Abstract
Introduction
Multiple sclerosis (MS) is a common inflammatory disorder of the central nervous system (CNS) characterized by demyelization, gliosis, axonal damage and progressive neurological dysfunction. It is one of the most incapacitating diseases in young people.
The pathogenic mechanisms of the disease are not yet clearly identified; however, one proposal is an activation of the lymphocytes that cross the blood–brain barrier (BBB) directly into the interstitial matrix. T cells are then reactivated by fragments of the myelin antigens exposed in the context of human leukocyte antigen (HLA) molecules of the surface of the antigen presenting cells.
Aetiologically, MS is a complex disease in which both genetic and environmental factors are involved. As of today, seven genome wide association studies (GWASs) have been performed in MS in order to discover the genetic factors involved in the susceptibility to suffer the disease. The strongest genetic association with MS in Northern Europeans is found to be with extended MHC haplotypes, especially those containing HLA-DRB1*15:01. 1 Out of this previous known association, GWASs have found several genetic associations. Recently a common network of different variants has been proposed for the autoimmune diseases and the model of a cooperative network of SNPs seems to be the way to understand the complex MS genetics. 2 However, each one of these susceptibility genes appears to contribute little to overall risk 1 and MHC is accepted to be the key susceptibility locus in MS.
Despite the great consistency that the *15:01 MS association has demonstrated since its discovery 30 years ago,3,4 the molecular mechanisms that lie behind it remain unclear. Several works have studied the role of DRB1* and DRB5* in the immune response of MS disease. Studies in EAE in mice show both the implication of both DRB1*15:01 and DRB5*0101 restricted encephalitogenic T cells which would modulate the primary T response. 5 Other studies in humans suggest that the differential expression of DRB1*15:01 and DRB5*0101 or DQA1*0102 genes of the extended haplotype associated with MS could modulate the clinical ‘phenotype’ of the disease. 6 In this sense, several genetic variants have been postulated as modulators of the risk conferred by the *15:01 allele, such as vitamin D receptor gene (VDR) variants and sex.7–9
Several studies have found a possible association between VDR variants and MS. However, the results of these studies have been contradictory. While some works have not revealed any association between VDR variants and MS, 10 some others have found a clear link. According to these, the ff allele of VDR seems to have a protective effect;11,12 in contrast, Taq I and Apa I restriction marker variants of the gene have been found to be linked to a higher risk of suffering the disease. 13 Moreover, a recent study has shown that the activation of several members of the nuclear receptor family, VDR among them, is suppressed in the pre-disease state of MS, which impairs apoptosis mediated depletion of activated T cells. 14
In the context of genetic association studies, we have to remember that allele frequencies are known to vary widely within and between populations, irrespective of disease status. Consequently, population stratification15,16 must be taken into account in this kind of study and the characterization of their distribution in the specific populations may provide valuable information.
The aim of this work is to test the association of MS with *15:01 in a sample set that contains a subgroup of Basque population, to check the influence of VDR and sex in this association and to study whether all of these variables affect HLA II gene expression.
Methodology
Blood sample collection
Three hundred and sixty-four MS patients and 513 controls were included in this study (Figure 1). All MS patients were diagnosed with multiple sclerosis according to Polman. 17 Peripheral blood of patients and healthy controls was obtained in the Neurology Department of Hospital Donostia after informed consent was given. Blood extraction was always performed in the early morning and RNA extraction was carried out no more than 2 h after the blood was collected and during this time was kept at 4°C. In all of the cases, 10 ml of blood was collected in EDTA tubes by venipuncture. All procedures have been approved by the hospital’s ethic committee.
Workflow and sample description.
Ethnic origin determination
Ethnic origin determination of patients and controls was carried out based on the linguistic root of the first two surnames. People whose first two surnames had a Basque linguistic root were considered to have a Basque origin. The rest of the people were classified as Spaniards (Otaegui et al, 2004).
DNA and RNA extraction
DNA extraction from white blood cells was carried out following a manual protocol (all samples). 5Prime’s Perfect Pure RNA Blood Kit was used for total white blood cell RNA extraction according to the manufacturer’s instructions (170 patients and 140 controls).
