Complex nature of SNP genotype effects on gene expression in primary human leucocytes
- Graham A Heap1,
- Gosia Trynka†2, 3,
- Ritsert C Jansen†2, 4,
- Marcel Bruinenberg2,
- Morris A Swertz2,
- Lotte C Dinesen5,
- Karen A Hunt1,
- Cisca Wijmenga2, 3,
- David A vanHeel1 and
- Lude Franke1, 2, 3Email author
© Heap et al; licensee BioMed Central Ltd. 2009
Received: 03 June 2008
Accepted: 07 January 2009
Published: 07 January 2009
Genome wide association studies have been hugely successful in identifying disease risk variants, yet most variants do not lead to coding changes and how variants influence biological function is usually unknown.
We correlated gene expression and genetic variation in untouched primary leucocytes (n = 110) from individuals with celiac disease – a common condition with multiple risk variants identified. We compared our observations with an EBV-transformed HapMap B cell line dataset (n = 90), and performed a meta-analysis to increase power to detect non-tissue specific effects.
In celiac peripheral blood, 2,315 SNP variants influenced gene expression at 765 different transcripts (< 250 kb from SNP, at FDR = 0.05, cis expression quantitative trait loci, eQTLs). 135 of the detected SNP-probe effects (reflecting 51 unique probes) were also detected in a HapMap B cell line published dataset, all with effects in the same allelic direction. Overall gene expression differences within the two datasets predominantly explain the limited overlap in observed cis-eQTLs. Celiac associated risk variants from two regions, containing genes IL18RAP and CCR3, showed significant cis genotype-expression correlations in the peripheral blood but not in the B cell line datasets. We identified 14 genes where a SNP affected the expression of different probes within the same gene, but in opposite allelic directions. By incorporating genetic variation in co-expression analyses, functional relationships between genes can be more significantly detected.
In conclusion, the complex nature of genotypic effects in human populations makes the use of a relevant tissue, large datasets, and analysis of different exons essential to enable the identification of the function for many genetic risk variants in common diseases.
Human gene expression levels have a strong heritable component [1–4]. At some genes, the variance in gene expression levels is an order of magnitude greater between unrelated individuals, than between identical twins . Quantitative mRNA levels are key regulators of phenotype and represent a link between genetic variation and phenotypic alterations. A term first introduced by Jansen & Nap , genetical genomics aims to identify the genetic variants that affect gene expression. By treating gene expression as a quantitative trait it is possible to correlate gene transcript expression levels with genomic locations such that expression quantitative trait loci (eQTLs) can be identified . In the human genome, cis associations, where a genetic variant affects a transcript that maps to the same locus, have been predominantly reported [3, 8]. Trans effects, where the genetic variant is distant to the transcript loci, are much harder to convincingly identify due to inherent multiple testing problems. Analysis of trans effects involves several magnitudes more statistical tests than for cis effects. Although individual studies have reported human trans associations, no effects have been convincingly replicated in multiple studies identified for the same transcript and variant [1, 9, 10].
Quantitative transcript expression and genotype relationships can be investigated via linkage or association based methodologies. Linkage studies use a genome wide series of markers in recombinant in-bred lines or families, to follow the heritability of a trait, whilst association studies usually compare large number of single nucleotide polymorphisms (SNPs) to transcript levels from unrelated individuals. Despite its extensive use in plant [11, 12], mouse [13, 14] and rat  models, transcript expression and genotype correlation studies have only recently been performed in humans [3, 5, 8].
Human association studies have centered on RNA obtained from leukocytes, predominantly Epstein-Barr virus (EBV) transformed B cell lines from HapMap individuals  or unrelated trios . A large scale genetical genomics linkage study has recently identified more than 1000 cis regulatory loci across the genome, in primary cryopreserved human leukocytes , although these are broad genomic regions rather than variants in more precisely defined LD blocks. However, the former studies are limited by their ability to assess transcripts that are significantly expressed at the point of RNA isolation. Cell extraction methods, cryopreservation and EBV transformation all affect individual mRNA expression levels. These variations in RNA analysis make the choice of analysis tissue of paramount importance.
Here we present an association based genetical genomics study using primary cell RNA from peripheral blood sampled from patients in remission from an immune-mediated disease. Immediate RNA preservation during blood sampling (using the PAXgene system) represents nearly in vivo human physiological gene expression. Celiac disease is a common (1% prevalence), inflammatory condition of the small intestine induced by intake of gluten in wheat, rye and barley. A strong genetic component has been established for disease with a monozygotic concordance of 75%  and 90% of cases possessing the HLA haplotype HLA-DQ2.5  and the remainder mostly have HLA-DQ8 . Despite the role of the HLA, the risk of disease is still greater in HLA matched monozygotic twins compared to HLA matched dizygotic twins . We recently performed a genome wide association and replication study using single nucleotide polymorphisms (SNPs) and identified an additional eight susceptibility loci that predispose to celiac disease [20, 21]. A surprising finding from this study was that seven of the eight regions contained promising candidate genes expressed in leucocytes of the immune system. This suggested the feasibility of genetical genomics approaches using peripheral blood to assess the biological function of the celiac disease associated risk variants. One goal of genetical genomics is to uncover previously unknown biological pathways. If genetic variation affects the expression of a gene in trans, this suggests a biological relationship exists between the two loci. To assess this, considerable amounts of effort have been devoted to the development and application of statistical frameworks that are capable of detecting these trans-eQTLs. However, detection of trans-eQTLs in human populations has proven less successful than in mouse, rat and plant recombinant inbred lines [13, 22]. It has been suggested that the extensive genetic and environmental diversity between human individuals masks many of the existing trans-effects.
