Journal of Infectious Diseases Advance Access published October 15, 2014
1
TNFSF8 regulatory variants are associated with excessive
cr ipt
inflammatory responses but not leprosy per se
Vinicius M. Fava1,2, Aurélie Cobat1,2, Nguyen Van Thuc3, Ana Carla P. Latini4, Mariane M. A. Stefani5, Andrea F. Belone4, Nguyen Ngoc Ba3, Marianna Orlova2, Jérémy
Erwin Schurr1,2,§ 1
Program in Immunology and Infectious Diseases in Global Health, The Research Institute of
2
an
the McGill University Health Centre, Montreal, H3G 1A4, Quebec, Canada
The McGill International TB Centre, Departments of Human Genetics and Medicine, McGill
M
University, Montreal, H3G 1A4, Quebec, Canada
Hospital for Dermato-Venerology, Ho Chi Minh City, District 3, Vietnam
4
Lauro de Souza Lima Institute, Bauru, 17034-971, São Paulo, Brazil
5
Tropical Pathology and Public Health Institute, Federal University of Goiás, Goiânia,
pt ed
3
74001-970, Goiás, Brazil 6
Core for Advanced Molecular Investigation, Pontifical Catholic University of Paraná,
Curitiba, 80215-901, Paraná, Brazil
Laboratory of Human Genetics of Infectious Diseases, Necker Branch, Institut National de
ce
7
la Santé et de la Recherche Médicale, U1163, Paris, 75015, France University Paris Descartes, Imagine Institute, Paris, 75015, France
9
St Giles Laboratory of Human Genetics of Infectious Diseases, Rockefeller Branch, The
Ac
8
Rockefeller University, New York, NY 11065, USA 10
Unité de Recherche Clinique, Centre d’Investigation Clinique, Necker and Cochin
Hospitals, Paris 75015, France
© The Author 2014. Published by Oxford University Press on behalf of the Infectious Diseases Society of America. All rights reserved. For Permissions, please e‐mail:
[email protected].
Downloaded from http://jid.oxfordjournals.org/ at University of Waterloo on December 15, 2014
us
Manry1,2, Marcelo T. Mira6, Vu Hong Thai3, Laurent Abel7,8,9, Alexandre Alcaïs7,8,9,10,
2 §
Corresponding author: Dr. Erwin Schurr, Montreal General Hospital Research Institute,
Room L11-521, 1650 Cedar Avenue, Montreal, Quebec, Canada, H3G 1A4, E-mail:
44513, Fax: +1-514-934-8238
Abstract
us
Background Type-1 Reactions (T1R) affect a considerable proportion of leprosy patients. In T1R cases, the host immune response pathologically overcompensates for the actual
an
infectious threat, resulting in nerve damage and permanent disability. Based on the results of a genome-wide association study of leprosy per se, we investigated the TNFSF15 chromosomal region for a possible contribution to susceptibility to T1R.
M
Methods We performed a high resolution association scan of the TNFSF15 locus to evaluate the association with T1R in two geographically and ethnically distinct populations: a family-
pt ed
based sample from Vietnam and a case-control sample from Brazil, comprising a total of 1768 subjects.
Results In the Vietnamese sample, 47 SNPs overlapping TNFSF15 and the adjacent TNFSF8 gene were associated with T1R but not with leprosy. Of the 47 SNPs, 39 were cis-eQTL for TNFSF8 including SNPs located within the TNFSF15 gene. In the Brazilian sample, 18 of
ce
these cis-eQTL SNPs overlapping the TNFSF8 gene were validated for association with T1R. Conclusions Taken together, these results indicate TNFSF8 and not TNFSF15 as an
Ac
important T1R susceptibility gene. Our data support the need for infection genetics to go beyond genes for pathogen control to explore genes involved in a commensurate host response.
Downloaded from http://jid.oxfordjournals.org/ at University of Waterloo on December 15, 2014
cr ipt
[email protected], E-mail:
[email protected], Tel.: +1-514-934-1934 extension
3
Background
cr ipt
Leprosy is an infectious disease with a demonstrated strong contribution of host
genetic factors to susceptibility. Employing candidate gene (e.g. TNF, TLR1 and IL10) and positional cloning approaches (PARK2/PACRG, LTA, HLA-C and genes in region 10p12-
p13) leprosy per se susceptibility genes have been identified and validated across multiple
us
identified SNPs located in the vicinity of six genes (NOD2, HLA-DR, CCDC122-LACC1, RIPK2 and TNFSF15) in association with leprosy per se (p < 10-7) [10, 11]. Interestingly,
an
four of the six genes implicated in leprosy susceptibility are validated Crohn’s disease susceptibility genes suggesting mechanistic similarities in the pathogenesis of leprosy and
M
Crohn’s disease [12].
In a previous study, we replicated HLA-DR, CCDC122-LACC1, RIPK2, and NOD2 as
pt ed
leprosy susceptibility genes, but we and others failed to replicate the locus at TNFSF15 [13, 14]. However, we noted that more than 85% of the Chinese leprosy patients in the discovery sample of the GWAS presented with disability. Type-1 Reactions (T1R) are the most frequent type of leprosy reaction and present as the major cause of permanent disability. T1R
ce
are characterized by exacerbated cell-mediated immune responses, leading to painful inflammation, and if untreated, irreversible peripheral nerve damage [15]. Critically, T1R can
Ac
occur during the chronic phase of leprosy or even after microbiological cure of the disease, pointing to a host immune process that is dissociated from the actual pathogenic threat.
