Hum Genet (2013) 132:265–273 DOI 10.1007/s00439-012-1244-5

ORIGINAL INVESTIGATION

Identification of genetic associations of SP110/MYBBP1A/RELA with pulmonary tuberculosis in the Chinese Han population

Lei Cai • Shao-Li Deng • Li Liang • Hui Pan • Jia Zhou • Mei-Yan Wang • Jun Yue • Chun-Ling Wan • Guang He • Lin He

Received: 18 June 2012 / Accepted: 17 October 2012 / Published online: 6 November 2012 Ó Springer-Verlag Berlin Heidelberg 2012

Abstract Genetic factors play important roles in the meta-analysis of rs9061 in East Asian populations showed development of tuberculosis (TB). SP110 is a promising that the rs9061 T allele conferred significant risk for TB candidate target for controlling TB infections. However, [P = 0.002, pooled odds ratio (OR), 1.24, 95 % confidence several studies associating SP110 single nucleotide poly- interval (CI) = 1.08–1.43]. The MYBBP1A GTCTTGGG morphisms (SNPs) with TB have yielded conflicting haplotype and haplotypes CGACCG/TGATTG within results. This may be partly resolved by studying other SP110 were found to be markedly and significantly asso- genes associated with SP110, such as MYBBP1A and ciated with TB (P = 2.00E-06, 5.00E-6 and 2.59E-4, RELA. Here, we genotyped 6 SP110 SNPs, 8 MYBBP1A respectively). Gene-based analysis also demonstrated that SNPs and 5 RELA SNPs in 702 Chinese pulmonary TB SP110 and MYBBP1A were each associated with TB patients and 425 healthy subjects using MassARRAY and (Pcor = 0.011 and 0.035, respectively). The logistic SNaPshot methods. Using SNP-based analysis with Bon- regression analysis results supported interactions between ferroni correction, rs3809849 in MYBBP1A [Pcorrected SP110 and MYBBP1A, indicating that subjects carrying a (cor) = 0.0038] and rs9061 in SP110 (Pcor = 0.019) were GC/CC genotype in MYBBP1A and CC genotype in SP110 found to be significantly associated with TB. Furthermore, possessed the high risk of developing TB (P = 1.74E-12). Our study suggests that a combination of SP110 and MYBBP1A gene polymorphisms may serve as a novel Electronic supplementary material The online version of this article (doi:10.1007/s00439-012-1244-5) contains supplementary material, which is available to authorized users.

L. Cai (&) C.-L. Wan G. He L. He L. Liang J. Yue (&) Bio-X Center, Key Laboratory for the Genetics Shanghai Key Laboratory of Mycobacterium Tuberculosis, of Developmental and Neuropsychiatric Disorders Shanghai Pulmonary Hospital, Tongji University School (Ministry of Education), Shanghai Jiaotong University, of Medicine, 507 Zhengmin Rd, Shanghai 200433, China 800 Dongchuan Road, Shanghai 200030, China e-mail: [email protected] e-mail: [email protected] H. Pan L. Cai J. Zhou Department of Immunology and Infectious Diseases, Gordon Life Science Institute, 13784 Torrey Del Mar Drive, Harvard School of Public Health, Boston, USA San Diego, CA 92130, USA M.-Y. Wang L. Cai J. Zhou Fudan University School of Life Sciences and Technology, Key Laboratory of Systems Biomedicine, Advanced Institute of Translational Medicine, Shanghai, China Shanghai Centre for Systems Biomedicine, Shanghai Jiaotong University, Shanghai, China

S.-L. Deng Department of Laboratory Medicine, Institute of Surgery Research, Daping Hospital, The Third Military Medical University, Chongqing, China

