All whole blood samples were collected under standard phlebotomy protocol into PAXgene tubes (PreAnalytix, Switzerland) and RNA extracted from the samples using the PAXgene Blood RNA Kit (PreAnalytix, Switzerland). All methods followed the protocols supplied in the kit 14.

The MessageAmp II Kit 15 was used to amplify the RNA for use with 10.5 K Peter MacCallum microarrays. These comprise ~10,500 elements representing 9,381 unique cDNAs (Unigene Build 144) and were printed onto superamine slides (Telechem) using a robotic arrayer (Virtek) by The Peter MacCallum Cancer Institute, Australia – full gene list published on the author's website 16. The SuperScript Indirect cDNA Labelling System (Invitrogen) was used for cDNA production and CyDye (Amersham Biosciences) labeling. The protocol supplied was followed 17. The labeled cDNA was hybridized at 65°C overnight, the arrays were washed and scanned using a 4000B scanner (AXON Instruments), then visualized using GenePix Pro 5.0 software. Each microarray was repeated using alternate CyDye combinations so that there were 2 arrays per patient. Each MS sample was compared to the female reference (pool of 20 females). Male controls were compared to this same reference. Genes different between the male controls and female reference were considered 'sex-specific' and excluded from the analysis of MS samples versus the reference sample.

Raw data files generated by the GenePix Pro software were loaded into the program R 18 and normalized using the freely available microarray analysis package Limma 18. The data was normalized using the print-tip loess method and all spots with an intensity value less than 100 were filtered out of the data sets as these were considered unreliable. To avoid type 2 errors (false rejection of association) we used the Significance Analysis of Microarrays (SAM) package 19, which uses modified t tests to compare results for each gene and uses an iterative procedure to evaluate the likelihood of the association occurring by chance. The SAM package was used in conjunction with Limma to identify the most statistically significant dysregulated genes. The GOstat program 20 was used to identify gene ontology groups which were highly represented in the dysregulated gene sets.

All primers were designed using Primer3 21. ALOX5 (H51574) forward primer sequence: 5'-CCACGGGGACTACATCGAGTT-3' and reverse primer sequence: 5'-CTTTACGTCGGTGTTGCTTGAG-3'. TGFβ1 (R36467) forward primer sequence: 5'-CGAGCCTGAGGCCGACTACTA-3' and reverse primer sequence: 5'-CTCGGAGCTCTGATGTGTTGAA-3'.

All RNA samples were quantified using RiboGreen reagent (Molecular Probes, Eugene, OR, USA) and fluorescence measured using a FluoroCount fluorescence microplate reader (Packard Bioscience, Meriden, CT, USA). Patient and control pool RNA was reverse-transcribed into cDNA by preparing 20 μl reverse transcription reaction containing: 500 ng RNA, 250 μg Oligo dT (Promega, Madison, WI, USA), 10 mM dNTPs (Promega, Madison, WI, USA). The reaction was incubated at 65°C for 5 min and cooled on ice, then 5× buffer (Promega, Madison, WI, USA), 40 U RNase inhibitor (Promega, Madison, WI, USA), and 400 U reverse transcriptase (Promega, Madison, WI, USA) was added and the reaction incubated at 42°C for 60 min then 80°C for 15 min.

SYBR Green (Applied Biosystems Foster City, CA, USA) was used to quantify cDNA by real-time PCR. The PCR reaction contained cDNA derived from 5 ng of MS or control total RNA, SYBR Green master mix (Applied Biosystems Foster City, CA, USA), and 100 ng of each primer. The reaction was amplified in a Corbett Rotor-Gene 2000 real-time PCR cycler (Corbett Research, NSW, Australia) using the following cycling profile: 1 cycle of 95°C for 10 min; 5 cycles (95°C for 30s, 64°C for 30s, 72°C for 30s); 35 cycles (95°C for 30s, 60°C for 30s, 72°C for 30s); then 72°C for 5 min. Melt curve analysis was then performed between 75–99°C in 1°C increments. RT-PCR products were electrophoresed on 2% 1× TBE agarose gels, stained with ethidium bromide (0.5 μg/ml) to visualize the product. Normalization to total RNA, or the 2^-ΔCt method 22 was used to determine differences in expression between the MS individuals and the control pool.

