Gene expression changes in peripheral blood mononuclear cells from ...

Journal of the Neurological Sciences 258 (2007) 52 – 59 www.elsevier.com/locate/jns

Gene expression changes in peripheral blood mononuclear cells from multiple sclerosis patients undergoing β-interferon therapy M.K. Singh a , T.F. Scott b , W.A. LaFramboise c , F.Z. Hu a , J.C. Post a , G.D. Ehrlich a,⁎ a

Center for Genomic Sciences, Allegheny Singer Research Institute, 320 E. North Avenue, Pittsburgh, PA 15212, USA b Department of Neurology, Allegheny General Hospital, Pittsburgh, PA 15212-4772, USA c Department of Pediatrics, Allegheny General Hospital, Pittsburgh, PA 15212-4772, USA Received 20 January 2005; received in revised form 12 September 2006; accepted 16 February 2007 Available online 30 April 2007

Abstract Objective: Multiple sclerosis (MS) is a disabling idiopathic inflammatory disorder with evidence of immune dysfunction. Current therapies for MS include preparations of β-interferon (βIFN). We studied the gene expression patterns in peripheral blood mononuclear cells from relapsing–remitting MS patients undergoing weekly βIFN-1a therapy (Avonex™; 30 mg intramuscular) to identify biomarkers for βIFN responsiveness. Methods: Oligonucleotide microarrays were used for the comparative analysis of gene expression patterns from longitudinal PBMC samples taken from five patients undergoing βIFN therapy. Results: On the basis of two-fold changes in expression levels and statistical analyses we selected a candidate diagnostic set of 136 genes that were differentially expressed between pretreatment and IFN-β-1a-treated MS patients. When we applied this gene set to cluster the specimens according to their expression profiles, the pretreatment samples clustered in one branch, and acute and chronic samples following treatment clustered in another branch. However, the chronic samples from the single clinical non-responder clustered with the pretreatment branch, suggesting that a possible reversal of βIFN-induced gene expression may be contributing to the poor clinical response. Conclusions: These 136 genes represent potential targets for new MS therapeutics and the basis for lack of βIFN response. © 2007 Elsevier B.V. All rights reserved. Keywords: Multiple sclerosis; Interferon; Microarrays; VCAM-1; Ephrin

1. Introduction The pathophysiology of MS involves an immune mediated attack on the central nervous system [1–3]. The clinical course of MS, the timing of relapses, and the rate of disability progression cannot be predicted. β-interferon (βIFN) is the most common therapy to control exacerbations in relapsing–remitting MS (RRMS), but it is only partially effective [4,5]. There are no biochemical or molecular markers that adequately predict response to the treatment. Currently disease progression is monitored based on the relapse rate, neurological deterioration, and evidence of disease activity on brain magnetic resonance imaging (MRI) ⁎ Corresponding author. Tel.: +1 412 359 4228; fax: +1 412 359 6995. E-mail address: [email protected] (G.D. Ehrlich). 0022-510X/$ - see front matter © 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.jns.2007.02.034

scans. However, radiological manifestations are only weakly correlative with the patient's clinical course. Microarray analyses simultaneously measure the transcript levels for thousands of genes [6]. The target organs for MS are the cellular adaptive immune system and the central nervous system (CNS). Microarray-based brain tissue studies have identified genes that are putatively associated with local pathogenesis in acute or chronic lesions [7–9]. Unlike the CNS, the cellular arm of the immune system can be readily accessed. Wandinger et al. using DNA microarrays to examine the effects of β-interferon therapy on PBMCs identified IL-12 receptor beta and CCR5 as being up-regulated both in vitro and in vivo [10]. Moreover, abnormal T-cell populations have repeatedly been observed in the peripheral blood of patients with MS [11–13]. Bomprezzi et al. determined that gene expression patterns