123 266 Hum Genet (2013) 132:265–273 marker for identifying the risk of developing TB in the structurally related inducible transcription factors (Bonizzi Chinese Han population. and Karin 2004). NF-jB, which is known to interact with IFN-c during TB development, regulates the expression of a wide variety of human genes, including factors regulating Introduction apoptosis and inducers of proliferation (Chan et al. 2001; Pahl 1999). Tuberculosis (TB) is a re-emerging disease epidemic in This study aimed to determine whether any genetic many parts of the world. As of 2010, China has the second- associations exist between SP110/MYBBP1A/RELA and largest number of incident cases worldwide (WHO 2011). pulmonary TB. We performed multiple analyses including Despite the availability of effective treatment over the past SNP-based, haplotype-based and gene-based analyses on an few decades, TB remains the second leading cause of association study using 1,127 Chinese subjects. The SNP- deaths from infectious diseases worldwide (WHO 2011). based analysis results were further evaluated using meta- Multiple factors contribute to the risk of infection and analysis and protein 3D structure analysis. The logistic development of TB, including host–pathogen interactions regression analysis was performed to evaluate the gene– and environmental factors. Furthermore, there is evidence gene interaction. Our work suggests a combination of SP110 to indicate that the risk of developing TB in human is and MYBBP1A gene polymorphisms is significantly asso- strongly influenced by genetic factors (Bellamy 2003; ciated with TB, which provides an effective novel bio- Cooke and Hill 2001). marker for TB susceptibility in the Chinese Han population. SP110 is emerging as a promising candidate for its mouse homologue controlling Mycobacterium tuberculosis (MTB) infections (Pan et al. 2005). SP110 contains Materials and methods nuclear-characterized structural motifs, such as the chro- matin-interacting SAND domain and the LXXLL nuclear Study subjects hormone interaction motif. Following induction by inter- feron-a and interferon-c, SP110 is preferentially expressed Patients with TB were recruited from Shanghai Pulmonary in leukocytes and spleen cells (Bloch et al. 2000). The Hospital and Chongqing Daping Hospital in southern mouse homologue of SP110 has been found to be strongly China between June 2009 and September 2011. All patients expressed in lung lesions from TB and macrophages of were clinically and radiologically diagnosed with pul- susceptibility to tuberculosis 1 (sst1)-resistant mice, but not monary TB; sputum smears and cultures were used to expressed in sst1-susceptible mice. Moreover, mouse confirm M. tuberculosis infection. All patients were HIV- SP110 has the ability to control MTB growth in macro- negative and none presented with other infectious diseases phages and induce apoptosis in infected macrophage cells or immunosuppressive conditions. The characteristics of (Pan et al. 2005). study subjects were listed in Table 1. Healthy control Several SNP-based studies have reported associations subjects were recruited from blood donors, who had neg- between SP110 and susceptibility to pulmonary TB in ative tuberculin skin tests with indurations of \5mmon different populations, however, these have yielded con- the fourth day after inoculation, no history of TB and no flicting results (Babb et al. 2007; Cong et al. 2010; Liang evidence of prior TB noted in chest radiographies. Both et al. 2011; Png et al. 2012; Szeszko et al. 2007; Tosh et al. cases and controls were selected with following procedures 2006). We note that inconsistencies arising from popula- to exclude the possibility of immunosuppressive condi- tion differences are more readily resolved by use of a gene- tions. Firstly, they were investigated that they had no his- based or multi-gene-based approach rather than using tory of an immunosuppressive condition (i.e., human either a SNP-based or a haplotype-based approach (Neale immunodeficiency virus (HIV) infection, leukemia, lym- and Sham 2004). Thus, these conflicting results may be phoma, diabetes mellitus, or renal failure) or of taking partly resolved by studying other genes associated with immunosuppressive drugs within the last 3 months. Sec- SP110, such as MYBBP1A and RELA, which have not been ondly, HIV antibody and CD4? cell counts were assessed investigated in previous works. MYBBP1A has been to exclude HIV infection. Thirdly, these CD3?, CD4? and reported to bind with SP110 in mice, and it has been CD8? cell counts were assessed to exclude cell immune demonstrated that SP110 induces cells apoptosis by inter- deficiency. Finally, measurements of quantitative immu- acting with MYBBP1A (Cai et al. 2010). MYBBP1A is a noglobulin levels, including IgG, IgA, and IgM, and transcription factor critical for hematopoietic cell proli- complement levels were used to exclude antibody and feration and differentiation and is involved in the NF-jB complement deficiency. Informed consent was obtained pathway through its ability to bind with RELA (Owen et al. from all TB patients and controls participating in this 2007). RELA is a member of the NF-jB family of study. This study protocol was approved by the Ethics 123 Hum Genet (2013) 132:265–273 267