We show that numerous cis-eQTLs can be identified through an expression analysis of peripheral blood RNA. We also show many of these are only detectable in peripheral blood RNA, and not in EBV-transformed B cell lines. Through a meta-analysis of these two datasets we identified numerous additional cis-eQTLs. We show that for some cis-eQTLs genetic variation does not lead to overall gene expression changes, but rather leads to shifts in the types of different splice isoforms that are produced. The complex nature of genotypic effects on gene expression limits our ability to fully elucidate the effect on RNA expression level or sequence and combined with the significant multiple testing problems prevented the accurate identification of trans effects.
115 UK celiac disease individuals were recruited from Barts and the London NHS Trust and the Oxford Radcliffe Hospitals NHS Trust after informed consent and with ethical approval. Individuals all had a small bowel endoscopic biopsy diagnosis of celiac disease, median age of 51 (23–88), a median age at diagnosis of 42 (1–75), a male to female sex ratio of 1:3, and a median length of treatment on a gluten free diet of 9.4 years (1–47). Celiac individuals responding to a gluten free diet typically show no detectable inflammation (which we confirmed at the mRNA level in peripheral blood). We also enrolled 22 healthy unrelated UK controls, with a male to female sex ratio of 1:1.2, but of unknown age.
PAXgene RNA Extraction
2.5 ml of peripheral blood was collected into a PAXgene tube (Becton Dickinson, UK, 762165). PAXgene vials were chosen to prevent density gradient centrifugation, immortalization or in vitro cell culture artifacts changing mRNA profiles.
PAXgene tubes were mixed gently and incubated at room temperature for two hours. After collection, tubes were frozen at -20°C for at least 24 hours followed by storage at -80°C. RNA was isolated using the PAXgene Blood RNA isolation kit (Qiagen, UK, 762174). RNA was quantified using the Nanodrop (Nanodrop Technologies, USA). Total RNA integrity was analyzed using an Agilent Bioanalyzer (Agilent Technologies, USA).
Anti-sense RNA synthesis, amplification, purification and hybridization to Illumina expression chips
Anti-sense RNA was synthesized, amplified and purified using the Ambion Illumina TotalPrep Amplification Kit (Ambion, USA) following the manufacturers' protocol. Complementary RNA was hybridized to Illumina HumanRef-8 v2 Whole Genome BeadChips and scanned on the Illumina BeadArray Reader. Data was handled through the Illumina BeadStudio Gene Expression module v3.2. All gene expression (MIAME compliant) and genotype data has been made freely available by submission to GEO under GSE11501.
Five celiac disease samples were excluded from subsequent analysis due to poor median probe intensity correlation with all other samples or incorrect sex assignment, based on an analysis of all the probes that mapped to the non-pseudoautosomal region of chromosome Y. A dataset of 110 celiac disease patients and 22 controls was then used for analysis.
Expression probes were mapped to the cDNA sequence from Ensembl v45_36g  and the NCBI build 36 genome assembly if necessary. Probes that had less than 96% sequence homology or that mapped to multiple loci were removed. Subsequent analyses were confined to autosomal probes, in order to prevent sex specific effects on gene expression. After removal of probes that map to sex chromosomes, data was quantile-quantile normalized .
Celiac disease sample genotypes
All celiac disease patients were genotyped as previously described  using Illumina Infinium HumanHap300v1.0 BeadChips.
Peripheral blood eQTL association analysis and false discovery rate control
257,013 autosomal SNPs were tested for association with expression levels in the 110 celiac disease samples that met analysis criteria of minor allele frequency (MAF) > 0.1, exact Hardy-Weinberg equilibrium P-Value > 0.0001 and call-rate > 0.95. Analyses were confined to those probe-SNP pairs for which the distance from probe genomic midpoint to SNP genomic location was less than 250 kb or 500 kb, depending on the analysis performed. To prevent spurious associations due to outliers, a non-parametric Spearman's rank correlation analysis was performed. In order to correct for multiple testing we controlled the false discovery rate (FDR) . The distribution of all the observed p-values was used to calculate the FDR, by comparing this distribution to a null-distribution, obtained from an identical analysis where the expression phenotypes, relative to the genotypes had been permuted. Through 1,000 permutations the Spearman's rank correlation P-value threshold could be determined that corresponded to an FDR of 0.01 or 0.05.
Validation panel: HapMap CEU samples
We compared the identified cis-eQTLs in the celiac peripheral blood dataset to a published human genetical genomics dataset . We reanalyzed expression data from EBV-transformed B cell lines (further described as HapMap B cell line dataset) for 90 CEU HapMap samples . Analyses were performed as described for the celiac disease samples. To enable a comparison between the celiac peripheral blood dataset and the HapMap B cell line dataset, only SNPs were tested that had been successfully called within HapMap and that were present on the Illumina HumanHap300 platform (257,013 SNPs). Although this is only a subset of all the SNPs that have been called for these HapMap samples, this subset of SNPs is known to capture most genetic Caucasian variation well .
Analysis of over- and underrepresented biological processes and function
We investigated over- or underrepresentation of certain biological processes or functions through an analysis of all significant cis-eQTL genes using the Panther Classification System  (Binomial P-Value, Bonferroni corrected).
Many cis-eQTLs have been detected in human datasets, but only a few trans-eQTLs have been found [1, 3, 8, 10]. These trans-eQTLs imply a biological relationship between the trans-locus and the trans-gene and as such can provide valuable biological insight. If several of these were to be identified, they would allow for the reconstruction of gene networks.