M.K. Singh et al. / Journal of the Neurological Sciences 258 (2007) 52–59

can distinguish patients with MS from controls and some of the differences identified were derived from activated T-cells [14]. Weinstock-Guttman et al. analyzed the acute transcriptional response of nearly 4000 genes in PBMC to βIFN [15]. They identified increases in known βIFN-inducible genes, and in genes involved in antiviral activity and βIFN signaling. Sturzebecher et al. identified some potential gene expression signatures that distinguished βIFN responders from non-responders [16]. Achiron et al. analyzed the expression of 12,000 genes in PBMC from patients with relapsing–remitting MS. Gene expression patterns distinguished patients with MS from controls as well as relapse from remission [17]. In this report we used oligonucleotide microarrays for the comparative analysis of gene expression patterns from longitudinal PBMC samples taken from patients before and after undergoing βIFN therapy. 2. Materials and methods 2.1. Patients We studied five RRMS patients with ages ranging from 32 to 46 years who were naïve to βIFN therapy at the outset of the trial who received weekly IFN-β-1a (Avonex™; 30 mg intramuscular) treatments beginning after the collection of a baseline venous blood specimen. Pretreatment specimens were drawn the day prior to initiation of therapy, acute post-treatment specimens were drawn 24 h after the initial therapeutic dose of interferon, and the chronic posttreatment specimens were obtained approximately six months after the initiation of weekly therapy just prior to receiving their weekly dose. All pre- and post-treatment blood samples were drawn into yellow top Vacutainer tubes, containing acid citrate dextrose as an anticoagulant. The chronic samples were drawn one day before the weekly treatment regimen approximately six months following the initiation of therapy. 2.2. Isolation of PBMC RNA Mononuclear cells were purified from the fresh peripheral blood specimens using lymphocyte separation media (FicollPaque) employing a density step gradient for the separation of the white blood cells. The mononuclear cells were trapped at the interface of the aqueous layer and the Ficoll following a 30 min centrifugation step. These peripheral blood mononuclear cells (PBMC) were recovered from the interface and washed twice with Hank's balanced salts solution (HBSS). For RNA isolation the PBMC were subjected to lysis in RLT buffer and purified using the Qiagen column purification procedure per the manufacturer's instruction (Qiagen, Valencia, CA). Each RNA sample was tested for purity and integrity by microcapillary electrophoresis using the Agilent 2100 Bioanalyzer. Any contaminating DNA in the isolated RNA, detected by PCR amplification of housekeeping genes, was removed by DNAse treatment.

53

2.3. Gene expression analyses using microarrays We utilized the CodeLink Expression Bioarrays (GE Biosciences) for the analysis of gene expression in PBMC following a preliminary comparative analyses among three commercial vendors to identify the most reproducible system. The CodeLink arrays contain oligonucleotides corresponding to 10,000 expressed human genes. Complementary RNA for microarray analysis from each PBMC specimen was synthesized from two micrograms of total RNA using oligo-dT/T7 promoter primer and reverse transcriptase followed by second strand cDNA synthesis using E. coli DNA polymerase. Labeled cRNA probes were made using T7 RNA polymerase and biotinylated rNTPs, hybridized in duplicate for all specimens to the microarrays, and developed with Streptavidin-Alexa. The microarrays were read in a Lumonix-Packard scanner and the data analyzed using Imagene, Genelinker and Statistical Analysis of Microarrays (SAM) software packages. The duplicate data sets for each specimen are used by the SAM algorithm which takes into account the variation in the expression of the entire gene set over the individual gene expressions, minimizing the false discovery rate and permitting a focus on a critical set of significant genes [18]. In a typical differential expression analysis the fold change parameter for significance is set at two-fold (upor down-modulated) which yields some 500 genes of interest. These genes are then grouped into functional subsets such as immune and inflammation response, cell– cell signaling, signal transduction, protein phosphorylation, apoptosis, transcription and translation and ion transport (cluster analysis). This approach allowed us to focus on critical genes exhibiting quantitative differences in gene expression between responders and non-responders. We also used software tools linked with a gene interaction data-base (Ingenuity Systems) in order to rapidly correlate gene names with the cellular functions of their cognate proteins. MIAME compliant microarray data has been deposited with the GEO DataSets database at the NCBI (http://www.ncbi.nlm.nih.gov/) under the reference series number GSE5574. 3. Results The RNAs from pretreatment, one-day post treatment (acute), and chronic treatment phase PBMCs were analyzed in duplicate for the expression of 10,000 genes by oligonucleotide microarrays. Acute treatment samples showed modulation of a large number of genes compared with the pretreatment samples including a number of known interferon-inducible genes. In t-tests we found a highly significant change in expression of 99 genes at P < 0.001, without a Bonferroni correction for multiple tests. The expression data sets were next analyzed using the program-Statistical Analysis of Microarrays (SAM). SAM takes into account both inter sample variability and overall variability on a microarray,

54


Table 1 Genes up-regulated in multiple sclerosis patients following β-interferon therapy GenBank

Symbol

Gene name

Expression after βIFN

Fold change

Mean change

SD

Interferon inducible and interferon pathway AF135187 MX1 Myxovirus (influenza virus) resistance 1, interferon-inducible protein BC001356 IFI35 Interferon-induced protein 35 NM_001548 IFIT1 Interferon-induced protein with tetratricopeptide repeats 1 NM_001549 IFIT4 Interferon-induced protein with tetratricopeptide repeats 4 NM_002198 IRF1 Interferon regulatory factor 1 NM_002463 MX2 Myxovirus (influenza virus) resistance 2 (mouse) NM_002535 OAS2 2′–5′-oligoadenylate synthetase 2, 69/71 kDa NM_002759 PRKR Protein kinase, interferon-inducible double stranded RNA dependent NM_003641 IFITM1 Interferon induced transmembrane protein 1 (9–27) NM_003733 OASL 2′–5′-oligoadenylate synthetase-like NM_004031 IRF7 Interferon regulatory factor 7 NM_004120 GBP2 Guanylate binding protein 2, interferon-inducible NM_005101 G1P2 Interferon, alpha-inducible protein (clone IFI-15K) NM_006084 ISGF3G Interferon-stimulated transcription factor 3, gamma 48 kDa NM_006417 IFI44 Interferon-induced protein 44 NM_016816 OAS1 2′,5′-oligoadenylate synthetase 1, 40/46 kDa NM_022873 G1P3 Interferon, alpha-inducible protein (clone IFI-6-16) NM_032643 IRF5 Interferon regulatory factor 5