Genotyping
HLA-DRB1*15:01 genotyping of the samples was carried out as previously described 19 using a 7300 Real Time PCR System (Applied Biosystems). Amplification detection of DRB1 was considered as a synonym of the existence of at least one *15:01 copy. Even though unable to distinguish between heterozygous and homozygous carriers of the haplotype, this method was chosen because we considered that the information obtained by this technique was sufficient for the achievement of our goals; also the method was cheaper and faster than other alternatives.
The deeper DRB1 and DRB5 genotyping was performed by both exon 2 PCR and high definition (HD) reverse SSO microbead arrays making use of Lab type kits (One Lambda, Inc.) and a Luminex system, respectively. The presence of DRB5 was tested by PCR-SSP, using the MicroSSP generic class II DNA typing tray (One Lambda, Inc.).
The VDR ApaI and TaqI polymorphisms (Table 1) were identified using PCR-RFLP. A 740-bp fragment generated by PCR was digested with the restriction endonuclease TaqI (Takara Bio) to yield 490 bp and 250 bp long fragments for the ‘T’ allele and 290, 250 and 200 bp long fragments for the ‘C’ allele. After digestion with ApaI, the same PCR product was cut into 515 and 225 bp fragments for the ‘a’ allele, whereas the ‘A’ allele was undigested. The PCR products and the restriction fragments were separated in a 3% agarose gel stained with ethidium bromide, and visualized by a Gene Flash Syngene Bio Imaging system (Syngene). The amplifications were performed using the following primer pair: F-CAGAGCATGGACAGGGAGCAAG; R-GCAACTCCTCATGGGCTGAGGTCTCA.
A summary of the genotyped VDR variants and the nomenclature used to designate them.
Gene expression
cDNA synthesis from total white blood cell RNA was performed using the High Capacity cDNA Reverse Transcription Kit (Applied Biosystems) following the manufacturer’s instructions. For gene expression measurement, DRB1 (Hs99999917_m1), DRB5 (Hs03046116_m1) and DQA1 (Hs03007426) specific Applied Biosystems taqman probes were used. qPCRs were performed in a volume of 10 µl, using the following mix: 5 µl 2x Universal Master Mix (Applied Biosystems), 0.5 µl Specific TaqMan Assay, 2.5 µl H2O and 2 µl of 25 ng/µl cDNA (50 ng). The amplifications were done employing a 7900 Real Time PCR System under the following thermal programme: stage 1, 50°C/2 min; stage 2, 95°C/10 min; stage 3, (95°C/15 s, 60°C/1 min) × 40 repeats. 18S was used as the endogenous control gene, as was previously proven to show great stability among our samples. Ct values were obtained using SDS 2.2.2 software. To ensure that the high sequence variability of the HLA II genes had not impaired our gene expression measurements, we performed a qPCR efficiency assay and found that efficiencies were similar among the groups under comparison for each of the genes making the expression value of the different groups comparable.
Data analysis
Frequencies of each genotype (DRB1*15:01, VDR ApaI and VDR TaqI) were calculated for MS patients and controls and a chi-square analysis was performed in order to find out whether differences in genotype frequencies were marked enough to reach statistical significance and calculate the OR for MS.
dCt values for gene expression were calculated using Microsoft Excel. Statistical analyses were performed by the PASW Statistics 18 (SPSS Inc.). For the statistical analysis, a data distribution normality Kolmogorov–Smirnov test was performed first and, as none of the variables showed a normal distribution, they were subjected to a Mann–Whitney U test (a non-parametric variable distribution comparing analysis) in order to find whether DRB1, DRB5 and DQA1 expression could be affected by *15:01.
Results
Genotyping
Our genotyping data (Table 2) confirms the *15:01 MS association and a higher risk of developing the illness for *15:01 carriers is seen (odds ratio, OR = 1.364; 95% CI = 1.107–1.681). However, no association has been found between VDR variants and the disease when considering all samples together. After data subdivision based on ethnic origin, instead, a statistical tendency of association has been observed between ApaI genotypes and MS (χ2= 5.535; p = 0.063) in Basques.
