rs3087243 but is in linkage equilibrium (R^2 =
0.5). rs231770-T is similarly associated with
decreasedCTLA4mRNA expression in CD8+
T cells with expression of S100B (CD8S100B)
T cells and is associated with the autoimmune
condition myasthenia gravis ( 59 ).
By linking allelic effects to changes in the
expression of genes known to be implicated
in autoimmune disease, we can support estab-
lished hypotheses and identify previously un-
characterized examples of cellular mechanisms
that underlie conditions and control immune
regulation. By focusing on genes involved in
autoimmune diseases, we evaluate how allelic
effects vary across cell types, highlighting
genes that encode membrane, nuclear, cyto-
plasmic, or endoplasmic reticulum (ER) pro-
teins (Fig. 5, C and D). One example isBACH2,
an essential transcription factor involved in
differentiating memory B and T cells ( 60 ). We
identify rs10944479, which was previously as-
sociated with thyroid peroxidase antibody pos-
itivity and hyperthyroidism ( 61 )andhasan
eQTL effect onBACH2in CD8NCcells. We iden-
tify eQTLs forBACH2in CD4NC, CD8NC, and
BMemcells, although the loci controlling ex-
pression in each cell type are independent of
one another (R^2 = 0 to 0.11; Fig. 5C). We dem-
onstrate that rs60849819-T is associated with a
significant down-regulation ofBACH2in in-
dividuals who are homozygous for the T allele
in BMemcells, and rs207253-A has a similar ef-
fect in CD4NCcells (Fig. 5C).
Another example that provides insight into
autoimmunediseaseisBLK.FiveeSNPswere
identified as being associated withBLKexpres-
sioninCD4NC, CD8+T cells with an effector
memory phenotype (CD8ET), CD8NC,BMem,
and BINcells (Fig. 5D) and are associated with
RA,SLE,Sjögren’s syndrome, and systemic
scleroderma ( 41 , 58 , 62 , 63 ) (Fig. 5D). One of
these loci, rs2736336, results in the differen-
tial expression ofBLKin BMemcells. The dy-
namic eQTL analysis shows that allelic effects
vary significantly across the B cell lineage, with
the largest genetic effects observed in the quan-
tiles of the memory B cells. rs2736336 is asso-
ciated with SLE ( 41 ), and carrying copies of the
autoimmune risk allele has been implicated in
hyperactivation of B cells, with enhanced T cell
costimulatory capacities ( 64 ). These results
suggest that an allelic variation at rs2736336
contributes to interindividual variation in
maintaining tolerance of B lymphocytes. Src
family tyrosine kinases, such asBLK, are crit-
ical components of the signaling pathways
that act downstream of the antigen recep-
tor and determine the strength of the signal
that a cell receives as a consequence of antigen
engagement.
