and Immunity (2007) 8, 590–603 & 2007 Nature Publishing Group All rights reserved 1466-4879/07 $30.00 www.nature.com/gene

ORIGINAL ARTICLE Genomic view of IFN-a response in pre-autoimmune NZB/W and MRL/lpr mice

Q Lu, N Shen, XM Li and SL Chen Joint Molecular Rheumatology Laboratory of Institute of Health Sciences and Shanghai RenJi Hospital, Shanghai Institutes for Biological Sciences, Chinese Academy of Sciences and Shanghai JiaoTong University School of Medicine, Shanghai, China

Interferon (IFN)-a is involved in the pathogenesis of systemic erythematosus. Studies in murine lupus models have revealed that type I IFN exerts either a protective effect in MRL/lpr, or can detrimentally impact disease progression, as in NZB/ W mice. To understand this paradox, we examined the kinetic global expression in pre-autoimmune NZB/W-, MRL/lpr- and normal BALB/c-derived splenic mononuclear cells following ex vivo IFN-a treatment. Analysis of IFN-a-induced patterns revealed genes associated with antiproliferative activity of IFN-a including CDKN1A, GADD45B, pituitary tumor-transforming 1, SCOTIN, ataxia telangiectasia-mutated homolog and calcyclin-binding were upregulated in MRL/ lpr and/or BALB/c mice. Of IFN-a-induced genes differentially expressed in NZB/W vs BALB/c and MRL/lpr mice at 3 h time point, enhanced expression of CCND1, D2, matrix metalloproteinase 13 and a panel of and chemokines and impaired expression of negative inflammatory regulators CD69 and an Src family kinase hemopoietic cell kinase were notable. Interestingly, the splenic mononuclear cells from the NZB/W not MRL/lpr lupus-prone mice at the pre-autoimmune stage before ex vivo IFN-a treatment, have increased expression of many known IFN-regulated genes. These results provide a unique genomic view of ex vivo IFN-a response in two lupus-prone models, and help to have an insight into the role of IFN-a in lupus pathogenesis Genes and Immunity (2007) 8, 590–603; doi:10.1038/sj.gene.6364421; published online 30 August 2007

Keywords: systemic lupus erythematosus; microarray gene expression analysis; NZB/W mice; MRL/lpr mice; IFN-a

Introduction anemia, anti-DNA autoantibodies, kidney disease and mortality.7 In contrast to normal BALB/c mice, Systemic lupus erythematosus (SLE) is a prototypic continuous in vivo treatment of IFN-a in pre-autoim- systemic autoimmune disease characterized by a break mune NZB/W mice from 8 weeks of age precipitates the of immune tolerance to self antigens, resulting in autoimmune process and kidney damage, leading to inflammation and damage to a range of organ systems. premature death from severe immune complex glomer- The exact pathoetiology of SLE remains elusive. Elevated ulonephritis. This result provides a direct evidence that serum levels of IFN-a have long been observed and IFN-a is implicated in the pathogenesis of SLE.8 Thus, correlated with disease activity. A series of studies using the type I IFNs have emerged as a dominant target in the microarray gene expression analysis in patient periph- pathogenesis of SLE. eral blood cells demonstrated an ‘IFN signature’ in SLE, However, a study of congenic lupus-prone MRL/ which correlated with renal and hematological involve- CD95lpr/lpr (MRL/lpr) mice, which lack the type I IFN ment of phenotypes.1–6 Supporting evidence has also or both type I and type II IFN receptors, arrived come from studies on murine lupus. The (New Zealand at countering conclusions. Deficiency of the type I IFN-R Black (NZB) Â New Zealand White (NZW))F1 hybrid surprisingly worsened lymphoproliferation, autoanti- female mice (NZB/W) develop a disease closely resem- body production and end-organ disease, thus demon- bling human SLE, characterized by the production of strating that type I IFN exerted a protective effect on antinuclear antibodies, severe glomerulonephritis and MRL/lpr mice.9 Initiating IFN-b treatment in MRL/lpr early mortality. In the NZB parental strain, which also mice with mild and advanced disease highly effective in develops a lupus-like syndrome, homozygous IFN-a/b prolonging survival and ameliorating the clinical (renal R-deleted NZB mice had significantly reduced antiery- function, proteinuria, and skin lesions), throcyte autoantibodies, erythroblastosis, hemolytic serologic (autoantibodies and cytokines) and histologic parameters of the lupus-like disease;10 however, IFN-a Correspondence: Professor N Shen, Joint Molecular Rheumatology has been demonstrated to play a role in the disease Laboratory of Institute of Health Sciences and Shanghai Renji pathogenesis in MRL/lpr mice by inducing the over- Hospital, Shanghai Institutes for Biological Sciences, Chinese expression of Fas on MRL/lpr lymphocytes and Academy of Sciences and Shanghai JiaoTong University School of on their spontaneous Fas-mediated cytotoxic potential, Medicine, 145 Shandong Mid. Road, 200001, Shanghai, China. E-mail: [email protected] which could be responsible for a chronic, nonantigen- Received 9 April 2007; revised 9 May 2007; accepted 20 July 2007; specific autoimmune attack on organs expressing low published online 30 August 2007 levels of Fas.11 MRL/lpr mice, in susceptible genetic Global analysis of lupus mice treated by IFN-a in vitro QLuet al 591 backgrounds, such as MRL, exhibit a lymphoprolifera- treatment relative to the untreated control culture in tive disorder that evolves into severe systemic auto- each strain were first assessed using BeadStudio soft- immunity due to defective Fas-induced death.12 The ware (Figure 1a), 2115, 1429 and 981 genes and expressed initiation of SLE is considered to be the effect of many sequence tags (ESTs) whose mRNA expression levels environmental triggers acting on individuals predis- changed significantly (Po0.01) after 3 h IFN-a stimula- posed towards SLE. IFN-a-based treatments are widely tion were identified in BALB/c-, MRL/lpr- and NZB/W- used for viral hepatitis and carcinoid tumors; however, derived splenic mononuclear cells, respectively. The several case reports have emerged describing autoim- expression of about half of these genes and ESTs changed mune conditions that have developed during IFN-a less than twofold in each strain. The expression that therapy.13–15 These observations in human and murine changed more than five, three and twofold was 205, 379 lupus implicate a complicated role of the type I IFN and 766, respectively, for BALB/c; 191, 330 and 584 for involved in lupus pathogenesis. We hypothesize that MRL/lpr; and 28, 161 and 418 for NZB/W mice. To inappropriate activation of type I interferon (IFN) increase the reliability of our microarray data and to signaling pathways, combined with a genetic predis- optimize the experimental procedure and validation position, may be important in SLE pathogenesis. To test method, genes that met both of the following criteria in this hypothesis and explore the molecular mediators three strains respectively were identified as IFN-a- underlying the acceleration of lupus phenotype in NZB/ regulated: (1) P-values o0.01 analyzed by the BeadStu- W not BALB/c mice by continuous in vivo IFN-a dio software for the difference in expression level relative treatment from 8 weeks of age (Mathian Alexis et al.8 to the untreated control; and (2) greater than a twofold study), we performed oligonucleotide microarray (Illu- change when compared with untreated control. We mina, San Diego, CA, USA) kinetic analysis of mRNA extracted a common set of 315 transcripts modulated in expression changes in the pre-autoimmune NZB/W and all three strains, 166 transcripts induced in both BALB/c MRL/lpr lupus-prone mice-derived splenic mononuc- and MRL/lpr mice, as well as a set of 288, 119 and 74 lear cells following ex vivo IFN-a treatment and using transcripts uniquely induced in BALB/c, MRL/lpr and BALB/c mice as a normal control. MRL/lpr mice exhibit NZB/W mice, respectively (Figure 1b). signs of B-cell activation and serum autoantibodies at Second, comparison of the difference in gene-expres- approximately 6 weeks of age and half the mice will die sion level between different strains were performed. The of glomerulonephritis by 6 months of age.16 In female following criteria were applied to generate genes that NZB/W mice, beginning at the age of B3–4 months, were differentially expressed between either two strains autoantibodies, including anti-DNA antibodies, are including NZB/W with respect to MRL/lpr; NZB/W developed and, by 7–8 months, immune complex with respect to BALB/c; and MRL/lpr with respect to deposits form in the kidneys. Mice die of the disease BALB/c and three subsets of genes which expressed within 1 year.17 In this study, MRL/lpr mice (females, 4–5 differentially in either strain with respect to the other two weeks of age), NZB/W mice (females, 8 weeks of age) strains at 3 h time point were extracted; (1) P-values and BALB/c mice (females, 8 weeks of age) were used. o0.01; (2) X1.5-fold difference in gene-expression level. For both lupus-prone murine models, serum IgG double- Of these differential-expressing genes between different stranded DNA antibody titers at the time of ex vivo IFN-a strains, we focused on the difference in expression level stimulation were assayed to assure these mice in the pre- of IFN-a-regulated genes. autoimmune state. Through the global analysis of The expression of these genes at the 6 and 24 h time signaling pathways, we observed that several genes points were tracked. Comparison of the gene expression implicated in signaling pathway and associated with at different time points in different strains was the antiproliferative activity of IFN-a were upregulated performed on normalized data. in MRL/lpr and BALB/c not NZB/W mice. By compar- ing the expression of IFN-stimulated genes (ISGs) in IFN-a-induced genes involved in signaling pathways signal intensities, a series of genes associated with There was a major subset of signaling molecules inflammation and proliferation were found to be over- increased concomitantly following IFN-a treatment in expressed in NZB/W mice. Notably, a significant IFN-a- all three strains, as well as a subset of signaling responsive gene-expression signature in the absence of molecules uniquely induced in the three strains. We IFN-a treatment in vitro was identified in pre-autoim- used the pathway analysis software CRSD (a compre- mune NZB/W mice at 8 weeks of age, but not in 4- to 5- hensive web server for composite regulatory signature week-old lpr and normal BALB/c mice. Of these genes, discovery) to investigate the global effect of IFN-a on the we propose that a panel of chemokines and cytokines, signaling pathways regulated in NZB/W and MRL/lpr especially CXCL10, as molecular mediators of IFN-a mice respectively. The top 17 signal pathways were action, might play an important role in disease develop- summarized in Table 1. The signaling pathways shared ment and exacerbation in genetically susceptible, pre- between the three strains with Po0.001 were mainly autoimmune NZB/W mice. composed of the Toll-like receptor signaling pathway, /signal transducers and activators of tran- scription (JAK/STAT) signaling pathway, IFN-a and Results MAPKinase signaling pathway, IL-2 receptor b chain in T-cell activation, network, antigen processing Profiling IFN-a-inducible gene expression in the pre- and presentation and regulation of transcriptional autoimmune NZB/W-, MRL/lpr- and BALB/c mice-derived activity by promyelocytic leukemia (PML). Several genes splenic mononuclear cells, respectively related to the Prion Pathway and Prion disease, antisense As we focused on the primary IFN-a response genes, pathway, Ahr pathway and Porphyr- changes in gene expression after 3 h ex vivo IFN-a in and chlorophyll metabolism were induced only in

