Content uploaded by Mrinal K. Sarkar
Author content
All content in this area was uploaded by Mrinal K. Sarkar on Oct 17, 2017
Content may be subject to copyright.
15 2 VOLUME 18 NUMBER 2 FEBRUARY 2017 nature immunology
ARTICLES
In addition, there is evidence that postmenopausal hormonal
therapy does not increase disease activity or the risk of major flares
in women with SLE5,8,9.
Skin is the biggest organ in humans; it is the front line of immune
protection and is a sensitive indicator of immunological dysregu-
lation10. Skin changes are prominently manifested in autoimmune
diseases such as SLE. Of the eleven criteria for the diagnosis of SLE,
four are cutaneous in nature: malar rash (butterfly-shaped rash across
the cheeks and nose), discoid rash (raised red patches), photosensi-
tivity (skin rash resulting from an unusual reaction to sunlight) and
mucosal ulcers. Collectively, skin involvement is present in 72–85% of
patients with SLE11. Systemic sclerosis, an autoimmune disease with
a female:male prevalence of 11:1, is characterized by skin symptoms
that include thickening and itching12.
To understand the cause of female-biased susceptibility to autoim-
mune diseases in humans, we investigated the sexual dimorphisms of
human skin. We identified a female-biased molecular signature that
was significantly associated with susceptibility to autoimmune dis-
eases. Sex differences extended beyond the signature to genome-wide
co-expression networks involving processes such as complement acti-
vation and phagocytosis. We further identified VGLL3 (‘vestigial-like
Autoimmune diseases are characterized by immune responses to self
antigens that result in tissue damage. It is estimated that autoimmune
diseases affect 7.5% of the US population, compared with a frequency
of 2.8% for cancer and 6.9% for heart diseases, and are among the lead-
ing causes of death and disability. Currently there is no cure, and com-
monly used immunosuppressant treatments can lead to devastating
side effects, such as serious infections and cancer1–3.
Many autoimmune diseases, ranging from systemic disorders, such
as systemic lupus erythematosus (SLE), to organ-specific diseases,
such as Grave’s disease, feature a greater prevalence in females than
in males (female:male, 9:1 (SLE) and 7:1 (Grave’s disease))2,3, whereas
the risk of contracting infectious diseases is higher in men4. Overall,
78% of the people affected with autoimmune diseases are women2,3.
Sex hormones are among the most-studied factors for contributions
to this sex bias. The role of sex hormones has been best studied in
mouse models of SLE, in which treatment with androgen is protec-
tive, whereas treatment with estrogen accelerates disease5. In humans,
however, the relationship between sex hormones and autoimmunity
seems to be more complicated. For example, when SLE occurs in
men, the disease is often more severe, and many autoimmune diseases
commonly have their onset before puberty or after menopause5–7.
1Department of Dermatology, University of Michigan, Ann Arbor, Michigan, USA. 2Department of Computational Medicine and Bioinformatics, University of Michigan,
Ann Arbor, Michigan, USA. 3Department of Biostatistics, School of Public Health, University of Michigan, Ann Arbor, Michigan, USA. 4Department of Internal
Medicine, Division of Nephrology, University of Michigan, Ann Arbor, Michigan, USA. 5Department of Pathology, University of Michigan, Ann Arbor, Michigan, USA.
6Ann Arbor Veterans Affairs Hospital, Ann Arbor, Michigan, USA. 7Department of Internal Medicine, Division of Rheumatology, University of Michigan, Ann Arbor,
Michigan, USA. Correspondence should be addressed to J.E.G. (johanng@med.umich.edu).
Received 2 August; accepted 22 November; published online 19 December 2016; doi:10.1038/ni.3643
A gene network regulated by the transcription factor
VGLL3 as a promoter of sex-biased autoimmune
diseases
Yun Liang1, Lam C Tsoi1–3, Xianying Xing1, Maria A Beamer1, William R Swindell1, Mrinal K Sarkar1,
Celine C Berthier4, Philip E Stuart1, Paul W Harms1,5, Rajan P Nair1, James T Elder1,6, John J Voorhees1,
J Michelle Kahlenberg7 & Johann E Gudjonsson1
Autoimmune diseases affect 7.5% of the US population, and they are among the leading causes of death and disability.
A notable feature of many autoimmune diseases is their greater prevalence in females than in males, but the underlying
mechanisms of this have remained unclear. Through the use of high-resolution global transcriptome analyses, we demonstrated
a female-biased molecular signature associated with susceptibility to autoimmune disease and linked this to extensive
sex-dependent co-expression networks. This signature was independent of biological age and sex-hormone regulation and was
regulated by the transcription factor VGLL3, which also had a strong female-biased expression. On a genome-wide level,
VGLL3-regulated genes had a strong association with multiple autoimmune diseases, including lupus, scleroderma and Sjögren’s
syndrome, and had a prominent transcriptomic overlap with inflammatory processes in cutaneous lupus. These results identified
a VGLL3-regulated network as a previously unknown inflammatory pathway that promotes female-biased autoimmunity.
They demonstrate the importance of studying immunological processes in females and males separately and suggest new
avenues for therapeutic development.
© 2017 Nature America, Inc., part of Springer Nature. All rights reserved.
nature immunology VOLUME 18 NUMBER 2 FEBRUARY 2017 153
family member 3’), one of the sex-biased transcription factors uncovered
in our analyses, as a critical regulator of the female-biased genes encod-
ing inflammatory products, including TNFSF13B (‘tumor necrosis
factor superfamily member 13b’; encoding the B-cell-stimulatory
molecule BAFF) and ITGAM (encoding the integrin αM), which are
a therapeutic target13 and a genetic risk factor for SLE14, respectively.
On a genome-wide level, the targets of VGLL3 had a strong associa-
tion with multiple autoimmune diseases, including lupus, systemic
sclerosis and Sjögren’s syndrome (SS), and had a prominent tran-
scriptomic overlap with inflammatory processes in cutaneous lupus.
VGLL3 was also required for the optimal response to interferons
in monocytes and salivary gland cells. Our results uncovered a sex-
hormone-independent mechanism that predisposes females to
autoimmune diseases, and they provided a foundation for the
development of novel, targeted treatment measures.
RESULTS
Sex differences in human skin
We analyzed 31 female and 51 male skin biopsy samples from healthy
donors by whole-genome RNA-sequencing (RNA-seq) analyses.
We identified 661 genes that were expressed differentially by the two
sexes (false-discovery rate (FDR), ≤ 0.1) (Supplementary Table 1).
268 genes were upregulated in males (i.e., were male biased), including
26 genes on the Y chromosome and six genes on the X chromosome.
393 genes were upregulated in females (i.e., were female biased),
including 55 genes on the X chromosome (Fig. 1a). As expected,
known sex-biased gene expression, such as that of the long noncoding
RNA XIST and ZFY (zinc-finger protein, Y-linked), was reproduced
in our data sets (Fig. 1b). Of the 55 genes that escaped X-inactivation,
seven have orthologs on the Y chromosome, and their expression
in males potentially enables dosage compensation (Supplementary
Fig. 1a–d). 48 genes did not have Y-linked orthologs (Supplementary
Fig. 1e–g), in support of the idea that incomplete X-inactivation
might contribute to sexually dimorphic traits.
Sex differences might extend beyond the 661 differentially
expressed genes (DEGs) to their associated networks. To test this, we
conducted gene–gene correlation analysis between the sex-specific
DEGs and all other genes in males and females, separately. Indeed,
sex-biased co-expression correlations were found from the gene-pair
level to the pathway-wide and genome-wide levels, and this included
genes encoding products involved in various immunological pro-
cesses such as phagocytosis and complement activation (Fig. 1c,d and
Supplementary Fig. 2a). In total, we identified 124,521 gene–gene
pairs that showed significant results in only female samples (FDR ≤
0.1) but not in male samples (P > 0.5); conversely, 158,303 gene–gene
pairs showed significance in only male samples (FDR ≤ 0.1) but not in
Upregulated
Chr1 M
a
c
b
d
80
60
40
20
20
10
80
60
40
20
0
100
30
100
80
60
40
20
0
20
10
0
80
60
40
20
0
VGLL3
ZFY
XIST
F
M
F
M
Male
4,000
11,000
10,000
9,000
8,000
SEPT2
7,000
6,000
3,500
ATRN
3,000
300
3,600
3,200
2,800
ATRN
2,400
400 600 800
ITGAM
400 500
ITGAM
600 700 25 50
PTX3
75
2,500
Female Female
10,000
9,000
8,000
7,000
SEPT2
6,000
30 60
PTX3
90
Male
F
Chr2
Chr3
Chr4
Chr5
Chr6
Chr7
Chr8
Chr9
Chr10
Chr11
Chr12
Chr13
Chr14
Chr15
Chr16
Chr17
Chr18
Chr19
Chr20
Chr21
Chr22
Chrx
Chry
0 50
C1S
CFB
C3
C1R
SLC11A1
DOCK2
C3
CD36
PTX3
FCER1G
CADM1
SLC11A1
FYN
DOCK2
ITGAM
100
Distance (Mb)
150 200 250
Female
Male
0.1 0.2 0.3 0.4 0.5 0.1 0.2 0.3 0.4 0.5 0.6 0.1 0.2 0.3 0.4
Figure 1 Identification of sex-biased genes from human skin biopsies. (a) Chromosomal locations of female- and male-biased genes. (b) Raw RNA-seq
reads for XIST, ZFY and VGLL3 in skin from females (F) and males (M). (c) Sex-biased co-expression correlation for genes encoding products in the
functional categories of complement activation (left), phagocytosis regulation (middle) and T cell proliferation (right). (d) Sex-specific co-expression
correlation for the ITGAM–ATRN and PTX3–SEPT2 gene pairs. Data are representative of two experiments b,c (in d, n = 51 males and n = 31 females).
