and Immunity (2003) 4, 177–186 & 2003 Nature Publishing Group All rights reserved 1466-4879/03 $25.00 www.nature.com/gene Analysis of expression profiles in human systemic lupus erythematosus using oligonucleotide microarray

G-M Han1, S-L Chen2, N Shen2,SYe2, C-D Bao2 and Y-Y Gu2 1Department of Rheumatology, Nanjing First Hospital, Nanjing Medical University, China; 2Shanghai Institute of Rheumatology, Second Medical University, Shanghai, China

Epidemiologic studies suggest a strong genetic component for susceptibility to systemic lupus erythematosus (SLE). To investigate the genetic mechanism of pathogenesis of SLE, we studied the difference in of peripheral blood cells between 10 SLE patients and 18 healthy controls using oligonucleotide microarray. When gene expression for patients was compared to the mean of normal controls, among the 3002 target genes, 61 genes were identified with greater than a two- fold change difference in expression level. Of these genes, 24 were upregulated and 37 downregulated in at least half of the patients. By the Welch’s ANOVA/Welch’s t-test, all these 61 genes were significantly different (Po0.05) between SLE patients and normal controls. Among these genes with differential expression, IFN-o and Ly6E (TSA-1/Sca-2) may play an important role in the mechanism of SLE pathogenesis. TSA-1 antigens may represent an important alternative pathway for T-cell activation that may be involved in IFN-mediated immunomodulation. Hierarchical clustering showed that patient samples were clearly separated from controls based on their gene expression profile. These results demonstrate that high-density oligonucleotide microarray has the potential to explore the mechanism of pathogenesis of systemic lupus erythematosus. Genes and Immunity (2003) 4, 177–186. doi:10.1038/sj.gene.6363966

Keywords: systemic lupus erythematosus; oligonucleotide microarray; gene expression profile

Introduction providing a format for identifying both alterations in individual genes and changes in transcriptional activity Systemic lupus erythematosus (SLE) is a chronic auto- on a whole genome scale.11,12 Microarrays may serve as immune disease with extremely heterogeneous clinical targets for hybridization of cDNA probes prepared from features, varying in pathogenicity from mild forms of the mRNA sample derived from cells or tissues, thus allowing disease to those that advance relentlessly leading to the analysis of differential gene expression. This approach progressive end organ damage. The etiopathogenesis has been used to obtain expression profile of complex of SLE is complex, involving both environmental and diseases and discover novel disease-related genes.13–16 genetic factors.1 Epidemiological studies suggest a In this paper, we have studied the difference in gene strong genetic component for susceptibility to SLE2,3 expression of peripheral blood white cells between SLE and multiple genes including those that affect immune patients and healthy controls using the oligonucleotide complex deposition, and the MHC have been implicated microarray in order to investigate the mechanism of in pathogenesis.4–6 Several whole-genome linkage stu- pathogenesis of systemic lupus erythematosus. dies have also been reported suggesting about 20 SLE susceptibility loci.7–10 Together, these studies suggest that SLE is a polygenic disorder with contributions from Results multiple genes, each one of which has a modest effect. Differential gene expression in SLE Gene expression studies in SLE may complement linkage and association studies. With the To identify genes expressed differentially between SLE sequence project nearly completed, most of the cDNA patients and normal controls, three selection criteria were sequences and physical locations of expressed human applied to each of the 3002 genes analyzed. (1) In total, genes are available. This is a critical resource for these 489 genes had intensities that were at least 10-fold (row expression approaches. New approaches in functional data 15 000/1500) greater than background in one or genomics, such as cDNA/oligonucleotide array, allow for more sample. (2) Out of 489 genes, 189 demonstrated such analysis. cDNA/oligonucleotide microarray allows 2.0-fold higher or lower expression in at least half of SLE monitoring of tens of thousands of genes and, eventually, patients than in the mean of a set of normal controls all expressed genes in the genome simultaneously, thus (NC1, NC2, and NC3). (3) Welch’s ANOVA/Welch’s t-test applied to the whole data set regardless of fold change, identified 602 genes that differed significantly between SLE patients and normal controls. A total of 61 genes met these three criteria. Correspondence: Dr G-M Han, Department of Rheumatology of Hospital Affiliated to Xuzhou Medical College, Xuzhou 221002, China. Of these 61 genes that were differentially expressed, 24 E-mail: [email protected] or hanguangmingdoctor@- were upregulated and 37 were downregulated (Tables 1 yahoo.com.cn and 2). Although most of these genes function in a ee n Immunity and Genes 178

Table 1 Genes upregulated in SLE patients

GenBank Noa Gene description (symbol) NC1 NC2 NC3 SLE1 SLE2 SLE3 SLE4 SLE5 SLE6 SLE7 SLE8 SLE9 SLE10 (ID) ratiob ratio ratio ratio ratio ratio ratio ratio ratio ratio ratio ratio ratio

AJ225089 10 20-50-Oligoadenylate synthetase-like 1.178 0.696 1.126 5.854 6.933 4.342 4.191 18.713 8.357 11.92 22.816 11.793 14.54 (OASL) U81800 10 Solute carrier family 16 (monocarboxylic 0.86 0.972 1.167 2.208 9.428 14.93 21.706 35.914 22.799 13.322 8.386 2.575 30.332 acid transporters), member 3 (SLC16A3) AF008442 10 RNA polymerase 1 subunit (RPA40) 0.896 0.923 1.181 2.086 2.482 2.007 2.503 4.765 2.793 8.102 43.818 2.463 14.546 M24594 10 Interferon-induced 56 (IFIT1) 1.263 0.699 1.038 8.9 7.128 4.521 3.003 9.648 6.582 9.223 24.305 18.791 10.947 X02669 9 Interferon, omega l (IFNWl) 0.628 0.749 1.623 10.83 4.888 2.574 0.595 18.645 7.325 16.134 148.794 6.622 46.146 M31165 9 Tumor necrosis factor, alpha-induced 1.087 0.693 1.22 10.08 13.871 10.522 1.733 16.656 19.273 19.878 20.501 24.247 9.775

protein 6 SLE human of profiles expression Gene (TNFAIP6) U56145 9 Lymphocyte antigen 6 complex, 1.077 0.893 1.03 1.015 12.615 4.503 10.156 10.116 11.161 21.796 24.217 19.857 24.686 E(LY6E) J05070 9 Matrix metalloproteinase 9 (gelatinase B, 0.667 0.607 1.727 0.815 11.215 5.319 5.621 4.116 9.35 16.935 5.801 11.7 14.455 92 kDa gelatinase, 92 kDa type-IV collagenase)