Given that T1R contributes to disability in leprosy and that we have been unable to
detect an association between TNFSF15 and leprosy per se, we wondered if TNFSF15 was a
Downloaded from http://jid.oxfordjournals.org/ at University of Waterloo on December 15, 2014
populations [1-9]. A genome-wide association study (GWAS) in a Chinese population
4 T1R susceptibility gene [13]. The present study analyzed the TNFSF15 chromosomal region for the presence of T1R risk factors and found that the SNP rs6478108 located in TNFSF15
cr ipt
was strongly associated with T1R but not with leprosy in the absence of T1R. Furthermore, this SNP in TNFSF15 was part of a bin of highly correlated SNPs that extended to the
proximity of the neighbouring TNFSF8 gene. Fine mapping of the TNFSF15-TNFSF8 SNP
bins identified a large number of SNPs significantly associated with T1R. However, the only
the TNFSF8 gene. SNPs associated with T1R but not T1R-free leprosy, have previously been reported as expression quantitative trait loci (eQTL) for TNFSF8 in monocytes and non-
Methods Subjects and study design
M
contribute to the risk of T1R in leprosy.
an
transformed peripheral blood cell [16, 17] suggesting that regulatory variants in TNFSF8
pt ed
The study was conducted according to the principles expressed in the declaration of Helsinki. Written informed consent was obtained for all adult subjects participating in the study. All minors assented to the study while a parent or guardian provided the informed consent on their behalf. The study was approved by the regulatory authorities and ethics
ce
committees of the participating centers.
For the current study we enrolled a total of 1,768 subjects from Vietnam and Brazil.
Ac
The Vietnamese sample was family-based while two Brazilian samples were populationbased case-control designs (Figure 1). Analysis of the linkage disequilibrium (LD) pattern in the studied chromosomal interval showed near identity for Brazilian samples (Supplementary
Downloaded from http://jid.oxfordjournals.org/ at University of Waterloo on December 15, 2014
us
SNPs that could be replicated in a second sample of Brazilian T1R patients were overlapping
5 Figure 1A and 1B). Hence, the two case-control samples were combined into a single group
cr ipt
of cases and controls.
The Vietnamese sample consisted of 356 leprosy affected offspring belonging to 327
nuclear families with no history of leprosy reactions other than T1R selected from the records available at the Dermato Venerology Hospital, Ho Chi Minh City, Vietnam as described
and 8 multiplex families with two T1R-affected sibs). The T1R-free subset contained the remaining 111 families (93 simplex families and 18 multiplex families of which 16 had two
an
leprosy-affected sibs and two had three and four leprosy-affect sibs each; Figure 1). The validation samples consisted of 347 leprosy-affected subjects with no history of leprosy
M
reactions other than T1R assembled from the records of the Reference Center for Diagnosis and Treatment, Goiânia and 411 leprosy-affected subjects recruited in the Jardim Guanabara Health center, Rondonópolis (Figure 1). Both enrolment centers are located in central-
pt ed
western Brazil and samples have been described elsewhere [2, 18]. The Brazilian and Vietnamese leprosy patients were recruited and followed up for at least three years to evaluate T1R outcomes. The risk of T1R is dependent on leprosy subtype and age-at onset of leprosy. To avoid confounding of results by these non-genetic risk factors T1R-free leprosy
ce
patients were used as controls at proportions that approximated the subtype distribution in the
Ac
T1R-affected patients (Table 1).
Genotyping
In the Vietnamese sample, four SNPs associated with leprosy per se in the Chinese
GWAS sample plus 36 tag SNPs targeting a 220kb region spanning TNFSF15-TNFSF8 were
genotyped as part of a larger effort using the high-throughput Illumina platform. After
Downloaded from http://jid.oxfordjournals.org/ at University of Waterloo on December 15, 2014
us
previously [13]. Of the 327 families 216 belonged to the T1R subset (208 simplex families
6 quality-control filtering, five of the 40 SNPs were non-informative and one deviated from Hardy-Weinberg equilibrium (HWE; p < 0.05). For the fine mapping of TNFSF8-TNFSF15
cr ipt
SNP bins associated with T1R in the discovery sample, 76 SNPs were selected and genotyped on the high-throughput SEQUENOM MassARRAY platform. We selected SNPs
that in the discovery phase showed p < 0.01 for association with T1R, and the four TNFSF15 SNPs associated with leprosy in the Chinese GWAS [10]. In addition, we selected known
genotyped in the Vietnamese and the two Brazilian samples. Of the 76 markers selected, 11 failed to be genotyped, seven presented minor allele frequency (MAF) < 0.02 and two had
an
call rate < 95% in at least one of the studied samples (Supplementary Table 1). Deviation from HWE was tested in 357 leprosy unaffected parents from the Vietnam T1R subset and in
Statistical analyses
M
384 T1R-free leprosy patients from Brazil.
pt ed
In the Vietnamese sample, evidence of association for SNP alleles was compared between T1R-affected and T1R-free families with an approximately matched proportion of leprosy clinical subtype. By contrasting evidence of association for the T1R-affected and T1R-free families it was possible to estimate the genetic heterogeneity. Only those
ce
associations that were found significant in the T1R families with significant evidence of heterogeneity compared to the T1R-free families were considered to represent T1R risk factors. In the Brazilian samples, T1R-free leprosy cases were considered controls and T1R-
Ac
affected cases were considered cases in a standard population based case-control association study following the same matching strategy for leprosy subtypes applied in the family-based sample.