ARTICLES
© 2017 Nature America, Inc., part of Springer Nature. All rights reserved.
15 4 VOLUME 18 NUMBER 2 FEBRUARY 2017 nature immunology
ARTICLES
female samples (P > 0.5). We further compared the correlation results
with those obtained from published microarray data sets of the skin15
and obtained high correlation concordance (Supplementary Fig. 2b).
This finding indicated the presence of sex-biased, genome-wide net-
works and suggested that the biological effect of the sex bias was much
greater than that anticipated from the initial list of DEGs.
Association between female-biased genes and autoimmunity
Analysis of biological functions that showed enrichment in the group
of DEGs revealed that the female-biased genes showed enrichment
for those encoding products involved in immunological and inflam-
matory processes, but the male-biased genes did not (Fig. 2a,b and
Supplementary Fig. 3a). In addition, network analysis organized
female-biased genes mainly into those encoding products involved
in complement-activation pathways known to be dysregulated in
autoimmune diseases (Fig. 2a).
The sex-specific upregulation, in females, of genes encoding prod-
ucts related to immunity led us to hypothesize that the female-biased
gene signature associates with high susceptibility to autoimmune dis-
eases. We detected significant overlap between female-biased genes
and common disease loci associated with SLE and systemic sclero-
sis, two female-dominant autoimmune diseases (P < 0.05; Fig. 2c).
Among female-biased genes, the female:male prevalence ratio was
significantly correlated with enrichment for disease-associated loci, as
measured by P values (Spearman coefficient (
ρ
) = 0.83; P = 1.5 × 10−2),
and with the observed-to-expected change (fold value;
ρ
= 0.88;
P = 7.2 × 10−3; Fig. 2c). There was no association between male-
biased genes and autoimmune diseases (data not shown). We also
implemented a sampling approach to estimate the empirical P values
for the enrichment, and the results were highly concordant with the
hypergeometric enrichment analysis (Supplementary Fig. 3b,c).
We confirmed the increased expression, in female skin, of the genes
encoding products related to immunity (Fig. 3a,b), including the gene
encoding BAFF (TNFSF13B; called ‘BAFF’ here), whose expression is
frequently increased in patients with SLE and that served as the first
approved target for a biologic therapy for SLE13, and ITGAM, whose
variants are associated with susceptibility to SLE14. Consistent with
the systemic feature of SLE symptoms, the female-biased pattern of
risk-gene expression was not restricted to skin but was also detected
to variable degrees in monocytes, B cells and T cells (Fig. 3c,d and
Supplementary Fig. 3d). We further observed higher expression
of the same female-biased risk genes in skin and monocytes from
patients with SLE than in that from sex-matched healthy control sub-
jects (Fig. 3e,f), which supported the proposal of the involvement of
their products in the pathogenesis of SLE. Collectively, these results
suggested that the female-biased inflammatory genes were associated
with high susceptibility to autoimmune processes.
Molecular mechanism for female-biased risk-gene expression
To search for the molecular mechanism underlying sex-biased risk-
gene expression, we used RNA-seq to assess the effects of physiological
or 100-fold-concentrated levels of estradiol or testosterone on gene
expression in primary human keratinocytes. Treatment with the sex
hormones did not alter the expressions of the female-biased genes
encoding products related to immunity (Fig. 3g,h). More broadly,
none of the 661 sex-specific DEGs were significantly regulated by
treatment with estradiol or testosterone in the settings assessed (data
not shown). To address the possibility that keratinocytes lose their
responsiveness to sex hormones after ex vivo culture, we turned to
our transcriptomics data of skin and reasoned that the expression of
sex-biased genes would decrease with age if they were regulated by
sex hormones. We observed no correlation between expression and
biological age for the genes investigated (Fig. 3i and Supplementary
Fig. 3e–i). Overall, we found no compelling evidence in support of the
direct regulation of female-biased risk genes by sex hormones.
Another potential mechanism for the regulation of the risk
genes would be via sex-biased transcription factors. We identified
eight putative female-biased transcription factors, on the basis of
their annotated function, from the 100 genes that were most sig-
nificantly female biased (ranked by FDR; Supplementary Tables 1
and 2). From transcriptomic analyses of primary keratinocytes
from three different female subjects, we found that six of the eight
genes were expressed in keratinocytes (Supplementary Table 2).
We were able to achieve efficient knockdown of five of the six genes by
RNA-mediated interference (RNAi) (Supplementary Fig. 4a–e and
Supplementary Table 2). We found that RNAi of VGLL3 decreased
the abundance of ITGAM and BAFF mRNA, but RNAi of KDM6A
(which encodes the lysine demethylase UTX; gene called ‘UTX’
here), ZFX (an X-linked gene that encodes the transcription factor
ZFX), FEZ1 (which encodes the adaptor FEZ1) or FHL1 (‘four and
a half Lin11, Isl-1 and Mec-3 (LIM) domains’) did not (Fig. 4a and
Supplementary Fig. 4a–e). Knockdown of VGLL3 did not affect the
expression of UTX or ZFX (Supplementary Fig. 4f), which suggested
that the effect of VGLL3 on the SLE-associated genes ITGAM and
BAFF was specific.
VGLL3 is a homolog of the Drosophila gene vg (‘vestigial’) , which
encodes a cofactor of Scalloped, the homolog of the transcription
CFB
CADM1
ITGAM PTX3 DOCK2
FYN
TNFSF13B
FCER2
FCER1G
C3
Complement and
humoral responses
T cell
proliferation
Phagocytosis
regulaton
HES1
HOXA7
SIX2
HOXA6
HOXA5
HOXA3 MEN1
OVOL2
FOXF1
System development
Morphogenesis
Factor(Z)
Atopic dermatitis
Basal cell carcinoma
Leprosy
Melanoma
Psoriasis
Systemic lupus erythematosus
Systemic lupus erythematosus
and systemic sclerosis
Systemic sclerosis
0 1 2 3 40
0
1
1
0
Enrichment in disease-
associated loci: (P value (–log))
2
3
4
5
Female/male prevalence
(log scale)
Female/male prevalence
(log scale)
1 2 3 4
c
T cell proliferation
0246
a b
P value (–log)
Triggering of complement
Adaptive immune response
Phagocytosis regulation
0 21 43 5
Enrichment in disease-
associated loci: (log fold)
Sequence-specific
DNA binding
Transcription
factor activity
P value (–log)
Regulation of
differentiation
Figure 2 Female-biased genes encode products associated with
autoimmune processes. (a) Enrichment of specific functional categories
among female-biased genes. (b) Enrichment of specific functional
categories among male-biased genes. (c) Correlation between enrichment
for disease-associated loci and female:male disease prevalence ratio
for female-biased DEGs. Data are representative of the analysis of
393 female-biased genes (a) and 268 male-biased genes (b), or are
representative of eight complex traits among 171 susceptibility loci (c).
© 2017 Nature America, Inc., part of Springer Nature. All rights reserved.
nature immunology VOLUME 18 NUMBER 2 FEBRUARY 2017 155
ARTICLES
enhancer TEF-1 (ref. 16), and it has sex-dependent dominance
in salmon17. The higher expression of VGLL3 (FDR = 7.2 × 10−4)
in females was confirmed in skin and in keratinocytes (Fig. 4b,c).
We found that, consistent with its transcriptional functions, VGLL3
had a phenotype of localization to the nucleus in healthy skin that was
more distinct in females than in males (Fig. 4d). In contrast, in lesional
skin from males and females with SLE, VGLL3 was concentrated
in the nuclei of cells from either sex (Fig. 4d), indicative of disease-
dependent regulation.
