incidence of severe arthritis and spondylitis in males, despite evidence of higher HLA–B27 folding efficiency and reduced ER stress ([22]). This phenotype differs from that seen in the 33-3–transgenic line as follows: the disease in (21-3 × 283-2)F1 rats is male-specific, peripheral arthritis is more severe and prevalent, spondylitis is much more prevalent, and gastrointestinal inflammation does not manifest.

We have previously shown that splenic DCs from (21-3 × 283-2)F1 male rats were as defective in their capacity to stimulate T cell proliferation as splenic DCs from 33-3 rats ([11]). Herein, we examined whether HLA–B27 expression would also be associated with defective IFN signaling in (21-3 × 283-2)F1 male rats. We found this to be the case: several genes associated with the IFN pathway (Cxcl10, Ifit2, Ifit3, and Irf7) were significantly down-regulated in (21-3 × 283-2)F1 male rats, as compared to genes from nontransgenic and/or hβ2m-transgenic rats of the 283-2 line (Figure 5A). Interestingly, among ER stress response genes, only Pdia4 was moderately overexpressed in (21-3 × 283-2)F1 rats (Figure 5B), a result consistent with the previously established reduction of HLA–B27 misfolding in this line. Finally, significant up-regulation of Socs3 and down-regulation of Il27 were evident in (21-3 × 283-2)F1 rats, as in rats of the 33-3 line (Figures 5A and C).

image

Figure 5. Expression of selected genes in (21-3 × 283-2)F1 B27-transgenic rat DCs after fluorescence-activated cell sorting (FACS). Quantitative reverse transcriptase–polymerase chain reaction (PCR) was used to calculate the expression of genes related to the interferon pathway and interleukin-10 (A), endoplasmic reticulum stress response (B), and cytokine suppression (C) on FACS-assessed CD103+CD4+ splenic DCs from 6–10-month-old arthritic HLA–B27/human β2-microglobulin (hβ2m)–transgenic male rats of the (21-3 × 283-2)F1 line. Nontransgenic Lewis rats and hβ2m single-transgenic rats of the 283-2 line were used as controls. PCR data were normalized to results obtained with Gapdh and Pgk1. Bars show the mean ± SEM of 3–4 rats per group. Statistical analysis was performed using Student's unpaired 2-tailed t-test. See Figure 1 for other definitions.

Download figure to PowerPoint

DISCUSSION

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. AUTHOR CONTRIBUTIONS
  7. Acknowledgments
  8. REFERENCES
  9. Supporting Information

Several hypotheses have been proposed to explain the striking association between SpA and HLA–B27. Investigations performed in B27-transgenic rat lines allowed us to address some of these associations. For example, the classic arthritogenic peptide hypothesis ([23]) was rendered unlikely, given that CD8+ T cells are not required for disease expression in these rats ([7, 24, 25]). Results of cell transfer experiments led us to speculate that APCs expressing high levels of HLA–B27 were required for disease induction, and that interaction with CD4+ T cells was also required ([6, 7]). It has also been proposed that proinflammatory Th17 cells are key effectors, since these cells accumulate in lymphoid and target organs in parallel with disease development ([12, 13, 26]). Such an interpretation is consistent with the results of recent experiments demonstrating that DCs from B27-transgenic rats favor a biased expansion of Th17 cells ([10, 13]). Moreover, B27-transgenic rat DCs seem to affect Treg cell differentiation and activation in such a way that they produce less IL-10 and more IL-17 ([14]). These results provide a plausible explanation as to why HLA–B27 may favor a Th17-driven disease—an interpretation consistent with other recent data on SpA ([27]).