Increase Increase Increase Increase Increase Increase Increase Increase Increase Increase Increase Increase Increase Increase Increase Increase Increase Increase

0.5–21.5 1.4–8.5 2.5–48.1 4.8–51.7 0.9–4.3 2.5–8.7 1.8–4.8 1.2–10.3 1.4–4.5 2.2–28.2 0.8–9.2 1.1–3.4 1.9–34.4 0.9–5.9 2–18.7 1.4–21.1 0.7–5.8 0.8–7.6

7.68 4.15 21.17 17.86 2.11 4.66 3.16 4.25 2.33 8.79 5.22 1.96 14.28 2.25 8.79 5.93 3.06 2.76

5 1.85 11.09 12.39 0.8 1.71 0.97 2.36 0.67 6.69 2.36 0.61 7.73 1.52 4.23 4.09 1.36 2.04

Transcriptional regulators NM_001421 ELF4 NM_003407 ZFP36 NM_003489 NRIP1 NM_003512 HIST1H2AC NM_003542 HIST1H4C NM_003542 HIST1H4F NM_012252 TFEC NM_016135 ETV7 NM_017523 HSXIAPAF1 NM_020152 C21orf7 NM_020830 WDFY1 NM_021170 Hes4 NM_021616 TRIM34 NM_022168 MDA5 NM_022750 ZC3HDC1 NM_030751 TCF8

E74-like factor 4 (ets domain transcription factor) Zinc finger protein 36, C3H type, homolog (mouse) Nuclear receptor interacting protein 1 Histone 1, H2ac Histone 1, H4c Histone 1, H4f Transcription factor EC ets variant gene 7 (TEL2 oncogene) XIAP associated factor-1 Chromosome 21 open reading frame 7 WD repeat and FYVE domain containing 1 bHLH factor Hes4 Tripartite motif-containing 34 Melanoma differentiation associated protein-5 Zinc finger CCCH type domain containing 1 Transcription factor 8 (represses interleukin 2 expression)

Increase Increase Increase Increase Increase Increase Increase Increase Increase Increase Increase Increase Increase Increase Increase Increase

0.8–3.1 0.7–4.2 0.7–4.4 0.8–5.8 0.1–2.5 0.1–2.5 0.7–28 0.8–39.1 0.3–14.4 0.7–10.4 0.7–3.8 0.7–4.6 0.9–4 1.6–8.7 0.4–5.7 0.1–50.3

1.75 1.84 1.77 2.36 0.84 0.84 6.68 11.26 4.49 2.71 2.04 2.76 2.16 3.58 2.8 4.72

0.68 0.87 1.07 1.26 0.68 0.68 8.84 9 3.96 2.41 0.87 1.09 0.87 1.98 1.38 8.44

Cell signalling AA595964 AB002313 AB023204 BF438425 M96824 NM_000389 NM_001493 NM_003120 NM_003335 NM_003681 NM_004417 NM_004574 NM_004635 NM_007315 NM_014314 NM_015907 NM_016323 NM_016381 NM_017414 NM_018120 NM_021105

NFkB inhibitor, alpha Plexin B2 Erythrocyte membrane protein band 4.1-like 3 v-Ha-ras Harvey rat sarcoma viral oncogene homolog Nucleobindin 1 Cyclin-dependent kinase inhibitor 1A (p21, Cip1) GDP dissociation inhibitor 1 Spleen focus forming virus (SFFV) proviral integration oncogene spi1 Ubiquitin-activating enzyme E1-like Pyridoxal (pyridoxine, vitamin B6) kinase Dual specificity phosphatase 1 Peanut-like 2 (Drosophila) Mitogen-activated protein kinase-activated protein kinase 3 Signal transducer and activator of transcription 1, 91 kDa DEAD/H (Asp-Glu-Ala-Asp/His) box polypeptide Leucine aminopeptidase 3 Cyclin-E binding protein 1 Three prime repair exonuclease 1 Ubiquitin specific protease 18 Armadillo repeat containing protein Phospholipid scramblase 1

Increase Increase Increase Increase Increase Increase Increase Increase Increase Increase Increase Increase Increase Increase Increase Increase Increase Increase Increase Increase Increase