(MMP9) Han G-M M17115 8 Interleukin 3 (colony-stimulating factor, 1.048 0.425 1.527 11.72 3.883 1.557 1.323 12.495 4.522 17.864 139.203 4.38 43.202 multiple) (IL3) 0 0 M87434 8 2 -5 -Oligoadenylate synthetase 2 0.951 0.871 1.179 0.593 3.137 1.726 6.069 11.26 10.857 4.463 14.094 17.734 13.758 al et (OAS2) U19251 8 Baculoviral IAP repeat-containing 1.346 0.788 0.866 1.999 5.542 2.091 2.835 1.508 2.146 3.748 4.778 6.537 4.792 (BIRC1) X04371 7 20,50-Oligoadenylate synthetase 1 (OAS1) 0.757 0.498 1.744 1.381 3.682 1.556 8.99 12.643 6.952 1.1 31.143 18.552 13.488 AF032387 7 Small nuclear RNA activating complex, 1.212 0.888 0.9 4.361 6.527 5.592 1.868 1.269 2.483 1.829 2.926 5.554 2.135 polypeptide 4, 190 kDa (SNAPC4) X52015 7 Interleukin 1 receptor antagonist (ILlRN) 1.124 0.835 1.042 2.446 1.416 0.865 0.541 4.271 5.373 4.442 3.005 2.112 3.854 AF083470 7 Interferon-induced protein with 1.294 0.85 0.857 1.626 29.817 0.531 5.062 1.158 36.433 13.294 17.208 40.827 32.701 tetratricopeptide repeats 4, between WI- 10247 and WI- 6075 (IFIT4) X59770 7 Interleukin 1 receptor, type II 1.187 0.595 1.218 0.913 6.705 9.667 29.444 3.812 1.308 11.051 0.54 22.987 6.273 U43672 6 Interleukin 18 receptor 1 (IL18R1) 1.502 0.708 0.79 1.963 8.02 20.441 1.742 1.228 2.785 1.061 32.986 4.451 2.907 AF057160 6 ADP-ribosyltransferase (NAD+; poly 0.808 1.057 1.135 1.297 3.09 2.752 1.802 1.793 1.805 3.448 4.004 3.35 2.508 (ADP-ribose) polymerase)-like l (ADPRTLl) U70981 6 Interleukin 13 receptor, alpha 2 1.126 1.017 0.857 1.211 0.975 1.35 1.969 3.958 2.241 2.589 8.581 6.048 3.441 (IL13RA2) X82434 6 Emerin (Emery–Dreifuss muscular 1.027 1.159 0.814 3.989 1.455 1.487 1.576 2.672 2.16 3.819 6.224 2.29 1.93 dystrophy) (EMD) Ml4660 6 Interferon-induced protein 54 (IFIT2) 1.149 0.998 0.853 1.103 0.678 1.097 1.36 4.415 2.59 4.212 9.548 5.793 3.468 AF051151 5 Toll-like receptor 5 (TLR5) 1.229 0.923 0.848 1.269 1.204 0.955 2.786 1.808 3.696 1.616 4.367 2.295 6.045 U32849 5 N-myc (and STAT) interactor (NMI) 0.856 1.118 1.026 0.85 3.002 2.61 1.421 1.304 2.506 1.879 3.134 2.549 1.723 M98045 5 Folylpolyglutamate synthase (FPGS) 0.973 0.927 1.1 1.878 1.667 1.715 2.228 2.525 1.609 1.075 6.633 4.627 2.766

a No: the number of the patients whose ratios are highter than 2.0. b Ratio: normalize all of the samples to the mean of a set of control samples (NCl, NC2, and NC3). Table 2 Genes downregulated in SLE patients

GenBank (ID) NOa Gene description (symbol) NC1 ratiob NC2 ratio NC3 ratio SLE1 ratio SLE2 ratio SLE3 ratio SLE4 ratio SLE5 ratio SLE6 ratio SLE7 ratio SLE8 ratio SLE9 ratio SLE 10 ratio

M63509 9 Glutathione S-transferase M2 1.248 0.927 0.825 0.414 0.458 0.598 0.156 0.167 0.08 0.208 0.4 0.259 0.143 (GSTM2) U69668 9 Translocated promoter region (to 1.102 0.889 1.008 0.142 0.34 0.352 0.871 0.211 0.096 0.317 0.209 0.3 0.265 activated MET oncogene)(TPR) U02619 9 General transcription factor IIIC, 1.422 0.71 0.868 0.172 0.371 0.788 0.17 0.005 0 0.118 0.106 0.1 0.114 polypeptide 1 (alpha subunit, 220 kDa )(GTF3C1) AJ007314 9 Hypermethylated in 0.913 1.165 0.922 0.022 0.488 0.252 0.801 0.127 0.053 0.232 0.095 0.09 0.204 1(HIC1) U62891 8 dUTP pyrophosphatase 1.265 1.134 0.601 0.096 0.689 0.678 0.317 0.062 0.121 0.314 0.254 0.236 0.311 AF058056 8 Solute carrier family 16 1.314 0.878 0.808 0.635 0.491 1.255 0.414 0 0.001 0 0.175 0.141 0.025 (monocarboxylic acid transporters), member 7(SLC16A7) X60489 8 Eukaryotic elongation 0.948 0.785 1.267 0.039 0.286 0.513 0.551 0.268 0.089 0.463 0.061 0.133 0.395 factor 1 beta 2(EEF1B2) X59870 8 Transcription factor 7 (T-cell 1.05 0.85 1.101 0.148 0.543 0.327 0.326 0.235 0.069 0.746 0.068 0.234 0.399 specific, HMG-box)(TCF7) L05148 8 Zeta-chain (TCR) associated 0.714 0.871 1.415 0.007 0.407 0.677 0.72 0.117 0.015 0.231 0.149 0.368 0.275 protein kinase (70 kDa)(ZAP70) M26062 8 Interleukin 2 receptor, 0.88 0.894 1.226 0.041 0.288 0.838 0.253 0.119 0.004 0.756 0.245 0.319 0.269 beta(IL2RB) L31581 8 Chemokine (C-C motif) receptor 0.881 0.913 1.206 0.227 0.324 0.745 0.78 0.405 0.156 0 0.145 0.126 0.135