RNA-seq analysis of female primary human keratinocytes showed
that, in addition to downregulating BAFF and ITGAM, knockdown
of VGLL3 downregulated seven of ten female-biased genes encoding
2.0
a c
d
e
h
i
fg
Estradiol
**
**
*
*
***
*
*
*
*
*
*
**
*
***
*
1.5
1.0
0.5
0.0
Expression (relative)
2.0
2.5
1.5
1.0
0.5
0.0
2.0
2.5
0 nM
1 nM
100 nM
1.5
1.0
0.5
0.0
2
3
1
0
N
SLE
N
SLE
Testosterone
0 nM
300
200
100
0
20 40
Age (years)
60 80
20 nM
2 µM
120
160
00
5
10
15
10
20
30
40
80
Expression (relative)
Expression (relative)
Expression (relative)
Expression (relative)
Expression (relative)
2.0
1.5
1.0
0.5
0.0
Eexpression (relative)
BAFF expression (FPKM)
BAFF
F
MF
M
F
M
BAFF
BAFF
BAFF
BAFF
BAFF
BAFF
ITGAM
ITGAM
ITGAM
ITGAM
ITGAM
ITGAM
ITGAM
C3
C3
C3
C3
C3
CFB
CFB
CFB
CFB KRT4
KRT4
CFB
CFB
DOCK2
bBAFF
Female
Male
C3 DOCK2
DOCK2
DOCK2
DOCK2
FCER1G
FCER1G
FCER1G
FCER1G
Figure 3 Expression of female-biased genes encoding products related to immunity is dependent on SLE disease states but not on sex-hormone levels.
(a) Quantitative RT–PCR (qRT–PCR) analysis of female-biased genes encoding products related to immunity in whole skin of healthy humans (n = 5 per sex);
results are presented relative to the mean of the expression levels in females. (b) Immunohistochemistry of the expression of genes as in a in the skin of
healthy humans. Scale bars, 50 µm. (c) qRT–PCR analysis of genes as in a in monocytes of healthy humans (n = 9 per sex); results presented
as in a. (d) qRT–PCR analysis of the products of genes as in a in B cells of healthy humans (n = 9 female; n = 8 male); results presented as in a.
(e) qRT–PCR analysis of genes as in a in the skin of patients with SLE and of healthy subjects (N) (n = 5 per group); results are presented relative to
the mean of the expression levels in healthy subjects. (f) qRT–PCR analysis of genes as in a in monocytes of patients with SLE and of healthy subjects
(n = 3 per group); results presented as in e. (g,h) RNA-seq analysis of genes as in a in primary human keratinocytes treated with various concentrations
(key) of estradiol (g) or testosterone (h); results are presented relative to the mean of the expression levels in untreated cells. (i) RNA-seq analysis of
BAFF in skin biopsies obtained from humans of various age (horizontal axis), presented as fragments per kilobase of exon per million fragments mapped
(FPKM). Each symbol (a–c,e–h) represents an individual donor; small horizontal lines indicate the mean (± s.e.m.). *P < 0.05 (two-tailed Student’s
t-test). Data are representative of three independent experiments (b), show results from three independent experiments (g,h) or show analysis for the
indicated number of subjects (a,c–f,i).
© 2017 Nature America, Inc., part of Springer Nature. All rights reserved.
15 6 VOLUME 18 NUMBER 2 FEBRUARY 2017 nature immunology
ARTICLES
0.0
ITGAM BAFF C3
0.5
1.0
1.5
a
b
d
e h
c
Expression (relative)
Scr Ri
VGLL3 Ri
FEZ Ri
UTX Ri
ZFX Ri
FHL Ri
***
0.0
F M F M
0.5
1.0
1.5
VGLL3 expression (relative)
*
0.0
0.5
1.0
1.5
VGLL3 expression (relative)
*
EOMES
IL24
TXNIP
CDK6
CSF1
CA9
KLRAP1
SPINK5
SAA1
IL7R
IL32
S100A9
BMPER
FMOD
CASP14
TP53INP1
TNFSF18
AEBP1
DSC1
TNC
CD248
LMO2
IL17RE
COL1A2
TNFSF13B
P2RY1
ICAM1
TET1
IL20
TSPAN7
CSPG4
ETS1
ADAMTS4
NGFR
LSP1
C3AR1
PDGFD
IRAK3
PLAT
C1S
MME
PIK3R1
PSTPIP1
ITGAM
POU3F1
IL1RAP
OLR1
TNFSF15
SDC2
IL33
IL6R
RUNX2
MYB
PROS1
TNFSF4
IL7
MMP9
MMP3
APLN
2.0
1.5
1.0
0.5
0.0
–2.0 –1.0 0.0 1.0 2.0
Density
Expression (log2 fold)
SCLE genes
Non-SCLE
genes
**
Expression (log
2
fold)
–3 0
Female Male Female Male
Normal SLE
f
Fold change
1.00
0.13
Kinase
Protein
Receptor
Co-factor
2 genes are associated by co-citation
2 genes are associated by expert curation
Gene A activates gene B
Gene A inhibits gene B
Gene A has a known transcription
factor binding site matrix and gene
B has a corresponding binding site
in one of its promoters
g
N SCLE VGLL3 Ri
Expression (log
2
fold)
–2
2
BAFF
VGLL3 Ri-1 VGLL3 Ri-2
5.01 × 10
–10
6.55 × 10
–5
3.84 × 10
–4
3.80 × 10
–2
1.19 × 10
–13
4.29 × 10
–2
4.77 × 10
–3
9.73 × 10
–4
q-Value
FCER1G
ITGAM
C1S
CD36
C3
CADM1
C1R
CFB
DOCK2
FYN
Figure 4 VGLL3 regulates genes associated with autoimmune diseases. (a) qRT–PCR analysis of ITGAM, BAFF and C3 in primary human keratinocytes
after RNAi with small interfering RNA (siRNA) with a scrambled sequence (Scr Ri) or siRNA targeting VGLL3, UTX, ZFX, FEZ or FHL (Ri; key); results
are presented relative to the mean of the expression levels in cells transfected with Scr Ri. (b) qRT–PCR analysis of VGLL3 in skin from healthy
females and males (n = 4 per group); results are presented relative to the mean of the expression levels in females. (c) qRT–PCR analysis of VGLL3 in
primary human keratinocytes (n = 4 donors per group); results presented as in b. (d) Immunohistochemistry of VGLL3 in skin from healthy subjects
(Normal) and patients with SLE (SLE); second row of each pair is an enlargement of the area outlined above. Scale bars, 50 µm (main images) or 4 µm
(magnified images). (e) RNA-seq analysis of the ten female-biased immunological transcripts (left margin) in primary human keratinocytes after RNAi
of VGLL3, presented as expression (log2 fold) and q value (right margin). (f) Literature-based network analysis of VGLL3-regulated genes encoding
products related to autoimmune disease. (g) Expression (log2 fold) of VGLL3 targets in the skin of healthy subjects (N) and patients with SCLE (SCLE),
as well as in keratinocytes after RNAi of VGLL3 (right). (h) Density plot of the expression (log2 fold) of genes upregulated in SCLE (SCLE genes) and
genes not upregulated in SCLE (Non-SCLE genes), assessed after knockdown of VGLL3. Each symbol (b,c) represents an individual donor; small
horizontal lines indicate the mean (± s.e.m.). *P < 0.05 and **P = 2.53 × 10−8 (two-tailed Student’s t-test (a–c) or Mann–Whitney–Wilcoxon test (h)).
Data are results from three (a) or two (e) independent experiments, are results from the indicated number of subjects (b,c,g) or from the analysis of 208
genes (f), or are representative of three (d) or five (h) independent experiments.