A different, regularly used approach for reconstructing gene networks is by systematically assessing co-expression between pairs of genes. We performed such a co-expression analysis for each probe that comprised a cis-eQTL and contrasted this to an analysis where the genetic component of the cis-eQTL on the probe intensity level had been removed.
We first investigated whether the identified probe that comprised a cis-eQTLs (probe distance of 250 kb, FDR of 0.05) showed co-expression with all other probes that mapped to different chromosomes for both the celiac peripheral blood dataset and the HapMap B cell line dataset. We correlated the measured intensity levels with all other probes that mapped on different chromosomes, through a Spearman's rank correlation analysis. By applying Bonferroni correction to account for multiple testing, sets of significantly correlated transcript-pairs could be identified.
Subsequently we used partial-correlations, enabling us to redo this co-expression analysis while effectively removing the genotypic effect on the probe intensity level for the probes that constitute the significant cis-eQTLs. Again, by applying Bonferroni correction, sets of significantly correlated transcript-pairs could be identified.
Celiac disease is an immune mediated disease, dominated by TH1 cytokine response. Given the importance of tissue specific RNA profiles, we felt that peripheral blood was an appropriate medium to study to investigate celiac disease associated genetic variants. Given the known genetic contribution to disease, we enrolled treated celiac disease individuals rather than healthy controls, leading to an enrichment for celiac disease associated genetic variants and underlying causal variants [20, 28]. For 110 celiac disease samples that passed quality control, both expression at 19,867 transcripts and genotype data for 257,013 SNPs, were analyzed.
Gene expression in celiac disease versus healthy control samples
To obtain the most accurate reflection of mRNA levels in peripheral blood leukocytes, whole blood RNA was immediately fixed during venepuncture in PAXgene vials, giving a reflection of in vivo RNA expression from whole blood. One hundred and fifteen treated celiac patients, all of whom were successfully treated and compliant with a gluten free diet for at least six months, were enrolled. Seventy percent of celiac disease patients were female, (66.6% of adult cases diagnosed with celiac disease in the population are female ). In addition, 22 random healthy control samples were enrolled to obtain a background RNA expression profile, enabling us to determine whether the patients indeed did not show ongoing inflammation. In a sex matched comparison no known inflammatory disease associated cytokines, including IFNG  and IL2  showed significantly increased expression (sex matched Wilcoxon Signed-Ranks Test P value < 0.05) in celiac versus control samples, as was expected since these patients had been treated with a gluten free diet.
Cis-associations of gene expression with SNPs
Summary of cis-eQTL findings from celiac peripheral blood and HapMap B cell line data sets
110 celiac disease samples
(Celiac peripheral blood dataset)
90 Caucasian HapMap samples
(HapMap B cell datasets)
Illumina HumanRef-8 v2 Whole
Genome BeadChip (19,867 mapped transcripts)
Illumina HumanRef-6 v1 Whole
Genome BeadChip (44,791 mapped transcripts)
Illumina Infinium HumanHap300 BeadChip (257,013 SNPs assessed)
Subset of all HapMap Genotypes present on Illumina Infinium HumanHap300 BeadChip (257,013 SNPs assessed)
SNP-Probe midpoint distance
FDR Spearman's Correlation P-Value Threshold
p < 1.67 × 10-5
p < 6.57 × 10-6
p < 1.36 × 10-4
p < 5.47 × 10-5
p < 1.08 × 10-5
p < 4.20 × 10-6
p < 8.79 × 10-5
p < 3.47 × 10-5
Number of performed tests
Number of detected different
Number of unique probes
Number of unique genes
Number of unique SNPs
An identical analysis was performed on publicly available EBV-transformed B cell line expression and genotype data for 90 CEU HapMap individuals . Using this dataset and a SNP-probe distance of 500 kb and FDR of 0.01 or 0.05, 1,799 and 2,839 different SNP-probe effects were detected, respectively. These reflect 536 and 821 unique probes, and 1,601 and 2,464 unique SNPs, respectively (see Table 1). Details of all detected individual cis-eQTLs are available in Additional file 1.
Distance between SNPs and probes that constitute the cis-eQTLs
Meta-analysis celiac peripheral blood and HapMap B cell lines data sets
135 identical SNP-Probe effects (reflecting 135 different SNPs and 51 different probes) had been identified in both the 110 celiac disease samples and in the 90 HapMap B cell line samples (FDR = 0.05). In all cases, the combined meta-P-Value for each of these shared cis-eQTLs was more significant due to the larger sample size, indicating allelic effects in the same direction. A comparison between cis-eQTLs that had been detected in a large scale linkage based genetical genomics study using RNA obtained from peripheral blood mononuclear cells , and the cis-eQTLs we detected, indicates that 57.3% (distance 250 kb, 0.05FDR) of probes, displaying a genotypic effect in the celiac peripheral blood dataset, also show a linkage cis-eQTL signal. Of note is that CCR3 and IL18RAP are both identified as cis-eQTLs in both these datasets.