Downloaded from http://jid.oxfordjournals.org/ at University of Waterloo on December 15, 2014
us
quantitative trait loci (eQTL) located in the studied region [16]. This set of markers was
7 The MAF, HWE and LD were estimated in Haploview v.4.2 in 357 leprosy unaffected parents for the discovery sample and in 384 T1R-free leprosy cases for the validation sample
cr ipt
[19]. The family-based association tests for individual SNPs were performed with the transmission disequilibrium test (TDT) for additive dominant and recessive genetic models as implemented in FBAT v.2.0.4 [20]. The smallest p values for the models tested are displayed as the best p value. Genetic heterogeneity between T1R and T1R-free subgroups was tested
testing for heterogeneity of the allele transmission rates in two subgroups can be done in the FBAT framework by simply pooling the two subgroups and coding the phenotypes T1 =
an
1/V1 for T1R individuals and T2 = -1/V2 for T1R-free individuals, where V1 and V2 denote the variance of the FBAT statistic for the T1R and T1R-free samples respectively. In
M
practice, this was achieved by using the offset option “-o” of the FBAT software while coding T1 = 1 and T2 = 0 . The “–o” option calculates an offset μ used to transform the phenotypic values in T1 = 1-μ and T2 = -μ that minimizes the variance of the statistics.
pt ed
Gaschignard et al. show that under an additive genetic model this is equivalent to coding T1 = 1/V1 and T2 = -1/V2. Conditional logistic regression was performed in SAS v.9.3 (SAS, Cary, NC, USA) to (i) estimate the OR for the Vietnamese family sample using the untransmitted allele as a pseudo control [21], (ii) do multivariate analyses of SNPs, and (iii)
ce
conduct an analysis of the validated SNPs in the combined Vietnamese and Brazilian sample. Analyses stratified by age-of-onset in the Vietnamese families did detect variability in support of association across age groups but this effect failed to be statistically significant.
Ac
The statistical power for the TDT analysis was calculated using genetic power calculator [22]. The case-control analyses were performed by logistic regression for three genetic models (additive, dominant and recessive) adjusting by sex and age at leprosy diagnosis employing Plink v.1.07 [23]. For the validation sample we performed a one-sided test with
Downloaded from http://jid.oxfordjournals.org/ at University of Waterloo on December 15, 2014
us
by means of the FBAThet statistic, as proposed by Gaschignard et al [submitted]. Briefly,
8 the alternative hypothesis that the T1R-risk alleles were enriched in the Brazilian T1R cases. For the replication study, we performed 1-sided statistical tests using the alternative
with an increased risk of T1R in the Brazilian sample
Results
us
SNP association scan of TNFSF15 and TNFSF8 in the Vietnamese sample
We first genotyped four SNPs (rs10982385, rs4574921, rs10114470, rs6478108) in
an
the Vietnamese sample that had been significantly associated with leprosy per se in Chinese patients. These four SNPs are located in the vicinity of the TNFSF15 gene. None of the four SNPs was associated with T1R-free leprosy, however SNP rs6478108, located in TNFSF15
M
intron 1, was significantly associated with T1R (p = 0.0002; odds ratio (OR) for recessive Gallele = 2.13; 95% confidence interval (CI): 1.45 – 3.13; G-allele frequency (f) = 0.58; Figure
pt ed
2A; Table 2). Since this SNP had been reported as a cis-eQTL for the neighbouring TNFSF8 gene we selected 36 SNPs in a 220 kb interval extending from TNFSF15 to TNFSF8. Among the 30 SNPs that passed quality control, nine presented significant evidence for association with T1R (p < 0.01; Figure 2A; Table 2). The most significant association was observed for rs7863183 (p < 0.0002; OR for additive T-allele = 2.08; 95% CI: 1.47 – 3.03; T-allele f =
ce
0.80; Figure 2A; Table 2). Although the statistical power to detect an association under the same conditions as for the T1R-subset (same genetic model, allele frequency and risk effect)
Ac
for rs6478108 and rs7863183 was robust (71% and 80%, respectively), we failed to detect even a trend for association of both SNPs in the T1R-free subset. In fact, none of the 34 SNPs (4 GWAS and 30 selected) displayed strong evidence (i.e. p < 0.01) of association with leprosy in T1R-free patients (Figure 2B; Table 2). The associations of ten SNPs with T1R were significantly heterogeneous between T1R-affected and T1R-free patients (p values
Downloaded from http://jid.oxfordjournals.org/ at University of Waterloo on December 15, 2014
cr ipt
hypothesis that the risk allele identified in the Vietnamese sample was the allele associated
9 ranging from 0.001 to 0.02; Table 2) supporting their role in T1R but not in leprosy per se.
cr ipt
These 10 SNPs could be grouped into four r 2 ≥ 0.5 SNP bins (data not shown).
Fine mapping of TNFSF15 and TNFSF8 SNP bins in Vietnam
To fine map the pattern of association, 76 SNPs were genotyped in all Vietnamese families (Supplementary Table 1). Of the 56 SNPs that passed quality control, 47 were
with T1R (Figure 3). For the significant SNPs in the T1R subset, the common allele was associated with risk of T1R (Figure 3). However, associated alleles were not restricted to a
an
narrow allele frequency interval and extended from 0.54 to 0.89 (Figure 3). Assuming moderate LD of r2 ≥ 0.5 the 47 SNPs associated with T1R could be grouped into four bins
M
which was consistent with the initial scan and the selection criteria used for the fine mapping step (Figure 3). Next, we conducted a multivariate analysis including one SNP per bin associated with T1R (r2=0.5; rs6478108, rs998410, rs7863183 and rs3181348). This analysis
pt ed
detected a single signal of association with T1R of the TNFSF8 cis-eQTL rs7863183 (pmv = 0.0003) while the remaining SNPs lost significance. The results are largely consistent with the overall linkage disequilibrium pattern in the targeted chromosomal interval (data not shown). In addition, haplotype analysis including the T1R risk alleles of rs6478108,
ce
rs998410, rs7863183 and rs3181348 (G-C-T-G respectively) did not result in stronger association signals with T1R as compared to univariate analyses T1R (G-C-T-G f = 0.53 vs
A-A-C-A f = 0.08; p = 0.0002) reinforcing the non-independence of these markers. Single
Ac
association signals were also detected when employing different multivariate strategies.
