causal loci are shared between an eQTL and
GWAS risk loci. This observation can be ex-
plained by either a causal effect or pleiotropy.
Mendelian randomization takes this one step
further, addressing these alternative hypothe-
ses to provide evidence of the direction of
causal effect (i.e., DNA to RNA to disease).
Single-cell eQTL analyses have several ad-
vantages over alternative methods that are
used to map the allelic architecture of tran-
scriptional regulation, such as cellular decon-
volution from bulk RNA-seq data. For example,
scRNA-seq–based approaches can identify
previously uncharacterized and rare cell types,
which are challenging to detect using decon-
volution methods ( 22 , 76 ). scRNA-seq also ac-
curately quantifies transcriptional abundance,
because amplified libraries can be collapsed
back to the level of individual transcript
molecules using unique molecular identifier
(UMI) barcodes. Nevertheless, ongoing work
investigating trans-acting variants and gene-
environment interactions at single-cell reso-
lution is required, particularly in the immune
system, where exposure to antigens or cyto-
kines can trigger changes in the transcrip-
tional profile of cells.
This work brings together genetic epidemi-
ology with scRNA-seq to uncover drivers of
interindividual variation in the immune sys-
tem. Our results demonstrate how segregating
genetic variation influences the expression of
genes that encode proteins involved in critical
immune regulatory and signaling pathways in
a cell type–specific manner. Understanding
the genetic underpinnings of immune system
regulation will have broad implications in the
treatment of autoimmune diseases and infec-
tions, transplantation, and cancer.Materials and methods summary
We collected peripheral blood from 1104 in-
dividuals. After DNA extraction, samples were
genotyped using the Illumina Infinium GlobalScreening Array. Poor genotyping quality,
cryptic relatedness, and ethnic outliers were
removed, yielding 1034 participants. Imputation
was performed using the Michigan Imputation
Server ( 24 ). PMBCs were isolated through
density-gradient centrifugation from heparin-
ized whole blood (8-ml cell preparation tubes;
BD Biosciences Australia; catalog no. 362753),
with live cells isolated with the Miltenyi Dead
Cell Removal Kit (Miltenyi; catalog no. 130-
090-101). Live cells were subsequently pooled
with 12 to 14 participant samples per pool,
which underwent single-cell RNA capture
and barcoding with the Single Cell 3′Library
and Gel Bead Kit (10x Genomics) to target the
capture of 20,000 cells per well. Library prepa-
ration and multiplex sequencing using an
Illumina NovaSeq 2000 generated 49 billion
reads. Reads underwent processing using the
Cell Ranger Single Cell Software Suite (v 2.2.0;
10x Genomics) into FASTQ files, followed by
demultiplexing into their respective pools, andYazaret al.,Science 376 , eabf3041 (2022) 8 April 2022 12 of 14
Table 1. A summary of significant evidence of causation between overlapping GWAS loci and identified eQTLs for autoimmune diseases.The
number of cell types in which causal effect is identified is given in parentheses. For loci with causal effects acting in multiple cell types, multiple independent
eQTLs are often present (table S18). Significance threshold of FDR < 0.5.Disease Loci Genes
Systemic lupus erythematosus 19 AIF1(3),BLK(3),BTN2A1(3),BTN2A2(1),BTN3A2(9),
C6orf48(4),FAM167A(1),HLA-B(2),HLA-C(4),HLA-DMA(1),
HLA-DQA1(2),HLA-DQB1(6),HLA-DQB2(2),HLA-DRB1(3),HLA-DRB5(9),
............................................................................................................................................................................................................................................................................................................................................MICB(1),UBE2L3(3),XXbac-BPG181B23.7(3),ZFP57(5)
Rheumatoid arthritis 37 AIF1(3),ANKRD55(3),B3GALT4(1),C2(1),C6orf48(1),CTLA4(5),
DDX6(3),HLA-A(9),HLA-B(6),HLA-C(12),HLA-DMA(1),HLA-DOB(2),
HLA-DPA1(2),HLA-DPB1(5),HLA-DQA1(11),HLA-DQA2(13),HLA-DQB1(12),
HLA-DQB2(4),HLA-DRB1(9),HLA-DRB5(12),HSD17B8(1),HSPA1B(1),IL6ST(1),
LST1(4),MDC1(1),MICA(1),MMEL1(1),RP11-279F6.3(1),RP11-973H7.4(1),
SKIV2L(2),SYNGR1(6),TAP1(1),TAPBP(4),UQCC2(1),XXbac-BPG181B23.7(2),
............................................................................................................................................................................................................................................................................................................................................XXbac-BPG299F13.17(14),ZFP57(5)
CrohnÕs disease 18 ADCY7(2),BRD7(3),C5orf56(1),CCDC101(1),CTD-2260A17.2(1),CYLD(1),
ERAP2(13),GSDMB(1),IP6K2(1),IRF1(4),ORMDL3(1),RNASET2(9),
............................................................................................................................................................................................................................................................................................................................................SLC22A5(1),SLC2A4RG(1),SNX20(1),SPNS1(3),TUFM(2),UQCRQ(1)
Inflammatory bowel disease 47 ADCY7(3),BRD7(3),C5orf56(1),CCDC101(2),CYLD(1),EGR2(1),ETS2(1),
FCGR3B(1),FYB(1),GMEB2(1),GPANK1(1),GPX1(2),GSDMB(1),
HCG23(1),HLA-DOB(3),HLA-DQA1(8),HLA-DQA2(13),HLA-DQB1(7),
HLA-DQB2(3),HLA-DRB1(4),HLA-DRB5(8),LAMB1(8),LST1(4),
MICB(1),NDUFS2(1),ORMDL3(5),PAPD5(1),PEX13(1),PNMT(1),
RBM6(1),RNASET2(7),RP11-229P13.20(1),RP11-324I22.4(1),RP11-94L15.2(1),
SLC22A5(1),SLC2A4RG(2),STMN3(2),TCTA(1),TNFRSF9(1),
............................................................................................................................................................................................................................................................................................................................................TUFM(8),UBE2L3(4),UQCRQ(1)
Multiple sclerosis 39 AHI1(8),AIF1(5),C2(1),CD6(1),CLEC2D(1),CLECL1(2),DDX39B(1),
DDX6(1),EAF2(2),FCRL3(2),HLA-A(13),HLA-B(14),HLA-C(9),
HLA-DMA(1),HLA-DOB(2),HLA-DPA1(1),HLA-DQA1(11),HLA-DQA2(13),
HLA-DQB1(13),HLA-DQB2(6),HLA-DRB1(9),HLA-DRB5(13),HLA-F(8),
HLA-G(6),HSPA1B(1),LST1(2),MAST3(2),MICA(1),MMEL1(1),MPV17L2(1),
PLEK(2),PSMB9(5),RPS18(9),SKIV2L(1),TYMP(3),VARS2(1),
............................................................................................................................................................................................................................................................................................................................................XXbac-BPG181B23.7(3),XXbac-BPG299F13.17(9),ZNRD1(2)
Ankylosing spondylitis 10 AIF1(2),C6orf48(3),HLA-A(1),HLA-B(9),HLA-C(9),HLA-DQA1(1),
............................................................................................................................................................................................................................................................................................................................................LST1(3),MICB(1),NCR3(2),XXbac-BPG181B23.7(4)
Type 1 diabetes mellitus............................................................................................................................................................................................................................................................................................................................................ 4 HLA-DQA1(9),HLA-DQA2(13),HLA-DQB1(9),HLA-DQB2(3)RESEARCH | RESEARCH ARTICLE