While it can be assumed that for most of the detected cis-eQTLs indeed the probe expression is affected by genetic variation, it can also be that SNPs, mapping to regions to which the probe hybridizes, may affect hybridization efficacies and result in cis-eQTLs  that are not due to expression differences. For the celiac peripheral blood dataset, 10.0% of all Illumina HumanRef-8 v2 probes map to regions that contain known dbSNP polymorphisms. For the HapMap B cell line dataset, 20.5% of all Illumina HumanRef-6 v1 probes map to known SNPs. For the probes that make up the identified cis-eQTLs (distance 250 kb, FDR = 0.05) this percentage is significantly higher for both the celiac peripheral blood analysis (12,5%, Fisher's Exact test P = 0.02) and even more pronounced for the HapMap B cell line dataset (29.1%, P = 7.76 × 10-11). Numbers within brackets in table 1 denote the number of cis-eQTLs, probes or genes that are potentially due to these primer polymorphisms. If these primer polymorphisms are responsible for different hybridization efficacies, these SNPs should be in LD with the SNPs that make up the cis-eQTLs. We could assess this for those cis-eQTLs where genotype data was available within HapMap Phase II for both the SNP that makes up the eQTL and for the SNPs that map within the probe that constitutes this particular eQTL. We compared the resulting distribution of r2 values to a null distribution, generated by calculating the LD between the SNP that makes up an eQTL and the SNP that mapped immediately adjacent to the primer polymorphism SNP. While the mean LD (mean r2 = 0.27) in the celiac peripheral blood was not significantly higher than the mean LD in the null distribution (mean r2 = 0.25, Wilcoxon Mann-Whitney P-Value = 0.86), this was observed for the HapMap B cell line dataset (mean r2 = 0.46, mean r2 of null distribution = 0.30, Wilcoxon Mann-Whitney P-Value = 6.3 × 10-8).
Differential gene expression between tissue sample types influences cis-eQTL detection
Overrepresented biological pathways
The genes, comprising the significant cis-eQTLs (distance 250 kb, FDR = 0.05), showed an overrepresentation of hydrolase and transferase functions for both datasets, but 'immunity and defense' cis-eQTLs genes were more predominantly detected in the celiac disease peripheral blood dataset than in the HapMap B cell line dataset (see Additional file 2), as might be expected from the differential RNA profiles of the cells under investigation in each dataset.
Significant opposite allelic directions for probes, mapping within the same genes
Genes containing multiple probes that are affected by SNPs that also affect other probes in the same gene, but with opposite allelic directions
Unique Illumina Identifier*
Number of dbSNP polymorphisms known within probe region
Identification of transeffects and conditioned co-expression analysis
We used the celiac disease dataset to assess whether significant trans-eQTLs could be identified (Spearman's rank correlation test P-Value < 1 × 10-12, FDR = 0.05). Two trans-eQTLs were detected: rs2318331 (mapping within COL22A1) with GI_48093066 (NBPF3) and rs12634559 (mapping downstream of IL1RAP) with GI_16306505 (CASP8AP2). However, these trans-eQTLs were not detected in the HapMap B cell line dataset, even after relaxing the nominal Spearman rank correlation P-value to 0.05.
Subsequently, we redid this co-expression analysis after having removed the genotypic effect on the probe intensity level for the probes that constitute the significant cis-eQTLs (Figure 4b). We observed a considerably increased number of significantly correlated transcript-pairs: 54,773 (an increase of 7.8%) and 258,874 (an increase of 53.8%), respectively.
To validate whether this increase in co-expressed transcripts reflects known biology, we assessed a collection of 80,350 known biological interactions (derived on 17 April 2007 from KEGG , BioGrid , Reactome , BIND , HPRD  and IntAct ). For the celiac peripheral blood dataset 91 of the 50,821 (0.179%) significantly correlated transcript-pairs that had been identified in the unconditioned co-expression analysis are known to interact. In the conditioned co-expression analysis 106 of the 54,773 (0.194%) transcript pairs were known to interact, indicating an 8.0% increase in the proportion of known biological interactions that was identified. Comparable results were obtained for the HapMap B cell line dataset: For the unconditioned co-expression analysis 334 of the 168,292 (0.195%) significant transcript-pairs are known to interact. In the conditioned co-expression analysis 634 of the 258.874 (0.245%) significant transcript-pairs are known to biologically interact, representing an increase of 23.4% in the proportion of known interactions among the most significantly co-expressed transcript-pairs.
Figure 4c provides an example for Ubiquitin (UBA52), for which a strong cis-eQTL was observed for probe GI_15451941 with rs2314664 (Spearman's correlation coefficient P-Value = 1.11 × 10-16) in the HapMap B cell line dataset. Within this probe three other polymorphism map (rs6554, rs34040670 and rs3209501), of which rs6554 is in near perfect LD with rs2314664 (r2 = 0.98, HapMap CEU population), suggesting this cis-eQTL does not reflect a real expression difference, but rather a difference in hybridization efficacy. Co-expression with other genes is already present in an unconditioned co-expression analysis, because the distribution of co-expression for UBA52 with all other genes differs significantly from a theoretical null-distribution (Wilcoxon-Mann Whitney P-Value < 1 × 10-12). However, in the conditioned co-expression analysis, this difference is more pronounced (see figure 4c). When confining this analysis to a set of 156 known interactions for UBA52, overall co-expression was significantly stronger for these pairs of genes in the conditioned (mean absolute Spearman's correlation coefficient = 0.39) than in the unconditioned co-expression analysis (mean absolute Spearman's correlation coefficient = 0.20, Wilcoxon-Mann Whitney P-Value < 10-50).