Frequencies of each genotype in patients and controls, chi-square analysis results and MS odds ratio between 15:01+ and 15:01–
Modulating factors of the *15:01-MS association
On the other hand, the *15:01-MS association was separately tested on each VDR genotype with the aim of detecting possible modulation phenomena. Optimal modulation detection was reached when putting AA and Aa genotypes (A+) together in Apa I and TT and TC (T+) in Taq I (Table 3). A certain degree of modulation of ORs has been detected in both markers (A+ = 1.361 vs. A– = 0.974; T+ = 1.265 vs. T– = 0.874). However, the statistical significance of the *15:01-MS association was lost in all cases.
*15:01–MS association modulation by VDR variants and sex.
15:01+ and 15:01- sample frequencies in both patients and controls, chi-square analysis results and odds ratio values are shown for each subgroup.
A+: AA/Aa, A-: aa, T+: TT/TC, T-: CC, F: females, M: males
The same analysis was performed after sex-based subdivision and a much clearer modulation was found. In fact, in our data, the *15:01-MS association seems to be female specific, with an OR of 1.656, and drops from statistical significance in males (p = 0.784).
Finally, the effect of ethnic origin on *15:01 distribution and the *15:01-MS association has been tested (Figure 2). A significantly tendency of presenting a higher *15:01 frequency was found in Basques when compared with non-Basques (45.1% vs. 34.4%; p = 0.02), a tendency maintained both in patients and controls. However, that tendency did not reach statistical significance (p = 0.087 and p = 0.059). In addition, no remarkable modulation of ORs has been detected (1.392 vs. 1.246) and the *15:01-MS association lost statistical significance in both subgroups (p = 0.159 and p = 0.102), probably due to sample size decrease.
*15:01 distribution differences between Basques and non-Basques both in patients and controls. Chi-square analysis results and odds ratio values are also shown.
Gene expression
Gene expression data of DRB1, DRB5 and DQA1 obtained from qPCR experiments was converted to dCt values using Microsoft Excel. In order to visualize the data and extract more biologically relevant conclusions, all expression values (-dCt-s) of the three genes have been represented in a plot after *15:01 and condition-based data organization (Figure 3). All expression values can visually be grouped into three main clusters: cluster A, high expression values for the three genes that match with *15:01 positive samples; cluster B, high expression values of DRB1 and DQA1 in some *15:01 negative samples; and cluster C, low expression values of DRB1 and DQA1 in some *15:01 negative samples and of DRB5 in all *15:01 negative samples. However, some exceptions to that rule can be seen in Figure 3 (encircled in red: two *15:01- samples with high DRB5 expression values).
Gene expression values after data organization based on *15:01 and condition. The dashed line divides *15:01 positive and *15:01 negative samples.
After the statistical analysis, *15:01 positive samples showed a significant overexpression of DRB1 (p < 0.001), DRB5 (p < 0.001) and DQA1 (p = 0.004) in *15:01+ patients (Table 4). No other variable (age, sex, VDR variants and disease state (relapse/remission) has been found to be linked to HLA II gene expression.
RQ values and Mann–Whitney U test results for a DRB1, DRB5 and DQA1 gene expression comparison between *15:01 positive and *15:01 negative samples both in patients and controls.
p < 0.05; ** < 0.001. RQ; Relative quantification (2-(mean dCt1 - mean dCt2))
We then wondered whether the disease could have an effect in HLA II gene expression independent from and masked by the *15:01 effect. In order to isolate the *15:01 variable, we segmented the data based on it and performed again a Mann–Whitney U test (Table 5). Patients showed a significant underexpression of DRB5 (p = 0.02) in the *15:01 positive group. In contrast, when focusing on *15:01 negative samples, DQA1 was the only gene showing a significant underexpression (p = 0.006) in patients and DRB5 was found to be overexpressed (p = 0.043).