However, the molecular mechanism by which B27-transgenic rat DCs may influence the differentiation of CD4+ T cells leading to this proinflammatory bias remains to be understood. It was previously reported that DCs from disease-prone rat lines exhibited altered functions ([4]). Consistent across disease-prone rat lines, the decreased capacity to support T cell stimulation, even in antigen-independent assays, was an abnormal feature ([8]). This phenomenon was linked to impaired formation of an immunologic synapse between B27-transgenic rat DCs and CD4+ T cells, which could itself be attributed to several molecular abnormalities, including a defective engagement of costimulatory molecules expressed on DCs with their partners on T cells, altered cytoskeletal dynamics, and reduced expression of class II MHC molecules ([8]). We have also found evidence that the expression of the HLA–B27 molecule on the DC surface was likely responsible for these molecular abnormalities ([8]). Similar defective capacity for T cell stimulation has now been shown among monocyte-derived DCs from AS patients ([28]).

In the present study, we further investigated DC dysregulation linked to HLA–B27 expression. Our study focused on ex vivo–sorted CD103+CD4+ splenic DCs, since this population has been implicated in T cell activation ([29]). Our findings of transcriptome analysis revealed coordinate aberrant gene expression patterns in DCs from B27-transgenic rats of the 33-3 line, with striking evidence of a “reverse” IFN signature. This result was confirmed in several different settings: among rats of various ages (starting at age 2 months, i.e., at disease onset), among highly purified DCs, and among rats of another HLA–B27 transgenic line, i.e., among (21-3 × 283-2)F1 rats, in which the phenotype closely mimics AS without gut inflammation. Moreover, we found that the signature in B27-transgenic rat DCs corresponded to the signature found in monocyte-derived macrophages from AS patients. This strengthened our observation and provided further evidence that the B27-transgenic rat is a faithful model of SpA.

In B27-transgenic rats of the 33-3 line, we further observed an overexpression of genes involved in ER stress processes (i.e., Calr, Pdia3, Pdia4, and Pdia6). This corroborated results of our previous proteomic study of CD103+ splenic DCs from the same line of rats, wherein we found enhanced expression of calreticulin and protein disulfide isomerase A3 ([9]). This is consistent with cells having survived an unfolded protein response, such as the response initially described in bone marrow–derived macrophages from 33-3 rats ([22]). HLA–B27 misfolding and the ensuing unfolded protein response have been proposed as triggering events explaining HLA–B27 pathogenicity in B27-transgenic rats by facilitating the transcription of IFNβ and/or IL-23p19 ([22]). However, such a mechanism would not directly account for the aberrant DC function studied herein, since we found no evidence of up-regulation of those cytokines. Moreover, the evidence of ER stress was much weaker in DCs from arthritic (21-3 × 283-2)F1 rats, consistent with previous experiments, results of which showed an attenuated unfolded protein response in this line, presumably as a consequence of enhanced hβ2m expression ([22]).

A group of innate immunity genes was also up-regulated. However, when FACS was used instead of magnetic selection to improve DC purification, this signature disappeared, indicating that this signature was secondary to contamination of DC samples by other cells. Similar signatures, attributed to tissue infiltration by neutrophils, macrophages, and mast cells, have been shown in human SpA studies ([30]), suggesting that it reflected the ongoing disease process and could likely be attributed to Th17-mediated inflammation.

Thus, the down-regulated IFN pathway (including the gene for IFNγ itself) was the DC signature that was most relevant to SpA pathogenesis. This, combined with decreased IL-10 production by B27-transgenic rat DCs, might be a contributing factor to the drive behind the differentiation of CD4+ T cells into pathogenic Th17 cells, and might also affect Treg cells by decreasing the ratio of IL-10 to IL-17 production, thereby contributing to disease induction ([31, 32]). Accordingly, enhanced IL-17–mediated inflammation has been shown to result from IFNγ signaling deficiency in DCs through reciprocal modulation of osteopontin and IL-27 ([21]). Thus, IFNγ induces IL-27, and conversely, it inhibits osteopontin production, with both effects contributing to the suppression of IL-17 production and the induction of IL-10 from T cells. Consistent with this interpretation, we observed a decreased expression of mRNA for Il27 in B27-transgenic rat DCs. Moreover, increased osteopontin was reported in AS ([33]), and the IL27A gene has recently been shown to be associated with AS susceptibility ([34]).