© 2017 Nature America, Inc., part of Springer Nature. All rights reserved.
nature immunology VOLUME 18 NUMBER 2 FEBRUARY 2017 157
ARTICLES
products related to immunity that were expressed in proliferating or
post-confluent keratinocytes (Fig. 4e). At a threshold of q < 0.05 and
|log2 fold| > 0.5, there were a total of 208 genes whose expression was
decreased by VGLL3-specific RNAi in keratinocytes (Supplementary
Table 3). To investigate whether genetic variants that affect the function
or expression of VGLL3 would also affect the expression of the targets
of VGLL3, we conducted expression quantitative trait loci (eQTL) anal-
ysis surrounding the region within 1 Mb (upstream or downstream)
of VGLL3. We observed the strongest cis-eQTL (i.e., those that are
within 1 Mb of the gene’s boundaries) signal at chromosome 3, posi-
tion 87902673 (P = 4 × 10−5; Supplementary Fig. 5a). Furthermore,
we identified nine targets of VGLL3 that were substantially
–0.5
0.0
0.5
Expression
(log fold) in lSSc
–1
0
1
Expression
(log fold) in morphea
–0.5 0.0 0.5 1.0 1.5
10
20
30
40
Target expression
P = 0.0101
–0.5 0.0 0.5 1.0
Target expression
lSSc
N
Morphea
N
5
10
15
20
25
30
35
40
45
0.0
0.2
0.4
0.6
0.8
1.0
a
d
g h
e f
b
c
Density
0.18
–0.05 0 0.05 0.1 0.15 0.2
P value (mean signed log
10
)
0.0
0.2
0.4
0.6
0.8
1.0
Density
–0.2 –0.1 0 0.1 0.2 0.3
P value (mean signed log
10
)
Targets Targets
0.292
Morphea-Rd
N-Rd
lSSc-Rd
N-Rd
–0.5 0.0 0.5 1.0
5
10
15
20
25
Non-target expression
–0.5 0.0 0.5 1.0 1.5
10
20
30
40
Non-target expression
Figure 5 VGLL3 targets encode products involved in multiple autoimmune
conditions. (a,b) Density plot of the null distribution for the P value (mean
signed log10) of the mean expression of VGLL3 targets (arrows) in limited
scleroderma (a) and morphea (b). (c,d) Expression of VGLL3 targets
(horizontal axes; ranked by expression (low (left) to high (right)) in samples
form healthy donors) in limited scleroderma (lSSc) (c) and morphea (d).
(e–h) Expression of VGLL3 targets (e,f) and genes that are not targets of
VGLL3 (Non-target) (g,h) in skin from donors with lSSc (e,g) or morphea
(f,h) or from healthy donors (N). P < 0.001 (a,b), P = 0.0412 (e),
P = 0.0101 (f), P = 0.8686 (g) and P = 0.6449 (h) (Mann–Whitney
U-test). Data are representative of the analysis of 16 patients with lSSc
and 15 healthy subjects (a) or of five patients with morphea and 15
healthy subjects (b) (2,000 simulation trials were used to generate the
null distributions shown in a,b) or are representative of five independent
analyses (c–h).
associated with the VGLL3-associated cis-eQTLs (Supplementary
Fig. 5b), indicative of a trans-eQTL effect. We observed considerable
enrichment for VGLL3 targets among the female-biased genes (P = 7.7
× 10−7) but not among the male-biased genes. VGLL3 targets showed
enrichment for ten pathways related to immunity (Supplementary
Fig. 5c), and autoimmune diseases were among the top three disease
states that showed enrichment for VGLL3 targets (97 genes (47% of
VGLL3 targets); P = 3.63 × 10−12). Network analysis of the 97 genes
revealed additional nodes of autoimmune pathogenesis (Fig. 4f).
Levels of the matrix metallopeptidase MMP-9 are greater in patients
with SLE, systemic sclerosis, multiple sclerosis (MS), SS, polymyositis
or rheumatoid arthritis (RA) than in healthy people18. Genome-wide
association studies have shown that variants of ETS1, which encodes
a transcription factor, confer susceptibility to SLE, RA and ankylosing
spondylitis19. Excess interleukin 7 (IL-7) is present in patients with
SS, MS or RA, and IL-7 promotes autoimmunity in lupus mice and
can be used to predictsclinical response to interferon-β (IFN-β) in
patients with MS20–22. Expression of the adhesion molecule ICAM-1
is upregulated in the brains of patients with MS and in mice with lupus
nephritis and arthritis, and its use has been advocated for controlling
autoimmune diseases23–25. Collectively, these lines of evidence sup-
port the proposal of a role for VGLL3 in promoting the expression of
multiple genes related to autoimmune disease.
VGLL3 targets in autoimmune diseases
If VGLL3 contributes to higher susceptibility to autoimmune diseases
in females, then one prediction would be that its targets are associ-
ated more tightly with such female-biased diseases than with diseases
that do not exhibit significant sex differences. To investigate this, we
performed transcriptomic analysis of skin biopsies from patients with
subacute cutaneous lupus erythematosus (SCLE), morphea or systemic
sclerosis. SCLE is a female-biased, lupus-specific eruption that fea-
tures prominent skin involvement. We first attempted to identify the
subset of VGLL3 targets whose expression was upregulated in SCLE,
using less-stringent criteria at the identification stage. Analysis of the
VGLL3 targets (genes decreased by RNAi of VGLL3 at a threshold
of q < 0.05 and a change in expression of <0.8-fold) and the gene sets
dysregulated in SCLE (genes upregulated in SCLE at a threshold of
q < 0.05 and a change in expression of ≥1.5-fold) revealed a significant
overlap in these genes (P = 2.9 × 10−5); we found 51 genes whose expres-
sion was increased in SCLE that were positively regulated by VGLL3
(Fig. 4g and Supplementary Table 4). The overlap included the true
type I interferon (IFN-α and IFN-β)–response genes LY6E, OAS1,
MX1 and IFI44 (refs. 26,27) (Fig. 4g and Supplementary Table 4),
consistent with the proposal of a central role for type I interfer-
ons in the pathogenesis of SLE28,29. Similarly, we found that after
knockdown of VGLL3, genes that were normally upregulated in
SCLE (Supplementary Fig. 6a and Supplementary Table 5) were
significantly more downregulated genome wide than ‘non-SCLE’
genes (genes not upregulated in SCLE) (P = 2.53 × 10−8) (Fig. 4h). In
contrast, genes whose expression was increased in plaque psoriasis
(Supplementary Table 5), a chronic inflammatory skin condition
that has no sex bias30, showed minimal correlation with regulation by
VGLL3 (Supplementary Fig. 6b,c). Likewise, genes whose expression
was increased in SCLE showed minimal correlation with regulation
by FEZ1 (Supplementary Fig. 6d), as knockdown of its expres-
sion did not reduce the expression of female-biased genes encoding
products related to autoimmunity (Fig. 4a). Genes whose expression
was increased in SCLE also showed minimal correlation with those
regulated by FYN, which is not known to exhibit transcription fac-
tor activities (Supplementary Fig. 6e) (Supplementary Fig. 6d,e);
© 2017 Nature America, Inc., part of Springer Nature. All rights reserved.
15 8 VOLUME 18 NUMBER 2 FEBRUARY 2017 nature immunology
ARTICLES
this suggested that the regulation of genes whose expression was
increased in SCLE was specific to VGLL3. Of note, sex differences
in the expression of VGLL3 and the targets of VGLL3, including
BAFF and ITGAM, were less apparent in comparisons of male and
female patients with SCLE (Supplementary Fig. 6f), consistent with
VGLL3’s being a sex-biased risk factor before disease manifestation
and a general regulator that is brought to comparable functional levels
in the two sexes as autoimmune conditions arise. Consistent with that
scenario was the similar patterns with which VGLL3 localized to the
nucleus in skin lesions from males and females with SCLE (Fig. 4d).
Similarly, VGLL3-regulated genes had significantly higher expression
in skin lesions from patients with morphea (female:male, 4:1 (ref. 31);
P = 3 × 10−4) and limited scleroderma, a subtype of systemic sclerosis
(female:male, 4:1 (ref. 32); P = 1.4 × 10−2), than in skin from healthy
subjects (Fig. 5a,b and Supplementary Fig. 6g). In a gene-by-gene
analysis of the highest-ranked targets of VGLL3 that are expressed in
patients with scleroderma and morphea33, we found that expression
of a majority of the targets was higher in skin from patients with these
diseases than in skin from healthy subjects (Fig. 5c–h).
To address the role of VGLL3 in a sex-biased autoimmune condi-
tion not located mainly in the skin, we extended our analyses to SS, an
autoimmune condition characterized by inflammation of salivary and
lacrimal glands with a reported female:male ratio as high as 20:1 (ref. 34).
Our examination of a published data set35 showed that the expres-
sion of VGLL3 mRNA was higher in the parotid tissue from patients
with primary SS than in that from healthy control subjects (Fig. 6a).
Concurrently, expression of the VGLL3 ‘node’ targets MMP9, ETS1, IL7
and IL7R (IL-7 receptor) was also higher in patients with SS (Fig. 6b).
Notably, the IL-7 axis has been shown to be pivotal in the pathogen-
esis of SS, with both IL-7 and its receptor being overexpressed in
this condition36,37. IL-7 enhances the T helper type 1 response and
T-cell-dependent activation of monocytes and B cells, and it promotes
lymphocyte infiltration of target organs mediated by IFN-γ and the
ligand of the chemokine receptor CXCR3 (refs. 37,38). Furthermore,
IL-7 has been shown to be a successful therapeutic target in this syn-
drome38. In a gene-by-gene view, we found that VGLL3 target genes
had higher expression in inflamed parotid tissue than in normal tissue
(Fig. 6c), a trend not observed for non-target genes (Fig. 6d).