RQ values and Mann–Whitney U test results for an HLA II gene expression comparison between patients and controls in both *15:01+ and *15:01– samples.
p < 0.05; ** < 0.001. RQ; Relative quantification (2-(mean dCt1 - mean dCt2))
Discussion
Our results confirm the already well established HLA-DRB1*15:01–MS association1,20 as we obtained a 1.364 OR for *15:01 haplotype carriers. However, this value lies well below those published in previous works as they range between 2 and 3, approximately.20–24 It is known that population stratification may create a smaller bias in association studies such as the ones involving HLA genes, as they present great allele frequency variability among different ethnic groups. 15 We checked this possibility in our samples and we found no significant change in the OR when comparing Basques with non-Basques (1.39 vs. 1.25).
However, we have to take into account that the ethnic background between cases and controls is very different and that could explain the low OR obtained for the *15:01 haplotype. As Basques have a clearly higher *15:01 frequency and the proportion of Basques is much higher in the controls (41.3% vs. 27.2%), this makes the difference of *15:01 frequency between patients and controls much lower and, thus, the power of the association analysis is decreased. This, in fact, could also explain the relatively low value obtained in the neighboring Biscayan province.24,25
As mentioned, the present study has also revealed a tendency towards a higher *15:01 carrier frequency both in Basque patients and controls when compared with non-Basques, although that tendency did not reach statistical significance. Indeed the *15:01 carrier frequencies both in patients (53%) and controls (41%) are higher than these observed for other European populations such as Swedish (39% and 16%), Serbian (34.6% and 19.7%), German (14.6% in controls) and Spaniards (18% in controls).20,26–28
Our data also shows that *15:01 confers risk for MS only in females. We wondered whether these results could be a consequence of a sampling artefact and tested whether *15:01 distribution is biased by sex in our sample. No statistical differences have been found in *15:01 distribution between females and males (χ2 = 0.007; p = 1.0) and, thus, the *15:01–MS association modulation by sex has been accepted as a reliable conclusion. These results strongly suggest that either *15:01 has no effect on disease appearance in males or its effect is completely diluted by other variables. Thus, it can be concluded that in our data all of the susceptibility linked to the *15:01-MS association comes from females.
Multiple sclerosis is a disease that affects differently males and females, 29 two-thirds of the patients being female. Several causes have been suggested to explain the gender issue such as sex hormones, genetic factors, immune bias and environment, 30 but the underlying cause remains elusive. With regard to that problem, several studies have appeared in recent years pointing to immunological differences as a plausible cause. While some researchers have seen a bigger inflammatory component in females, 30 others have focused their efforts on trying to elucidate the relationship between the HLA genes, sex and MS and have shed some light on the issue. The *15:01 containing HLA-DR2,DQ6 haplotype, which confers risk of undergoing MS, has been found to be more frequent in female than in male patients, 8 the statistically significant difference in *15:01 frequency between MS patients and controls seems to be female specific 7 and the HLA-DR15 phenotype has been found to be associated to sex in MS. 9 Moreover, this modulation by sex of the risk conferred by HLA genes has been reported in other diseases such as narcolepsy-cataplexy 31 and type 1 autoimmune hepatitis, 31 and oestrogen receptor gene polymorphisms have also been found to alter the *15:01-MS association. 23
Concerning the VDR analysis, we have not been able to confirm an association between variants in ApaI or TaqI markers of VDR and the disease when considering all samples together, but a statistical tendency of association between ApaI variants and MS has been observed in Basques. These results highlight, once again, the strong effect of ethnic origin on genetic associations and demonstrate that when ethnic stratification of samples is suspected it must be taken into account to achieve correct interpretation of the results, despite the consequent loss of statistical power. In fact, ethnic origin along with environmental factors such as sunlight exposure and vitamin D intake could help to explain the controversy on the association between VDR variants and MS. VDR variants have also been proposed to modulate the *15:01-MS association.33–35 Niino et al. observed a higher risk of suffering the disease in samples containing both the ‘A’ allele of ApaI and the *15:01 haplotype, compared with ‘A’ negative and *15:01 positive samples. We also made that observation and, in addition, our data suggests that TaqI seems to perform a similar modulation, where *15:01 confers risk only in ‘T’ positive individuals, which is in contradiction to the protective role of the ‘T’ allele proposed by Agliardi et al.