Genes and Immunity Global analysis of lupus mice treated by IFN-a in vitro QLuet al 592

Figure 1 Gene-expression profiling of splenic mononuclear cells from pre-autoimmune NZB/W and MRL/lpr lupus-prone mice and normal BALB/c mice in response to 3 h IFN-a stimulation. (a) Splenic mononuclear cells from three strains were left untreated or stimulated with IFN-a for 3 h. RNA was prepared and processed for hybridization to the Sentrix Mouse-6 Expression BeadChip. Data were analyzed with the BeadStudio software, using the algorithm described in Materials and methods. The graphs showed genes with signal intensities changed X2- fold in response to 3 h IFN-a stimulation. For each comparison (3 h IFN-a stimulation relative to untreated in three strains respectively), the number of genes induced X2/3/5-fold is presented in the upper left corner and the number of genes repressed X2/3/5-fold is presented in the lower right corner. (b) The diagram indicates the number of overlapping IFN-a-regulated genes that were expressed in three strain- derived splenic mononuclear cells in response to 3 h IFN-a stimulation. Three hundred and fifteen transcripts were shared by all three strains; 166 transcripts induced in both BALB/c and MRL/lpr mice, as well as a set of 288, 119 and 74 transcripts induced uniquely in BALB/c, MRL/lpr and NZB/W mice, respectively.

BALB/c mice. Figure 2 illustrated a subset of genes NZB/W mice detected by microarray, the time course of involved in the above-mentioned signaling pathways. expression levels of IP10, XCL1, MIP1-a and IL-15 in The complex crosstalk among multiple signaling mole- three strains were assessed by real-time PCR and thereby cules and involved pathways underlying IFN-a treat- confirming the microarray data (Figure 3). ment were suggested. Of a panel of cytokines and chemokines induced in all three strains, MIP-1a and Effects on and death MCP-5 (CCL12), ATAC(XCL1), CXCL10 (IP10), IL15 and As auto-reactive T- and B-cell activation and prolifera- IL10 were significantly overexpressed in NZB/W vs tion play key roles in the development of the lupus MRL/lpr and BALB/c mice. It has been reported that syndrome, genes involved in and cell-cycle IFN-a priming confers a proinflammatory gain of regulation induced by IFN-a in the three strains were function effect on IL-10, leading to IL-10-induced summarized in Table 2. The common and unique- expression of IRF1, CXCL9 and CXCL10 that may induced proapoptotic genes in three strains mirrored contribute to its pathogenic role in SLE.18 Chemokines the activation of diverse apoptosis-related signaling MIP-1a (CCL3), MIP-1b (CCL4), RANTES (CCL5) and pathways. Analysis using CRSD software highlighted ATAC were known not only as chemoattractants but also several apoptosis-related signaling pathway headings, as coactivators of macrophages, constituting together including regulation of transcriptional activity by PML in with IFN-g a group of type 1 cytokines to act as a all three strains; the D4–GDI signaling pathway, role of functional unit, used by NK cells and CD8 þ T cells in the mitochondria in apoptotic signaling and caspase cascade immune defense.19 in apoptosis in NZB/W mice; and the ceramide signaling To validate further the enhanced proinflammatory pathway in BALB/c and lpr mice. The Fas signaling expression pattern induced by IFN-a predominantly in pathway was shared by NZB/W and BALB/c, but not

Genes and Immunity Global analysis of lupus mice treated by IFN-a in vitro QLuet al 593 Table 1 ISG expression patterns at 3 h

BALB/c MRLlpr NZB/NZW

Signal pathway P-value Signal pathway P-value Signal pathway P-value

Toll-like receptor signaling 2.18EÀ08 Toll-like receptor signaling 7.38EÀ11 IL-2 receptor b chain in T-cell 1.24EÀ06 pathway pathway activation 1 JAK–STAT signaling pathway 2.20EÀ08 JAK–STAT signaling pathway 2.10EÀ08 JAK–STAT signaling pathway 1.96EÀ06 IL-2 receptor b chain in T-cell 3.20EÀ07 MAPK signaling pathway 6.47EÀ08 Toll-like receptor signaling 3.40EÀ06 activation pathway Cytokine– 5.28EÀ07 IL-2 receptor b chain in T-cell 3.36EÀ06 IL-6 signaling pathway 4.23EÀ05 interaction activation Regulation of transcriptional 9.52EÀ06 Keratinocyte differentiation 1.15EÀ05 Keratinocyte differentiation 6.49EÀ05 activity by PML Keratinocyte differentiation 1.69EÀ05 p38 MAPK signaling pathway 7.63EÀ05 MAPK signaling pathway 7.08EÀ05 MAPK signaling pathway 2.19EÀ05 Oxidative stress-induced gene 8.26EÀ05 IFN-a signaling pathway 7.91EÀ05 expression via Nrf2 IL-6 signaling pathway 3.46EÀ05 IL-6 signaling pathway 8.26EÀ05 FAS signaling pathway ( CD95 ) 1.46EÀ04 Oxidative stress-induced gene 3.46EÀ05 Cytokine–cytokine receptor 9.65EÀ05 D4–GDI signaling pathway 2.32EÀ04 expression via Nrf2 interaction Ceramide signaling pathway 4.31EÀ05 Ceramide signaling pathway 9.87EÀ05 Stress induction of HSP 3.52EÀ04 regulation Apoptosis 4.84EÀ05 B-cell receptor signaling 1.07EÀ04 IL-7 signal transduction 4.24EÀ04 pathway Antisense pathway 4.89EÀ05 TPO signaling pathway 1.17EÀ04 Regulation of transcriptional 4.24EÀ04 activity by PML IL-10 anti-inflammatory 8.12EÀ05 IFN-a signaling pathway 1.32EÀ04 Role of mitochondria in 9.29EÀ04 signaling pathway apoptotic signaling Bone remodelling 1.08EÀ04 Antigen processing and 1.79EÀ04 Cytokine network 9.29EÀ04 presentation Cytokines and inflammatory 1.56EÀ04 Bone remodelling 4.77EÀ04 Caspase cascade in apoptosis 0.00106 response Cadmium induces DNA 1.79EÀ04 Neurodegenerative disorders 5.10EÀ04 IL12- and Stat4-dependent 0.00106 synthesis and proliferation in signaling pathway in Th1 macrophages development 17 Effect of METS on 2.79EÀ04 Apoptosis 6.11EÀ04 Cell adhesion molecules 0.00113 macrophage differentiation (CAMs)

Abbreviation: ISGs, IFN-stimulated genes. ISGs in three strains respectively were analyzed with the CRSD software based on KEGG and Biocarta data base. Shown are the top 17 signal pathways in three strains respectively. All genes upregulated by IFN-a involved in above signal pathways met the criteria of Po0.01 with more than twofold change in signal intensities (see details in Figure 2).