7(CCR7) SLE Han human G-M of profiles expression Gene M83653 8 Acid phosphatase 1, 0.84 0.99 1.17 0.12 0.414 0.46 0.816 0.428 0.232 1.494 0.112 0.332 0.472 soluble(ACPl)

AF108945 7 Signal peptidase complex 1.266 0.935 0.799 0.145 0.423 0.677 1.347 0.336 0.193 0.409 0.359 0.59 0.438 al et (18 kDa)(SPC18) M22815 7 Myosin, light polypeptide 2, 1.175 0.844 0.981 0.44 1.424 1.095 1.07 0.007 0 0 0 0 0 regulatory, cardiac, slow(MYL2) AF082516 7 Imidazoline receptor candidate 1.329 0.836 0.835 0.277 0.501 0.714 0.707 0.212 0.07 0.419 0.248 0.269 0.263 (1-1) AF072242 7 Methyl-CpG binding domain 1.034 0.857 1.109 0.163 0.437 0.573 1.364 0.117 0.258 0.254 0.362 0.468 0.617 protein 2(MBD2) M29065 7 Heterogeneous nuclear 0.793 1.206 1.001 0.115 0.441 0.399 0.498 0.213 0.386 0.646 0.36 0.566 0.735 ribonucleoprotein A2/ B1(HNRPA2B1) Y17975 7 ATPase, H+ transporting, 1.016 0.895 1.089 0.092 0.419 0.604 2.474 0.256 0.338 0.417 0.25 0.462 0.669 lysosomal (vacuolar proton pump) membrane sector associated protein M8- 9(APT6M8-9) U00672 7 Interleukin 10 receptor, 0.897 0.777 1.326 0.137 0.478 0.405 0.442 0.309 0.592 1.245 0.148 0.271 0.62 alpha(ILl0RA) U62962 7 initiation 1.115 0.936 0.949 0.137 0.472 0.401 0.915 0.414 0.134 0.679 0.134 0.298 0.502 factor 3, subunit 6 ee n Immunity and Genes (48 kDa)(EIF3S6) X1 5260 7 T-cell receptor, delta 0.942 1.035 1.023 0.01 0.159 0.523 0.257 0.151 0.234 0.615 0.111 0.242 0.516 (V,D,J,C)(TCRD) M92303 7 Calcium channel, voltage- 0.922 0.608 1.47 0.64 1.592 2.432 0.449 0.003 0.011 0.012 0.006 0.01 0.014 179 ee n Immunity and Genes 180

Table 2 (continued)

GenBank (ID) NOa Gene description (symbol) NC1 ratiob NC2 ratio NC3 ratio SLE1 ratio SLE2 ratio SLE3 ratio SLE4 ratio SLE5 ratio SLE6 ratio SLE7 ratio SLE8 ratio SLE9 ratio SLE 10 ratio

dependent, beta 1 subunit(CACNBl) M12959 7 T-cell receptor, alpha 0.861 0.963 1.176 0.064 0.606 0.701 0.586 0.228 0.177 0.271 0.045 0.172 0.24 (V,D,J,C)(TCRA) Y00711 6 Lactate dehydrogenase B(LDHB) 1.115 0.942 0.942 0.045 0.521 0.799 1.133 0.259 0.273 0.61 0.172 0.467 0.429 M28699 6 Nucleophosmin (nucleolar 1.002 0.786 1.212 0.144 0.587 0.944 0.699 0.387 0.368 0.545 0.163 0.428 0.464 phosphoprotein B23, numatrin)(NPMl) AF028832 6 Heat shock 90 kDa protein 1, 1.072 0.83 1.098 0.187 0.685 0.729 1.463 0.439 0.378 0.111 0.22 0.433 0.813 alpha(HSPCA)

M62880 6 Integrin, beta 7(ITGB7) 1.122 1.07 0.808 0.061 1.058 0.831 0.832 0.141 0.019 0.692 0.141 0.15 0.122 SLE human of profiles expression Gene M22538 6 NADH dehydrogenase 0.962 0.972 1.067 0.068 0.458 0.769 1.38 0.381 0.355 0.788 0.208 0.246 0.655 (ubiquinone) flavoprotein 2(24 kDa)(NDUFV2) M32721 6 ADP-ribosyltransferase (NAD+; 0.95 0.864 1.187 0.133 0.637 0.82 1.024 0.347 0.333 0.59 0.033 0.435 0.37 poly (ADP-ribose) polymerase)(ADPRT) X69295 6 msh (Drosophila) homeo box 1.121 1.108 0.771 0.409 0.644 0.591 0.479 0.156 0.152 0.288 0.667 0.84 0.151 homolog 2(MSX2) Han G-M AF015608 6 RNA-binding protein S1, serine- 0.757 1.054 1.189 0.474 0.869 0.628 0.258 0.053 0.28 0.6 0.314 0.645 0.289 rich domain(RNPSl) tal et U57099 6 Nuclear protein, marker for 1.286 0.854 0.86 0.221 0.785 1.262 0.796 0.037 0.04 0.166 0.872 0.291 0.084 differentiated aortic smooth muscle and downregulated with vascular injury(APEGl) U12255 5 Fc fragment of IgG, receptor, 1.076 0.814 1.11 0.115 0.417 0.755 0.845 0.351 0.363 0.287 0.736 0.578 0.695 transporter, alpha(FCGRT) D43682 5 Acyl-coenzyme A 0.93 0.908 1.162 0.182 0.97 0.973 0.833 0.296 0.363 0.789 0.33 0.466 0.868 dehydrogenase, very long chain(ACADVL) AF004709 5 -activated protein kinase 1.062 0.901 1.036 0.779 0.831 1.234 0.839 0.194 0.118 0.556 0.468 0.213 0.322 13(MAPK13) X57346 5 Tyrosine 3-monooxygenase/ 1.179 0.989 0.832 0.131 0.489 0.587 1.157 0.291 0.189 0.425 0.602 0.663 0.611 tryptophan 5-monooxygenase activation protein, beta polypeptide(YWHAB) D16480 5 Hydroxyacyl-coenzyme A 0.893 1.085 1.021 0.157 0.333 0.56 1.512 0.311 0.348 0.56 0.228 0.539 0.686 dehydrogenase/3-ketoacyl- coenzyme A thiolase/enoyl- coenzyme A hydratase (trifunctional protein), alpha subunit(HADHA)