Genomic DNA pools consisting of DNA samples from 217 HLA-DR2 positive MS patients, 155 HLA-DR2 negative MS patients, 169 probands from MS multiplex families, and from 187 unrelated healthy controls were used for promoter region sequencing. PCR primers were designed to amplify the 5' untranslated promoter region (approximately 500 bp upstream of the 1^st exon) using Primer3 20. ALOX5 (H51574) forward primer sequence: 5'-AGCCTCTGTGCTCCAGAATC-3' and reverse primer sequence: 5'-GGCTGAGGTAGATGTAGTCGTCA-3'. TGFβ1 (R36467) forward primer sequence: 5'-GGGAGGTGCTCAGTAAAGG-3' and reverse primer sequence: 5'-CTCGCTGTCTGGCTGCTC-3'. PCR reactions were prepared using 2× PCR Mastermix (MBI Fermentas, Vilinus, Lithuania), 100 ng of each PCR primer and 50 ng of pooled DNA samples or 0.1–1 μg of control individual genomic DNA. Reactions were amplified in the Mastercycler thermocycler (Eppendorf AG, Hamburg, Germany) using the following 64° -60°C "touchdown" temperature profile: 1 cycle of 95°C for 2 min 30 s; then 5 cycles (95°C for 30 s, 64°C for 30 s, 72°C for 30 s), then 30 cycles of (95°C for 30 s, 60°C for 30 s, 72°C for 30 s); then 1 cycle of 72°C for 5 min. For GC-rich sequences in the ALOX5 and TGFβ1 promoter regions, 5% DMSO and 0.02 U/μl Phusion DNA polymerase (Finnzymes, Finland) was also added to the reaction and a 62° -58°C "touchdown" temperature profile with a 98°C denaturation temperature, was used: 1 cycle of 98°C for 2 min 30 s; then 5 cycles (98°C for 30 s, 62°C for 30 s, 72°C for 30 s), then 30 cycles of (98°C for 30 s, 58°C for 30 s, 72°C for 30 s); then 1 cycle of 72°C for 5 min. PCR products were electrophoresed on 2% 1 × TBE agarose gels, stained with ethidium bromide (0.5 μg/ml) to visualize PCR products. The PCR products were then purified using 1 μl Exo-SAP-IT (exonuclease I/Shrimp alkaline phosphatase) (USB Corporation, Cleveland, OH, USA) incubated at 37°C for 15 min followed by 80°C for 15 min.

Unidirectional DNA sequencing was performed using an ABI Prism Big Dye terminator sequencing kit version 3.0 (Applied Biosystems, Foster City, CA, USA). PCR products were either sequenced in the forward or reverse direction using an internal primer. ALOX5 (H51574) reverse internal primer sequence: 5'-CATCTAGCGCCGCAGC-3'. TGFβ1 (R36467) forward internal primer sequence: 5'-CTCAGTAAAGGAGAGCAATTC-3'. Reactions were incubated at 95°C for 5 min, followed by 25 cycles (95°C for 30 s, 50°C for 15 s and 60°C for 4 min). Reactions were then electrophoresed on an ABI PRISM 3100 Genetic Analyzer (Applied Biosystems, Foster City, CA, USA) by the Westmead Millenium Institute DNA Sequencing Facility. DNA sequences were viewed using ABI Prism Editview software (Applied Biosystems, Foster City, CA, USA).

Up to 388 trio families, recruited by the Institute for Immunology and Allergy Research, were utilized for this genotyping study. Since appropriate matching of controls is essential to avoid false positive associations resulting from population stratification or non-random mating, genetically unrelated family members of the MS patients are usually used as controls. However, confounding due to hidden stratification can still occur. For this reason we have used trio families in this study. When a patient and both parents (a "trio family") are considered, the non-transmitted parental alleles form the controls. In these trio families association was tested using the transmission disequilibrium test (TDT) 23.