We have demonstrated the use of peripheral blood RNA samples for the detection of cis-eQTLs and have shown that there is strong allelic concordance with cis-eQTLs that also had been detected in a HapMap B cell line dataset. These results indicate that a meta-analysis with larger sample size and hence statistical power results in a considerable increase in the detected cis-eQTL even though the arrays that had been used were different. Some cis-eQTLs can be observed across multiple tissue types, in all cases in the same allelic direction, suggesting a major conserved function. However most of the detected cis-eQTLs in these datasets were only detected in one of the two tissues, suggesting that more insight can be gained in the functional consequences of genetic variation by performing genetical genomics studies using different types of cells and tissues. This point is particularly relevant for identifying the function of risk variants for common diseases: for example study immune tissues for immune-mediated diseases, adipose tissue for obesity traits , brain tissue for neurological traits . Genetical genomics experiments performed in outbred human populations are additionally complex and in addition to tissue specific RNA profiles, care should be taken to investigate alternative RNA sequence events, primer polymorphisms that generate false eQTLs and strong cis-eQTLs that obscure weaker trans effects.
It is attractive to assume most of the observed cis-eQTLs reflect overall gene expression level alterations. However, we did observe 14 genes (Table 2) where different probes showed significant opposite allelic effects. For at least five out of the 14 genes (POFUT2, PTER, MYOM2, ADI1 and HLA-DRB5) polymorphisms within the probe are known to exist in dbSNP. For each of these genes, it may be that one cis-eQTL for instance reflects a real expression difference, whereas the other reflects a hybridization effect due to the presence of these SNPs within the probe. However, recently Kwan et al  provided another potential explanation, as they showed for three of the 14 genes (IRF5, MRPL43 and PTER), using Affymetrix GeneChip Human Exon 1.0 ST Arrays, that different genetic variants can result in premature 3' termination events. The SNPs that make up these cis-eQTLs are all in strong linkage disequilibrium with the SNPs Kwan et al described (IRF5: rs7808907 and rs6969930: D' = 1, R2 = 0.74 rs2863095, MRPL43: rs2863095 and rs12241232: D' = 0.89, R2 = 0.75, PTER: rs7909832 and rs1055340: D' = 1, R2 = 1), supporting our observations. Kwan et al examined different exonic effects through an independent validation using quantitative RT-PCR (out of a total of 25 validated genes) and estimated that only 39% of the detected cis-eQTLs influence overall gene expression levels. For the remaining cis-eQTLs genetic variation results in preliminary terminated transcripts (18%), not initiated transcripts (11%), transcripts that are spliced differentially (26%) or a combination of these (6%). We acknowledge that our use of oligonucleotide arrays, predominantly targeting the 3' end of genes, gives a more limited picture of splicing since only 3' termination events can be seen. However, the data presented above and alluded to by Kwan et al  suggest that SNP genotypes can have a significant effect on alternatively spliced transcript isoforms. This additional layer of complexity should be examined in future genetical genomics experiments to fully elucidate the genotypic effects on RNA.
Although the two trans-effects identified (IL1RAP and CASP8AP2) are interestingly both involved in an apoptosis pathway, these trans-eQTLs were not detected in the HapMap B cell line dataset. These trans-eQTL results are in strong contrast to the number of cis-eQTLs we detected within the two data sets. This results primarily from the limited statistical power to detect trans-eQTLs, due to the number of statistical tests that need to be performed. Additionally, in order to cause a trans-effect, more intermediate biological steps are required that introduces additional biological noise, lessening the correlation between genotype and trans-gene expression. Cis-eQTLs can have a significant impact on the co-expression of genes; this genotypic variation may hinder the identification of significant trans effects. This adds an additional layer of complexity to the analysis of genotypic effects on gene expression in outbred populations.
As it has been one of the goals of genetical genomics to identify biological relationships we suggest that the conditioned co-expression analysis we carried out here, might help to uncover these: We have shown that more known biological relationships can be identified when using genetical genomics to perform co-expression analyses that have been conditioned on genotype. An explanation for this observation is that cis-eQTLs sometimes convolute co-expression, as is exemplified for UBA52, where primer polymorphisms probably affect hybridization characteristics.
This study shows that PAXgene isolated peripheral blood RNA is a powerful resource for investigating functional consequences of genetic variation. We have shown that for some of the cis-eQTLs the functional consequences are more complex than previously assumed. Additionally, these findings imply that biological relationships can be extracted in outbred populations, although in a somewhat different manner than what is commonly used to detect biological relationships through trans-eQTLs in inbred model organisms.
As this study has only combined genetics with genomics, we envision more extensive integrative approaches, incorporating e.g. epigenetics and proteomics, will help to improve the detection of previously unknown biological pathways.
Single Nucleotide Polymorphisms
Expression Quantitative Trait Locus
We thank UK clinicians who collected samples, and sample donors. We thank Erik Sluiters for mapping all probe identifiers. Some statistical analyses were performed using the Genetic Cluster Computer in Amsterdam http://www.geneticcluster.org, which is financially supported by the Netherlands Organization, for Scientific Organization (NWO, grant 480-05-003). We acknowledge funding from Coeliac UK (to D.A.v.H); the Netherlands Organization for Scientific Research (NWO, VICI grant 918-66-620 to C.W.); the Celiac Disease Consortium (an innovative cluster approved by the Netherlands Genomics Initiative and partly funded by the Dutch government (grant BSIK03009 to C.W.)); the Netherlands Genomics Initiative (grant 050-72-425 and fellowship grant to L.F.); the Wellcome Trust (GR068094MA Clinician Scientist Fellowship to D.A.v.H.)