a No: the number of the patients whose ratios are lower than 0.5. b Ratio: normalize all of the samples to the mean of a set of control samples (NC1, NC2, and NC3). Gene expression profiles of human SLE G-M Han et al 181 variety of processes/pathways, several related groups of profiles were heterogeneous as demonstrated by differ- genes could be identified. From the interferon pathways, ent branches. IFN-o, IFIT1 (interferon-induced protein 56), IFIT2 (interferon-induced protein 54), IFIT4 (interferon- Verification of microarray hybridization induced protein with tetratricopeptide repeats 4), OAS1 For verification of hybridization signals through the (20,50-oligoadenylate synthetase 1), OAS2 (20,50-oligoade- 0 0 5’TaqMan fluorogenic quantitative PCR assay using nylate synthetase 2), OASL (2 -5 oligoadenylate synthe- gene-specific primers and probes, we selected three tase-like), and LY6E (lymphocyte antigen 6 complex, differentially expressed genes (Ly6E, OAS2, and locus E) had higher expression in SLE patients. From CEBPD). Quantitative TaqMan PCR analysis was per- the T-cell receptors (TCRs) pathways, TCR a, TCR d, formed for Ly6E, OAS2, CEBPD, and GAPDH. GAPDH and Zap-70(zeta-chain (TCR) associated protein kinase was used as an internal control. When comparing the 70 kDa) had lower expression in SLE patients. gene expression ratios between the patients and controls obtained by oligonucleotide microarrays to those ob- Two-way cluster analysis tained by fluorogenic quantitative PCR, similar expres- To define patterns of gene expression in controls and SLE sion pattern were obtained for Ly6E, OAS2, and CEBPD. patients, we performed a two-way hierarchical cluster Figure 2 shows the comparison between the two analysis of both samples and genes using GeneSpring quantitative methods for each of the three genes. In all software. Figure 1 presents the results of clustering. Gene cases, the expression pattern as measured by microarray expression data were shown in a matrix format with each quantitation is similar to those measured by TaqMan column representing hybridization results for a single analysis, indicating that quantitation of gene expression gene, and each row representing the measured expres- by oligonucleotide microarray hybridization was sion level for 61 differential expression genes in a single accurate. sample, red and blue colors indicating transcript levels above and below normal. There were 10 SLE patients, two normal control groups, NC1 and NC2 from the same Discussion control group, and NC3 from another control group. As shown in the dendrogram, patient samples were clearly In this study, we used global expression profiling with separated from controls based on their gene expression oligonucleotide microarray to systematically characterize profile. Within the SLE patient cluster, gene expression gene expression of peripheral blood cells from SLE

6161

- NC1

- NC2

- NC3

- SLE1

- SLE2

- SLE6

... - SLE9

- SLE10

- SLE7

- SLE5

- SLE3

- SLE4

- SLE8

Figure 1 Hierarchical cluster analysis of gene expression profile. Each row corresponds to a single sample, and each column corresponds to a single gene (61 differentially expressed genes). The branch lengths indicate the correlation with which genes or samples were joined, with longer branches indicating a lower correlation. Color represents different transcript levels, red represents higher gene expression, blue represents lower gene expression, yellow represents no different gene expression. NC: normal control, NC1 and NC2 from the same control group, NC3 from another control group. SLE: systemic lupus erythematosus, SLE1 and SLE2 represent the same patient of SLE.

Genes and Immunity Gene expression profiles of human SLE G-M Han et al 182 TaqMan analysis (Ly6E) Microarray (Ly6E) 25 40 20 30 15 20 10 Fold-change

Fold-change 10 5 0 0 NC RA NC RA PM PM SLE-1 SLE-2 SLE-3 SLE-4 SLE-5 SLE-6 SLE-7 SLE-8 SLE-1 SLE-2 SLE-3 SLE-4 SLE-5 SLE-6 SLE-7 SLE-8

Taqman analysis (OAS2) Microarray (OAS2) 16 16 14 14 12 12 10 10 8 8 6 6 Fold-change Fold-change 4 4 2 2 0 0 NC RA PM NC RA PM SLE-1 SLE-2 SLE-3 SLE-4 SLE-5 SLE-6 SLE-7 SLE-8 SLE-1 SLE-2 SLE-3 SLE-4 SLE-5 SLE-6 SLE-7 SLE-8

TaqMan analysis (CEBPD) Microarray (CEBPD) 8 5 4.5 7

e 4 6 3.5 5 3 4 2.5 3 2 Fold-chang

Fold-change 1.5 2 1 1 0.5 0 0 NC RA PM NC RA PM SLE-1 SLE-2 SLE-3 SLE-4 SLE-5 SLE-6 SLE-7 SLE-8 SLE-1 SLE-2 SLE-3 SLE-4 SLE-5 SLE-6 SLE-7 SLE-8 Figure 2 Verification of microarray quantification. Relative mRNA levels (Fold-change y-axis) of the indicated genes Ly6E, OAS2, and CEBPD were measured using the TaqMan 5’nuclease fluorigenic quantitative PCR assay (left column) in same samples (x-axis) as those used to prepare probes for the microarray hybridizations (right column). Data from TaqMan 5’nuclease fluorigenic quantitative PCR assay were normalized to internal control gene of GAPDH, data from microarray hybridizations were normalized as stated in Materials and methods. Quantitative measurements from both assays produced similar expression pattern for the genes measured.