MRL/lpr mice, verifying the accuracy of this pathway apoptosis in response to cellular stress. PTTG1 was analysis system (Figure 2). identified as a transcriptional activator of the p53 Progression through the cell cycle is regulated by signaling pathway and causes p53-dependent and cyclin/cyclin-dependent kinase (CDK) complexes and -independent apoptosis.21,22 Ataxia telangiectasia-mu- CDK inhibitors. Cyclin D2 (CCND2) in complex with tated homolog as one of the master modulators of the CDK4 or CDK6 phosphorylates retinoblastoma family cell-cycle checkpoint signaling pathways are required for , thereby causing the de-repression of regulation of a wide variety of downstream proteins, 20 23 transcription to promote G1/S cell-cycle transition. including the tumor suppressor protein p53. Calcyclin- CCND2 was upregulated in three strains and CCND1 binding protein (SIP) as a ubiquitin E3 ligase component was upregulated in two lupus-prone mice, both of two of the p53-induced b-catenin degradation pathway D-type overexpressed in NZB/W mice. Schlafen functions as an essential checkpoint for cell-cycle family members SLFN1, SLFN2 and SLFN8 and retino- progression.24 CDKN1A binds to and inhibits the activity blastoma family retinoblastoma-like 1 (p107) are negative of cyclin-CDK2 or -CDK4 complexes to induce G1-phase cell-cycle regulators found to be upregulated in three arrest.25,26 GADD45-b negatively regulates cell growth by strains. Several genes related to the p53 signaling interacting with cdc2 and cyclin B, thus causing G2/M pathway were transcriptionally upregulated in BALB/c cell-cycle arrest.27 Both p21WAF1 and GADD45-b have and/or MRL/lpr, but not in NZB/W mice. These been appointed a functional role in mouse lupus included pituitary tumor-transforming 1 (PTTG1), ataxia development.28–30 CDKN1A and PTTG1 was selected telangiectasia-mutated homolog, calcyclin-binding pro- and the difference in fold induction between two lupus- tein (CACYBP) and the p53-targeted genes, cyclin- prone mice was confirmed by real-time PCR (Figure 3). dependent kinase inhibitor CDKN1A (p21WAF1), SCOTIN These data suggest that two lupus-prone models differ in and growth arrest and DNA-damage-inducible 45 b their response to the antiproliferative activity of IFN-a, (GADD45B). The tumor suppressor protein p53 is a well- which is interesting given the protective role that IFN-a known that induces growth arrest or may have in the MRL/lpr model.

Genes and Immunity Global analysis of lupus mice treated by IFN-a in vitro QLuet al 594

Figure 2 ISGs within the signal transduction network. IFN-a-inducible gene-expression profiles in the three strains were subjected to signaling pathway analysis using the CRSD software. The expression of ISGs involved in signaling pathways were visualized with the software HCE 3.0. The input data were transformed as described in Materials and methods. ISGs, IFN-stimulated genes.

ISGs differentially expressed at the 3 h time point in NZB/W CCND2 was, in part, induced by the ‘IL-2 receptor b vs MRL/lpr and BALB/c mice chain in T-cell activation signaling pathway’, which

In Figure 4a, among ISGs which was shared by three enables cells to progress from G1 to S phase to promote strains and overexpressed in NZB/W mice, CXCL10 was T-cell proliferation.32 notable with a 4.5-fold difference in signal intensities Cbp/p300-interacting transactivator CITED2, micro- compared to lpr and a 3.2-fold change compared to tubule-associated protein 7 (MTAP7), matrix metallopro- BALB/c at 3 h time point. Others mainly included IL-15, teinase-13 (MMP-13) and -rich motor neuron 1 CCND1, CCND2, cyclin-dependent kinase-like 2 (CRIM1) were induced and overexpressed in a unique (CDKL2), secretin (SCT), a gastrointestinal peptide also way in the NZB/W mice. MMP-13 is responsible for type recently substantiated as a neuron peptide. IL-15 II collagen degradation, which has an obvious role in the resembles IL-2 in its biological activities, stimulating joint tissue destruction that is a major feature of T-cell and NK-cell proliferation and activation as well as inflammatory joint diseases such as rheumatoid arthritis enhancing B-cell expansion and Ab production.31 and osteoarthritis.33 CITED2 interacts with the ubiqui-

Genes and Immunity Global analysis of lupus mice treated by IFN-a in vitro QLuet al 595

Figure 3 Comparison of gene expression detected on the microarray and using real-time PCR methods. Pooled RNA was prepared from splenic mononuclear cells derived from five of NZB/W, MRL/lpr and BALB/c mice, respectively, following time course of IFN-a treatment for 3, 6 and 24 h. Both the results of microarray and real-time PCR were presented as fold induction relative to the untreated BALB/c mice. tously expressed transcriptional co-activators CBP/p300, 23 (H2-T23), IFN-activated gene 203 (IFI203), IFN- which have the histone acetyltransferase activity and induced protein 44 (IFI44), CD69, lymphocyte antigen 6 participate in the activity of hundreds of different complex LY6C, 5-, cytosolic III (NT5C3) and transcription factors, including STATs.34,35 Both real-time membrane-spanning 4-domains, subfamily A, member PCR and microarray agreed, indicating a transient 4C (MS4A4C) were upregulated by IFN-a in all three upregulation pattern of expression unique in NZB/W strains, but displayed lower expression levels in NZB/W mice, which peaked at 3 h IFN-a stimulation for CITED2 mice compared to BALB/c and MRL/lpr mice at 3 h. (Figure 3). The glandular kallikrein gene family KLK5, CD69 was reported to play a key role in decline of KLK27 and kallikrein 1-related peptidase b4 (KLK1B4, autoimmune reactivity and inflammation by synthesis of NGFA), and IFN-induced transmembrane protein 2 TGF-b, using a model of collagen-induced arthritis (CIA) (IFITM2) were identified initially as primary responsive in WT and CD69-deficient mice.36 A member of the Src genes in BALB/c and MRL/lpr mice at the 3 h time family of tyrosine kinase HCK (hemopoietic cell kinase) point, however, maximal induction of KLK5, KLK27, was upregulated by IFN-a with 5.2-fold in BALB/c, 2.4- KLK1B4, IFITM2 were seen at 6 h in all three strains and fold in MRL/lpr and 1.5-fold in NZB/W mice, and shown significant overexpression in NZB/W mice. The overexpressed in BALB/c and MRL/lpr mice signifi- glandular kallikrein gene family encodes specific pro- cantly at 3 h time point comparing to NZB/W mice. teases involved in the processing of biologically active Using hck-/-fgr-/-mice, HCK has been demonstrated as peptides. OAS (20–50 oligoadenylate synthetases) are a a downregulator of myeloid cell chemokine signaling by group of involved in antiviral response. maintaining the tonic of paired-Ig-like Interestingly, expression of OAS1G and OAS1C (OASL2) receptor B (PIR-B).37 mRNA were induced by IFN-a stimulation at 3 h and peaked at 6 h in both lupus-prone mice but were ISGs differentially expressed at the 3 h time point in lpr vs unaffected in IFN-a-treated splenic mononuclear cells BALB/c and NZB/W mice from BALB/c mice. In Figure 4b, ISGs that were overexpressed in lpr vs Ubiquitin-specific protease 25 (Usp25), N- (and BALB/c and NZB/W mice could be grouped into five STAT) interactor (NMI), histocompatibility 2, T-region major families: (1) immune response (lymphocyte anti-

Genes and Immunity 596 ee n Immunity and Genes

Table 2 Comparison of response to apoptosis and antiproliferative action of IFN-a

Gene name Accession ID BALB/c MRL/lpr NZB/NZW

Fold induction by Signal intensity after 3 h Fold induction by Signal intensity after 3 h Fold induction by Signal intensity after 3 h IFN-a IFN-a stimulation IFN-a IFN-a stimulation IFN-a IFN-a stimulation

Apoptosis CASP4 NM_007609.1 4.9 1442.1 4.8 1285.6 2.8 1190.4 lblaayi flpsmc rae yIFN- by treated mice lupus of analysis Global CASP8 NM_009812.1 2.5 602.1 2.8 647.7 2.1 876.9 TNFSF10 NM_009425.1 26.9 51.2 17.2 63.6 14.4 53.3 DAXX NM_007829.1 8.4 965.6 9.0 887.5 2.8 907 CLIC4 XM_124389.1 3.3 3436 2.7 3159.9 2.5 3440.9 PARP14 NM_145481.1 15.5 525.3 19.9 388.5 2.4 470.6 PML NM_008884.2 6.1 2842.4 7.2 2111.7 2.7 2464.2 BCL2 NM_009741.2 2.3 256.6 2.3 220.1 2.3 237.1 SCOTIN NM_025858.1 2.7 8362.5 2.7 7379.8 1.5 6902.1 CASP1 NM_009807.1 2 1945.2 1.9 2108 1.6 1833.4 TNFRSF5 NM_170704.1 2.3 350.7a 2.1 322.9a 1.5 310.9a PIM2 NM_138606.1 2.5 272.4 1.8 157.4 1.5 205.6 TEGT NM_026669.2 2.1 2925.4 1.8 2602.2 1.3 2339.3