BAFF IL7 MMP9
LY6E OAS1 MX1 IFI44
e g
h
f
0.0 Control SS
0.5
1.0
1.5
2.0
a
VGLL3 expression (relative)
*
MMP9 ETS1 IL7 IL7R
b
0
5
10
15
VGLL3 targets expression
(relative)
Control SS
*
*
**
BAFF ITGAM FCER1G
2.0
1.5
1.0
0.5
0.0
–1.0 –0.5 0.0 0.5 1.0
Density
Expression (log
2
fold)
SS genes
Non-SS genes
0.0
0.5
1.0
1.5
Expression (relative)
Scr Ri-1 VGLL3 Ri-1 Scr Ri-2 VGLL3 Ri-2
*****
*
0.0
0.5
1.0
1.5
200
400
600
Expression (relative)
UT, Scr Ri-1
UT, VGLL3 Ri-1
UT, VGLL3 Ri-2
IFN-α + IFN-
β
, Scr Ri-1
IFN-α + IFN-
β
, VGLL3 Ri-1
IFN-α + IFN-
β
, VGLL3 Ri-2
*
*
*
*
*
*
*
*
**
*
0
5
10
15
Expression (relative)
UT, Scr Ri
UT, VGLL3 Ri
IFN-
α
, Scr Ri
IFN-
α
, VGLL3 Ri
IFN-
α
+ IFN-
γ
, Scr Ri
IFN-
α
+ IFN-
γ
, VGLL3 Ri
*
**
*
*
*
c
–4
–2
0
2
4
Expression (log fold)
Targets
d
–2
0
2
Expression (log fold)
Non-targets
Figure 6 VGLL3 regulation of genes altered in SS. (a,b) Expression of VGLL3 mRNA (a) and of MMP9, ETS1, IL7 and IL7R mRNA (b) in parotid tissue
from patients with SS (n = 24) and control subjects (n = 25); results are presented relative to those of control subjects, set as 1. (c,d) Expression of
VGLL3 targets (c) and randomly selected genes that are not targets of VGLL3 (d) in patients with SS (genes (horizontal axis) ranked as in Fig. 5c,d).
(e) Density plot of the expression (log2 fold) of ‘SS genes’ (expression increased in SS at a threshold of a change in expression of 1.5-fold and q < 0.05)
and ‘non-SS genes’ (expression not increased as described above), assessed after knockdown of VGLL3. P < 2.2 × 10−6 (Mann–Whitney–Wilcoxon
test). (f) qRT–PCR analysis of BAFF, ITGAM and FCER1G in monocytes treated by RNAi with siRNA with a scrambled sequence (Scr Ri-1 and
Scr Ri-2) or siRNA targeting VGLL3 (VGLL3 Ri-1 and VGLL3 Ri-2). (g) qRT–PCR analysis of LY6E, OAS1, MX1 and IFI44 in monocytes treated with
siRNA as in f in the presence (IFN-α + IFN-β) or absence (UT) of treatment with IFN-α and IFN-β; results are presented relative to the mean of the
expression levels in untreated or Scr-Ri-treated cells. (h) Expression of BAFF, IL7 and MMP9 mRNA in cultured salivary gland cells treated with
siRNA as in f and not treated with interferons (UT) or treated with IFN-α alone or together with IFN-γ (key); results presented as in g. Each
symbol (f–h) represents an individual donor; small horizontal lines indicate the mean (± s.e.m.). *q < 0.05 (a,b) or *P < 0.05 (Student’s t-test; f–h).
Data are representative of analysis of 24 patients with SS and 25 healthy subjects (a,b; mean + s.e.m.), or of five independent analyses (c–e) or
are results from three independent experiments (f–h).
© 2017 Nature America, Inc., part of Springer Nature. All rights reserved.
nature immunology VOLUME 18 NUMBER 2 FEBRUARY 2017 159
ARTICLES
Collectively, the expression of VGLL3 target gene was higher than that
of non-target genes in patients with SS (Supplementary Fig. 6h), and
the expression of genes upregulated in SS decreased significantly after
knockdown of VGLL3 (Fig. 6e). Consistent with our observations of
tissue from patients with SCLE, VGLL3 was mainly localized in the
cell nucleus in affected tissue from patients with SS (data not shown).
Collectively, we observed higher expression of VGLL3 target genes
in tissue from patients with the four autoimmune diseases assessed
than in non-diseased tissue, and this increase was not observed for
non-target genes. Therefore, the VGLL3-regulated genes were linked
to multiple female-biased autoimmune diseases.
To investigate whether VGLL3 regulates genes encoding products
that promote autoimmunity in cell types other than keratinocytes,
we studied the response of three female-biased genes induced in
SLE, BAFF, ITGAM and FCER1G (which encodes the Fc fragment of
receptor for immunoglobulin E), to disruption of VGLL3 expression
in monocytes. Alterations in monocytes are hallmarks of SLE, includ-
ing increased production of BAFF, which is involved in B cell dif-
ferentiation and T cell activation28. We observed that VGLL3 was
required for optimal expression of BAFF and FCER1G in monocytes
from female subjects (Fig. 6f), which indicated that VGLL3 partici-
pated in promoting the expression of female-biased genes encoding
products related to autoimmunity in monocytes. We further examined
a potential role for VGLL3 in regulating type I interferon responses
in both monocytes and cultured salivary gland cells, given the cen-
tral role of type I interferons in the pathogenesis of SLE28 and SS39
and our observation of interferon-response genes among the VGLL3
target gene set. By focusing on the expression of the true type-I-inter-
feron-response genes LY6E, OAS1, MX1 and IFI44 in peripheral blood
mononuclear cells26,27, we confirmed their induction by IFN-α and
IFN-β in monocytes and found that their maximal induction required
VGLL3 expression (Fig. 6g). Similar to what we observed in mono-
cytes, VGLL3 was required for the induction of pro-inflammatory-
product-encoding genes BAFF, IL7 and MMP9 by IFN-α, with or
without co-stimulation by IFN-γ, in salivary gland cells (Fig. 6h),
consistent with published findings showing that cytokine induction
can be dependent on both type I interferons and IFN-γ in these cells40.
This observation indicated that VGLL3 might promote inflammation
events via supporting type I interferon responses.
DISCUSSION
There is a critical need to understand the biological differences
between men and women, including how they influence the mani-
festation of different diseases and the response to therapy41,42.
Autoimmune diseases represent one of the most prominent examples
of sexually dimorphic human diseases, with a notable predominance
in females. Our data demonstrated that even in healthy people, there
were widespread sex-dependent differences in the activity of multi-
ple immunological pathways. The sex-biased genes identified here
overlapped genetic risk variants that have been previously identified
for autoimmune diseases, including SLE and systemic sclerosis, and
their expression was increased in sites of involvement. This find-
ing suggested that these sex-biased genes contributed to not only
increased disease susceptibility but possibly also heightened disease
activity. In this context, we note that being female is the strongest
risk factor for the development of autoimmunity, and it dwarfs the
identified autoimmune genetic risk variants. Thus, these results
provided novel insights into how sex contributes to autoimmune
disease etiology. Furthermore, they suggest possibilities for the iden-
tification of high-risk populations using biomarkers based on these
risk-associated genes.
In contrast to the enrichment for genes encoding pro-autoimmune
factors among female-biased genes, male-biased genes were specifi-
cally associated with those encoding transcription factors, some of
which have been linked to anti-inflammatory processes. For instance,
skin grafts engineered to produce the transcription factor HOXA3
confer diminished inflammatory responses43. HOXA5 is induced
by the anti-inflammatory agent colchicine44, and the transcription
factor SIX2 is repressed in chronic inflammation45. Diminished
abundance of the transcription factor FOXF1 is associated with pul-
monary inflammation, and loss FOXF1 enhances production of the
chemokine CXCL12 and inflammation46. The transcription factor
HES1 suppresses CXCL1 expression induced by Toll-like-receptors47.
Further studies will be needed to investigate the potential roles of the
factors encoded by these genes, as well as other male-biased genes, in
protection from autoimmunity.