patients compared to controls. As a screening approach, intensity. Therefore, we suspect that such false-negative we used a high-density oligonucleotide microarray, results were due to the relatively low sensitivity of the representing 3002 known human genes. After ensuring oligonucleotide microarray method, resulting in a bias of the reproducibility and reliability of the microarray the array data toward abundant transcripts.18 Of methodology, we identified 61 genes with differential the many differentially expressed genes found in our expression between SLE patients and controls by an study, changes in protein expression levels have been expression ratio greater than 2.0-fold in at least half of previously reported including IFN-o,19 OAS series the 10 patients studied. By the Welch’s ANOVA/Welch’s enzymes,20,21 ADPRT,22 and TCR a.23 t-test, all these 61 genes were significantly different A noteworthy finding in our study, is the interferon (Po0.05) between SLE patients and normal controls. pathway. The first documented cytokine abnormality in Among these differentially expressed genes, IFN-o and SLE patients was an increased serum level of interferon, Ly6E (TSA-1/Sca-2) may play an important role in the subsequently characterized as IFN-a.24 Owing to a mechanism of pathogenesis of SLE. Our discussion of about 60–70% between IFN-o focuses primarily on the pathway of the interferon and IFN-a, IFN-o was originally considered a subfamily system. of IFN-a (thus its old designation as IFN-a II).25 Later, Most of the genes identified in this study as differen- IFN-o was found in the sera of SLE, and it may also be a tially expressed, such as IL-1RII, MMP9, CCR7, IL-18R, major IFN species produced by peripheral blood cells in IL1RN, showed expression patterns similar to those SLE patients.19,20,26–28 In naturally occurring SLE, both reported previously using membrane microarrays.17 circulating IFN-a and increased levels of IFN-a-inducible Other results reported previously in SLE patients, such , such as OAS1, OAS2, and OASL, have been as increased IL-10, IL-6, and bcl-2, were not confirmed reported.19–21 Because long-term IFN-a therapy of pa- by our microarray data.17 The fluorescent intensity of tients with nonautoimmune disorders can induce anti- these genes was equal to the background fluorescence nuclear antibodies, antibodies to native DNA and also

Genes and Immunity Gene expression profiles of human SLE G-M Han et al 183 an SLE-like syndrome,29,30 the endogenous production study are differentially expressed, yielding results of IFN-a in SLE may play a role in the development similar to those reported by others using many of this autoimmune disease. This possibility is also individual normal controls.17 In addition, this pooling supported by findings that IFN-a, beside its antiviral strategy appears to have been effective in previous activity, also has several immunoregulatory functions. studies.15 For example, IFN-a stimulates immunoglobulin produc- Of additional concern, in the current study the samples tion by PBMC in vitro,31 can keep activated T cells alive,32 contained unseparated leukocytes. In contrast to malig- inhibits B cell receptor-mediated apoptosis,33 and in- nant diseases, which involve the transformation of duces dendritic cell differentiation in SLE.34 As a result, single cell type, SLE involves immune abnormalities IFN-a may play a pivotal role in the etiopathogenesis in a wide variety of cell populations to include B and of SLE.35 T lymphocyte, monocytes, and natural killer cells. Ly6E, also named TSA-1 (thymic shared antigen-1)/ A decrease or increase in gene expression detected Sca-2 (stem cell antigen-2), is a member of the Ly-6 family in a mixed cell population may be because of the of molecules.36 Although the biologic roles for those differences in cell populations (such as lymphopenia) molecules are not well understood, there is mounting between the SLE patients and normal controls. Future evidence that they participate in intercellular adhesion studies using purified cells will be necessary to elucidate and signaling.36 Initially, TSA-1 was found on all changes in gene expression in specific cell types immature thymocyte subsets (CD3À4À8À, CD3À4À8+, further. However, these may be more subject to artifacts CD3À4+8À, CD3À4+8+, CD3low4+8+), but not found caused by cell purification methods. Although the on CD3high4+8À and CD3high4À8+ thymocytes, biological significance of these differentially expressed early thymic migrants and peripheral T cells. So TSA-1 genes discussed above are not yet completely under- was taken as a novel thymocyte marker discriminat- stood, the current observations may serve as necessary ing immature from mature thymocyte subsets. In platform to explore relevant mechanisms of pathogenesis the periphery, TSA-1 was detected on B lymphocytes.37 of SLE. Our work shows that high-density oligonucleo- In addition, there was a direct inverse relationship tide microarray have the potential to explore the between TSA-1 and CD3 expression on thymocytes. mechanism of pathogenesis of complex diseases such There were many reports about TCR/CD3 defects in as SLE. SLE patients.23,38,39 In our study, we also found the TCRs pathways, TCRa, TCRd, and Zap-70 associated protein kinase 70 kDa) had lower expression in SLE Materials and methods patients. So, an inverse relationship between TSA-1 and CD3 expression on thymocytes may reflect Patients and controls abnormal activated lymophocytes in SLE patients. We collected 10 Chinese patients all fulfilling the Noda et al 40 found protection from anti-TCR/CD3- American College of Rheumatology classification criteria 44 induced apoptosis in immature thymocytes by a for SLE. These patients included one male patient and signal though TSA-1/Sca-2, and modulation of TCR- nine female patients, 17 to 39 years old, averaging 29 mediated signaling pathway by TSA-1/Sca-2.41 The years old. Data on the age, clinical characteristics, disease surface expression of Ly-6 antigens on T cells in vivo activity, and current medication are presented in Table 3. and in vitro can be dramatically enhanced by IFN-a.42 Disease activity in SLE patients was determined using 45 TSA-1 mRNA levels in normal human PBMCs were also SLEDAI score. There were two groups of Chinese enhanced strongly by IFN-a and slightly by IFN-g.36 normal controls, one group included 10 normal controls, These studies suggested that TSA-1 antigens may five male controls and five female controls. The other represent an important alternative pathway for T-cell included eight normal controls, four male controls and activation in SLE patients involved in IFN-mediated four female controls, 20 to 36 years old, with an average immunomodulation. age of 30 years. Finally, a number of the observed differences in gene expression in SLE patients could be attributed to the Total RNA extraction and preparation of target immunosuppressive medication rather than the disease biotinylated cRNA itself, such as IL-1RII gene upregulated by glucocorti- Heparinized venous peripheral blood was collected. coids.17 Hierarchical clustering showed that patient Erythrocytes were then lysed with a lyse solution of samples were clearly separated from controls based on amine oxalate. Total RNA extraction was performed their differential gene expression profile. These differ- using the TRIzol reagent (Life Technologies, Inc.). All of ential expression genes can be taken as characteristic total RNA from the two groups of normal controls was signatures of gene expression in SLE patients for clinical separately pooled, as two normal controls of patient of diagnosis and differential diagnosis.43 However, profil- SLE (NC1and NC2 from one group of normal controls; ing a larger cohort of patients will be required to confirm NC3 from the second group of normal controls). these results. Conversion of 20 mg total RNA to single-strand cDNA This new and comprehensive approach has some and double-strand cDNA template was done by reverse limitations. For instance, the pooling strategy may transcription using a cDNA synthesis kit (Roche). In vitro mask some degree of gene expression heterogeneity in transcription was performed using T7 MEGAscript normal control. We separately pooled total RNA from transcription kit (Ambion), in the presence of biotin- the two normal control groups, in an attempt to reduce labeled CTP (biotin-14-CTP Life Technologies) and the variations. Analyzing many of individual normal unlabeled ATP, CTP, GTP, and UTP. Finally, we purified controls would have made this study economically biotin-labeled cDNA using phenol–chloroform–isoamyl prohibitive. Most of the genes identified in this alcohol extraction method.