- Dixon AL, Liang L, Moffatt MF, Chen W, Heath S, Wong KC, Taylor J, Burnett E, Gut I, Farrall M, Lathrop GM, Abecasis GR, Cookson WO: A genome-wide association study of global gene expression. Nature genetics. 2007, 39 (10): 1202-1207. 10.1038/ng2109.View ArticlePubMedGoogle Scholar
- Goring HH, Curran JE, Johnson MP, Dyer TD, Charlesworth J, Cole SA, Jowett JB, Abraham LJ, Rainwater DL, Comuzzie AG, Mahaney MC, Almasy L, MacCluer JW, Kissebah AH, Collier GR, Moses EK, Blangero J: Discovery of expression QTLs using large-scale transcriptional profiling in human lymphocytes. Nature genetics. 2007, 39 (10): 1208-1216. 10.1038/ng2119.View ArticlePubMedGoogle Scholar
- Stranger BE, Forrest MS, Clark AG, Minichiello MJ, Deutsch S, Lyle R, Hunt S, Kahl B, Antonarakis SE, Tavare S, Deloukas P, Dermitzakis ET: Genome-wide associations of gene expression variation in humans. PLoS genetics. 2005, 1 (6): e78-10.1371/journal.pgen.0010078.View ArticlePubMedPubMed CentralGoogle Scholar
- Stranger BE, Forrest MS, Dunning M, Ingle CE, Beazley C, Thorne N, Redon R, Bird CP, de Grassi A, Lee C, Tyler-Smith C, Carter N, Scherer SW, Tavare S, Deloukas P, Hurles ME, Dermitzakis ET: Relative impact of nucleotide and copy number variation on gene expression phenotypes. Science. 2007, 315 (5813): 848-853. 10.1126/science.1136678.View ArticlePubMedPubMed CentralGoogle Scholar
- Cheung VG, Conlin LK, Weber TM, Arcaro M, Jen KY, Morley M, Spielman RS: Natural variation in human gene expression assessed in lymphoblastoid cells. Nature genetics. 2003, 33 (3): 422-425. 10.1038/ng1094.View ArticlePubMedGoogle Scholar
- Jansen RC, Nap JP: Genetical genomics: the added value from segregation. Trends Genet. 2001, 17 (7): 388-391. 10.1016/S0168-9525(01)02310-1.View ArticlePubMedGoogle Scholar
- Alberts R, Fu J, Swertz MA, Lubbers LA, Albers CJ, Jansen RC: Combining microarrays and genetic analysis. Briefings in bioinformatics. 2005, 6 (2): 135-145. 10.1093/bib/6.2.135.View ArticlePubMedGoogle Scholar
- Emilsson V, Thorleifsson G, Zhang B, Leonardson AS, Zink F, Zhu J, Carlson S, Helgason A, Walters GB, Gunnarsdottir S, Mouy M, Steinthorsdottir V, Eiriksdottir GH, Bjornsdottir G, Reynisdottir I, Gudbjartsson D, Helgadottir A, Jonasdottir A, Jonasdottir A, Styrkarsdottir U, Gretarsdottir S, Magnusson KP, Stefansson H, Fossdal R, Kristjansson K, Gislason HG, Stefansson T, Leifsson BG, Thorsteinsdottir U, Lamb JR: Genetics of gene expression and its effect on disease. Nature. 2008, 452 (7186): 423-428. 10.1038/nature06758.View ArticlePubMedGoogle Scholar
- Birney E, Stamatoyannopoulos JA, Dutta A, Guigo R, Gingeras TR, Margulies EH, Weng Z, Snyder M, Dermitzakis ET, Thurman RE, Kuehn MS, Taylor CM, Neph S, Koch CM, Asthana S, Malhotra A, Adzhubei I, Greenbaum JA, Andrews RM, Flicek P, Boyle PJ, Cao H, Carter NP, Clelland GK, Davis S, Day N, Dhami P, Dillon SC, Dorschner MO, Fiegler H: Identification and analysis of functional elements in 1% of the human genome by the ENCODE pilot project. Nature. 2007, 447 (7146): 799-816. 10.1038/nature05874.View ArticlePubMedGoogle Scholar
- Myers AJ, Gibbs JR, Webster JA, Rohrer K, Zhao A, Marlowe L, Kaleem M, Leung D, Bryden L, Nath P, Zismann VL, Joshipura K, Huentelman MJ, Hu-Lince D, Coon KD, Craig DW, Pearson JV, Holmans P, Heward CB, Reiman EM, Stephan D, Hardy J: A survey of genetic human cortical gene expression. Nature genetics. 2007, 39 (12): 1494-1499. 10.1038/ng.2007.16.View ArticlePubMedGoogle Scholar
- Keurentjes JJ, Fu J, de Vos CH, Lommen A, Hall RD, Bino RJ, Plas van der LH, Jansen RC, Vreugdenhil D, Koornneef M: The genetics of plant metabolism. Nature genetics. 2006, 38 (7): 842-849. 10.1038/ng1815.View ArticlePubMedGoogle Scholar
- Keurentjes JJ, Fu J, Terpstra IR, Garcia JM, Ackerveken van den G, Snoek LB, Peeters AJ, Vreugdenhil D, Koornneef M, Jansen RC: Regulatory network construction in Arabidopsis by using genome-wide gene expression quantitative trait loci. Proceedings of the National Academy of Sciences of the United States of America. 2007, 104 (5): 1708-1713. 10.1073/pnas.0610429104.View ArticlePubMedPubMed CentralGoogle Scholar