NSMAF NM_010945.1 2 350.5 2.5 373.8 1.3 411 QLu a

MYD88 NM_010851 2.8 347.2 2.7 310.4 1.4 263.3 vitro in SH3GLB1 NM_019464.1 1.8 248.7 2.0 344 1.4 185.9 al et GZMA NM_010370.1 5 200.3 1.4 66.6 1.8 42.1 TNFRSF6 NM_007987.1 2.3 76.1 2.1 7.7 1.9 70.6 EBAG9 NM_019480.3 2.6 162.7 1.8 151.1 1.4 122.8 CASP3 NM_009810.1 1.6 340.1 1.4 542.5 2.2 187.8

Cell proliferation CCND1 NM_007631.1 2.6 37.6 3.5 90.6 2.1 188.6 CCND2 NM_009829.2 4.0 632.2 5.2 709 2.5 1232.4 RBL1 NM_011249.1 4.9 387.1 5.1 401.2 3.8 373.5 SLFN1 NM_011407.1 5.3 1056.3 4.5 961.6 2.2 1045.5 SLFN2 NM_011408.1 2.2 1442.9 2.4 1778.6 2.1 1179.5 MAP3K8 NM_007746.1 2.7 158.5 2.1 146.9 2.4 176.9 CDKN1A NM_007669.2 2.9 215.1 2.0 267.2 1.5 264.5 ATM NM_007499.1 2.4 119.3 2.6 136 1.8 109.7 MAFK NM_010757.1 4.4 264.1 2.6 281.2 1.8 263.5 PTTG1 NM_013917.1 5.6 1070 3.2 1179.6 1.1 190.8 MYC NM_010849.2 2.8 71.5 2.6 55 1.4 96.2 CAMK2D NM_023813.2 2.0 56.4 2.2 98.3 1.5 65.2 FOS NM_010234.2 2.0 97.3 1.9 107.4 1.3 95.8 GADD45B NM_008655.1 2.4 722.3 1.5 390.5 1.3 415.4 UBE2C NM_026785.1 3.2 146.6 0.8 237.5 1.0 62.5 PLK2 NM_152804.1 2.2 120 1.2 74.4 0.7 20.6 CACYBP NM_009786.1 2.8 784.6 1.7 825.5 1.5 440.2 CUL2 AK004875 1.6 44.3 2.8 70 1.7 56.4

Gene expression was measured with oligonucleotide microarrays performed on normalized data. Data represent genes involved in apoptosis and cell-cycle regulation, whose expression was differentially induced with more than twofold by IFN-a and peaked at 3 h time point in splenic mononuclear cells of either strain with Po0.01 (BeadStudio software). a Peaked at 6 h stimulation. Global analysis of lupus mice treated by IFN-a in vitro QLuet al 597 a

b

c

Figure 4 IFN-a-regulated genes differentially expressed between three strains at 3 h time point. Genes were selected for this analysis if they met the following criteria: (1) identified as IFN-a-regulated genes in either strain in response to 3 h IFN-a stimulation; (2) X1.5-fold difference in gene-expression levels between three strains at 3 h time point and Po0.01: (a) IFN-a-regulated genes expressed differentially in NZB/W vs MRL/lpr and BALB/c mice; (b) IFN-a-regulated genes expressed differentially in MRL/lpr vs NZB/W and BALB/c mice; (c) IFN-a- regulated genes expressed differentially in BALB/c vs MRL/lpr and NZB/W mice. Hierarchical clustering analysis of the genes (rows) with the HCE 3.0 software was performed as described in Materials and methods. These genes on (Chr.) and loci (cM) involved in the biological process were also summarized.

Genes and Immunity Global analysis of lupus mice treated by IFN-a in vitro QLuet al 598 gen 6 complex LY6F, LY6D, IFN-induced with helicase C enza virus) resistance 1 (MX1), OAS1G, 20–50 oligoa- domain 1 (IFIH1)); (2) complement and coagulation denylate synthetase 2 (OAS2), , IFN- (complement factor B (H2-BF) and fibrinogen-like pro- inducible double-stranded RNA-dependent (PRKR), tein 2 (FGL2)); (3) metabolism (histidine decarboxylase STAT1, transporter 1, ATP-binding cassette, sub-family (HDC), transporter solute carrier family 25, member 12 B (TAP1), guanylate nucleotide-binding protein 1 (GBP1) (SLC25A12) and xanthine dehydrogenase (XDH)); (4) and IFN, a-inducible protein (G1P2), as well as a set of growth-regulated (SLFN4 and maternal embryonic transcription factors involved in the transcriptional kinase (MELK)); and (5) miscellaneous regulation of IFN-inducible genes, including STAT1, (Phosphatase INPP5B, transcription activator nucleo- STAT2, IRF1 and IRF7. The IFN regulatory factors IRF-1 some-binding protein 1 (NSBP1) and cysteine-type and IRF-7, which are known as transcriptional activators endopeptidase Rhesus blood group-associated A glyco- of the type I IFN genes, and Pit-1, Oct-1, and Oct-2 and protein (RHAG)). the Caenorhabditis elegans gene unc-86 (POU) domain, Serine peptidase inhibitor, clade A, member 3F class 2, transcription factor 1 (POU2f1), a transcriptional (SERPINA3F), SLFN8, guanylate nucleotide-binding repressor, were up and downregulated, respectively, protein (GBP) 6, Z-DNA-binding protein 1 (ZBP1), in NZB/W mice compared to lpr and BALB/c mice suppressor of cytokine signaling 3 (SOCS3), IFN-induced (Table 3). The upregulated expression of many IFN- protein 35 (IFI35) and tryptophanyl-tRNA synthetases responsive genes and transcriptional activators of the (WARs) showed lower expression levels in MRL/lpr type I IFN genes and ISGs in the splenic mononuclear compared to BALB/c and NZB/W mice at the 3 h time cells from pre-autoimmune NZB/W mice in the absence point; all of these genes were upregulated by IFN-a of IFN-a treatment, reflected the comparative activation treatment in all three strains. Amyotrophic lateral state of the IFN-a signaling pathway and an amplifica- sclerosis 2 (juvenile) homolog (human) (ALS2) and tion loop of IFN-a production in the pre-autoimmune histocompatibility 2 (H2)-Q7 were induced in MRL/lpr, NZB/W mice. To confirm the expression data from showing lower expression level compared to the other oligonucleotide microarray studies, IRF7, MX1 and two strains. ISG20 were chosen and the expression were analyzed by real-time PCR; both of the methods detected the ISGs differentially expressed at the 3 h time point in BALB/c vs similar expression patterns of expression (Figure 3). lpr and NZB/W mice In Figure 4c, ISGs overexpressed in BALB/c mice could be grouped into four major families: (1) immune Discussion response (GBP1, GBP5); (2) transcription factor (, CCHC domain-containing ZCCHC11, aryl hydro- We have used oligonucleotide arrays to study the IFN-a- carbon receptor nuclear translocator (ARNT) and growth induced changes in mRNA expression in splenic mono- arrest and DNA-damage-inducible 45 b(GADD45b)); nuclear cells from pre-autoimmune NZB/W and MRL/ (3) metabolism-regulated (phosphoglycerate mutase lpr lupus-prone mice to understand better the involve- 1(PGAM1), and carboxypeptidase A3, mast cell ment of IFN-a in SLE pathogenesis at the genomic level. (CPA3)); (4) apoptosis regulators (granzyme A (GZMA) Three hours of IFN-a stimulation differentially increased and heat-shock 70 kDa protein 1B (HSPA1B)); and (5) expression of many signaling molecules within the signal organization and biogenesis (thymo- transduction network, strongly suggesting the crosstalk sin, b 10 (TMSB10) and cysteine and glycine-rich protein among multiple signaling pathways, with a considerable 1 (CSRP1)); (6) miscellaneous (RNA polymerase 1–3 number of signaling molecules shared by all three strains (RPO1-3), cytotoxic T lymphocyte-associated protein 2 a examined here, thus reflecting the important and core (CTLA2A)). function of these genes and their associated signaling IFN-induced protein with tetratricopeptide repeats 3 pathways in modulating the function of IFN-a. (IFIT3), torsin A-interacting protein 2 (TORIAP2), gua- A major finding of this report is an IFN-a signature in nine nucleotide-binding protein, b 4(GNB4), gene model the splenic mononuclear cells of pre-autoimmune NZB/ 440(GM440), neuron-specific gene family member 2 W mice at 8 weeks age that is not established in 4- to 5- (NSG2) and macrophage-expressed gene 1(MPEG1) were week-old lpr or normal BALB/c mice in the absence of ex commonly induced in all three strains, but were vivo IFN-a treatment. Nine of 25 IFN-regulated genes expressed at a lower level in BALB/c compared to the that were increased significantly in NZB/W vs BALB/c other two strains. Cathepsin E (CSTE) and histocompat- and lpr mice prior to ex vivo IFN-a treatment were type I ibility 2, K region locus 1 (H2-K1) were induced in IFN-induced genes, including IRF7, MX1, MX2, IFIT2, BALB/c mice. H2-K1 showed significantly lower expres- IFIT4, PRKR, ISG20, G1P2 and OAS; 3 of 25, CXCL9, sion level compared to the other two strains. GBP2 and IGTP were preferentially induced by IFN-g and IFN-a may trigger the production of IFN-g.2,6 This IFN signature in pre-autoimmune NZB/W mice at 8 weeks may suggest that the primary driver for the IFN- of age regulated gene-expression signature in the splenic mono- Of the 311 transcripts that were identified as differen- nuclear cells of pre-autoimmune NZB/W mice before ex tially expressed genes in splenic mononuclear cells of vivo IFN-a treatment might be type I IFN. In accordance pre-autoimmune NZB/W mice at 8 weeks of age with a series of valuable microarray studies on human compared to 4 to 5-week-old lpr mice and 8-week-old lupus demonstrating an IFN-a signature in SLE, our data BALB/c mice in the absence of IFN-a treatment, most imply that increased expression of IFN-a-regulated genes notable were significantly upregulated IFN-responsive in the splenic mononuclear cells of pre-autoimmune genes with IFN-responsive elements, such as CXCL10, NZB/W mice before ex vivo IFN-a treatment may present GBP2, IFN regulatory factor 1 (IRF1), myxovirus (influ- in a much earlier, pre-clinical stage of lupus development