One hundred and two trio families were genotyped for the ALOX5 6 bp (GGGCGG) deletion. PCR primers were designed to amplify the transcription factor binding region, containing the repeated motif GGGCGG, using Primer3 21. Forward primer: 5'-AGGTCCCGCCCAGTC-3' and reverse: 5'GGTCTGGCTCCAGGCT-3'. PCR reactions were prepared using 2× ImmoMix PCR Mastermix (Bioline), 100 ng of each PCR primer, 0.25 μl of 100% DMSO and 50 ng of DNA samples. Reactions were amplified in the Mastercycler thermocycler (Eppendorf AG, Hamburg, Germany) using the following 68° -64°C "touchdown" temperature profile: 1 cycle of 96°C for 10 min; then 5 cycles (96°C for 30 s, 68°C for 30 s, 72°C for 30 s), then 25 cycles of (96°C for 30 s, 64°C for 30 s, 72°C for 30 s); then 1 cycle of 72°C for 5 min. PCR products were electrophoresed on 15% acrylamide gels made with: 40% acrylamide; 10× TBE buffer; ammonium persulfate (100 mg/ml); TEMED; and H₂O. The gels were stained with ethidium bromide (0.5 μg/ml) to visualize PCR products. Transmission disequilibrium tests (TDT) were performed using the Haploview program 24.

Three hundred and eighty eight trio families were genotyped for the TGFβ1 -508 C>T SNP. Genotyping of the TGFβ1 SNP was performed by the Australian Genome Research Facility (AGRF) (Queensland, Australia). The AGRF custom SNP genotyping service utilized the homogenous MassExtend (hME – single base extension) analysed on the Sequenom Autoflex Mass spectrometer and the Samsung 24 pin nanodispensor. SNP assays were designed by AGRF, PCR oligos were obtained and then processed in multiplex format, following PCR, on a mass spectrometer. The mass spectrometer allowed different SNPs to be identified with a high level of sensitivity. Transmission disequilibrium tests (TDT) were performed as above. The same genotyping strategy was employed for an independent cohort of 250 trios collected by Victorian institutions 25.

Table 1 lists the statistically significant (p ≤ 0.05) over-represented pathways, in the relapse and remission upregulated gene sets, according to the GOstat program. There were no over- or under-represented pathways in the downregulated gene sets for either group. The over-represented pathways in the relapse upregulated gene set were predominantly inflammation/apoptosis-related, whilst the pathways in the remission upregulated gene set were quite distinct, being related instead to protein transport and localization.

Prior to 2007, the two largest genomic screens have been that of the GAMES collaborative using > 4000 microsatellite markers in a case controls study 26, and the affected family study using 400 SNPs in 1000 individuals 27. The latter identified 5q33, 17q23 and 19p13 as having the highest LOD scores of the non-HLA region, and several genes from the most dysregulated set in relapse and remission were encoded in these genomic regions (Table 2). The meta-analysis from the GAMES study found 3 microsatellite markers associated with MS, D11S1986, D19S552, and D20S894. Only 3 genes were encoded within 1 MB, the likely maximum linkage distance, of any of these markers: TGFβ1, CEACAM1 and the glial maturation factor (all from 19q13.2). Only TGFβ1 is immune cell specific 28, so this gene was selected for further study. A more recent screen by the international multiple sclerosis consortium (IMSGC), has provided a more comprehensive coverage of the genome 10, and identified 38 genes with primary evidence for association. Of these 4 were dysregulated in our study: CD58, DBC1, TAF1A, and FHIT (Table 2). The genetic association of CD58 and DBC1 were confirmed in the IMSGC's replication study.

In the TGFβ1 promoter sequence, one known SNP at position -508 T>C (rs1800469) was confirmed in both the MS and control populations (Table 3). No other previously identified or common novel SNPs were found. This SNP was not associated with MS overall in the Westmead cohort (380 trios), but as there was a trend (P < 0.13), and the statistical power may not have been sufficient, we increased the cohort size to 620 trios. However, in the combined cohort there was still no significant association.