- Petretto E, Mangion J, Dickens NJ, Cook SA, Kumaran MK, Lu H, Fischer J, Maatz H, Kren V, Pravenec M, Hubner N, Aitman TJ: Heritability and tissue specificity of expression quantitative trait loci. PLoS genetics. 2006, 2 (10): e172-10.1371/journal.pgen.0020172.View ArticlePubMedPubMed CentralGoogle Scholar
- Schadt EE, Monks SA, Drake TA, Lusis AJ, Che N, Colinayo V, Ruff TG, Milligan SB, Lamb JR, Cavet G, Linsley PS, Mao M, Stoughton RB, Friend SH: Genetics of gene expression surveyed in maize, mouse and man. Nature. 2003, 422 (6929): 297-302. 10.1038/nature01434.View ArticlePubMedGoogle Scholar
- Hubner N, Wallace CA, Zimdahl H, Petretto E, Schulz H, Maciver F, Mueller M, Hummel O, Monti J, Zidek V, Musilova A, Kren V, Causton H, Game L, Born G, Schmidt S, Muller A, Cook SA, Kurtz TW, Whittaker J, Pravenec M, Aitman TJ: Integrated transcriptional profiling and linkage analysis for identification of genes underlying disease. Nature genetics. 2005, 37 (3): 243-253. 10.1038/ng1522.View ArticlePubMedGoogle Scholar
- Greco L, Romino R, Coto I, Di Cosmo N, Percopo S, Maglio M, Paparo F, Gasperi V, Limongelli MG, Cotichini R, D'Agate C, Tinto N, Sacchetti L, Tosi R, Stazi MA: The first large population based twin study of coeliac disease. Gut. 2002, 50 (5): 624-628. 10.1136/gut.50.5.624.View ArticlePubMedPubMed CentralGoogle Scholar
- van Heel DA, Hunt K, Greco L, Wijmenga C: Genetics in coeliac disease. Best Pract Res Clin Gastroenterol. 2005, 19 (3): 323-39. 10.1016/j.bpg.2005.01.001.View ArticlePubMedGoogle Scholar
- Karell K, Louka AS, Moodie SJ, Ascher H, Clot F, Greco L, Ciclitira PJ, Sollid LM, Partanen J: HLA types in celiac disease patients not carrying the DQA1*05-DQB1*02 (DQ2) heterodimer: results from the European Genetics Cluster on Celiac Disease. Human immunology. 2003, 64 (4): 469-477. 10.1016/S0198-8859(03)00027-2.View ArticlePubMedGoogle Scholar
- Nistico L, Fagnani C, Coto I, Percopo S, Cotichini R, Limongelli MG, Paparo F, D'Alfonso S, Giordano M, Sferlazzas C, Magazzu G, Momigliano-Richiardi P, Greco L, Stazi MA: Concordance, disease progression, and heritability of coeliac disease in Italian twins. Gut. 2006, 55 (6): 803-808. 10.1136/gut.2005.083964.View ArticlePubMedPubMed CentralGoogle Scholar
- van Heel DA, Franke L, Hunt KA, Gwilliam R, Zhernakova A, Inouye M, Wapenaar MC, Barnardo MC, Bethel G, Holmes GK, Feighery C, Jewell D, Kelleher D, Kumar P, Travis S, Walters JR, Sanders DS, Howdle P, Swift J, Playford RJ, McLaren WM, Mearin ML, Mulder CJ, McManus R, McGinnis R, Cardon LR, Deloukas P, Wijmenga C: A genome-wide association study for celiac disease identifies risk variants in the region harboring IL2 and IL21. Nature genetics. 2007, 39 (7): 827-829. 10.1038/ng2058.View ArticlePubMedPubMed CentralGoogle Scholar
- Hunt KA, Zhernakova A, Turner G, Heap GA, Franke L, Bruinenberg M, Romanos J, Dinesen LC, Ryan AW, Panesar D, Gwilliam R, Takeuchi F, McLaren WM, Holmes GK, Howdle PD, Walters JR, Sanders DS, Playford RJ, Trynka G, Mulder CJ, Mearin ML, Verbeek WH, Trimble V, Stevens FM, O'Morain C, Kennedy NP, Kelleher D, Pennington DJ, Strachan DP, McArdle WL: Newly identified genetic risk variants for celiac disease related to the immune response. Nature genetics. 2008, 40 (4): 395-402. 10.1038/ng.102.View ArticlePubMedPubMed CentralGoogle Scholar
- Potokina E, Druka A, Luo Z, Wise R, Waugh R, Kearsey M: Gene expression quantitative trait locus analysis of 16 000 barley genes reveals a complex pattern of genome-wide transcriptional regulation. Plant J. 2008, 53 (1): 90-101.View ArticlePubMedGoogle Scholar
- Hubbard TJ, Aken BL, Beal K, Ballester B, Caccamo M, Chen Y, Clarke L, Coates G, Cunningham F, Cutts T, Down T, Dyer SC, Fitzgerald S, Fernandez-Banet J, Graf S, Haider S, Hammond M, Herrero J, Holland R, Howe K, Howe K, Johnson N, Kahari A, Keefe D, Kokocinski F, Kulesha E, Lawson D, Longden I, Melsopp C, Megy K: Ensembl 2007. Nucleic Acids Res. 2007, D610-617. 10.1093/nar/gkl996. 35 Database
- Bolstad BM, Irizarry RA, Astrand M, Speed TP: A comparison of normalization methods for high density oligonucleotide array data based on variance and bias. Bioinformatics (Oxford, England). 2003, 19 (2): 185-193. 10.1093/bioinformatics/19.2.185.View ArticleGoogle Scholar
- Hochberg Y, Benjamini Y: More powerful procedures for multiple significance testing. Statistics in medicine. 1990, 9 (7): 811-818. 10.1002/sim.4780090710.View ArticlePubMedGoogle Scholar