Genes and Immunity Gene expression profiles of human SLE G-M Han et al 184 Table 3 Characteristics of SLE patients

No Age (birth data)/sex Diagnosis data Diagnosis criteria for SLE SLEDAI score Trementa

SLE-l 1968/female 2002–06 Butterfly rash 2 Pred 30 mg/day Pericarditis and pleuritis for 2 weeks Persistent proteinuria Anti-dsDNA (Ab+) IFANA(+)b SLE-2 1968/female 2002–06 Butterfly rash 7 Pred 30 mg/day Pericarditis and pleuritis for 2 months Persistent proteinuria Anti-dsDNA Ab(+) IFANA(+) SLE-3 1985/female 2000–06 Butterfly rash 4 Pred 25 mg/day Arthritis Persistent proteinuria Anti-dsDNA Ab(+) SLE-4 1967/female 1999–01 Arthritis 0 Pred 10 mg/day Leukopenia MTX 5 mg/week Thrombocytopenia CQ 250 mg/day Anti-dsDNA Ab(+) IFANA(+) SLE-5 1976/female 2000–08 Butterfly rash 9 Pred 20 mg/day Oral ulcers Arthritis Leukopenia Thrombocytopenia Anti-dsDNA Ab(+) IFANA(+) SLE-6 1963/female 1990–03 Oral ulcers 17 No treated for Leukopenia 2 month Thrombocytopenia Anti-dsDNA Ab(+) IFANA(+) SLE-7 1982/female 2001–07 Butterfly rash 8 No treated Photosensitivity Arthritis Thrombocytopenia IFANA(+) SLE-8 1973/female 1998–11 Butterfly rash 6 Pred 40 mg/day Arthritis HCQ 400 mg/day Pericarditis Leukopenia Thrombocytopenia Anti-dsDNA Ab(+) IFANA(+) SLE-9 1963/female 2001–05 Butterfly rash 2 Prednisolone Photosensitivity 10 mg/day Oral ulcers Arthritis IFANA(+) SLE- 10 1978/male 2001–06 Arthritis 6 Pred 30 mg/day Pericarditis for 9 months Anti-dsDNA Ab(+) IFANA(+)

a current therapies at the time of sampling: pred, prednisone; HCQ, hydroxychloroquine; CQ, chloroquine; MTX, methotrexate. b IFANA, immunofluorescence antinuclear antibody.

Hybridization of biotin-labeled cRNA to by the ExpressionChip DNA Microarray System kit oligonucleotide microarrays and detection (Mergen Ltd). We used Mergen ExpressChipTM DNA Microarray System Human HO4 (cat# HO4-001) that contained 104 Image and data acquisitions and normalization negative control spots, 88 positive control spots and 2–3 The image of gene expression of microarray was scanned duplicate spots for 164 genes. So there were 3002 genes with a ScanArray 5000 (GSI Lumonics). Then we which distributed over subarrays A–D. The complete list converted the scanned image information into numeric of genes with accession numbers is published at http:// data though QuantArry 2.0 image analysis software (GSI www.mergen-ltd.com. Hybridization of biotin-labeled Lumonics). In order to transform raw data to normalized cRNA with oligonucleotide microarray and detection data, we subtract the median value of the negative was done according to the operating program provided controls. The formula used here was