As another putative consequence of a reverse IFN signature in DCs, B27-transgenic rats might have an altered capacity to raise an appropriate immune response to infectious agents ([35]). Indeed, heightened vulnerability to pathogenic Listeria monocytogenes was previously shown ([36]). Such a mechanism could also account for the dysregulated microbiota that has been documented in the 33-3 line. Although it remains to be determined in this rat line whether it is a cause or consequence of gut inflammation ([37]), recent data indicate that (21-3 × 283-2)F1 rats, which show no gut inflammation, also harbor a microbiome that is distinctly different from that in nontransgenic rats (ref.[38], and Taurog JD, et al: unpublished observations).

We examined Socs1 and Socs3 expression levels, since both are essential inhibitors of Toll-like receptors (TLRs) and cytokine receptor cascades, including IFN receptor–mediated signal transduction. STAT-1 is hyperactivated, and Ifng and IFN-regulated genes are up-regulated in SOCS-1–deficient DCs. Conversely, SOCS-3–transduced DCs express low levels of IFNγ ([39]). We found no difference in Socs1 expression between DCs from B27-transgenic rats and DCs from control rats, but we did observe overexpression of Socs3 in B27-transgenic rat DCs, thereby providing a possible explanation for the reverse IFN signature. Consistently, our transcriptome analysis showed moderate but significant decreased expression of several other genes associated with TLR and cytokine signaling (Irak2, Mp4k4, and Traf3) in DCs from B27-transgenic rats (data not shown).

Whether there is a causal link between HLA–B27 expression in DCs and Socs3 induction remains to be established. SOCS are induced in response to TLR and cytokine signaling by a negative feedback loop that engages signaling by the Tyro 3/Axl/Mer family of receptor tyrosine kinases ([40]). In addition to their canonical function of antigen presentation, class I MHC molecules can be signaling receptors, mediating reverse signaling via association with other receptors or directly through aggregation, and can exert nonclassic functions ([41, 42]). Interestingly, class I MHC molecules expressed in APCs are involved, via reverse signaling, in negative regulation of TLR-triggered inflammatory responses mediated by the signaling molecule tumor necrosis factor receptor–associated factor 6 ([43]). Thus, the singular biochemical properties of HLA–B27, including a tendency to form oligomers of heavy chains at the cell surface, as previously shown in B27-transgenic rat DCs ([44]), could theoretically affect MHC-dependent feedback and influence downstream events, resulting in heightened SOCS-3 induction.

In summary, our findings indicate that expression of HLA–B27 in APCs may lead to a defect in IFN signaling, secondary to Socs3 induction. Given the critical role of the IFN pathway in the control of both inflammatory and regulatory responses by APCs, we postulate that this deficiency may have fundamental consequences and implications for SpA pathogenesis and treatment.

AUTHOR CONTRIBUTIONS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. AUTHOR CONTRIBUTIONS
  7. Acknowledgments
  8. REFERENCES
  9. Supporting Information

All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Breban had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study conception and design. Fert, Glatigny, Colbert, Taurog, Chiocchia, Araujo, Breban.

Acquisition of data. Fert, Glatigny, Letourneur, Jacques, Smith, Colbert, Araujo.

Analysis and interpretation of data. Fert, Cagnard, Letourneur, Jacques, Colbert, Chiocchia, Araujo, Breban.

Acknowledgments

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. AUTHOR CONTRIBUTIONS
  7. Acknowledgments
  8. REFERENCES
  9. Supporting Information

We thank Muriel Andrieu and Karine Labroquère (Institut Cochin) for technical support regarding fluorescence-activated cell sorting. We also thank Gaëlle Charlon, Ludovic Maingault, and Sabria Allithi (Institut Jacques Monod, Paris, France) for managing and genotyping the rat colonies.

REFERENCES

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. AUTHOR CONTRIBUTIONS
  7. Acknowledgments
  8. REFERENCES
  9. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. AUTHOR CONTRIBUTIONS
  7. Acknowledgments
  8. REFERENCES
  9. Supporting Information

Additional Supporting Information may be found in the online version of this article.

FilenameFormatSizeDescription
ART_38318_sm_SupplData.docx3599KSupplementary Data

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.