Downloaded from http://jid.oxfordjournals.org/ at University of Waterloo on December 15, 2014
us
significantly associated (p < 0.05) and eight were highly significantly associated (p < 0.001)
10 Validation of T1R SNP risk alleles in Brazilian patients Since initially the LD structure in the Brazilians was unknown we decided to
cr ipt
genotype the Brazilian samples for the same SNPs used to fine map the TNFSF15-TNFSF8 locus in the Vietnamese families. We replicated (punilateral < 0.05) 18 SNPs in the Brazilian
samples, all of them being part of the same Vietnamese bin (Figure 3). In the Brazilians, 15 SNPs belonged to the same r 2 ≥ 0.8 bin while the remaining three markers were single SNP
corresponding bins mapped to the vicinity of the TNFSF8 gene. Except the SSB SNPs (rs1322067, rs3181195 and rs1555457), all validated SNPs displayed a switch of MAF from
an
the Vietnamese to the Brazilian sample. Nevertheless, all risk alleles were consistent between Vietnamese and Brazilian patients (Figure 3). The multivariate analysis of the
M
Brazilian sample included sex, age at leprosy diagnosis and one SNP per bin associated with T1R (r2=0.5, rs3181348, rs1555457 and rs1322067). Consistent with the results obtained in the Vietnamese sample, the best multivariate models in Brazil included one SNP, sex and
pt ed
age. However, statistically the tested SNPs were equivalent. Of note, rs1322067 was found to be a cis-eQTL for TNFSF8 in monocytes and whole blood cells while rs3181348 and rs1555457 are cis-eQTL for TNFSF8 in whole blood cells (Figure 4). In the combined Vietnamese and Brazilian sample the two strongest effects for the replicated SNPs were
ce
observed for rs3181348 (pcombined = 0.001) and rs1555457 (pcombined = 0.0007; Figure 3).
Ac
Discussion
T1R and leprosy offer a prime example of how susceptibility to a pathogen and
excessive host responsiveness triggered by the pathogen but unfolding even in its absence can be dissected. The incidence of T1R varies substantially across populations, ranging from 6% to 67% of all leprosy patients in different leprosy settings [24]. In approximately one-third of
Downloaded from http://jid.oxfordjournals.org/ at University of Waterloo on December 15, 2014
us
bins (SSB; Figure 3). All SNPs with significant evidence for association and their
11 cases T1R diagnosis is co-incident with leprosy diagnosis. The vast majority of the remaining T1R cases occur within 2 years after onset of anti-leprosy drug treatment. The
cr ipt
identity of T1R risk factors, the reasons for the pronounced population variability in incidence and the cause of variable time-to-onset are unknown. Studies of patients
undergoing acute T1R observed high serum levels of Th1-type associated cytokines, including IFNγ, TNF, CXCL10 and IL6 [25-27]. However, none of these cytokines
transcriptomics studies employing a prospective design have shown a strong impact of constitutive host genetic factors for risk of T1R [24]. Here, we describe TNFSF8 regulatory
an
variants as genetic risk factors for T1R but not for leprosy per se and provide an example for the distinct genetic control of susceptibility to infectious disease and excessive host
M
responsiveness triggered by the causative pathogen. T1R is unlikely to be a confounder for other described leprosy risk factors since none of the published leprosy per se associated SNPs showed significant evidence for association with T1R in our sample.
pt ed
Albeit the linkage disequilibrium pattern of associated SNPs in the Vietnamese sample was complex, multivariate analyses detected only a single signal strongly arguing against the independent contribution of TNFSF8 and TNFSF15 to T1R risk. While it was impossible to assign the genetic effect to either TNFSF15 or TNFSF8 in the Vietnamese
ce
sample, the association results in the Brazilian sample only implicated TNFSF8 in T1R. Nevertheless, the strength of the p values for association in the Brazilian sample was weaker as compared to the Vietnamese discovery sample. Several factors acting independently or
Ac
jointly may underlie the weaker association in Brazilian patients. First, a substantial drop in the T1R-risk allele frequencies in the Brazilian population reduced effective sample size and weakened power to detect an association under the same genetic model as employed in the Vietnam sample. Given this very substantial change in allele frequencies, it is interesting that
Downloaded from http://jid.oxfordjournals.org/ at University of Waterloo on December 15, 2014
us
displayed constitutive serum levels that are risk factors for T1R in leprosy patients. Recent
12 the same risk alleles were detected in both ethnic groups. Second, it is possible that diagnostic criteria may have varied slightly between Vietnam and Brazil. While we made
cr ipt
every effort to apply uniform diagnostic criteria it is important to realize that there are no objective physiological assays that determine T1R and we cannot rule out small deviations in case identification across enrolment sites [28]. A question remains if we missed SNPs that are T1R risk factors in the Brazilian sample. A brief survey of the CEU, YRI and MEX
in LD (r2 = 0.5) with the Vietnamese tag SNPs in these three populations had been tested in the Brazilian sample. Nevertheless, considering the large differences in epidemiology,
an
geography and ethnicity of the Vietnamese and Brazilian samples our results support an important role of TNFSF8 in the pathogenesis of T1R. The reason why the TNFSF15 SNPs
M
were initially detected as leprosy per se risk factors is not known. However, it is well known that associations in large samples can be traced to subsets of the overall sample. Hence, the most parsimonious explanation is that the proportion of T1R cases in the GWAS discovery
pt ed
sample was large enough to result in a significant result for the overall sample.