Finally, we sought to evaluate the impact of
eQTLs on cellular composition within the
OneK1K cohort. For each eSNP 1 ,wetested
for the association between an individual’sYazaret al.,Science 376 , eabf3041 (2022) 8 April 2022 8 of 14
CCR6Plasma DCCD4
NCETSOX4CD8
ETNCS100BNK
NKRB
MemINMono
CNCSurface or membraneCDBTN3A1
CD247CD27
CD37CD6
CD63CD83
CLEC2DCLECL1
CRHR1CTLA4
FCRL3DSE
GPR18GNG8
IL12RB2IFNGR2
IL18R1IL2RA
LRRC37A2ITGA4
MMEL1LY 9
PTGIRRGS1
SCAMP3SLC15A2
TMEM258SLC44A2
TNFRSF14UBE2D3Nuclear, cytoplasm or ERp=1.36×10-5p=6.91×10-6rs207253 rs10944479BACH2rs60849819GG GT TT0.00.6GG GT TT
0.00.6GG GT TT0.00.4−0.25GG GT TT0.50GG GT TT−0.10.3−0.25GG GT TT0.75AA GA GG AA GA GG AA GA GG AA GA GG0.000.05−0.10.2−0.100.10
0.000.05AA GA GG AA GA GG AA GA GG−0.250.750.00.1
−0.050.10
CD4NC CD4ET CD4SOX4 CD8ET CD8NC CD8S100B NKNKR Plasma BMem BIN MonoC MonoNC DCp=5.20×10-5-0.1Plasma DCCD4
NCETSOX4CD8
ETNCS100BNK
NKRB
MemINMono
CNCA C TA 2
AHI1BACH2
BATF3CCDC85B
CENPUCENPW
CTSWDDX6
DGKQDRAP1
ETS1ETV7
FIBPG ATA 3
GPX1HHEX
IRF7JAZF1
LBHLY S T
MPHOSPH9NCKIPSD
NUTF2PHF5A
PLCL1PPP5C
PRKCBRGS14
RPS26SESN3
SHMT1SKAP2
SLC2A4RGSNRPC
SP140UBASH3A
UBE2L3ULK3
XBP1ZFP36L1
ZFP90ZNF652ABSELL (rs4987360; CD62L)p=2.29×10-11 p=6.45×10-30
AA AG GG AA AG GG AA AG GG AA AG GG AA AG GG AA AG GG AA AG GGAA AG GG AA AG GG AA AG GG AA AG GG AA AG GG AA AG GG AA AG GG−0.20.6−0.51.0−0.50.50.000.75−0.40.2−0.51.0−0.20.4−0.60.3−11−0.40.4−0.30.6−0.40.8−0.20.2−1.51.0CD4NC CD4ET CD4SOX4 CD8ET CD8NC CD8S100B NKNKR Plasma BMem BIN MonoC MonoNC DCCTLA4rs3087243rs3087243 rs231770p=1.04×10-102p=2.32×10-35 p=1.53×10-14 p=1.42×10-36 p=3.36×10-7AA GA GG AA GA GGAA GA GG AA GA GG AA GA GG−0.05AA GA GG0.150.000.090.00.30.000.250.00.60.000.05CD4NC CD4ET CD4SOX4 CD8ET CD8NC CD8S100B NKNKR Plasma BMem BIN MonoC MonoNC DCCC CT TT CC CT TT0.000.030.000.15BLKp=2.07×10-39 p=9.94×10-25p=1.56×10-22 p=1.04×10-5 p=9.46×10-6
rs2736336 rs2409780rs62489069 rs4841546
CD4NC CD4ET CD4SOX4 CD8ET CD8NC CD8S100B NKNKR Plasma BMem BIN MonoC MonoNC DCBLK
rs922483AA AG GG AA AG GG0.00.2
0.000.04CC CT TT0.000.08CC CT TT CC CT TT CC CT TT0.000.06
0.00.2
0.000.05GG GT TT GG GT TT−0.250.25
0.01.0CC TC TT CC TC TT CC TC TT CC TC TT
0.00.90.00.60.00.4−0.20.2Fig. 5. Genetic variation leads to cell typeÐspecific immune regulation.Cell type–specific eQTLs for genes
known to play a role in immune function and common autoimmune diseases were identified. (A) The eQTL
forSELL(CD62L) exhibits different allelic directions between the lymphoid and myeloid lineages; the effects
for the eSNP rs4987360 are shown. (B) Allelic plots for the inhibitory receptorCTLA4.(C) With regard to
nuclear, cytoplasm, or ER genes, we highlightBACH2, which showed significant and independent eQTLs
in three cell types tagged by eSNPs rs207253, rs10944479, and rs60849819. (D) Cell type–specific eQTLs
for cell surface receptors or membrane-associated proteins implicated in common autoimmune diseases.
Focusing onBLK, we identified cell type–specific eQTLs and expression patterns across cell types.pvalues
are from Spearman’s rank correlation testing. Red lines indicate the allelic effect of significant eQTLs
identified at an FDR less than 0.05.
RESEARCH | RESEARCH ARTICLE