0.1–4.2 0.2–5.1 0.9–15.3 0.2–2.3 0.8–5 1.1–20 0.4–5.5 1–10 0.8–3.1 0.8–7.8 0.3–5.4 1.4–10.9 0.7–4.2 1.9–7.3 1.2–6.1 1.7–15.1 3.2–18.4 1.1–5.7 1.3–9.3 0.3–1.7 1–9.8

1.75 2.4 3.87 0.81 1.9 5.79 1.77 3.83 1.73 2.43 2.06 3.94 1.94 3.72 3.11 5.69 7.24 2.69 4.98 0.71 4.95

1.11 1.15 4.3 0.58 1.25 5.04 1.58 2.78 0.72 1.74 1.35 2.66 0.83 1.27 1.36 3.33 3.56 1.05 2.37 0.39 2.58

NFKBIA PLXNB2 EPB41L3 HRAS NUCB1 CDKN1A GDI1 SPI1 UBE1L PDXK DUSP1 PNUTL2 MAPKAPK3 STAT1 RIG-I LAP3 CEB1 TREX1 USP18 FLJ10511 PLSCR1


55

Table 1 (continued) GenBank

Symbol

Gene name


Cell signalling NM_021991 NM_030671 NM_033027 NM_033123 U70451

JUP PTPRO AXUD1 PLCZ1 MYD88

Junction plakoglobin Protein tyrosine phosphatase, receptor type, O AXIN1 up-regulated 1 Phospholipase C, zeta 1 Myeloid differentiation primary response gene (88)

Increase Increase Increase Increase Increase

1.4–5.8 0.4–5.2 0.8–5.1 1.1–5.6 0.5–4.6

3.22 2.19 2.71 2.52 1.94

1.07 1.12 1.22 0.97 1.19

Growth factors, chemokines and receptors AJ224864 CMRF-35H Leukocyte membrane antigen AK002186 TIEG2 TGFB inducible early growth response 2 AW963062 TNFSF13B Tumor necrosis factor (ligand) superfamily, member 13b BE895761 MAGEC1 Melanoma antigen, family C, 1 NM_000211 ITGB2 Integrin, beta 2 NM_000265 NCF1 Neutrophil cytosolic factor 1 NM_000433 NCF2 Neutrophil cytosolic factor 2 NM_000566 FCGR1A Fc fragment of IgG, high affinity Ia, receptor for (CD64) NM_001250 TNFRSF5 Tumor necrosis factor receptor superfamily, member 5 NM_001295 CCR1 Chemokine (C-C motif) receptor 1 NM_001565 CXCL10 Chemokine (C-X-C motif) ligand 10 NM_002029 FPR1 Formyl peptide receptor 1 NM_002189 IL15RA Interleukin 15 receptor, alpha NM_002258 KLRB1 Killer cell lectin-like receptor subfamily B, member 1 NM_002346 LY6E Lymphocyte antigen 6 complex, locus E NM_002620 PF4V1 Platelet factor 4 variant 1 NM_003263 TLR1 Toll-like receptor 1 NM_003810 TNFSF10 Tumor necrosis factor (ligand) superfamily, member 10 NM_004054 C3AR1 Complement component 3a receptor 1 NM_004236 TRIP15 Thyroid receptor interacting protein 15 NM_006291 TNFAIP2 Tumor necrosis factor, alpha-induced protein 2 NM_006889 CD86 CD86 antigen (CD28 antigen ligand 2, B7-2 antigen) NM_007115 TNFAIP6 Tumor necrosis factor, alpha-induced protein 6 NM_009587 LGALS9 Lectin, galactoside-binding, soluble, 9 (galectin 9) NM_022168 MDA5 Melanoma differentiation associated protein-5 NM_024021 MS4A4A Membrane-spanning 4-domains, subfamily A, member 4 NM_024408 NOTCH2 Notch homolog 2 (Drosophila) S72487 ECGF1 Endothelial cell growth factor 1 (platelet-derived) U36759 PTCRA Pre T-cell antigen receptor alpha

Increase Increase Increase Increase Increase Increase Increase Increase Increase Increase Increase Increase Increase Increase Increase Increase Increase Increase Increase Increase Increase Increase Increase Increase Increase Increase Increase Increase Increase

0.9–5.7 0.8–5.4 0.8–14.2 0.7–4.5 0.7–3.4 0.9–7 0.8–4.7 1.4–27.3 0.6–5.9 0.6–11.9 4.9–96.2 1.1–3.1 1.3–4.1 0.3–2.2 2.5–15.4 1.2–8.3 0.5–4.5 0.4–33.2 0.8–10.1 0.2–7.2 0.7–5 0.8–11.7 0.7–5.1 1.5–7.8 1.6–8.7 2.9–62.8 1.2–3.7 0.5–4.7 1.2–7.5