Of the SNPs investigated 47 had previously been shown to be eQTL for TNFSF8 (Figure 4). Of the initial 56 SNPs, genotypes of 19 SNPs had been reported to correlate with
ce
TNFSF8 expression levels in monocytes while 43 SNPs were eQTL for TNFSF8 in a metaanalysis of untransformed peripheral blood cells (Figure 4). Interestingly, all 18 SNPs validated in the Brazilian sample were among the eQTL for TNFSF8 and six SNPs had been
Ac
reported by two independent studies in different cell types and multiple cohorts (Figure 4). It is possible that the high number of eQTL for TNFSF8 in the studied interval is a reflection of
the highly correlated nature of the SNP alleles. However, a proportion of the correlated SNPs possibly represent genomic variations with independent gene regulatory function. For
Downloaded from http://jid.oxfordjournals.org/ at University of Waterloo on December 15, 2014
us
HapMap populations argues against this possibility since more than 80% of all SNPs that are
13 example, SNP rs1322067 is located under a peak of a H3K27ac site (Supplementary Figure 2A). Acetylation of K27 at histone H3 promotes opening of chromatin and enhanced
cr ipt
regulatory activity. Equally interesting, for the variant rs3181348 the ancestral A-allele is highly conserved across 30 eutherian mammals (Supplementary Figure 2B). The rs3181348
G T1R risk allele is a CpG methylation site according ENCODE data (Supplementary Figure 2B). Methylation of CpGs sites in promoter regions favors gene silencing while the absence
rs12337739 locate to the region of a series of transcription factor binding sites and may modulate binding of key transcriptional regulators (Supplementary Figure 2C). The most
an
significant SNP in Vietnam, rs7863183 T-allele is an important element in the binding motif of Hic1 and the T-allele which promotes binding was a risk factor for T1R (Supplementary
M
Figure 2D). Hence, these variants could potentially impact on TNSFS8 gene expression.
The TNFSF8 gene is a member of the tumor necrosis superfamily and encodes a
pt ed
surface antigen expressed in a subpopulation of T and B cells [29]. Functionally, TNFSF8 (aliases CD30L or CD153) fits the pathological characteristics of T1R. In humans, TNFSF8 has been shown to impact on the balance between Th1 – Th2 cells [30] and T1R occurs almost exclusively in borderline leprosy subtypes which are characterized by an unstable Th1
ce
– Th2 response [31]. Moreover, our results indicated a possible role for neutrophils in the pathogenesis of T1R since the T1R-risk alleles of SNPs rs2295800, rs1555457 and rs3181348 are associated with elevated neutrophil counts [32]. Previous studies suggested
Ac
that increased neutrophil counts may favour the emergence of a pathogenic proinflammatory loop of chemoattraction mediated by CCL20, CCL2, CXCL10 and CXCL8 [33]. CCL20 and CCL2 were part of a T1R signature gene expression set observed in patients that were destined to undergo T1R [24] while CXCL10 is over-expressed in patients with acute T1R
Downloaded from http://jid.oxfordjournals.org/ at University of Waterloo on December 15, 2014
us
of methylation promotes the gene expression. Additionally, markers rs1322054 and
14 [26, 34]. Neutrophils have also emerged as a key component in tuberculosis pathogenesis [35]. Hence, the identification of TNFSF8 as global T1R risk factor adds further evidence for
cr ipt
the importance of neutrophils in human inflammatory diseases.
In summary, our data show that SNPs in the TNFSF15-TNFSF8 locus previously
associated with leprosy per se are in fact T1R risk factors. Key associated SNPs are eQTLs
Brazilian patients only implicated TNFSF8 in T1R risk. These results indicate TNFSF8 as T1R susceptibility gene in two geographically and ethnically distinct populations. Future
an
research will need to address if TNFSF8 is a specific mediator of mycobacterial-triggered
Conflict of interests
M
excessive host responses or functions as general controller of inflammatory responses.
pt ed
The author’s declare that they have no conflict of interests
Funding
This work was supported by grants from the Canadian Institutes of Health Research (CIHR) to Erwin Schurr; MALTALEP from l’Ordre de Malte to Alexandre Alcaïs and Erwin Schurr;
ce
and L’Agence Nationale de la Recherche to Alexandre Alcaïs; Vinicius Medeiros Fava was supported by a fellowship [Proc. 9074/11-2] of Coordenação de Aperfeiçoamento de Pessoal
Ac
de Nível Superior (CAPES) Ministry of Education of Brazil, Brasilia, Brazil; Aurélie Cobat was the recipient of a Banting postdoctoral fellowship from the Government of Canada, and Jérémy Manry was supported by a postdoctoral fellowship from CIHR.
Downloaded from http://jid.oxfordjournals.org/ at University of Waterloo on December 15, 2014
us
for TNFSF8 directly implicating this gene in T1R. Similarly, results of a replication study in
15
Acknowledgement We thank all family members and leprosy patients who participated in this study. We thank
cr ipt
all the members of the Schurr lab, and the members of the Necker and Rockefeller branches of the laboratory for Human Genetics of Infectious Diseases for useful discussions and
an M pt ed ce Ac
Downloaded from http://jid.oxfordjournals.org/ at University of Waterloo on December 15, 2014
us
suggestions.