Genes and Immunity Global analysis of lupus mice treated by IFN-a in vitro QLuet al 599 Table 3 Upregulation of IFN-responsive genes in NZB/W mice comparing to MRL/lpr and BALB/c mice in the absence of IFN-a treatment

Gene name Accession ID Description Signal intensity

Untreated 3 h IFN-a stimulation

BALB/c MRL/lpr NZB/W BALB/c MRL/lpr NZB/W

Signaling molecules Irf7 NM_016850.1 Interferon regulatory factor 7 64.9 47.4 295.3 1114.2 912.1 939.3 Stat1 NM_009283.2 Signal transducer and activator of transcription 1 99.8 94.4 307.7 605.2 780.2 746 Stat2 NM_019963.1 Signal transducer and activator of transcription 2 29.8 32.4 90.7 257.5 270.5 290.1 Pou2f1 NM_011137.1 POU domain, class 2, transcription factor 1 241.2 283.9 90.5 266.8 166.2 103.7

Interferon-responsive genes with IFN-responsive elements Cxcl10 NM_021274 Chemokine (C-X-C motif) ligand 10 11.8 19.7 749.9 969.5 689.8 3113.1 Gbp2 NM_010260.1 Guanylate nucleotide-binding protein 2 77.6 99.9 337.5 1486.3 1141.7 900.3 Irf1 NM_008390.1 Interferon regulatory factor 1 195.9 162.2 422 779.8 617.1 659 Mx1 NM_010846.1 Myxovirus ( virus) resistance 1 8.1 À4.5 107.7 457.4 582.1 598 Oas1g NM_011852.2 2–5 oligoadenylate synthetase 1A À9.6 4.5 146.1 À5.6 436.4 643.5 Oas2 NM_145227.1 2–5 oligoadenylate synthetase 2 À0.6 À11.8 67.1 429.4 399.2 407.9 Stat1 NM_009283.2 Signal transducer and activator of transcription 1 99.8 94.4 307.7 605.2 780.2 746 Tap1 NM_013683.1 Transporter 1, ATP-binding cassette, subfamily B 64.4 56.9 123.5 235.1 253.3 266.7 Prkr NM_011163.2 Protein kinase, IFN-inducible dsRNA-dependent 75.5 51.8 319.3 1352.8 1276 1166.7 G1p2 NM_015783.1 Interferon, a-inducible protein 78.6 72.6 549.1 2208.8 2348.9 2362.8

Miscellaneous Ifitm2 NM_030694.1 Interferon-induced transmembrane protein 2 413.6 598.1 4119.9 534.3 615.9 5362.6 Ifitm1 NM_026820 Interferon-induced transmembrane protein 1 286.5 503.8 2104.6 2168.5 2767.3 3987.9 Ifi47 NM_008330.1 Interferon-ginducible protein 701.8 653.2 1931.6 6264.4 5580.5 5732.5 Ifit3 NM_010501.1 IFN-induced protein with tetratricopeptide repeats 3 18.8 10.7 488.6 1141 1571 2346.8 Ifi205 NM_172648 Interferon-activated gene 205 676.6 549.8 1422.2 4511.6 4435.9 4673.9 Ifi1 NM_008326.1 Interferon-inducible protein 1 355.1 243.8 1039.2 3789.3 2726.2 3225.9 Ifit2 NM_008332.2 IFN-induced protein with tetratricopeptide repeats 2 69.8 78.5 719.5 2906 2953.7 3298.2 Mx2 NM_013606 Myxovirus (influenza virus) resistance 2 41 7.7 200 1121.3 857.1 765.8 ISG20 NM_020583.4 Interferon-stimulated protein 75.4 130 333.6 1386.6 1204.8 1267.1 Socs1 NM_009896.1 Suppressor of cytokine signaling 1 75.1 62.2 196.5 1005 697.3 794.2 Daxx NM_007829.1 Fas death domain-associated protein 114.4 98.7 328.5 965.6 887.5 907 Igtp NM_018738.2 Interferon-g-induced GTPase 271 235.7 984.2 2552.2 2295.8 3162.3 Oasl1 NM_145209.2 2–5 oligoadenylate synthetase-like 1 35.4 9.9 161.9 601.2 531.8 485.3

These genes were selected because their difference in expression was significant (Po0.01) with X2-fold change in the splenic mononuclear cells of 8-week-old NZB/NZW mice as compared to the normal 8-week-old BALB/c and 4 to 5-week-old lpr mice. and require a genetically susceptible background. The signal transduction through the type I IFNAR. In turn, NZB/W and the MRL/lpr strain mice are two classic IFNAR signal transduction increases IRF7 expression, spontaneous lupus-prone models that share severe which is an IFN-a/b-inducible gene. Increased expres- systemic autoimmunity and immune complex-mediated sion and activation of IRF7 leads to enhanced expression fatal glomerulonephritis. Lupus susceptibility in the of IFN-b and many of the IFN-a genes. Accordingly, a NZB/W model is predominantly due to genetic interac- significant upregulation of TLR7, ISGs and transcrip- tions between alleles of H2z from NZB and H2d from tional activators of the type I IFN genes and ISGs (IRF7, NZW. In addition to the MHC loci, several lupus- IRF1, STAT1, STAT2) in the splenic mononuclear cells susceptibility loci derived from the NZB and NZW derived from preautoimmune NZB/W mice in the strains have been identified, including Sle1-Sle6, Nba1- absence of IFN-a treatment suggested that an amplifica- Nba4, Nwa1-Nwa2 and Lbw1-Lbw8.38–42 However, the tion loop of type I IFN production was in operation. lpr mutation in an MRL susceptibility background is a In this report, we have identified a set of significantly key element in the penetrance of disease pathogenesis in overexpressed genes induced by IFN-a in the splenic the MRL/lpr strain. mononuclear cells of NZB/W mice compared to BALB/c The significant overexpression of IP10 in the splenic and MRL/lpr mice after 3 h treatment including MMP-13 mononuclear cells of pre-autoimmune NZB/W mice in and CITED2. The cytokine-activated JAK–STAT pathway the absence of IFN-a treatment appears to be mediated, is perhaps the most extensively studied IFN-dependent at least in part, by a Toll-like receptor signaling pathway signaling pathway. In this pathway, cytokines bind to secondary to IFN-b production43,44 and is controlled at their cognate receptors and initiate a signal resulting the transcriptional level by the transcription factors IFN- in the activation of JAK and other signal transduction stimulated gene factor 3 complex (ISGF3), IRF3, nuclear and transcriptional activator proteins, leading to the factor-kB (NF-kB).45,46 The stimuli remain to be defined; synthesis of target genes. Activation of the JAK–STAT however, IFN-a was reported expressing at low levels in signaling pathway can be downregulated by the sup- normal or axenic mice.47 Rapid stimulation of the Toll- pressor of cytokine signaling (SOCS) proteins. MMP-13 is like receptor causes IRF3 activation, IFN-b synthesis and one of a group of zinc-dependent matrix metalloprotei-