Notably, our results suggest that sex differences in immunological
regulation extend beyond the DEGs identified in this study to exten-
sive genome-wide co-expression networks. For example, in females,
expression of the SLE risk factor ITGAM is positively correlated with
that of ARTN (artemin), the monocyte counterpart of the T cell, B cell
and natural killer cell dipeptidyl peptidase IV (CD26), whose inhibi-
tors are promising drugs for various autoimmune diseases48, whereas
this is not seen in males. Similarly, in females there is a positive
correlation between expression of PTX3, which regulates clearance
of apoptotic cells49, and that of SEPT2, a GTP-binding, cytoskeleton-
interacting protein and a putative autoantigen in systemic sclerosis50,
but this is not observed in males. Therefore, ITGAM-ARTN and
PTX3-SEPT2 might be regulated and their products might function
in a common pathway in females but not in males. The presence of
such DEG sets on a genome-wide level indicates additional layers of
sexually dimorphic immune regulation beyond mRNA levels.
VGLL3 is a putative transcription factor16, and it had a prominent
female-biased expression pattern in our data. We identified VGLL3
as a previously unrecognized inflammatory pathway in autoimmunity
and a critical regulator of female-biased inflammatory processes. Of
note, it has been shown that in salmon, VGLL3 exhibits sex-dependent
dominance, promoting later maturation in females17. The findings
that VGLL3 promotes the expression of several existing autoimmune-
disease drug targets and genes encoding inflammatory molecules,
including BAFF (SLE), MMP9 (SLE, systemic sclerosis, MS, SS and
polymyositis), IL-7 (SLE, SS, MS and RA) and ICAM-1 (SLE, MS and
RA), and that it influences type I interferon responses in immune
and non-immune cell populations positions it at the intersection of
multiple autoimmune pathways for potential therapeutic targeting.
Notably, we demonstrated that in males affected by cutaneous lupus,
expression of VGLL3 was similar to that seen in females and that this
was associated with translocation of VGLL3 to the nucleus in actively
inflamed tissue, consistent with VGLL3’s role as a transcription factor;
this indicated that in affected males, the VGLL3-regulated pathway
became activated. This would make VGLL3 an attractive therapeutic
target because it is present in diseased tissue of both females and males
and a reduction in functional VGLL3 to levels observed in healthy
males would be considered to be unlikely to cause serious side effects.
Further studies are needed to address the regulation of VGLL3 and the
mechanisms involved in its activation with disease onset in males.
In summary, our results have identified transcriptomic differ-
ences between females and males that were associated with extensive
genome-wide co-expression gene networks that influenced various
immunological processes, including various autoimmune processes.
Furthermore, these results identified a VGLL3-regulated gene network
as a previously unrecognized inflammatory pathway that promoted
© 2017 Nature America, Inc., part of Springer Nature. All rights reserved.
16 0 VOLUME 18 NUMBER 2 FEBRUARY 2017 nature immunology
female-biased autoimmunity, and they demonstrated the importance
of studying immunological processes in females and males separately.
Because many of these diseases are inadequately controlled with exist-
ing treatments, identifying a unifying molecular basis underlying
multiple autoimmune diseases might have far-reaching implications
for the development of novel therapeutics.
METHODS
Methods, including statements of data availability and any associated
accession codes and references, are available in the online version of
the paper.
Note: Any Supplementary Information and Source Data files are available in the
online version of the paper.
ACKNOWLEDGMENTS
We thank A.A. Dlugosz for critical discussions and reading of the manuscript;
S. Stoll, Y. Xu, T. Quan, Y. Li, L. Wolterink and L. Reingold for technical help; and
A. Libs for help with biopsy samples and files. Supported by the
US National Institutes of Health (K08-AR060802 and R01-AR069071to J.E.G.;
and R03-AR066337 and K08-AR063668 to J.M.K.), an A. Alfred Taubman
Medical Research Institute Kenneth and Frances Eisenberg Emerging Scholar
Award (J.E.G.), the Doris Duke Charitable Foundation (2013106 to J.E.G.)
and a Pfizer Aspire Award (J.E.G.).
AUTHOR CONTRIBUTIONS
Y.L., J.E.G., J.T.E., J.M.K. and J.J.V. designed the study and wrote the manuscript;
Y.L., X.X., M.A.B., P.W.H., P.E.S., M.K.S., R.P.N. and C.C.B. collected and
analyzed data; and L.C.T. and W.R.S. analyzed data. All authors reviewed and
commented on the manuscript.
COMPETING FINANCIAL INTERESTS
The authors declare no competing financial interests.
Reprints and permissions information is available online at http://www.nature.com/
reprints/index.html.
1. American Autoimmune Related Diseases Association. The cost burden of autoimmune
disease: the latest front in the war on healthcare spending. (American Autoimmune
Related Diseases Association, Eastpointe, Michigan, USA, 2011).
2. Fish, E.N. The X-files in immunity: sex-based differences predispose immune
responses. Nat. Rev. Immunol. 8, 737–744 (2008).
3. Whitacre, C.C. Sex differences in autoimmune disease. Nat. Immunol. 2, 777–780
(2001).
4. Klein, S.L. The effects of hormones on sex differences in infection: from genes to
behavior. Neurosci. Biobehav. Rev. 24, 627–638 (2000).
5. Holroyd, C.R. & Edwards, C.J. The effects of hormone replacement therapy on
autoimmune disease: rheumatoid arthritis and systemic lupus erythematosus.
Climacteric 12, 378–386 (2009).
6. Seligman, V.A., Lum, R.F., Olson, J.L., Li, H. & Criswell, L.A. Demographic
differences in the development of lupus nephritis: a retrospective analysis.
Am. J. Med. 112, 726–729 (2002).
7. Mackay, I.R. Science, medicine and the future: tolerance and autoimmunity.
Br. Med. J. 321, 93–96 (2000).
8. Sánchez-Guerrero, J. et al. Menopause hormonal therapy in women with systemic
lupus erythematosus. Arthritis Rheum. 56, 3070–3079 (2007).
9. Mok, C.C. et al. Safety of hormonal replacement therapy in postmenopausal
patients with systemic lupus erythematosus. Scand. J. Rheumatol. 27, 342–346
(1998).
10. Nestle, F.O., Di Meglio, P., Qin, J.Z. & Nickoloff, B.J. Skin immune sentinels in
health and disease. Nat. Rev. Immunol. 9, 679–691 (2009).
11. Cervera, R. et al. Systemic lupus erythematosus: clinical and immunologic patterns
of disease expression in a cohort of 1,000 patients. Medicine (Baltimore) 72,
113–124 (1993).
12. Cooper, G.S. & Stroehla, B.C. The epidemiology of autoimmune diseases.
Autoimmun. Rev. 2, 119–125 (2003).
13. Vincent, F.B., Morand, E.F., Schneider, P. & Mackay, F. The BAFF (APRIL)
system in SLE pathogenesis. Nat. Rev. Rheumatol. 10, 365–373 (2014).
14. Harley, J.B. et al. Genome-wide association scan in women with systemic lupus
erythematosus identifies susceptibility variants in ITGAM, PXK, KIAA1542 and other
loci. Nat. Genet. 40, 204–210 (2008).
15. Grundberg, E. et al. Mapping cis- and trans-regulatory effects across multiple tissues
in twins. Nat. Genet. 44, 1084–1089 (2012).
16. Maeda, T., Chapman, D.L. & Stewart, A.F. Mammalian vestigial-like 2, a cofactor
of TEF-1 and MEF2 transcription factors that promotes skeletal muscle
differentiation. J. Biol. Chem. 277, 48889–48898 (2002).
17. Barson, N.J. et al. Sex-dependent dominance at a single locus maintains variation
in age at maturity in salmon. Nature 528, 405–408 (2015).
18. Ram, M., Sherer, Y. & Shoenfeld, Y. Matrix metalloproteinase 9 and autoimmune
diseases. J. Clin. Immunol. 26, 299–307 (2006).
19. Yang, W. et al. Genome-wide association study in Asian populations identifies
variants in ETS1 and WDFY4 associated with systemic lupus erythematosus. PLoS
Genet. 6, e1000841 (2010).
20. Lee, L.F. et al. IL-7 promotes TH1 development and serum IL-7 predicts clinical
response to interferon-β in multiple sclerosis. Sci. Transl. Med. 3, 93ra68 (2011).
21. Pickens, S.R. et al. Characterization of interleukin-7 and interleukin-7 receptor in
the pathogenesis of rheumatoid arthritis. Arthritis Rheum. 63, 2884–2893 (2011).
22. Bikker, A. et al. Increased expression of interleukin-7 in labial salivary glands of
patients with primary Sjögren’s syndrome correlates with increased inflammation.
Arthritis Rheum. 62, 969–977 (2010).
23. Wuthrich, R.P., Jevnikar, A.M., Takei, F., Glimcher, L.H. & Kelley, V.E. Intercellular
adhesion molecule–1 (ICAM-1) expression is upregulated in autoimmune murine
lupus nephritis. Am. J. Pathol. 136, 441–450 (1990).
