An autoimmune disease risk variant has atransmaster regulatory effect mediated by IRF1 under immune stimulation

Abstract

Functional mechanisms remain unknown for most genetic loci associated to complex human traits and diseases. In this study, we first mapped trans-eQTLs in a data set of primary monocytes stimulated with LPS, and discovered that a risk variant for autoimmune disease, rs17622517 in an intron of C5ORF56, affects the expression of the transcription factor IRF1 20 kb away. The cisregulatory effect on IRF1 is active under early immune stimulus, with a large number of trans-eQTL effects across the genome under late LPS response. Using CRISPRi silencing, we showed that the SNP locus indeed functions as an IRF1 enhancer with widespread transcriptional effects. Genome editing by CRISPR further indicated that rs17622517 is indeed a causal variant in this locus, and recapitulated the LPS-specific trans-eQTL signal. Our results suggest that this common genetic variant affects inter-individual response to immune stimuli via regulation of IRF1. For this autoimmune GWAS locus, our work provides evidence of the causal variant, demonstrates a condition-specific enhancer effect, identifies IRF1 as the likely causal gene in cis, and indicates that overactivation of the downstream immune-related pathway may be the cellular mechanism increasing disease risk. This work not only provides rare experimental validation of a masterregulatory trans-eQTL, but also demonstrates the power of eQTL mapping to build mechanistic hypotheses amenable for experimental follow-up using the CRISPR toolkit.

Co-expression analysis reveals interpretable gene modules controlled by trans-acting genetic variants

“…In case of SLC39A8 there seemed to be a temporal delay with the cis -eQTL being active early in LPS response and trans -eQTL appearing much later after proposed accumulation of the ZIP8 protein and increase in intracellular zinc concentration. Temporal delay has similarly been reported for the trans -eQTLs at the INFB1 ( Fairfax et al, 2014 ) and IRF1 ( Brandt et al, 2020 ) loci. This suggests that if cis and trans effects are separated from each other either in time (early versus late response) or space (different cell types that interact with each other), then this might limit the power of methods that rely on genetically predicted gene expression levels to identify regulatory interactions ( Liu et al, 2018 ; Luijk et al, 2018 ; Wheeler et al, 2019 ) and infer causal models.…”

“…Although motif analysis is often underpowered, it can provide directly testable hypotheses about the trans -eQTL mechanism such as the MTF1 transcription factor that we identified at the SLC39A8 locus. Similar approaches have also been successfully used to characterise trans -eQTLs involving IRF1 and IRF2 transcription factors ( Brandt et al, 2020 ; Fairfax et al, 2014 ).…”

Co-expression analysis reveals interpretable gene modules controlled bytrans-acting genetic variants

“…In case of SLC39A8 there seemed to be a temporal delay with the cis-eQTL being active early in LPS response and trans-eQTL appearing much later after proposed accumulation of the ZIP8 protein and increase in intracellular zinc concentration. Temporal delay has also been reported for the trans-eQTLs at the INFB1 [18] and IRF1 [42] loci. Similarly, stimulation-specific cis-eQTL at CR1 locus in whole blood is not able to explain the full extent of the trans-eQTL observed at the same locus [47].…”

“…Although motif analysis is often underpowered, it can provide directly testable hypotheses about the trans-eQTL mechanism such as the MTF1 transcription factor that we identified at the SLC39A8 locus. Similar approaches have also been successfully used to characterise trans-eQTLs involving IRF1 and IRF2 transcription factors [18,42].…”

GREEN-DB: A framework for the annotation and prioritization of non-coding regulatory variants from whole-genome sequencing data

“…The combination of the information present in GREEN-DB with these prediction scores can effectively capture variants involved in human diseases, as shown by our ability to recapitulate known disease-associated variants from the literature. Indeed, considering a collection of 61 variants from 3 different publications [64][65][66] , representing both close and distant regulatory variants, our annotations allowed us to associate them with the correct controlled gene and classify them as "deleterious" considering a stringent threshold for one (52/61) or multiple (46/61) of the 10 best-performing prediction scores. Specifically, the proposed combination of NCBoost, FATHMM-MKL and ReMM was able to correctly classify as "deleterious" 42 (69%) and 60 (98%) variants using the stringent FDR50 and the more relaxed TPR90 thresholds, respectively.…”