Genes and Immunity Global analysis of lupus mice treated by IFN-a in vitro QLuet al 600 nases (MMPs). Many MMPs possess transcription factor- genomic level and may therefore gain a unique angle of binding motifs such as AP1, AP2, NF-kB, PEA3, Sp1 and insight into the genetic component of IFN-a involved in STAT elements in their 50-flanking regulatory DNA the pathogenesis of SLE. Study on the spontaneous sequences.48,49 As a member of the CITED family of lupus-predisposition models at the pre-autoimmune transcriptional co-activators, the promoter of CITED2 stage may permit characterization of immunologic contains multiple STAT-binding sites, consistent with its abnormalities before, during and after disease develop- responsiveness to different cytokines.50,51 In addition, ment. In this study, we have identified an IFN-a crosstalk between the IFN-a system and a member of the signature in pre-autoimmune NZB/W that is not found nuclear -peroxisome proliferator-acti- in MRL/lpr mice in the absence of IFN-a treatment, vated receptor a (PPARa) is suggested by CITED2, which suggesting that IFN-a signature may be an early is a known PPARa co-activator.52 The PPARs have been immunologic abnormality required by the genetic implicated in a wide range of biological processes, susceptibility background. At the genomic level, we including cell growth, lipid homeostasis, apoptosis and have observed different downstream effects on cell inflammation.53 Notably, an additional transcriptional proliferation and inflammation in the splenic mono- co-activator, NMI, which interacts with all STATs (except nuclear cells from two lupus-prone models by short-term STAT2) and potentiates their association with the co- in vitro exposure to IFN-a. These results confirmed the activators CBP/p30054 was upregulated in all three heterogeneous nature of SLE and suggested the require- strains by 3 h IFN-a stimulation and significantly over- ment of the genetic susceptibility background that expressed in MRL/lpr and BALB/c mice at 3 h time contributed to the effects of IFN-a on the pathogenesis point. Interestingly, NMI and CITED2 exhibited different of SLE. responses to IFN-a treatment in two lupus-prone strains. Dysregulation of chemokine and cytokine expression has a complex role in the pathogenesis of autoimmune Materials and methods diseases. It has been reported through a survey of the serologic proteome in human SLE that a set of IFN- Mice regulated inflammatory chemokines had highly co- Two NZB/W and MRL/lpr lupus-prone mouse models ordinated upregulation and their levels correlated were used for IFN-a-induced gene-expression profiling strongly with clinical and laboratory measures of disease studies. BALB/c mice were used as controls. Female activity, thus, may function as biomarkers for active MRL/lpr and normal BALB/c mice were purchased human SLE.55 In this study, the complex interactions of from the Shanghai Laboratory Animal Center, Chinese cytokines and chemokines underlying IFN-a treatment Academy of Science (Shanghai, China). Female NZB/W were suggested. Of a panel of chemokines and cytokines mice were obtained from the Model Animal Research overexpressed in NZB/W mice after IFN-a treatment in Center of Nanjing University (Nanjing, China). All mice vitro, the disparity in the basal and stimulated level of were maintained under specific pathogen-free conditions CXCL10 compared to the other two strains was striking. in an environmentally controlled clean room at the CXCL10 can be expressed locally in response to Animal Center of Medical College, Shanghai JiaoTong inflammatory and antigenic stimuli and may amplify University. Splenocytes were obtained from individual subsequent tissue reactions by virtue of activating and NZB/W, MRL/lpr and BALB/c mice, respectively, when selective chemotactic activities on activated CD4 þ both of NZB/W and BALB/c mice were 2 months of age lymphocytes, signaling through the chemokine receptor and MRL/lpr mice were 4–5 weeks of age. Sera were CXCR3. Therefore, CXCL10 might be a critical mediator collected for measurement of IgG antibodies to dsDNA. of CD4 þ lymphocyte migration and involved in CD4 þ T- cell-dependent immune-responses56 underlying the IFN- Preparation, culture and IFN-a treatment of splenic a treatment process in NZB/W mice. mononuclear cells The NZB/W model is generally considered to most Splenocytes from each mouse were obtained by gently closely reflect the properties of human SLE, in that this disrupting the between sintered glass slides in strain exhibits a strong female gender bias in suscept- RPMI 1640 medium (Gibco-BRL, Invitrogen Corp., ibility. This female gender bias is thought to be mediated, Carlsbad, CA, USA) containing 10% fetal calf serum in part, through estrogen and prolactin.57 Studies in male (Hyclone, Logan, UT, USA), supplemented with the and female NZB/W lupus-prone mice demonstrated that antibiotics penicillin (100 U mlÀ1) and streptomycin hyperprolactinemia induced by transplantation of pitui- (100 U mlÀ1). After contaminating RBCs were removed tary glands leads to disease exacerbation and early by treating cell suspensions with a 5 min exposure to 5 ml 58 mortality. Overexpression of human PTTG1 C-terminal per spleen of erythrocyte-lysing buffer (0.144 M NH4Cl, peptide, which contains PXXP motifs (proline–proline– 17 mM Tris pH 7.2), splenocytes were washed twice, and serine–proline), has been reported to suppress prolactin resuspended at 108 cells in 1–2 ml of phosphate-buffered expression in rat prolactin (rPRL)- and GH-secreting saline with 3% fetal calf serum and 2 mM ethylenedia- GH3 cells.59 The significant difference in expression level mine tetraacetic acid. Splenic mononuclear cells from of PTTG1 in NZB/W mice compared to lpr and BALB/c each mouse were isolated using NycoPrep 1.077 A (Axis- mice after 3 h IFN-a treatment in vitro might provide a Shield, Oslo, Norway) and independently incubated at potential subcellular therapeutic target for PRL hyperse- 2 Â 106 cells mlÀ1 of complete RPMI 1640 medium sup- cretion. plemented with 10% heat-inactivated fetal bovine serum In summary, using microarray technology, we have and antibiotics in the absence or presence of determined the ex vivo IFN-a response of splenic mono- 1 Â103 IU mlÀ1 of recombinant mouse IFN a A (PBL nuclear cells derived from pre-autoimmune NZB/W and Biomedical Laboratories, Piscataway, NJ, USA) for 3, 6 MRL/lpr lupus-prone mice and BALB/c mice at the and 24 h. The splenic mononuclear cells were harvested

Genes and Immunity Global analysis of lupus mice treated by IFN-a in vitro QLuet al 601 and cell viability was determined by trypan blue-dye computed from the background model characterizing the exclusion. chance that the target sequence signal was distinguish- able from the negative controls. The default setting of Total RNA extraction, complementary RNA preparation and ‘detection level’ is 40.99 (signal intensity 423.5). Fold hybridization changes for gene expression were calculated, and genes Total RNA from untreated splenic mononuclear cells and with X1.5-fold change in gene expression were selected cells stimulated for 3, 6 and 24 h with IFN-a from the for further characterization. three strains was extracted using the RNeasy mini kit In this microarray system, system controls cover every with on-column DNase digestion (Qiagen Inc., Valencia, aspect of the array experiment from the biological CA, USA). To avoid fluctuation between individuals, specimen (housekeeping controls), to sample labeling total RNA were pooled from five (NZB/W) and five (RNA spike), to hybridization (Cy3-Labeles hybridiza- (MRL/lpr) lupus-prone mice and five normal BALB/c tion controls, low/high stringency hybridization con- mice, at each time point (0, 3, 6 and 24 h), respectively. trols), to signal generation (Biotin control) and negative Biotinylated cRNA was prepared using the Illumina controls. TotalPrep RNA Amplification kit (Ambion Inc., Austin, Hierarchical clustering analysis and visualization in TX, USA) according to the manufacturer’s instruction tree view were performed in Hierarchical Clustering manual. Briefly, about 50 ng mlÀ1 total RNA in 11 ml Explore (HCE) 3.0 (Human-Computer Interaction Lab, nuclease-free water per sample was reverse-transcribed University of Maryland, College Park, MD, USA). Input into first-strand cDNA with an oligo(dT) primer bearing data were first standardized row by row (that is the a T7 promoter using Array-Script reverse transcriptase deviation from the mean was calculated and then the (RT). After the second-strand synthesis and clean-up, deviation was divided by s.d.) followed by hierarchical biotinylated cRNA was generated by in vitro transcrip- clustering (combination of clustering parameters of tion with T7 RNA polymerase supplemented with biotin Average Linkage Method and Personal Correlation uridine triphosphate. The Sentrix Mouse-6 Expression Coefficient Similarity and Distance Measure). BeadChip (Illumina Inc., San Diego, CA, USA) represents a set of six individual Genechip probe arrays, each Real-time PCR containing approximately 36 400 oligonucleotide probe Total RNA (100 ng) was processed directly to cDNA by sets representing a murine genome. The Sentrix Mouse-6 reverse transcription with Superscript II (Life Technolo- Expression BeadChips were then hybridized, washed gies, Carlsbad, CA, USA) according to the manufac- and scanned on an Illumina Sherlock Scanner according turer’s protocol in a total volume of 20 ml. All PCR to the manufacturer’s protocol at the National Engineer- primers (Table 4) were designed using software OLIGO 6 ing Center for Biochip in Shanghai. using published sequence data from the NCBI database. Primers were synthesized by Shanghai Sangon Biological Data analysis Engineering Technology. cDNA were amplified with the For the comparison between a group of samples (referred SYBR Premix Ex Taq (TaKaRa, Dalian). PCR amplifica- to as the condition group) to a reference group, tion reactions contained 0.5–1 ml cDNA, 2.5 ml of SyBR ‘Differential Expression assay’ was done using the premix ExTaq (2 Â ), 0.1 ml of Rox Reference Dye (50 Â ) ‘Illumina custom’ error model, the ‘rank invariant’ data and 0.1 ml of each of the specific primers. Primer normalization method and several filters including diff concentrations in the final volume of 5 ml were 200 nM, score, detection level and fold change which defaulted respectively. All reactions were performed in triplicate in by the BeadStudio package. P-values can be calculated an ABI PRISM 7700 system and the thermal cycling using the following approach: conditions were: 15 s at 951C followed by 40 cycles of 951C for 5 s, and 601C for 30 s. Purity of the amplified Diff Score ¼ 10sgn ðm À m Þ log ðPÞ cond ref 10 PCR products was determined by a heat-dissociation where P is the statistical confidence, and mcond and mref are protocol to enable detection of non-specific amplifica- the average intensity of condition and reference sample, tion. We used the expression of b-actin in splenic respectively; Genes with ‘DffScore 420’ were selected as mononuclear cells to normalize the expression data of P-values o0.01. Detection is defined as: 1—P-values each of the genes.