Because the gene association studies do not exclude regions as having associations, we examined the promoter region of ALOX5, which is upregulated in many MS peripheral blood gene expression arrays 711, including both the relapse and remission sets here. In the ALOX5 5'UTR sequence, one known SNP at position -557 T>C (rs12762303) was confirmed in both the MS and control populations, in addition to a deletion of one of the 6 bp tandem repeats (GGGCGG) of the consensus Sp1 binding sequence located between position -147 and -176 upstream from the ATG start site of the gene, as identified by In et al 29 (Table 3). This region also contains 5 Egr-1 binding motifs, located between position -148 and -180.

When 102 trio families were genotyped for the ALOX5 6 bp GGGCGG deletion, TDT testing showed no evidence of transmission distortion for the deletion with 38 long (normal sequence): 34 short (deletion) transmitted (p < 0.6). The short allele (minor allele) had a frequency of 0.19 in the MS population and 0.20 in the parental population. This deletion appears therefore not to be a significant risk factor in MS. We aimed to have the power to detect at least a 1.5- to 2-fold risk; however, a larger sample would be required for statistical power to detect a smaller risk factor. According to the McGinnis method, for trios, a 1.5-fold risk requires 266 families for an 80% chance of detecting association at a 5% significance level whilst a 2-fold risk requires 85 families. Due to the small sample size no stratification was performed. Both the -557 SNPs were equally represented in the control and MS (DR+, DR-, Familial) pools, and so was not examined further 30.

This study found a large number of dysregulated genes in peripheral blood during the relapse and remission phases of RRMS. When these sets of genes were compared against the total gene set contained on the array, there was a statistically significant over-representation of apoptosis- and inflammatory-related gene ontologies in the relapse up-regulated gene set. This was expected as MS relapse is associated with detrimental inflammation in which increases in inflammatory cell activation and the activation of inflammatory-related pathways is predominant. Recent studies have also found upregulation of inflammatory anti-apoptotic pathways in relapse 6, as a distinct MS cluster 31, or in MS compared to controls 323334. This is also consistent with the success of immunomodulatory therapies in RRMS 35. In the remission up-regulated gene set, there was an over-representation of protein transport and localization, and actin polymerisation ontologies. Since these processes are altered when cells change shape and motility, they may indicate that fluctuation in cell trafficking is important in remission.

To date, microarray studies using MS blood have focused on peripheral blood mononuclear cells. Recently, the involvement of other blood cells in MS disease pathogenesis, such as neutrophils, has raised interest due to increasing evidence suggesting that elements of the immune response classically associated with allergy, may contribute to the pathogenesis of autoimmune CNS demyelinating diseases, such as MS and its animal model EAE 36. We have identified the upregulation, both during relapse and remission, of arachidonate 5-lipoxygenase (ALOX5), an enzyme that catalyzes the initial steps in the conversion of arachidonic acid to biologically active leukotrienes in neutrophils 3738. Interestingly, we previously found an up-regulation of ALOX5 in both secondary progressive and primary progressive MS patients 7. Gene expression studies of MS and EAE lesions have also reported increased levels of ALOX5 11 in addition to other allergy-related mediators 394041.

Combining genomic and transcriptomic data has proved useful in identifying genetic associations in disease. In this way we earlier identified CD127 polymorphisms as associated with PPMS 7, an association now independently confirmed in two large cohorts 89. The general principal of finding SNPs affecting expression in the 5' UTR region has also now been established by combining expression phenotype and HapMap data 4 or from the ENCODE project 5. We have examined the putative promoter regions of TGFβ1 and ALOX5 for polymorphisms then tested the association of these variants with MS. A further advantage of this approach is that if promoter polymorphisms are associated with MS, then this provides support for the observed aberrant expression being pathogenic rather than homeostatic.

Genomic regions chosen for screening of dysregulated genes were based on the current largest genomic studies 102627. The genomic region from the affected family study was large, and this is reflected in the large number of genes selected using these criteria. Three genes were targeted from the more stringent criteria provided by the GAMES study 26, and TGFβ1 was selected for further study as it is the only one whose expression is confined to immune cells.