24. Bö, L. et al. Distribution of immunoglobulin superfamily members ICAM-1, ICAM-2,
ICAM-3 and the β2-integrin LFA-1 in multiple sclerosis lesions. J. Neuropathol.
Exp. Neurol. 55, 1060–1072 (1996).
25. Anderson, M.E. & Siahaan, T.J. Targeting ICAM-1–LFA-1 interaction for controlling
autoimmune diseases: designing peptide and small-molecule inhibitors. Peptides
24, 487–501 (2003).
26. Baechler, E.C. et al. Interferon-inducible gene expression signature in peripheral
blood cells of patients with severe lupus. Proc. Natl. Acad. Sci. USA 100,
2610–2615 (2003).
27. Obermoser, G. & Pascual, V. The interferon-α signature of systemic lupus
erythematosus. Lupus 19, 1012–1019 (2010).
28. Banchereau, J. & Pascual, V. Type I interferon in systemic lupus erythematosus and
other autoimmune diseases. Immunity 25, 383–392 (2006).
29. Ohl, K. & Tenbrock, K. Inflammatory cytokines in systemic lupus erythematosus.
J. Biomed. Biotechnol. 2011, 432595 (2011).
30. Gudjonsson, J.E. & Elder, J.T. Psoriasis: epidemiology. Clin. Dermatol. 25, 535–546
(2007).
31. Leitenberger, J.J. et al. Distinct autoimmune syndromes in morphea: a review
of 245 adult and pediatric cases. Arch. Dermatol. 145, 545–550 (2009).
32. Mayes, M.D. et al. Prevalence, incidence, survival and disease characteristics of
systemic sclerosis in a large US population. Arthritis Rheum. 48, 2246–2255
(2003).
33. Milano, A. et al. Molecular subsets in the gene expression signatures of scleroderma
skin. PLoS One 3, e2696 (2008).
34. Patel, R. & Shahane, A. The epidemiology of Sjögren’s syndrome. Clin. Epidemiol.
6, 247–255 (2014).
35. Horvath, S. et al. Systems analysis of primary Sjögren’s syndrome pathogenesis
in salivary glands identifies shared pathways in human and a mouse model. Arthritis
Res. Ther. 14, R238 (2012).
36. Hillen, M.R. et al. High soluble-IL-7 receptor expression in Sjögren’s syndrome
identifies patients with increased immunopathology and dryness. Ann. Rheum. Dis.
75, 1735–1736 (2016).
37. Bikker, A. et al. IL-7-activated T cells and monocytes drive B cell activation
in patients with primary Sjogren’s syndrome. Ann. Rheum. Dis. 70, A63 (2011).
38. Jin, J.O., Kawai, T., Cha, S. & Yu, Q. Interleukin-7 enhances the TH1 response to
promote the development of Sjögren’s-syndrome-like autoimmune exocrinopathy in
mice. Arthritis Rheum. 65, 2132–2142 (2013).
39. Yao, Y., Liu, Z., Jallal, B., Shen, N. & Rönnblom, L. Type I interferons in Sjögren’s
syndrome. Autoimmun. Rev. 12, 558–566 (2013).
40. Jin, J.O., Shinohara, Y. & Yu, Q. Innate immune signaling induces interleukin-7
production from salivary gland cells and accelerates the development of primary
Sjögren’s syndrome in a mouse model. PLoS One 8, e77605 (2013).
41. Kim, A.M., Tingen, C.M. & Woodruff, T.K. Sex bias in trials and treatment must end.
Nature 465, 688–689 (2010).
42. Woodruff, T.K. Sex, equity and science. Proc. Natl. Acad. Sci. USA 111,
5063–5064 (2014).
43. Kachgal, S., Mace, K.A. & Boudreau, N.J. The dual roles of homeobox genes in
vascularization and wound healing. Cell Adh. Migr. 6, 457–470 (2012).
44. Ben-Chetrit, E., Bergmann, S. & Sood, R. Mechanism of the anti-inflammatory
effect of colchicine in rheumatic diseases: a possible new outlook through microarray
analysis. Rheumatology 45, 274–282 (2006).
45. Umazume, T. et al. Lacrimal gland inflammation de-regulates extracellular matrix
remodeling and alters molecular signature of epithelial stem/progenitor cells. Invest.
Ophthalmol. Vis. Sci. 56, 8392–8402 (2015).
46. Kalin, T.V. et al. Pulmonary mastocytosis and enhanced lung inflammation in mice
heterozygous null for the Foxf1 gene. Am. J. Resp. Cell Mol. 39, 390–399 (2008).
47. Shang, Y., Karmen, A. & Xiaoyu, H. Transcription repressor Hes1 is a selective
regulator of TLR-induced CXCL1 expression and neutrophil responses. J. Immunol.
192, 62–69 (2014).
48. Steinbrecher, A. et al. Targeting dipeptidyl peptidase IV (CD26) suppresses
autoimmune encephalomyelitis and upregulates TGF-β1 secretion in vivo.
J. Immunol. 166, 2041–2048 (2001).
49. Rovere, P. et al. The long pentraxin PTX3 binds to apoptotic cells and regulates their
clearance by antigen-presenting dendritic cells. Blood 96, 4300–4306 (2000).
50. Hall, P.A. & Russell, S.E. The pathobiology of the septin gene family. J. Pathol.
204, 489–505 (2004).
ARTICLES
© 2017 Nature America, Inc., part of Springer Nature. All rights reserved.
nature immunology
doi:10.1038/ni.3643
ONLINE METHODS
Skin biopsies from normal tissue and SCLE lesions. All subjects provided
informed consent for biopsies from normal skin and SCLE skin lesions. Cases
of SCLE biopsies were identified from the University of Michigan Pathology
Database under Institutional Review Board (IRB) #HUM72843. Patients who
met both clinical and histologic criteria for SCLE were included. Fresh skin
samples were acquired according to IRB #HUM66116. All patient recruitment
and samples were treated according to the Declaration of Helsinki.
Identification of sex-biased genes. RNA-seq data of 82 normal skin sam-
ples (GSE63980) were used to identify genes that are differentially expressed
between the two sexes in the skin. The gender effect for expression of each
gene was modeled by linear regression, with the age of the patient at biopsy
as a covariate. Specifically, we used the RNA-seq data of normal skin samples
from our large cohort that studied the transcriptomes of psoriasis (GSE63980).
We obtained the sex and age of biopsy for 82 patients in the cohort, and we
used the pipeline and gene model described previously for RNA-seq analysis,
including read mapping, assembly and quantification of gene expression51.
We performed analysis for genes only expressed in at least 20% of the nor-
mal samples. The algorithm DESeq was used for expression normalization.
Rank-based inverse normalization was then applied on each gene’s normal-
ized expression values. Linear regression was used to model the gender effect
for expression each gene (i.e., differential expression analysis), and P values
were computed using the Wald test. We used the age of the patient at the time
of biopsy as a covariate. False discovery rate (FDR) was used to control the
multiple testing.
Gene–gene co-expression analysis and functional characterization. We
conducted gene–gene correlation analysis between sex-specific DEGs versus
all genes in males and females, separately. Gene–gene co-expression was meas-
ured by Spearman rank correlation coefficient (
ρ
); P values were computed
using algorithm AS89, and we computed FDR for multiple-testing comparison.
In total, there were 434,900 gene pairs with significant correlation (FDR ≤ 0.1)
in both male and female correlation analyses; we also identified 124,521 gene–
gene pairs that show significant correlation in only female samples (FDR ≤ 0.1)
but not in male samples (P > 0.5), and 158,303 gene–gene pairs show only
significant correlation in only male samples (FDR ≤ 0.1) but not in female
samples (P > 0.5). To assess whether the difference in gene–gene correlation
between sexes is significant, we devised a permutation approach. Specifically,
we first permuted the sex labels and calculated the difference in gene–gene
correlations between the pseudo-male and pseudo-female samples. The per-
mutation was performed 10,000 times to construct the null distribution for
the difference in correlation patterns. The significance was then estimated
by comparing the observed gene–gene gender correlation difference to the
null distribution for the same gene–gene pair. Functional enrichment analysis
was performed using hypergeometric distribution, with biological annota-
tions retrieved from the Gene Ontology, KEGG (Kyoto Encyclopedia of Genes
and Genomes) and Biocarta databases. Enriched functions were identified
using FDR ≤ 0.1. Spearman correlation (
ρ
) was used to study the difference
(
ρ
diff =
ρ
male −
ρ
female) in co-expression networks between the two sexes and to
study and compare the co-expression networks of the most notable enriched
biological functions/pathways between the two sexes.