Table 4 Primers used for real-time polymerase chain reaction

Gene Primer 1 (50-30) Primer 2 (50-30)

XCL1 TGTCACCAAACGAGGACTAA AGTTTCAGCCATGTTCTTTC CXCL10 ACGAACTTAACCACCATCTTCC TCTTCCATATACAATGCAAAGCA IL-15 AGTGACTTTCATCCCAGTTG GAGCACGTTTCTTACTGTTTC CDKN1A CTGTCTTGCACTCTGGTGTC CGCTTGGAGTGATAGAAATC CITED2 CCGTTCTGGATCAGGAAAAGT CACTGACGACATTCCACACC MX1 CTGAGGGCTCTGGGTGT GTAACAATACCACTGCCTCTG IRF7 CCTGATCCTGGTGAAGCTGG TGGGAGTTGGGATTCTGAGTC CCL3 TTCTCTGTACCATGACACTCTGC CGTGGAATCTTCCGGCTGTAG PTTG1 GCTAGAAAGGAAGCGCAGAA TAACAAACAGGCGGCAATTC ISG20 GCAGTACTACAGCCGAGTGT TCGCTGAGAGATTTTGTAGAG

Primers were designed using software OLIGO6 using sequence data from the NCBI database.

Genes and Immunity Global analysis of lupus mice treated by IFN-a in vitro QLuet al 602 Acknowledgements interferon therapy for myeloproliferative disorders. Clin Immunol Immunopathol 1992; 65: 70–74. We thank HF Wang for assistance in data analysis. The 16 Hahn BH. Animal models of systemic lupus erythematosus. work was supported by grants from the Shanghai In: Wallace DJ, Hahn BH (eds). Dubois’ Lupus Erythematosus, Commission of Science and Technology (06JC14050), 6th edn. Lippincott Williams and Wilkins: Philadelphia, 2002, Shanghai leading Academic Discipline project (T0203), pp 339–388. Chinese Natural Science Foundation Grant (30471613), 17 Theofilopoulos AN. Murine models of lupus. In: Lahita RG China Postdoctoral Science Foundation (20060390634) (eds). Systemic Lupus Erythematosus. Churchill Livingston: New York, 1992, pp 121. and Shanghai Postdoctoral Science Foundation 18 Sharif MN, Tassiulas I, Hu Y, Mecklenbrauker I, Tarakhovsky (06R214144). A, Ivashkiv LB. IFN-a priming results in a gain of proin- flammatory function by IL-10: implications for systemic lupus erythematosus pathogenesis. J Immunol 2004; 172: 6476–6481. References 19 Dorner BG, Scheffold A, Rolph MS, Hu¨ ser MB, Kaufmann SHE, Radbruch A et al. MIP-1a, MIP-1b, RANTES, and ATAC/ 1 Hooks JJ, Moutsopoulos HM, Geis SA, Stahl NI, Decker JL, lymphotactin function together with IFN-g as type 1 cyto- Notkins AL. Immune interferon in the circulation of patients kines. Proc Natl Acad Sci USA 2002; 99: 6181–6186. with autoimmune disease. N Engl J Med 1979; 301: 5–8. 20 Dyson N. The regulation of E2F by pRB-family proteins. Genes 2 Baechler EC, Batliwalla FM, Karypis G, Gaffney PM, Ortmann Dev 1998; 12: 2245–2262. WA, Espe KJ et al. Interferon-inducible gene expression 21 Hamid T, Kakar SS. PTTG/securing activates expression of signature in peripheral blood cells of patients with severe p53 and modulates its function. Mol Cancer 2004; 3: 18. lupus. Proc Natl Acad Sci USA 2003; 100: 2610–2615. 22 Yu R, Heaney AP, Lu W, Chen J, Melmed S. Pituitary tumor 3 Bennett L, Palucka AK, Arce E, Cantrell V, Borvak J, transforming gene causes aneuploidy and p53-dependent and Banchereau J et al. Interferon and granulopoiesis signatures p53-independent apoptosis. J Biol Chem 2000; 275: 36502– in systemic lupus erythematosus blood. J Exp Med 2003; 197: 36505. 711–723. 23 Pusapati RV, Rounbehler RJ, Hong S, Powers JT, Yan M, 4 Han GM, Chen SL, Shen N, Ye S, Bao CD, Gu YY. Analysis of Kiguchi K et al. ATM promotes apoptosis and suppresses gene expression profiles in human systemic lupus erythema- tumorigenesis in response to Myc. Proc Natl Acad Sci USA tosus using oligonucleotide microarray. Genes Immun 2003; 2006; 103: 1446–1451. 4: 177–186. 24 Fukushima T, Zapata JM, Singha NC, Thomas M, Kress CL, 5 Kirou KA, Lee C, George S, Louca K, Peterson MG, Crow MK. Krajewska M et al. Critical function for SIP, a ubiquitin E3 Activation of the interferon-alpha pathway identifies a ligase component of the beta-catenin degradation pathway, subgroup of systemic lupus erythematosus patients with for thymocyte development and G1 checkpoint. Immunity distinct serologic features and active disease. Arthritis Rheum 2006; 24: 29–39. 2005; 52: 1491–1503. 25 el-Deiry WS, Harper JW, O’Connor PM, Velcul escu VE, 6 Crow MK, Kirou KA, Wohlgemuth J. Microarray analysis of Canman CE, Jackman J et al. WAF1/CIP1 is induced in

interferon-regulated genes in SLE. Autoimmunity 2003; 36: p53-mediated G1 arrest and apoptosis. Cancer Res 1994; 54: 481–490. 1169–1174.