In addition to being close to the microsatellite in the GAMES meta analysis 26, in 19q13.2, in which TGFβ1 is encoded, this gene has been identified as being associated with MS susceptibility in other studies 424344454647. Dysregulated expression of TGFβ1 may have significant consequences in MS pathogenesis and potentially could be due to sequence variation in the TGFβ1 promoter region. Only one common SNP, -508 T>C, was detected in the 500 bp 5' UTR putative promoter sequence 48 (Table 3). From twin studies, Grainger et al 49 demonstrated TGFβ1 expression was under genetic control, and that the T allele is associated with higher expression. Two groups have studied the association of this SNP with MS 5051. We found evidence for an association when 102 trios were genotyped, and then increased the numbers to 388 trio families. This is well in excess of the number needed to have the statistical power to have an 80% chance of detecting a risk factor of 1.5 at the 0.05 significance level (212 families needed according to the McGinnis method). We identified a trend towards over-transmission of the T allele in MS in our patient cohort. We then tested the association in a second independent trio cohort, but found only non-significant over-transmission (Table 4).

Interestingly Green et al, in contrast to Weinshenker et al, found the wild-type C allele (or the "low-producer" genotype), which was under-transmitted in our study, was associated with a milder disease course. EAE and human drug response data suggest high TGFβ1 expression would be protective 5253545556. Green et al 50 also found strong linkage disequilibrium (LD) existing between the SNP at position -508 and the codon 10 SNP and partial LD with two other codon changing SNPs, which may result in an altered function of the translated protein, which may in turn be the cause of the association with susceptibility found in our study. Neither Green et al (2001), nor Weinshenker et al (2001), found an association between the -508 SNP and MS susceptibility (Table 5). A final assessment of the significance of TGFβ1 as a risk factor will be possible after larger, multiple cohorts have been studied, with informed stratification for clinical parameters, and after testing of epistatic interactions.

TGFβ1 is a multifunctional growth factor with demonstrable immunosuppressive effects on a number of cells including B cells, CD4+ T cells (Th1 and Th2), CD8+ cytotoxic T lymphocytes, natural killer cells and macrophages 57 and has been recently been implicated as a key factor in the differentiation of Th17 and T regulatory cells 58. The full significance of TGFβ1-mediated effects on immune responses is still incompletely understood, but have been implicated in MS pathogenesis, with some studies finding reduced TGFβ1 levels during a relapse and an increase during remission 59606162. However, it has also been reported that TGFβ1 levels increase during remission and even more so during relapse and in progressive patients 63.

This study is the first to investigate the ALOX5 promoter variations in MS patients. We did not identify any association between the variant 5'UTR alleles and MS, suggesting that these insertion/deletion (in/del) variants in the Sp1/Egr-1 binding sites are not pathogenic in MS, and may not be the cause of the increased ALOX5 expression measured in MS patients by microarray analysis. However, extensive quantitative studies would be needed to confirm this. As mentioned previously, our small sample size does not exclude this in/del as an appreciable risk factor. It is possible that in MS, as found in asthma 64, these promoter variations may affect disease modification rather than disease susceptibility – an idea worthy of further investigation if ALOX5 inhibitors were considered for MS treatment.

Of the 38 genes implicated in MS susceptibility on the first screen in the recently published whole genome screen, four were in the dysregulated gene sets identified here (Table 2). Two of these four genes were further supported as associated with MS susceptibility in their replication study: CD58 and DBC1. CD58, in common with the most associated genes from that and other studies 8910, IL7Rα and IL2Rα, affects regulatory T cell (Treg) differentiation and proliferation. TGFβ1 is essential for Treg differentiation. The up-regulation of CD58 in remission is consistent with the evidence that the protective allele is associated with higher CD58 expression, and that CD58 promotes Treg differentiation. Gene expression and genotyping data therefore further implicate Tregs in MS pathogenesis. Notably, the two most associated genes from the IMSGC study, IL7Rα and IL2Rα, were not detected as dysregulated in whole blood. DBC1 is anti-proliferative 65, so that the higher expression observed in remission is consistent with an immunosuppressive function.