Genes and Immunity Gene expression profiles of human SLE G-M Han et al 185 (signal strength of gene A in sample X)À(median signal of Statistical analysis the negative controls in sample X). Data were analyzed with GeneSpring software (Silicon Duplicated spots for each gene were averaged. Then Genetics).46 The detailed protocols are provided by the we normalized each sample to itself, to remove the manufacturer at http://www.sigenetics.com/cgi/SiG. differences in amount of exposure between samples, cgi/Products/GeneSpring. To identify those genes whose making different samples comparable to one another. expression levels changed differentially between SLE The formula used to do this was patients and normal controls, as well as to reduce the (the signal strength of gene A in sample X)/(the median all data set to a manageable size and prevent the inclusion of of the measurements taken in sample X). false positives, genes were only selected for further analysis if they met the following three criteria: (1) The Microarray reproducibility raw intensity was 10-fold higher than the background To assess array- and hybridization-based experimental nonspecific binding for that hybridization of at least one variability, we initially performed hybridization on sample of the 13 samples, (2) Welch’s ANOVA/ Welch’s duplicate microarrays using the same pool of control t-test were used to screen for statistically significant target RNA from the first group of normal control (see differences in expression level between SLE patients and supplementary data). One initial RNA sample was normal controls, the significance level was taken at a divided into two aliquots (NC1 and NC2), which were conventional 0.05 and (3) the gene expression change level then taken through cDNA synthesis and cRNA labeling in half of SLE patients (5/10 patients) were 2.0-fold higher in parallel, hybridization of biotin-labeled cRNA to two or lower than the mean of a set of normal controls (NC1, oligonucleotide microarrays. Using all 3002 genes as data NC2, and NC3). To group patient and control samples on points, a linear regression analysis was performed on the the basis of similarity in their gene expression, two-way fluorescence intensity difference per gene obtained in hierarchical cluster analysis of both statistically significant experiments 1 and 2. The two experiments were highly difference genes and samples was performed using reproducible: 98% of the genes showed o2-fold differ- GeneSpring software. ence in response, with a high correlation coefficient (r2 ¼ 0.9373). Upon further analysis, we also found that Taqman 5’nuclease fluorogenic quantitative PCR assay the false-positive results were mostly random. If we Quantitative TaqMan PCR analysis (Perkin Elmer) was repeated the experiment twice, the incidence of false performed for genes of Ly6E, OAS2, CEBPD, and positive was 0.04% (2 Â 2%), so repeating the experiment GAPDH. GAPDH was used as an internal control. The twice would avoid false positive as a result of impurity. primers and TaqMan probes were designed using the In addition, repeated experiments of different samples Primer Express software (Perkin Elmer), as follows: LY6E would overcome the error from using a single sample. (forward primer 5’-AGA CCT GTT CCC CGG CC-3’, These data demonstrated a high level of reproducibility reverse primer 5’-CAG CTG ATG CCC ATG GAA G–3’, of the technique. TaqMan probes FAM 5’-CCC ATC CCA GAA GGC GTC AAT GTT G–3’TAMRA), CEBPD(forward primer 5’-GCG Microarray reliability ACA AGG CCA AGC G-3’, reverse primer 5’-TCG TTC In our experiment, we used the ExpressChipTM DNA TCA GCC GAC AGC T-3’, TaqMan probes FAM5’-AAC Microarray System from Mergen Ltd. It is a simple-to- CAG GAG ATG CAG CAG AAG TTGGTG–3’TAMRA), use and cost-effective kit for high-throughput gene OAS2(forward primer 5’-TGA GAG CAA TGG GAA expression profiling. The system contained two identical ATG GG-3, reverse primer 5’-AGG TAT TCC TGG ATA glass slides spotted with oligonucleotide microarrays. AAC CAA CCC-3’, TaqMan probes FAM5’-CCC AGC The two slides allowed for the comparison of gene TGT CCT CGG TGC CTG–3’TAMRA), GAPDH(forward expression levels between two different sources. Each primer 5’-GAA GGT GAA GGT CGG AGTC-3’, reverse slide was arrayed with more than 3000 different primer 5’-GAA GAT GGT GAT GGG ATT TC-3’, TaqMan oligonucleotide sequences specific for human genes with probes FAM5’-CAA GCT TCC CGT TCT CAG CC- known biological roles. In addition, human housekeep- 3’TAMRA). The reaction was carried out in an ABI prism ing genes and non-human sequences were included on 7700 Sequence Detector (Perkin-Elmer) for 4 min at 951C each slide to serve as positive, negative, and normal- and 35 cycles of 15 s at 951C and 1 min at 601C. Results ization controls. The design of the oligonucleotide were obtained and analyzed using SDS software (Perkin- sequences followed a set of rigorously controlled criteria, Elmer). including unique match with GenBank’s human data- base, minimal variation in Tm and GC contents, low nucleotide repetition, size restriction (30-mer), and Acknowledgements consistent position within the gene sequences (relative to mRNA 3’ end). The arraying process was under This work was supported by the National Nature Science stringent control for superior printing quality. This Foundation of China (No. 39730430). We thank the ensured consistency, high sensitivity and optimal sig- patients for providing the blood samples. nal-to-background ratio. The oligonucleotides were covalently bound to the glass surface through a chemical reaction. From our results, the fluorescence signal intensity of blank spots and negative spots was very References low, where as the fluorescence signal intensity of positive 1 Hochberg MC. The epidemiology of systemic lupus erythe- spots was very high, and had saturation point and matosus. In: Wallace DJ, Hahn BH (eds). Dubois’ Lupus nonsaturation point, demonstrating the reliability of our Erythematosus. Williams and Wilkins: Baltimore, 1997, data. pp 49–65.