16 REFERENCE
cr ipt
us
an
M
pt ed
ce Ac
Downloaded from http://jid.oxfordjournals.org/ at University of Waterloo on December 15, 2014
1. Cardoso CC, Pereira AC, Brito-de-Souza VN, et al. TNF -308G>A single nucleotide polymorphism is associated with leprosy among Brazilians: a genetic epidemiology assessment, meta-analysis, and functional study. J Infect Dis 2011; 204:1256-63. 2. Marques Cde S, Brito-de-Souza VN, Guerreiro LT, et al. Toll-like receptor 1 N248S single-nucleotide polymorphism is associated with leprosy risk and regulates immune activation during mycobacterial infection. J Infect Dis 2013; 208:120-9. 3. Pereira AC, Brito-de-Souza VN, Cardoso CC, et al. Genetic, epidemiological and biological analysis of interleukin-10 promoter single-nucleotide polymorphisms suggests a definitive role for -819C/T in leprosy susceptibility. Genes Immun 2009; 10:174-80. 4. Alter A, Fava VM, Huong NT, et al. Linkage disequilibrium pattern and age-at-diagnosis are critical for replicating genetic associations across ethnic groups in leprosy. Hum Genet 2013; 132:107-16. 5. Alcais A, Alter A, Antoni G, et al. Stepwise replication identifies a low-producing lymphotoxin-[alpha] allele as a major risk factor for early-onset leprosy. Nat Genet 2007; 39:517-22. 6. Alter A, Huong NT, Singh M, et al. Human leukocyte antigen class I region singlenucleotide polymorphisms are associated with leprosy susceptibility in Vietnam and India. J Infect Dis 2011; 203:1274-81. 7. Alter A, de Leseleuc L, Van Thuc N, et al. Genetic and functional analysis of common MRC1 exon 7 polymorphisms in leprosy susceptibility. Hum Genet 2010; 127:337-48. 8. Grant AV, Cobat A, Van Thuc N, et al. CUBN and NEBL common variants in the chromosome 10p13 linkage region are associated with multibacillary leprosy in Vietnam. Hum Genet 2014. 9. Mira MT, Alcais A, Nguyen VT, et al. Susceptibility to leprosy is associated with PARK2 and PACRG. Nature 2004; 427:636-40. 10. Zhang FR, Huang W, Chen SM, et al. Genomewide association study of leprosy. N Engl J Med 2009; 361:2609-18. 11. Zhang F, Liu H, Chen S, et al. Identification of two new loci at IL23R and RAB32 that influence susceptibility to leprosy. Nat Genet 2011; 43:1247-51. 12. Schurr E, Gros P. A common genetic fingerprint in leprosy and Crohn's disease? N Engl J Med 2009; 361:2666-8. 13. Grant AV, Alter A, Huong NT, et al. Crohn's disease susceptibility genes are associated with leprosy in the Vietnamese population. J Infect Dis 2012; 206:1763-7. 14. Wong SH, Hill AV, Vannberg FO, India-Africa-United Kingdom Leprosy Genetics C. Genomewide association study of leprosy. N Engl J Med 2010; 362:1446-7; author reply 7-8. 15. Rodrigues LC, Lockwood D. Leprosy now: epidemiology, progress, challenges, and research gaps. Lancet Infect Dis 2011; 11:464-70. 16. Zeller T, Wild P, Szymczak S, et al. Genetics and beyond--the transcriptome of human monocytes and disease susceptibility. PLoS One 2010; 5:e10693. 17. Westra HJ, Peters MJ, Esko T, et al. Systematic identification of trans eQTLs as putative drivers of known disease associations. Nat Genet 2013; 45:1238-43. 18. Sousa AL, Fava VM, Sampaio LH, et al. Genetic and immunological evidence implicates interleukin 6 as a susceptibility gene for leprosy type 2 reaction. J Infect Dis 2012; 205:141724. 19. Barrett JC, Fry B, Maller J, Daly MJ. Haploview: analysis and visualization of LD and haplotype maps. Bioinformatics 2005; 21:263-5. 20. Horvath S, Xu X, Laird NM. The family based association test method: strategies for studying general genotype--phenotype associations. Eur J Hum Genet 2001; 9:301-6.