7 Santiago-Raber ML, Baccala R, Haraldsson KM, Choubey D, 26 Matsuoka M, Tani K, Asano S. Interferon-alpha-induced G1 Stewart TA, Kono DH et al. Type-I interferon receptor phase arrest through up-regulated expression of CDK deficiency reduces lupus-like disease in NZB mice. J Exp inhibitors, p19Ink4D and p21Cip1 in mouse macrophage. Med 2003; 197: 777–788. Oncogene 1998; 16: 2075–2086. 8 Mathian A, Weinberg A, Gallegos M, Banchereau J, Koutou- 27 Zhan Q, Antinore MJ, Wang XW, Carrier F, Smith ML, Harris zov S. IFN-a induces early lethal lupus in preautoimmune CC et al. Association with Cdc2 and inhibition of Cdc2/Cyclin (New Zealand Black  New Zealand White) F1 but not in B1 kinase activity by the p53-regulated protein Gadd45. BALB/c mice. J Immunol 2005; 174: 2499–2506. Oncogene 1999; 18: 2892–2900. 9 Hron JD, Peng SL. Type I IFN protects against murine lupus. 28 Santiago-Raber ML, Lawson BR, Dummer W, Barnhouseb M, J Immunol 2004; 173: 2134–2142. Koundouris S, Wilson CB et al. Role of cyclin kinase inhibitor 10 Schwarting A, Paul K, Tschirner S, Menke J, Hansen T, p21 in systemic autoimmunity. J Immunol 2001; 167: 4067–4074. Brenner W et al. Interferon-b: a therapeutic for autoimmune 29 Goulvestre C, Chereau C, Nicco C, Mouthon L, Weill B, lupus in MRL-Faslpr mice. J Am Soc Nephrol 2005; 16: Batteux F. A mimic of p21WAF1/CIP1 ameliorates murine 3264–3272. lupus. J Immunol 2005; 175: 6959–6967. 11 Hadj-Slimane R, Chelbi-Alix MK, Tovey MG, Bobe P. An 30 Liu L, Tran E, Zhao YN, Huang YC, Flavell R, Lu BF. Gadd45b essential role for IFN-a in the overexpression of Fas ligand and Gadd45g are critical for regulating autoimmunity. J Exp on MRL/lpr lymphocytes and on their spontaneous Med 2005; 202: 1341–1347. Fas-mediated cytotoxic potential. J Interferon Cytokine Res 31 Waldmann TA. Targeting the interleukin-15/interleukin-15 2004; 24: 717–728. receptor system in inflammatory autoimmune diseases. 12 Watanabe-Fukunaga R, Brannan CI, Copeland NG, Jenkins Arthritis Res Ther 2004; 6: 174–177. NA, Nagata S. Lymphoproliferation disorder in mice ex- 32 Moon JJ, Rubio ED, Martino A, Krumm A, Nelson BH. plained by defects in Fas antigen that mediates apoptosis. Permissive role for phosphatidylinositol 3-kinase in the Stat5- Nature 1992; 356: 314–317. mediated expression of cyclin D2 by the interleukin-2 13 Ronnblom LE, Alm GV, Oberg K. Autoimmune phenomena in receptor. J Biol Chem 2004; 279: 5520–5527. patients with malignant carcinoid tumors during interferon- 33 Konttinen YT, Ainola M, Valleala H, Ma J, Ida H, Mandelin J alpha treatment. Acta Oncol 1991; 30: 537–540. et al. Analysis of 16 different matrix metalloproteinases (MMP- 14 Niewold TB, Swedler WI. Systemic lupus erythematosus 1–MMP-20) in the synovial membrane: different profiles in arising during interferon-alpha therapy for cryoglobulinemic trauma and rheumatoid arthritis. Ann Rheum Dis 1999; 58: vasculitis associated with hepatitis C. Clin Rheumatol 2005; 24: 691–697. 178–181. 34 Yahata T, de Caestecker MP, Lechleider RJ, Andriole S, Roberts 15 Wandl UB, Nagel-Hiemke M, May D, Kreuzfelder E, Kloke O, AB, Isselbacher KJ et al. The MSG1 non-DNA-binding Kranzhoff M et al. Lupus-like autoimmune disease induced by transactivator binds to the p300/CBP coactivators, enhancing

Genes and Immunity Global analysis of lupus mice treated by IFN-a in vitro QLuet al 603 their functional link to the Smad transcription factors. J Biol 47 Gresser I, Vignaux F, Belardelli F, Tovey MG, Maunoury MT. Chem 2000; 275: 8825–8834. Injection of mice with antibody to mouse interferon a/b 35 Bhattacharya S, Eckner R, Grossman S, Oldread E, Arany Z, decreases the level of 20–50 oligoadenylate synthetase in d’Eckner A et al. Cooperation of Stat2 and p300/CBP in peritoneal macrophages. J Virol 1985; 53: 221–227. signaling induced by interferon-a. Nature 1996; 383: 344–347. 48 Qian L, Liu Y, Sun HB, Yokota H. Systems analysis of matrix 36 Sancho D, Gomez M, Viedma F, Esplugues E, Gordon-Alonso metalloproteinase mRNA expression in skeletal tissues. Front M, Garcia-Lopez MA et al. CD69 downregulates autoimmune Biosci 2002; 7: a126–a134. reactivity through active transforming growth factor-b pro- 49 Li WQ, Dehnade F, Zafarullah M. Oncostatin M-induced duction in collagen-induced arthritis. J Clin Invest 2003; 112: matrix metalloproteinase and tissue inhibitor of metallopro- 872–882. teinase-3 genes expression in chondrocytes requires 37 Zhang H, Meng F, Chu CL, Takai T, Lowell CA. The Src family Janus kinase/STAT signaling pathway. J Immunol 2001; 166: kinases Hck and Fgr negatively regulate neutrophil and 3491–3498. dendritic cell chemokine signaling via PIR-B. Immunity 2005; 50 Sun HB, Zhu YX, Yin T, Sledge G, Yang YC. MRG1, the 22: 235–246. product of a melanocyte-specific gene related gene, is a 38 JØrgensen TN, Gubbels MR, Kotzin BL. Links between type I cytokine-inducible transcription factor with transformation interferon and the genetic basis of disease in mouse lupus. activity. Proc Natl Acad Sci USA 1998; 95: 13555–13560. Autoimmunity 2003; 36: 491–502. 51 Leung MK, Jones T, Michels CL, Livingston DM, Bhattacharya 39 Vyse TJ, Drake CG, Rozzo SJ, Roper E, Izui S, Kotzin BL. S. Molecular cloning and chromosomal localization of the Genetic linkage of IgG autoantibody production in relation to human CITED2 gene encoding p35srj/Mrg1. Genomics 1999; lupus nephritis in New Zealand hybrid mice. J Clin Invest 61: 307–313. 1996; 98: 1762–1772. 52 Tien ES, Davis JW, Vanden Heuvel JP. Identification of the 40 Morel L, Rudofsky UH, Longmate JA, Schiffenbauer J, CREB-binding protein/p300-interacting protein CITED2 as a Wakeland EK. Polygenic control of susceptibility to murine peroxisome proliferator-activated receptor a coregulator. J Biol systemic lupus erythematosus. Immunity 1994; 1: 219–229. Chem 2004; 279: 24053–24063. 41 Mohan C, Alas E, Morel L, Yang P, Wakeland EK. Genetic 53 Delerive P, Fruchat JC, Staels B. Peroxisome proliferator- dissection of SLE pathogenesis Sle 1 on murine 1 activated receptors in inflammation control. J Endocrinol 2001; leads to a selective loss of tolerance to H2A/H2B/DNA 169: 453–459. subnucleosomes. J Clin Invest 1998; 101: 1362–1372. 54 Zhu M, John S, Berg M, Leonard WJ. Functional association of 42 Kono DH, Burlingame RW, Owens DG, Kuramochi A, Balderas Nmi with Stat5 and Stat1 in IL-2- and IFNgamma-mediated RS, Balomenos D et al. Lupus susceptibility loci in New Zealand signaling. Cell 1999; 96: 121–130. mice. Proc Natl Acad Sci USA 1994; 91: 10168–10172. 55 Bauer JW, Baechler EC, Petri M, Batliwalla FM, Crawford D, 43 Kawai T, Takeuchi O, Fujita T, Inoue JI, Mu¨ hlradt PF, Sato S Ortmann WA et al. Elevated serum levels of interferon- et al. Lipopolysaccharide stimulates the MyD88-independent regulated chemokines are biomarkers for active human pathway and results in activation of IFN-regulatory factor 3 systemic lupus erythematosus. PLoS Med 2006; 3: 2274–2284. and the expression of a subset of lipopolysaccharide-inducible 56 Loetscher M, Gerber B, Loetscher P, Jones SA, Piali L, Clark- genes. J Immunol 2001; 167: 5887–5894. Lewis I et al. Chemokine receptor specific for IP10 and Mig: 44 Toshchakov V, Jones BW, Perera PY, Thomas K, Cody MJ, structure, function, and expression in activated T-lympho- Zhang S et al. TLR4, but not TLR2, mediates IFN-b-induced cytes. J Exp Med 1996; 184: 963–969. STAT1a´/b-dependent gene expression in macrophages. Nat 57 Elbourne KB, Keisler D, McMurray RW. Differential effects of Immunol 2002; 3: 392–398. estrogen and prolactin on autoimmune disease in the NZB/W 45 Wu C, Ohmori Y, Bandyopadhyay S, Sen G, Hamilton T. mouse model of SLE. Lupus 1998; 7: 420–427. Interferon-stimulated response element and NFB sites co- 58 McMurray R, Keisler D, Kanuckel K, Izui S, Walker S. operate to regulate double-stranded RNA-induced transc. Prolactin influences autoimmune disease activity in the female J Interferon Res 1994; 6: 357–363. NZB/W mouse. J Immunol 1991; 147: 3780–3787. 46 Nakaya T, Sato M, Hata N, Asagiri M, Suemori H, Noguchi S 59 Horwitz GA, Miklovsky I, Heaney AP, Ren SG, Melmed S. et al. Gene induction pathways mediated by distinct IRFs Human pituitary tumor-transforming gene (PTTG1) motif during viral infection. Biochem Biophys Res Commun 2001; 283: suppresses prolactin expression. Mol Endocrinol 2003; 17: 1150–1156. 600–609.

Supplementary Information accompanies the paper on Genes and Immunity website (http://www.nature.com/gene)

Genes and Immunity