2.26 2.2 3.73 2.68 1.8 2.64 2.44 7.41 2.06 4.57 30.49 2 2.48 0.72 6.22 3.09 2.34 7.02 3.54 2.14 2.64 3.26 2.52 3.11 3.58 11.41 2.17 2.31 2.68

1.19 1.07 3.34 1.18 0.75 1.4 1.21 8.88 1.53 4.27 28.19 0.6 0.71 0.56 3.76 1.82 1.34 7.59 2.7 1.89 1.29 3.28 1.43 1.66 1.98 16.47 0.75 1.13 1.75

Immunoglobulin-like receptors AF025529 LILRA1 AF025529 LILRA2 AF025529 LILRB1 NM_005874 LILRB2 NM_005874 LILRB5 NM_006847 LILRB4 NM_006863 LILRA1 NM_006863 LILRA2 NM_006863 LILRB1 NM_006864 LILRB3 NM_006865 LILRA3

Increase Increase Increase Increase Increase Increase Increase Increase Increase Increase Increase

1.2–6.9 1.2–6.9 1.2–6.9 0.4–12.4 0.4–12.4 1.2–6.5 0.9–6.8 0.9–6.8 0.9–6.8 1.2–4.8 2–7.5

2.49 2.49 2.49 4.23 4.23 2.99 2.64 2.64 2.64 2.5 4.37

1.07 1.07 1.07 3.48 3.48 1.44 1.45 1.45 1.45 0.91 1.7

Increase Increase Increase Increase Increase Increase Increase Increase Increase Increase

1–5.4 0.2–2.1 0.4–2.8 10.2–69.6 0.6–3 1.5–14.6 0.6–4.2 0.2–2.7 1.3–9.6 1.5–19.4

2.57 0.96 1.35 35.6 1.6 6.08 1.78 1.13 3.42 7.71

1.31 0.57 0.79 16.92 0.66 3.6 1.12 0.61 2.2 5.06

Leukocyte immunoglobulin-like receptor, subfamily A, member 1 Leukocyte immunoglobulin-like receptor, subfamily A, member 2 Leukocyte immunoglobulin-like receptor, subfamily B, member 1 Leukocyte immunoglobulin-like receptor, subfamily B, member 2 Leukocyte immunoglobulin-like receptor, subfamily B, member 5 Leukocyte immunoglobulin-like receptor, subfamily B, member 4 Leukocyte immunoglobulin-like receptor, subfamily A, member 1 Leukocyte immunoglobulin-like receptor, subfamily A, member 2 Leukocyte immunoglobulin-like receptor, subfamily B, member 1 Leukocyte immunoglobulin-like receptor, subfamily B, member 3 Leukocyte immunoglobulin-like receptor, subfamily A, member 3

Membrane transport and other cellular functions AL161952 GLUL Glutamate-ammonia ligase (glutamine synthase) BM723478 SLC2A9 Solute carrier family 2 (facilitated glucose transporter), member 9 BM741997 SLC25A3 Solute carrier family 25 (mitochondrial carrier; phosphate carrier), member 3 NM_000062 SERPING1 Serine (or cysteine) proteinase inhibitor, clade G (C1 inhibitor), member 1 NM_000310 PPT1 Palmitoyl-protein thioesterase 1 (ceroid-lipofuscinosis, neuronal 1, infantile) NM_000355 TCN2 Transcobalamin II; macrocytic anemia NM_000434 NEU1 Sialidase 1 (lysosomal sialidase) NM_000519 HBD Hemoglobin, delta NM_000712 BLVRA Biliverdin reductase A NM_001710 BF B-factor, properdin

Fold change

Mean change

SD

(continued on next page)

56


Table 1 (continued) GenBank

Symbol

Gene name


Fold change

Mean change

SD

Membrane transport and other cellular functions NM_001757 CBR1 Carbonyl reductase 1 NM_001912 CTSL Cathepsin L NM_003827 NAPA N-ethylmaleimide-sensitive factor attachment protein, alpha NM_006924 SFRS1 Splicing factor, arginine/serine-rich 1 NM_015535 DNAPTP6 DNA polymerase-transactivated protein 6 NM_020201 NT5M 5′,3′-nucleotidase, mitochondrial NM_020980 AQP9 Aquaporin 9 NM_030807 SLC2A11 Solute carrier family 2 (facilitated glucose transporter), member 11

Increase Increase Increase Increase Increase Increase Increase Increase

1–3.3 0.3–7.6 1–6.1 0.3–1.5 1.8–31.2 0.5–3.3 0.8–6.5 0.1–1.2

2.25 3.73 2.42 0.73 7.24 1.88 2.78 0.72

0.67 1.72 1.31 0.35 6.01 0.72 1.57 0.31

Genes with known functional homologies AF086130 C6orf187 Chromosome 6 open reading frame 187 AF217974 TSRC1 Thrombospondin repeat containing 1 AL117452 FTHFSDC1 Formyltetrahydrofolate synthetase domain containing 1 NM_006820 C1orf29 Chromosome 1 open reading frame 29 NM_016410 C9orf83 Chromosome 9 open reading frame 83 NM_016489 NT5C3 5′-nucleotidase, cytosolic III NM_016619 PLAC8 Placenta-specific 8 NM_031458 BAL B aggressive lymphoma gene NM_032495 HOP Homeodomain-only protein