17
cr ipt
us
an
M
pt ed
ce Ac
Downloaded from http://jid.oxfordjournals.org/ at University of Waterloo on December 15, 2014
21. Schaid DJ, Rowland C. Use of parents, sibs, and unrelated controls for detection of associations between genetic markers and disease. Am J Hum Genet 1998; 63:1492-506. 22. Purcell S, Cherny SS, Sham PC. Genetic Power Calculator: design of linkage and association genetic mapping studies of complex traits. Bioinformatics 2003; 19:149-50. 23. Purcell S, Neale B, Todd-Brown K, et al. PLINK: a tool set for whole-genome association and population-based linkage analyses. Am J Hum Genet 2007; 81:559-75. 24. Orlova M, Cobat A, Huong NT, et al. Gene set signature of reversal reaction type I in leprosy patients. PLoS Genet 2013; 9:e1003624. 25. Scollard DM, Adams LB, Gillis TP, Krahenbuhl JL, Truman RW, Williams DL. The continuing challenges of leprosy. Clinical microbiology reviews 2006; 19:338-81. 26. Stefani MM, Guerra JG, Sousa AL, et al. Potential plasma markers of Type 1 and Type 2 leprosy reactions: a preliminary report. BMC Infect Dis 2009; 9:75. 27. Moraes MO, Sarno EN, Almeida AS, et al. Cytokine mRNA expression in leprosy: a possible role for interferon-gamma and interleukin-12 in reactions (RR and ENL). Scand J Immunol 1999; 50:541-9. 28. Lockwood DN, Nicholls P, Smith WC, et al. Comparing the clinical and histological diagnosis of leprosy and leprosy reactions in the INFIR cohort of Indian patients with multibacillary leprosy. PLoS Negl Trop Dis 2012; 6:e1702. 29. Oflazoglu E, Grewal IS, Gerber H. Targeting CD30/CD30L in oncology and autoimmune and inflammatory diseases. Advances in experimental medicine and biology 2009; 647:17485. 30. Pellegrini P, Berghella AM, Contasta I, Adorno D. CD30 antigen: not a physiological marker for TH2 cells but an important costimulator molecule in the regulation of the balance between TH1/TH2 response. Transplant immunology 2003; 12:49-61. 31. Ranque B, Nguyen VT, Vu HT, et al. Age is an important risk factor for onset and sequelae of reversal reactions in Vietnamese patients with leprosy. Clin Infect Dis 2007; 44:33-40. 32. Arruda-Olson AM, Roger VL, Chai HS, et al. Association of TNFSF8 polymorphisms with peripheral neutrophil count. Mayo Clinic proceedings Mayo Clinic 2011; 86:1075-81. 33. Pelletier M, Maggi L, Micheletti A, et al. Evidence for a cross-talk between human neutrophils and Th17 cells. Blood 2010; 115:335-43. 34. Scollard DM, Chaduvula MV, Martinez A, et al. Increased CXC ligand 10 levels and gene expression in type 1 leprosy reactions. Clin Vaccine Immunol 2011; 18:947-53. 35. Berry MP, Graham CM, McNab FW, et al. An interferon-inducible neutrophil-driven blood transcriptional signature in human tuberculosis. Nature 2010; 466:973-7. 36. Rosenbloom KR, Dreszer TR, Long JC, et al. ENCODE whole-genome data in the UCSC Genome Browser: update 2012. Nucleic Acids Res 2012; 40:D912-7. 37. Ward LD, Kellis M. HaploReg: a resource for exploring chromatin states, conservation, and regulatory motif alterations within sets of genetically linked variants. Nucleic Acids Res 2012; 40:D930-4. 38. Boyle AP, Hong EL, Hariharan M, et al. Annotation of functional variation in personal genomes using RegulomeDB. Genome Res 2012; 22:1790-7. 39. Flicek P, Ahmed I, Amode MR, et al. Ensembl 2013. Nucleic Acids Res 2013; 41:D4855.
Table 1. Leprosy clinical subtype across the studied populations Vietnam
Goiânia
Rondonópolis
rip t
19
Combined Brazil
T1R-affected
T1R-free
T1R-affected
T1R-free
T1R-affected
T1R-free
T1R-affected
T1R-free
TT
6
2.7%
__
__
__
__
18
8.7%
21
8.9%
22
12.5%
21
5.6%
40
10.4%
BT
51
22.8%
53
40.2%
69
49.6%
125
60.1%
72
30.6%
77
43.8%
141
37.7%
202
52.6%
BB
92
41.1%
49
37.1%
28
20.1%
16
7.7%
114
48.5%
60
34.1%
142
38.0%
76
19.8%
BL
66
29.5%
30
22.7%
42
30.2%
26
12.5%
20
8.5%
8
4.5%
62
16.6%
34
8.9%
LL
8
3.6%
__
__
__
__
23
11.1%
1
0.4%
__
__
1
0.3%
23
6.0%
I*
1
0.4%
__
__
__
__
0
0
7
3.0%
9
5.1%
7
1.9%
9
2.3%
Total
224
139
208
M an u
132
sc
R&J
235
176
374
384
*Indeterminate leprosy patients are not expected to develop reactions and the small percentage of indeterminate cases among T1R cases may reflect
ep t Ac c
T1R, Type-1 reaction
Downloaded from http://jid.oxfordjournals.org/ at University of Waterloo on December 15, 2014
T1R.
ed
patients that were classified as indeterminate at leprosy diagnosis but had progressed to one of the borderline forms of leprosy prior to the development of