- Steemers FJ, Gunderson KL: Whole genome genotyping technologies on the BeadArray platform. Biotechnology journal. 2007, 2 (1): 41-49. 10.1002/biot.200600213.View ArticlePubMedGoogle Scholar
- Mi H, Guo N, Kejariwal A, Thomas PD: PANTHER version 6: protein sequence and function evolution data with expanded representation of biological pathways. Nucleic Acids Res. 2007, D247-252. 10.1093/nar/gkl869. 35 Database
- Hunt KA, Zhernakova A, Turner G, Heap GA, Franke L, Bruinenberg M, Romanos J, Dinesen LC, Ryan AW, Panesar D, Gwilliam R, Takeuchi F, McLaren WM, Holmes GK, Howdle PD, Walters JR, Sanders DS, Playford RJ, Trynka G, Mulder CJ, Mearin ML, Verbeek WH, Trimble V, Stevens FM, O'Morain C, Kennedy NP, Kelleher D, Pennington DJ, Strachan DP, McArdle WL: Newly identified genetic risk variants for celiac disease related to the immune response. Nat Genet. 2008, 40 (4): 395-402. 10.1038/ng.102.View ArticlePubMedPubMed CentralGoogle Scholar
- van Heel DA, West J: Recent advances in coeliac disease. Gut. 2006, 55 (7): 1037-1046. 10.1136/gut.2005.075119.View ArticlePubMedPubMed CentralGoogle Scholar
- Anderson RP, van Heel DA, Tye-Din JA, Barnardo M, Salio M, Jewell DP, Hill AV: T cells in peripheral blood after gluten challenge in coeliac disease. Gut. 2005, 54 (9): 1217-1223. 10.1136/gut.2004.059998.View ArticlePubMedPubMed CentralGoogle Scholar
- Ciclitira PJ, Ellis HJ: In vivo gluten ingestion in coeliac disease. Digestive diseases (Basel, Switzerland). 1998, 16 (6): 337-340. 10.1159/000016887.View ArticleGoogle Scholar
- Alberts R, Terpstra P, Li Y, Breitling R, Nap JP, Jansen RC: Sequence polymorphisms cause many false cis eQTLs. PLoS ONE. 2007, 2 (7): e622-10.1371/journal.pone.0000622.View ArticlePubMedPubMed CentralGoogle Scholar
- Kanehisa M, Goto S, Kawashima S, Okuno Y, Hattori M: The KEGG resource for deciphering the genome. Nucleic acids research. 2004, D277-280. 10.1093/nar/gkh063. 32 Database
- Breitkreutz BJ, Stark C, Reguly T, Boucher L, Breitkreutz A, Livstone M, Oughtred R, Lackner DH, Bahler J, Wood V, Dolinski K, Tyers M: The BioGRID Interaction Database: 2008 update. Nucleic acids research. 2008, D637-640. 36 Database
- Vastrik I, D'Eustachio P, Schmidt E, Joshi-Tope G, Gopinath G, Croft D, de Bono B, Gillespie M, Jassal B, Lewis S, Matthews L, Wu G, Birney E, Stein L: Reactome: a knowledge base of biologic pathways and processes. Genome biology. 2007, 8 (3): R39-10.1186/gb-2007-8-3-r39.View ArticlePubMedPubMed CentralGoogle Scholar
- Alfarano C, Andrade CE, Anthony K, Bahroos N, Bajec M, Bantoft K, Betel D, Bobechko B, Boutilier K, Burgess E, Buzadzija K, Cavero R, D'Abreo C, Donaldson I, Dorairajoo D, Dumontier MJ, Dumontier MR, Earles V, Farrall R, Feldman H, Garderman E, Gong Y, Gonzaga R, Grytsan V, Gryz E, Gu V, Haldorsen E, Halupa A, Haw R, Hrvojic A: The Biomolecular Interaction Network Database and related tools 2005 update. Nucleic acids research. 2005, D418-424. 33 Database
- Mishra GR, Suresh M, Kumaran K, Kannabiran N, Suresh S, Bala P, Shivakumar K, Anuradha N, Reddy R, Raghavan TM, Menon S, Hanumanthu G, Gupta M, Upendran S, Gupta S, Mahesh M, Jacob B, Mathew P, Chatterjee P, Arun KS, Sharma S, Chandrika KN, Deshpande N, Palvankar K, Raghavnath R, Krishnakanth R, Karathia H, Rekha B, Nayak R, Vishnupriya G: Human protein reference database – 2006 update. Nucleic acids research. 2006, D411-414. 10.1093/nar/gkj141. 34 Database
- Kerrien S, Alam-Faruque Y, Aranda B, Bancarz I, Bridge A, Derow C, Dimmer E, Feuermann M, Friedrichsen A, Huntley R, Kohler C, Khadake J, Leroy C, Liban A, Lieftink C, Montecchi-Palazzi L, Orchard S, Risse J, Robbe K, Roechert B, Thorneycroft D, Zhang Y, Apweiler R, Hermjakob H: IntAct – open source resource for molecular interaction data. Nucleic acids research. 2007, D561-565. 10.1093/nar/gkl958. 35 Database
- Kwan T, Benovoy D, Dias C, Gurd S, Provencher C, Beaulieu P, Hudson TJ, Sladek R, Majewski J: Genome-wide analysis of transcript isoform variation in humans. Nature genetics. 2008Google Scholar
- The pre-publication history for this paper can be accessed here:http://www.biomedcentral.com/1755-8794/2/1/prepub
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.