Increase Increase Increase Increase Increase Increase Increase Increase Increase

1–11.7 1.1–5.3 0.9–2.9 1.6–44.7 0.9–5 0.9–6.4 1.3–6.8 0.4–9.1 0.3–3

3.85 3.04 1.83 20.26 2.49 3 2.9 3.79 0.72

2.57 1.19 0.55 9.46 0.92 1.6 1.38 2.07 0.74

EST, unknown biological function AB014543 KIAA0643 KIAA0643 protein AK023113 KIAA1618 KIAA1618 NM_015680 CGI-57 Hypothetical protein CGI-57 NM_018381 FLJ11286 Hypothetical protein FLJ11286 NM_018607 PRO1853 Hypothetical protein PRO1853 NM_024736 FLJ12150 Hypothetical protein FLJ12150 NM_031229 C20orf18 Chromosome 20 open reading frame 18 NM_031950 KSP37 Ksp37 protein U17077 BENE BENE protein

Increase Increase Increase Increase Increase Increase Increase Increase Increase

0.6–6.7 1.2–3.9 0.9–3.8 0.6–3.2 0.3–2.7 0.4–7.7 0.5–3.6 0.2–3.5 0.2–2.5

1.72 2.54 1.66 1.91 0.92 2.09 1.72 0.85 0.91

1.24 0.71 0.97 0.8 0.56 2.06 0.87 0.86 0.59

βIFN = β-interferon; fold change = the range (minimum and maximum) of ratios obtained by comparing each of the individual post-treatment specimens with the individual pre-treatment specimens for each gene; mean change = the arithmetic average of the ratios obtained by comparing all pre-treatment specimens with all post-treatment specimens values for each gene; SD = the standard deviation of the ratios obtained for each gene by comparing pre-treatment specimens with posttreatment specimens.

minimizing the false discovery rate and permitting us to focus on a critical set of significant genes [18]. We identified 136 characteristic genes modulated two-fold or more in acute samples over the pretreatment samples using SAM (Table 1). These included: genes coding for IFN-responsive genes (18 genes); transcriptional regulators (16 genes); cell signaling genes (26 genes); growth factors, cytokines and their receptors (29 genes); immunoglobulin receptors (11 genes); transport and other cellular proteins (18 genes); and ESTs and other genes with poorly defined products (18 genes). 3.1. Cluster analysis We used cluster analysis to examine the 136 genes identified by SAM for functional grouping of the samples (Fig. 1). As expected, the duplicate samples grouped in the same branch of the Clustal tree. In general, the post treatment samples clustered separately from the pretreatment samples. All acute samples clustered separately from the pretreated samples, and the chronic samples from four/five patients also clustered with the acute samples, however, one chronic patient clustered with the pretreatment samples. When the

code on the patient samples was broken it was determined that the single chronic post-treatment specimen that clustered with the pretreatment samples was obtained from the single patient who did not respond long-term to the βIFN treatment. This treatment nonresponsive patient experienced clinical exacerbations including optic neuritis and ataxia requiring steroid treatment, none of the other patients reported any progression of symptoms during the course of the study. Thus, it would appear that this cluster of 136 genes, or some subset thereof, may be diagnostic of responsiveness to βIFN therapy and that if these genes are not found to be differentially expressed that it may be indicative of treatment failure. 3.2. Focus genes We selected a subset of 202 genes with similar regulation patterns in all four responders that had at least a 1.5-fold change in expression to identify the biological network of interacting genes. Focus gene products interact with other gene products in the selected set and were identified using software tools linked with a gene interaction data-base


57

We identified three such networks containing 31 genes from our subset of 202 genes, including 29 up-regulated and two down-regulated genes (Fig. 2). 15 of these 31 genes are shared with our diagnostic set of 136 genes providing a cross validation of the two approaches. Three genes, VCAM-1, a mediator of immune cell passage through the blood brain barrier, Notch2, a gene controlling cell fate decisions, and EphrinB2, a known mediator of axonal guidance and T-cell signaling, make up a potential diagnostic set that were upregulated in all of the responders during both the acute and chronic phases of disease, but were down-regulated in the chronic specimen from the non-responder. 4. Discussion

Fig. 1. Clustal analysis of gene expression in PBMC from pretreatment, acute and chronic patients using 136-gene set.

(Ingenuity Systems). Networks are centered on the focus genes and include other genes that connect focus genes to each other or specifically interact with multiple focus genes.