20 Table 2. SNP association scan of TNFSF15 and TNFSF8 in the Vietnamese sample.
M/m
111 T1R free families
Risk
Risk
MAF a
(bp)
pb Allele
p pb
Allele
C/A
0.46
C
0.43
C
0.41
0.67
rs4574921
117538334
A/G
0.38
G
0.40
A
0.64
0.64
rs10114470
117547772
A/G
0.44
A
0.20
A
0.38
0.64
rs6478108
117558703
G/A
0.42
G
0.0002
G
0.30
0.02
rs6478106
117545666
G/A
0.18
G
0.04
G
0.64
0.13
rs7847158
117560265
A/G
0.06
A
0.76
A
0.09
0.45
rs6478109
117568766
T/C
0.43
T
0.0007
T
0.44
0.02
rs1407308
117570223
T/G
0.06
T
0.89
T
0.09
0.35
rs10982412
117570856
C/T
0.05
C
0.61
T
0.28
0.28
rs2145929
117581940
T/C
0.24
T
0.93
T
0.30
0.57
rs7868351
117584082
A/C
0.22
C
0.68
A
0.06
0.18
rs2006996
117592638
T/C
0.16
T
0.005
T
0.16
0.01
rs11554257
117605070
A/G
0.27
G
0.93
A
0.07
0.25
rs2093403
117611913
T/C
0.23
C
0.43
T
0.06
0.10
rs10114224
117613032
A/G
0.28
G
0.65
A
0.22
0.28
rs7866342
117627569
G/T
0.45
G
0.0009
G
0.43
0.02
117630043
G/A
0.18
G
0.0009
A
0.57
0.005
117636879
G/T
0.05
G
0.75
G
0.32
0.70
117642418
T/C
0.20
T
0.0002
C
0.32
0.001
rs7028891
117645015
A/G
0.16
A
0.001
A
0.73
0.02
rs2295800
117664211
T/C
0.22
T
0.005
C
0.24
0.02
rs3181366
117666776
C/T
0.23
C
0.12
T
0.77
0.28
rs3181362
117667443
T/C
0.06
T
0.12
C
0.04
0.02
rs1322055
117669585
T/C
0.21
C
0.45
C
0.24
0.62
rs1555457
117674320
C/T
0.15
C
0.004
T
0.55
0.01
rs2418326
117679474
G/A
0.05
A
0.53
G
0.44
0.35
rs7874896
Ac
ce
rs7863183
an
M
pt ed
rs4979467
us
117493017
Heterogeneity
Downloaded from http://jid.oxfordjournals.org/ at University of Waterloo on December 15, 2014
rs10982385
cr ipt
Position Marker
216 T1R families
21 117684547
C/T
0.21
C
0.02
T
0.45
0.14
rs3181360
117691558
C/T
0.06
C
0.12
T
0.08
0.02
rs3181348
117694184
G/A
0.26
G
0.003
A
0.19
0.007
rs2181033
117697831
A/G
0.21
A
0.02
G
0.26
0.13
rs17292115
117699745
A/C
0.07
C
0.03
A
0.74
0.05
rs13288168
117699786
C/T
0.04
T
0.04
T
0.30
0.50
rs10817686
117699850
A/G
0.28
G
0.52
A
0.02
0.26
rs10982461
117701408
T/G
0.48
G
0.12
T
0.03
0.03
us
MAF; minimum allele frequency, was estimated in 357 leprosy unaffected parents
b
p value refers to the best genetic model (Additive, Dominant or Recessive)
Ac
ce
pt ed
M
an
a
Downloaded from http://jid.oxfordjournals.org/ at University of Waterloo on December 15, 2014
M; Major allele. m; minor allele
cr ipt
rs1322058
22
Figure legends
cr ipt
Figure 1. Samples and Study design. The boxes represent the description of the family-based sample from Vietnam and the casecontrol samples from Goiânia and Rondonópolis. The Vietnamese sample was stratified
according T1R status. The median and mean age at leprosy diagnosis for all the groups is
the Vietnamese sample.
an
Figure 2. SNP association scan of the TNFSF15 and TNFSF8 chromosomal region in
A total of 34 SNPs mapping to a 220kb genomic interval extending from TNFSF15 to
M
TNFSF8 are shown. Results are shown as negative logarithm of p value (NLP) on the y axis supporting association of a SNP with T1R plotted against the SNP chromosomal position in mega base according to human build GRCh37 on the x axis. The black and grey arrows in
pt ed
the upper part of the graphs indicate the physical position and orientation of the TNFSF15 and TNFSF8 genes, respectively. The dotted line indicates the threshold of p = 0.01. A. 216 families from the T1R subset. B. 111 families from the T1R-free subset.
ce
Figure 3. Fine mapping of TNFSF8 and TNFSF15 SNP bins. Association results of 56 SNPs with T1R for the Vietnamese, Brazilian and combined samples are shown. The major/minor alleles, T1R risk allele frequencies, odds ratios for best
Ac
genetic model (dominant, recessive or additive) with 95% CI for the Vietnamese and Brazilian samples, as well as p values for the association with T1R are shown. For the Brazilian sample a one-sided test was performed. SNPs belonging to the same bin are linked
Downloaded from http://jid.oxfordjournals.org/ at University of Waterloo on December 15, 2014
us
shown in the lower part of the graph. T1R, type-1 reversal reaction; S.D., standard deviation.
23 and grouped by color M, Major allele; m, minor allele; RA, risk allele; q, frequency; OR,
cr ipt
odds ratio; CI, confidence interval; LD, linkage disequilibrium.
Figure 4. Abundance of eQTLs for TNFSF8 among SNPs validated as risk factor for T1R.
Chromosomal position in mega base (Mb) is indicated on the left side of the graph. The
gray and black arrow, respectively. Physical locations of 56 tested SNPs are indicated by short horizontal bars that are linked to the rs SNP identifiers. SNPs validated for the
an
association with T1R (p < 0.05) in the Brazilian sample are identified with an X. SNP bins in the HapMap CEU sample are shown since the eQTL mappings had been conducted in this
M
ethnic group. SNPs belonging to the same bin are linked and grouped by colors. SNPs identified as cis-eQTLs for TNFSF8 in circulating monocytes obtained from German subjects and non-transformed peripheral blood cells obtained from seven European cohorts are
Ac
ce
pt ed
indicated in the last two columns [16, 17].
Downloaded from http://jid.oxfordjournals.org/ at University of Waterloo on December 15, 2014
us
physical location and transcriptional orientation of TNFSF8 and TNFSF15 are represented as
ce
pt ed
M
an
us
cr ipt
Downloaded from http://jid.oxfordjournals.org/ at University of Waterloo on December 15, 2014
Ac
24
ce
pt ed
M
an
us
cr ipt
Downloaded from http://jid.oxfordjournals.org/ at University of Waterloo on December 15, 2014
Ac
25
ce
pt ed
M
an
us
cr ipt
Downloaded from http://jid.oxfordjournals.org/ at University of Waterloo on December 15, 2014
Ac
26
ce
pt ed
M
an
us
cr ipt
Downloaded from http://jid.oxfordjournals.org/ at University of Waterloo on December 15, 2014
Ac
27