Disease-association screening. We retrieved disease-associated genetic
signals available from the NHGRI catalog52 and Immunobase (https://www.
immunobase.org/), and we processed the data to consolidate the disease
names and maintain only signals that achieve genome-wide association
(P ≤ 5 × 10−8). Combining results from the two sources retained 9,599 vari-
ant-to-trait associations. We further merged genetic variants from same locus
by using a ±500-kb interval as a criterion to select the most significant marker
in each locus and focused on complex skin traits with at least five associated
loci. Enrichment of sex-biased genes was assessed by the hypergeometric test,
using all skin-expressing genes from the RNA-seq cohort as background.
The female-to-male prevalence ratios for skin-associated traits were retrieved
from previous literature data12,30 ,53,54.
In addition to the hypergeometric enrichment approach to compute the
significance between female-biased genes versus genes from complex skin
disease loci, we also devised a sampling strategy. For each of the complex
skin disease we investigated, we randomly selected the same number of loci
from the NHGRI disease catalog and enumerated the randomly selected loci
which comprised of female-biased genes. We repeated the process 10,000 times
and constructed the null distribution for the expected number of overlap.
We then estimated the significance by comparing the observed overlap against
the null distribution. This robust approach presented empirical P values for the
eight diseases which were highly concordant with what we observed using the
enrichment approach, replicating the findings that loci associated with SLE,
systemic sclerosis and atopic dermatitis are enriched with female-biased genes
(Supplementary Fig. 5a). Supplementary Figure 5b shows the null distribu-
tion for the expected overlap for random loci, and the red lines illustrate the
observed overlap results from SLE/SS (top) and AD (bottom), respectively.
Expressed quantitative trait loci (eQTL) analysis. To examine whether the
variants that affect the functions or expression of VGLL3 would also affect the
expression of the VGLL3 targets, we first identified the nonsynonymous or
splice site variants for VGLL3 using the phase 3 1000 Genomes. Among the
nine nonsynonymous and one splice variants identified from 1000 Genomes
project, all of them are rare variants or variants with low minor-allele fre-
quencies, thus we were not able to conduct eQTL analysis on these variants
as they are not well-imputed in our genetic cohorts. We then turned to mark-
ers that can influence the expression of VGLL3 by conducting a cis-eQTL
analysis surrounding the ±1-Mb region of VGLL3. The results are shown
in Supplementar y Figure 5 and illustrated the strongest cis-eQTL signal
at chr3:87902673 (P = 4 × 10−5 ). We then investigated whether this VGLL3
cis-eQTL would influence the expression of VGLL3 target genes (208 genes
decreased by VGLL3 knockdown in keratinocytes at the q < 0.05, |log2FC|
> 0.5 threshold; as in Supplementary Table 3), and notably, we identified
nine VGLL3 targets that are also significantly associated with the cis-eQTLs,
indicating a trans-eQTL effect (Supplementar y Fig. 5).
Cell culture, stimulation, RNAi and gene expression analyses. Normal human
keratinocytes (NHKs) were established from healthy adults as previously
described55 and grown in medium 154 CF (ThermoFisher Scientific). Informed
consent was obtained from all subjects. NHKs were used at passage 1 or 2.
For sex hormone stimulation, estradiol (Sigma E2578) or testosterone (Sigma
T1500) was applied to passage 1 cells for 24 h. Monocytes were isolated as indi-
cated below and maintained in RPMI (ATCC) with 10% FBS (ThermoFisher
Scientific). A253 salivary gland cells were obtained from ATCC and cultured
in McCoy’s 5a medium (ATCC) with 10% FBS (ThermoFisher Scientific).
siRNA was introduced by electroporation using Lonza 4D-nucleofector
following the manufacturer’s instructions. For IFN stimulation, IFN-α (R&D
systems) was used at 1,000 U/ml, IFN-β (R&D systems) at 1,000 U/ml and
IFN-γ (R&D systems) at 2,000 U/ml. Interferons were applied to cells that had
been electroporated with the indicated siRNA for 12 h, and RNA was collected
by the Qiagen RNeasy plus kit. qRT–PCR was performed on a 7900HT Fast
Real-time PCR system (Applied Biosystems) with TaqMan Universal PCR
Master Mix (ThermoFisher Scientific). RNA-seq libraries were prepared using
the Illumina Truseq RNA library prep kits and sequenced on the Illumina
HiSeq platform. Differential expression analyses were performed using EdgeR,
and functional enrichment and literature-based network analyses were per-
formed with the Genomatix software. To study whether genes upregulated in
SCLE are regulated by VGLL3 on a genome-wide level, we first defined genes
whose expression was increased in SCLE as those upregulated in SCLE relative
to their expression in normal tissue at a threshold of a change in expression of
≥2-fold and q < 0.01 (Supplementary Table 5) and defined ‘non-SCLE genes’
as the rest of genes. We then performed density plots of the ‘log2 fold’ change
in expression after VGLL3 knockdown for SCLE and non-SCLE genes, respec-
tively. Mean expression changes for the two groups of genes were calculated,
and the Mann–Whitney–Wilcoxon test was used for significance.
Peripheral blood mononuclear cell (PBMC) isolation. PBMCs from healthy
individuals and patients with SLE were obtained as part of this study, and
informed consent was obtained from all subjects. PBMCs were isolated from
whole blood using the Ficoll method. Monocyte, T cells and B cells were
isolated from PBMCs using MACS negative selection kits.
© 2017 Nature America, Inc., part of Springer Nature. All rights reserved.
nature immunology doi:10.1038/ni.3643
Immunohistochemistry. Formalin-fixed, paraffin-embedded specimens
on slides were heated for 30 min at 55 °C, rehydrated and epitope-retrieved
with Tris-EDTA, pH 9. Slides were blocked, incubated with primary anti-
body (anti-C3, Sigma HPA020432, 1:500 dilution; anti-VGLL3, Sigma
HPA054953, 1:250 dilution; anti-DOCK2, Sigma HPA036488, 1:300 dilu-
tion) overnight at 4 °C, washed, incubated with secondary antibody (anti-
rabbit-IgG, Vector Laboratories, BA-1000, 1:50 dilution), developed with DAB
(3,3′ diaminobenzidine, BD Biosciences 550850) and counterstained using
hematoxylin. Images presented are representative of three experiments.
Statistical analysis. Statistical tests used are described as in the individual
Online Methods subsections and in the figure legends. Student’s t-tests
are two-tailed. The exact P values, when not specified in the figures, are
as follows:
Figure 3a (left to right): 0.026, 0.020, 0.036, 0.012, 0.024, 0.033. Figure 3c
(left to right): 0.020, 0.014, 0.043. Figure 3d (left to right): 0.027, 0.025, 0.026,
0.025. Figure 3e (left to right): 0.024, 0.044, 0.036, 0.011. Figure 3f (left to
right): 0.021, 0.037, 0.043. Figure 3g: 0.041. Figure 3h: 0.043. Figure 4a (left
to right): 0.011, 0.001. Figure 4b: 0.001. Figure 4c: 0.010. Figure 6a: 0.049.
Figure 6b (left to right): 0.011, 0.0003, 0.0085, 0.004. Figure 6f (left to right):
0.032, 0.020, 0.045, 0.048. Figure 6g (left to right): 0.00003, 0.004, 0.0004,
0.010, 0.0065, 0.007, 0.007, 0.004, 0.021, 0.015, 0.016, 0.014. Figure 6h (left to
right): 0.022, 0.0004, 0.002, 0.022, 0.0003, 0.0005, 0.013.
Data availability. The data that support the findings of this study are available
from the corresponding author upon request. RNA-seq data are available in
the GEO database with the accession code GSE63980.
51. Tsoi, L.C. et al. Analysis of long noncoding RNAs highlights tissue-specific expression
patterns and epigenetic profiles in normal and psoriatic skin. Genome Biol. 16, 24
(2015).
52. Welter, D. et al. The NHGRI GWAS Catalog, a curated resource of SNP–trait
associations. Nucleic Acids Res. 42, D1001–D1006 (2014).
53. Carson, C.G. Risk factors for developing atopic dermatitis. Dan. Med. J. 60, B4687
(2013).
54. Scrivener, Y., Grosshans, E. & Cribier, B. Variations of basal cell carcinomas
according to gender, age, location and histopathological subtype. Br. J. Dermatol.
147, 41–47 (2002).
55. Elder, J.T. et al. Retinoic acid receptor gene expression in human skin. J. Invest.
Dermatol. 96, 425–433 (1991).
© 2017 Nature America, Inc., part of Springer Nature. All rights reserved.