There are relatively few studies of gene expression during βIFN therapy of MS patients. Weinstock-Guttman et al. analyzed the acute transcriptional response of nearly 4000 genes in PBMC to βIFN using spotted DNA arrays on nylon filters [15]. This study addressed less than a tenth of the human gene set and the sensitivity of nylon filter arrays is significantly lower than that of slide-based microarrays. Sturzebecher et al. studied RNA expression profiles of the PBMC of ten patients undergoing βIFN therapy using MiniLymphochip cDNA microarrays and reported differences among non-responder and responder phenotypes in their

Fig. 2. Relative expression of 26 potential focus genes identified by protein interaction mapping. Color code blue to red with red indicating highest level of expression.

58


gene expression profile [16]. We used 10,000 gene microarrays including most of the well defined genes for the serial analysis of gene expression patterns from longitudinal PBMC samples from five MS patients undergoing βIFN therapy. Using SAM we selected a set of 136 genes for detailed expression analyses, to compare transcript levels in pretreatment samples with those in acutely treated samples using a two-fold change in expression as a cut-off. When we applied this gene set to cluster the specimens according to their expression profiles, the pretreatment samples clustered in one branch, and acute and chronic samples clustered in another branch. However, the chronic samples from the single non-responder clustered with the pretreatment branch, suggesting that a possible reversal of IFNβ-induced gene expression may be contributing to the poor clinical response. Among the up-regulated genes vascular cell adhesion molecule-1 (VCAM-1) has been previously studied in MS patients [20]. Soluble VCAM-1 may prevent the interaction of the T-cells with endothelial cells of the blood brain barrier (BBB) by binding to the lymphocyte integrin. Increased VCAM-1 and soluble VCAM-1 levels have been associated with IFN-beta treatments and with better prognosis in MS patients [20,21]. EphrinB2 (EFNB2) is a member of the family of transmembrane anchored ligands of the ephrin receptor tyrosine kinase family, the largest subgroup of receptor tyrosine kinases (RTK) known [22]. EFNB2 expression is seen in most human tissues to varying degrees and appears highest in lung and lowest in lymphocytes. However, some lymphocyte activities have been shown to be modulated by ephrin receptors [23,24]. Upregulation of EFNB in PBMC of responders possibly modulates crucial signaling leading to IFNβ mediated amelioration of MS. Recently Branzini et al. parsed the microarray data in the public domain on βIFN-induced genes in MS patients and selected 70 potentially diagnostic genes for the drug response [25]. They followed 52 patients undergoing βIFN therapy at three-monthly intervals for expression of these genes by realtime PCR. Both our study and that of Branzini et al. showed an increase in expression of five members of the caspase family of apoptotic proteins, TRADD, and BAX for the poor responders. This is not surprising as MS has been associated previously with increased apoptosis [26]. The gene sets in common identified by these two studies include a number of genes that are associated with regulating a response to βIFN therapy [19,25] including the transcription factors IRF1, IRF5, IRF7, STAT1 and five other genes including MX1, NFKBIA, FLIP, a homolog of H-RAS and CD86. In summary we studied the gene expression patterns in PBMC from relapsing–remitting MS patients undergoing weekly βIFN-β-1a therapy. Our study included pretreatment, acute treatment, and chronic treatment specimens from which we may have begun to identify a diagnostic set of genes for βIFN responsiveness and nonresponsiveness, however, as the clinical decline in our nonresponder occurred prior to collecting the chronic treatment specimen