Expression
We found HLA II gene expression differences between patients and controls after *15:01-based data subdivision. We thought that these differences in HLA II gene expression may not be an effect of the disease itself but an effect of the treatments patients were receiving, as most of the drugs used in MS treatment work via immunomodulation. To assess that question, we compared gene expression values between treated and non-treated patients, but no statistically significant differences were found in any of the genes (data not shown). On the other hand, seeing that the trend differs between the *15:01+ and the *15:01- samples, a chi-square analysis was performed with the aim of testing whether treatment frequencies (the proportions of treated patients) were different between the two groups. That analysis did not report significant differences either (χ2 = 0.232; p = 0.789). Thus, treatment does not seem to be the underlying cause of the HLA II gene expression differences observed between MS patients and controls in our sample. However, specific studies are needed in larger samples in which the effect of different drugs may be analysed separately.
Although the *15:01-MS association was well established more than 30 years ago,3,4 the molecular mechanisms underlying this link are unknown. Ramagopalan et al. 34 hypothesized that, as the promoter of HLA-DRB1*1501 contains a vitamin D response element (VDRE), in a vitamin D deficient environment the gene expression of HLA-DRB1*1501 would be low enough to impair auto-reactive T cell depletion in the thymus and lead to MS development. Our results do not support this theory. According to our results, the gene expression of the three HLA II genes in the study, DRB1, DRB5 and DQA1, is consistently high in *15:01 positive samples, both in patients and controls. Moreover, VDR variants seem to have no effect on the expression of these genes.
Finally, we have found that DRB5 gene expression is nearly *15:01 specific. While all *15:01 positive samples showed a high DRB5 expression, in just two of the 117 *15:01 negative samples was DRB5 found to be overexpressed, being nearly absent in most of the samples. These results are consistent with the already well known fact that the DRB5 locus is carried exclusively on DRB1*15 and *16 haplotypes.35,36 In fact, DRB5 expression could be used as a DRB1*15:01 screening tool.
However, some exceptions to the general expression pattern described above have been observed. In two *15:01- negative samples (exception numbers 1 and 2) DRB5 has been found to be highly expressed. In order to clarify the basis of these exceptions, a deeper DRB1 and DRB5 allele determination was performed. For the first exception, the genotyping assays revealed that it is DRB1*0103/*0301, which is completely concordant with our previous *15:01- result, but no presence of the DRB5 locus could be found; even the expression value was clearly high. By contrast, the SSP analysis of the second exception revealed the presence of DRB5, which is compatible with the high expression result, and the DRB1 genotyping revealed it to be *1201/*1601, which is consistent both with the previous *15:01- result and with the literature, as the presence of the DRB5 locus has been described to be closely linked to the DRB1*16 haplotype.
In summary, this study confirms the *15:01-MS female specific association. Even though not directly connected with the disease, the vitamin D receptor seems to act synergistically with the *15:01 haplotype in the development of MS. More efforts must be directed in the future towards clarifying the role of VDR in MS pathogenesis. Finally, *15:01 has proven to be a haplotype consistently linked to a high HLA II gene expression, with a nearly exclusive expression of DRB5.
Footnotes
Acknowledgments
We wish to thank the MS patients and healthy controls that took part in the study. We also thank Asunción Iribarren, Nahikari Pastoriza, Naiara Telletxea and Vanessa Blazquez for their invaluable technical support.
Funding
This work was supported by the Basque Government (grant numbers BFI09.294 to HI and BFI09.206 to MMC), Rio-Hortega (grant number CM09/00129 to TC), the Ilundain Fundazioa, and Fundación 2000 and FIS (grant number PS09/02105).
Conflict of interest statement
The authors have no financial conflicts of interest.