Genes and Immunity Gene expression profiles of human SLE G-M Han et al 186 2 Wakeland EK, Liu K, Graham RR, Behrens TW. Delineating 25 Hauptmann R, Swetly P. A novel class of human type I the genetic basis of systemic lupus erythematosus. Immunity interferons. Nucleic Acids Res 1985; 13: 4739–4749. 2001; 15: 397–408. 26 Adolf GR. Monoclonal antibodies and enzyme immunoassays 3 Deapen DM, Escalante A, Weinrib L, Horwitz DA, Mack M. A specific for human interferon (IFN) o 1: evidence that IFN- o 1 revised estimate of twin concordance in systemic lupus is a component of human leukocyte IFN. Virology 1990; 175: erythematosus. Arthritis Rheum 1992; 35: 311–318. 410–417. 4 Vyse TJ, Todd JA. Genetic analysis of autoimmune diseases. 27 Cederblad B, Blomberg S, Vallin H et al. Patients with systemic Cell 1996; 85: 311–318. lupus erythematosus have reduced numbers of circulating 5 Arnett FC. The genetic basis of lupus erythematosus. In: natural interferon-alpha-producing cells. J Autoimmunol 1998; Wallace DJ, Hahn BH (eds). Dubois’ Lupus Erythematosus. 11: 465–470. Williams and Wilkins: Baltimore, 1997, pp 77–117. 28 Blomberg S, Eloranta ML, Cederblad B et al. Presence of 6 Wu J, Edberg JC, Redecha PB et al. A novel polymorphism of cutaneous interferon-a producing cells in patients with Fcg RIIIa(CD16) alters receptor function and predisposes to systemic lupus erythematosus. Lupus 2001; 10: 484–490. autoimmune disease. J Clin Invest 1997; 100: 1059–1070. 29 Ronnblom LE, Alm GV, Oberg KE. Possible induction of 7 Tsao BP, Cantor RM, Kalunian KC et al. Evidence for linkage of systemic lupus erythematosus by interferon-a treatment in a candidate 1 region to human systemic lupus patient with a malignant carcinoid tumour. J Intern Med 1990; erythematosus. J Clin Invest 1997; 99: 725–731. 227: 207–210. 8 Shai R, Quismorio FP, Li L et al. Genome-wide screen for 30 Ronnblom LE, Alm GV, Oberg KE. Autoimmunity after alpha- systemic lupus erythematosus susceptibility genes in multi- interferon therapy for malignant carcinoid tumour. Ann Intern plex families. Hum Mol Genet 1999; 8: 639–644. Med 1991; 115: 178–183. 9 Gaffney PM, Kearns GM, Shark KB et al. A genome-wide 31 Neubauer RH, Goldstein L, Rabin H, Stebbing N. Stimulation search for susceptibility genes in human systemic lupus of in vitro immunoglobulin production by interferon-a. erythematosus sib-pair families. Proc Natl Acad Sci USA 1998; J Immunol 1985; 134: 299–303. 95: 14875–14879. 32 Marrack P, Kappler J, Mitchell T. Type I interferon keep 10 Gaffney PM, Ortmann WA, Selby SA et al. Genome screening activated T cells alive. J Exp Med 1999; 189: 521–529. in human systemic lupus erythematosus: results from a 33 Su L, David M. Inhibition of B cell receptor-mediated second Minnesota cohort and combined analyses of sib-pair apoptosis by IFN. J Immunol 1999; 162: 6317–6321. families. Am J Hum Genet 2000; 66: 547–556. 34 Blanco P, Palucka AK, Gill M, Pascual V, Banchereau J. 11 Brown PO, Hartwell L. Genomics and human disease- Induction of dendritic cell differentiation by IFN-a in systemic variations on variation. Nature Genet 1998; 18: 91–93. lupus erythematosus. Science 2001; 294: 1540–1543. 12 Derisi JL, Lyer VR, Brown PO. Exploring the metabolic and 35 Ronnblom L, Alm GV. An etiopathogenic role for the type I genetic control of gene expression on a genomic scale. Science IFN system in SLE. Trends Immunol 2001; 22: 427–431. 1997; 278: 680–686. 36 Shan XC, Bourdeau A, Rhoton A et al. Characterization and 13 Schena M, Shalon D, Davis RW et al. Quantitative monitoring mapping to human chromosome 8q24.3 of Ly-6-related gene of gene expression patterns with a complementary DNA 9804 encoding an apparent homologue of mouse TSA-1. microarray. Science 1995; 270: 467–470. J Immunol 1998; 160: 197–208. 14 Heller RA, Schena M, Chai A et al. Discovery and analysis of 37 Godfrey DI, Masciantonio M, Tucek CL et al. Thymic shared inflammatory disease-related genes using cDNA microarrays. antigen-1, a novel thymocyte marker discriminating immature Proc Natl Acad Sci USA 1997; 94: 2150–2155. from mature thymocyte subsets. J Immunol 1992; 148: 15 Lawrance IC, Fiocchi C, Chakravarti S. Ulcerative colitis and 2006–2011. Crohn’s diseases: distinctive gene expression profiles and novel 38 Liossis SC, Ding XZ, Dennis GJ, Tsokos GC. Altered pattern of susceptibility candidate genes. Hum Mol Genet 2001; 10: 445–456. TCR/CD3-mediated protein-tyrosyl phosphorylation in T cells 16 Staudt LM. Gene expression physiology and pathophysiology from patients with systemic lupus erythematosus. J Clin Invest of the immune system. Trends Immunol 2001; 22: 35–40. 1998; 101: 1448–1457. 17 Rus V, Atamas SP, Shustova V et al. Expression of cytokin- and 39 Brundula V, Rivas LJ, Blasini AM et al. Diminished levels of T chemokine-related genes in peripheral blood mononuclear cell receptor z chains in peripheral blood T lymphocytes from cells from lupus patients by cDNA array. Clin Immunol 2002; patients with systemic lupus erythematosus. Arthritis Rheum 102: 283–290. 1999; 42: 1908–1916. 18 Taniguchi M, Miura K, Iwao H, and Yamanaka S. Quantitative 40 Noda S, Kosugi A, Saitoh SI, Narumiya S, Hamaoka T. assessment of DNA microarrayFcomparison with Northern Protection from anti-TCR/CD3-induced apoptosis in imma- blot analyses. Genomics 2001; 71: 34–39. ture thymocytes by a signal through thymic shared antigen-1/ 19 Von Wussow P, Jakschies D, Hartung K, Deicher H. Presence stem cell antigen-2. J Exp Med 1996; 183: 2355–2360. of interferon and anti-interferon in patients with systemic 41 Saitoh SI, Kosugi A, Noda S et al. Modulation of TCR- lupus erythematosus. Rheumatol Int 1988; 8: 225–230. mediated signaling pathway by thymic shared antigen- 20 Preble OT, Rothko K, Klippel JH, Friedman RM, Johnston MI. 1(TSA-1)/stem cell antigen-2(Sca-2). J Immunol 1995; 155: Interferon-induced 2’-5’adenylate synthetase in vivo and 5574–5581. interferon production in vitro by lymphocytes from systemic 42 Dumont FJ, Coker LZ. Interferon-a/b enhances the expression lupus erythematosus patients with and without circulating of Ly-6 antigens on T cells in vivo and in vitro. Eur J Immunol interferon. J Exp Med 1983; 157: 2140–2146. 1986; 16: 735–740. 21 Hylton W, Cayley J, Dore C, Denman AM. 2’,5’-Oligoadeny- 43 Golub TR, Slonim DK, Tamayo P et al. Molecular classification late synthetase induction in lymphocytes of patients with of cancer: class discovery and class prediction by gene connective tissue diseases. Ann Rheum Dis 1986; 45: 220–224. expression monitoring. Science 1999; 286: 531–537. 22 Tsao BP, Cantor RM, Grossman JM et al. PARP alleles within 44 Tan EM, Cohen AS, Fries JF et al. The 1982 revised criteria for the linked chromosomal region are associated with systemic the classification of systemic lupus erythematosus. Arthritis lupus erythematosus. J Clin Invest 1999; 103: 1135–1140. Rheumatosus 1982; 25: 1271–1277. 23 Raziuddin S, Al-Janadi MA, Alwabel AA. T-cell receptor a/b 45 Bombardier C, Gladman DD, Urowitz MB, Caron D, Chang chain-CD3 protein complex defect in systemic lupus erythe- CH. Derivation of the SLEDAI: a disease activity index for matosus: T-cell function. Am J Med 1992; 93: 461–466. lupus patients. The Committee on Prognosis Studies in SLE. 24 Ytterberg SR, Schnitzer TJ. Serum interferon levels in patients Arthritis Rheum 1992; 35: 630–640. with systemic lupus erythematosus. Arthritis Rheum 1982; 25: 46 Lockhart DJ, Winzeler EA. Genomics, gene expression and 401–406. DNA arrays. Nature 2000; 405: 827–836.

Genes and Immunity