we do not know if the reversion to a pretreatment gene expression profile is diagnostic. Future studies will need to include more patients and more frequent assay time points to determine these issues. On the basis of changes in gene expression levels and statistical analyses we selected a candidate diagnostic set of 136 genes that were differentially expressed between pretreatment and βIFN-1a-treated MS patients. The expression of the gene subsets identified both in our study and those of others [16,25] should be studied in a larger set of MS patients. Acknowledgements This work was supported by Allegheny Singer Research Institute, Allegheny General Hospital, Grants from the Health Resources and Services Administration, and grants from Biogen Corporation and the Pittsburgh Foundation. References [1] Noseworthy JH, Lucchinetti CF, Rodriguez M, Weinshenker BG. Multiple Sclerosis. N Engl J Med 2000;343:938–52. [2] Steinman L. Immune therapy for autoimmune diseases. Science 2004;305:212–5. [3] Sospedra M, Martin R. Immunology of multiple sclerosis. Annu Rev Immunol 2005;23:683–747. [4] Jacobs L, Salazar AM, Herndon R, Reese PA, Freeman A, Josefowicz R, et al. Multicentre double-blind study of effect of intrathecally administered natural human fibroblast interferon on exacerbations of multiple sclerosis. Lancet 1986;2:1411–3. [5] Arnason BG. Interferon beta in multiple sclerosis. Clin Immunol Immunopathol 1996;81:1–11. [6] Schena M. DNA microarrays: a practical approach. Oxford University Press; 1999. [7] Whitney LW, Becker KG, Tresser NJ, Caballero-Ramos CI, Munson PJ, Prabhu VV, et al. Analysis of gene expression in multiple sclerosis lesions using cDNA microarrays. Ann Neurol 1999;46:425–8. [8] Chabas D, Baranzini SE, Mitchell D, Bernard CC, Rittling SR, Denhardt DT, et al. The influence of the proinflammatory cytokine, osteopontin, on autoimmune demyelinating disease. Science 2001;294: 1731–5. [9] Lock C, Hermans G, Pedotti R, Brendolan A, Schadt E, Garren H, et al. Gene-microarray analysis of multiple sclerosis lesions yields new targets validated in autoimmune encephalomyelitis. Nat Med 2002;8: 500–8. [10] Wandinger KP, Sturzebecher CS, Bielekova B, Detore G, Rosenwald A, Staudt LM, et al. Complex immunomodulatory effects of interferonbeta in multiple sclerosis include the upregulation of T helper 1-associated marker genes. Ann Neurol 2001;50:349–57. [11] Santoli D, Moretta L, Lisak R, Gilden D, Koprowski H. Imbalances in T cell subpopulations in multiple sclerosis patients. J Immunol 1978;120:1369–71. [12] Huddlestone JR, Oldstone MB. T suppressor (TG) lymphocytes fluctuate in parallel with changes in the clinical course of patients with multiple sclerosis. J Immunol 1979;123:1615–8. [13] Hafler DA, Fox DA, Manning ME, Schlossman SF, Reinherz EL, Weiner HL. In vivo activated T lymphocytes in the peripheral blood and cerebrospinal fluid of patients with multiple sclerosis. N Engl J Med 1985;312:1405–11. [14] Bomprezzi R, Ringner M, Kim S, Bittner ML, Khan J, Chen Y, et al. Gene expression profile in multiple sclerosis patients and healthy controls: identifying pathways relevant to disease. Hum Mol Genet 2003;12:2191–9.

M.K. Singh et al. / Journal of the Neurological Sciences 258 (2007) 52–59 [15] Weinstock-Guttman B, Badgett D, Patrick K, Hartrich L, Santos R, Hall D, et al. Genomic effects of IFN-beta in multiple sclerosis patients. J Immunol 2003;171:2694–702. [16] Sturzebecher S, Wandinger KP, Rosenwald A, Sathyamoorthy M, Tzou A, Mattar P, et al. Expression profiling identifies responder and non-responder phenotypes to interferon-beta in multiple sclerosis. Brain 2003;126:1419–29. [17] Achiron A, Gurevich M, Friedman N, Kaminski N, Mandel M. Blood transcriptional signatures of multiple sclerosis: unique gene expression of disease activity. Ann Neurol 2004;55:410–7. [18] Tusher VG, Tibshirani R, Chu G. Significance analysis of microarrays applied to the ionizing radiation response. Proc Natl Acad Sci U S A 2001;98:5116–21. [19] Singh MK, Scott TF, LaFramboise WA, Hu FZ, Post CJ, Ehrlich GD. Gene expression changes in peripheral blood mononuclear cells from Multiple Sclerosis patients undergoing β-interferon therapy. 56th annual meeting of the American Academy of Neurology, San Francisco, CA; 2004. [Abstract #040]. [20] Graber J, Zhan M, Ford D, Kursch F, Francis G, Bever C, et al. Interferon-beta-1a induces increases in vascular cell adhesion

[21]

[22] [23]

[24] [25]

[26]

59

molecule: implications for its mode of action in multiple sclerosis. J Neuroimmunol 2005;161:169–76. Calabresi PA, Pelfrey CM, Tranquill LR, Maloni H, McFarland HF. VLA-4 expression on peripheral blood lymphocytes is downregulated after treatment of multiple sclerosis with interferon beta. Neurology 1997;49:1111–6. Pasquale EB. Eph-ephrin promiscuity is now crystal clear. Nat Neurosci 2004;7:417–8. Freywald A, Sharfe N, Rashotte C, Grunberger T, Roifman C. The EphB6 receptor inhibits JNK activation in T lymphocytes and modulates T cell receptor-mediated responses. J Biol Chem 2003;278:10150–6. Sharfe N, Freywald A, Toro A, Dadi H, Roifman C. Ephrin stimulation modulates T cell chemotaxis. Eur J Immunol 2002;32:3745–55. Baranzini SE, Mousavi P, Rio J, Caillier SJ, Stillman A, Villoslada P, et al. Transcription-based prediction of response to IFNâ using supervised computational methods. PLoS Biol 2005;3:166–76. Barnett MH, Prineas JW. Relapsing and remitting multiple sclerosis: pathology of the newly forming lesion. Ann Neurol 2004;55: 458–68.