Table 1 Characteristics of subjects in this study using 5.4 pmol of each primer extension probe, 50 mmol Characteristics Cases Controls of iPlex Termination Mix, and 0.5 U of iPLEX enzyme (Sequenom). The extension reactions were carried out at Number 702 425 94 °C for 30 s and then 94 °C for 5 s, followed by 5 cycles at Age, mean ± SD (years) 40.0 ± 15.6 41.2 ± 17.7 52 °C for 5 s and 80 °C for 5 s for a total of 40 cycles and Male (%) 68.2 66.7 finally at 72 °C for 3 min. Purified extension reaction prod- Number of patients with/without fever 254/448 ucts with cation exchange resin were spotted onto Spectro- Number of patients with cavities in 369 CHIPs and determined by MALDI-TOF mass spectrometry. chest radiograph The SNP rs9061 and two other SNPs (rs11556887 and SD standard deviation rs1135791) investigated to evaluate the genotyping above were studied using the SNaPshot Kit (Applied Biosystems, Committee of Shanghai Pulmonary Hospital and Chongq- Inc.). The details of the primers used are listed in Supple- ing Daping Hospital. mentary Table 2. PCR was performed in a 20-lL reaction In total, we had 702 patients with TB and 425 healthy volume including components similar to those listed above. control subjects in the present study. To detect alleles with For amplification, samples were denatured at 95 °C for C20 % frequency, and an odds ratio (OR) of 1.5, for an 15 min followed by 11 cycles of denaturation at 94 °C for uncorrected significance threshold of significance at 20 s, 62-0.5 °C/cycle for 40 s, 72 °C for 1.5 min and 26 P = 0.05, the sample size in this study has over 80 % cycles of annealing at 94 °C for 20 s, 56 °C for 30 s, 72 °C power to detect associations in both the allelic [1 degree for 1.5 min and a final extension at 72 °C for 5 min. Purified freedom (df)] and genotypic (2 df) models. This is based on PCR products with SAP and exonuclease I (ExoI) proces- the prevalence of TB at 112 cases per 100,000 people in sion were then used as templates for a minisequencing China (Purcell et al. 2003; WHO 2011). extension reaction using the SNaPshot Kit. Purified products of minisequencing extension reactions with SAP procession SNP selection and genotyping were then detected by capillary electrophoresis on an ABI3730xl DNA analyzer (Applied Biosystems, Inc). Eight SNPs within the MYBBP1A gene, six SNPs in the In the current study, the genotype agreement rate SP110 gene, and five SNPs in the RELA gene (Supple- between duplicate samples was 100 %, and the call rate per mentary Table 1) were selected as the tag SNPs, using r2 SNP was 100 %, except for rs1045845, rs751670 and C0.50 and a minor allele frequency (MAF) C0.05 in the rs7211078, which had call rates of 99.73, 99.91 and Chinese Han population as selection criteria, provided by 99.91 %, respectively. public SNP databases (http://www.ncbi.nlm.nih.gov/snp and http://www.hapmap.org). These tagging and poten- Database construction for meta-analysis tially functional SNPs were located within 10 kb upstream of the 50 un-translated region and 10 kb downstream of the A thorough search of HuGE Navigator (Yu et al. 2008), 30 un-translated region of each gene. PubMed, ISI Web of Science, EMBASE, and CNKI data- Genomic DNA was isolated from 0.2 mL EDTA anti- bases was performed, covering all papers published until May coagulated blood samples using standard methods (Cai et al. 30, 2012 with a combination of ‘‘TB’’, ‘‘tuberculosis’’, 2005). Genotyping of all SNPs, except for rs9061, was ‘‘genotype’’, ‘‘polymorphism’’ and ‘‘SP110’’. To identify performed using the MASSarray matrix-assisted laser other suitable papers that may have been missed in the initial desorption ionization-time-of-flight (MALDI-TOF) mass search, references of retrieved papers were also screened. The spectrometry platform (Sequenom; San Diego, CA). The following criteria were adopted for selection of meta-analysis details of the primers used for polymerase chain reaction studies: (1) papers concerning SP110 gene polymorphisms (PCR) and single base extension are shown in Supplemen- and the risk of TB; (2) case–control studies providing TB tary Table 2. Multiplex PCR was performed in a 5-lL diagnoses, the sources of cases and controls, using subjects reaction volumes containing 19 HotStar Taq buffer, lacking TB as controls; (3) papers listing sample size and 2.8 mM Mg2?, 0.1 U of HotStar Taq polymerase (Qiagen, genotype frequency for each SNP. The SNPs reported by at Inc.), 2 ng of whole-genome-amplified genomic DNA, least three papers were included in the meta-analysis. Sup- 0.5 pmol of each primer, and 0.5 mmol of dNTPs. Ther- plementary Table 3 lists the database used for these studies. mocycling was carried out at 94 °C for 15 min, followed by 45 cycles at 94 °C for 20 s, 56 °C for 30 s, and 72 °C for Statistical analysis 1 min, with a final incubation at 72 °C for 3 min. Unin- corporated dNTPs were deactivated using 0.3 U of shrimp A Chi-squared test and t test were used to analyze the alkaline phosphatase (SAP) followed by primer extension statistical difference in the characteristics between patients 123 268 Hum Genet (2013) 132:265–273 and controls. Fisher’s exact test was used to test the Hardy– were performed using R with the package SKAT v0.72 Weinberg equilibrium (HWE) for each SNP in control (Wu et al. 2010). subjects. For each individual genotype, the OR and 95 % confidence interval (CI) were estimated using logistic 3D structural modeling regression models adjusted for gender and age. The asso- ciation between the genotypes of individual SNPs and TB In this study, the I-TASSER (Iterative Threading was determined using the Cochran–Armitage trend test. To ASSEmbly Refinement) algorithm server was employed to correct for multiple testing for each SNP, both the Bon- construct the 3D (three-dimensional) structural models of ferroni method (Emily et al. 2009) and 10,000 permuta- the human SP110 SAND domain (Zhang 2009). The tions testing were performed. In Bonferroni method, the algorithm was based on the secondary-structure enhanced raw P value was multiplied by 19 tests. In permutation profile–profile threading alignment (PPA) and the iterative testing, 10,000 tests were performed for all SNPs and two implementation of the threading assembly refinement sets of empirical significance values were calculated, i.e. program. pointwise estimates of an individual SNPs significance and a value that controls for that fact that thousands of other SNPs were tested. This was achieved by comparing each Results observed test statistic against the maximum of all permuted statistics (over all SNPs) test for each single replicate. Patient characteristics For meta-analysis, the heterogeneity assumption was checked using the Chi-square-based Q-test. The heteroge- Among 702 TB patients and 425 controls, three patients neity was not considered significant when P [ 0.1, and the and two control subjects were missing genotypes for one pooled OR estimate of each study was calculated by the SNP from the genotyping results. However, they were fixed effects model (Mantel–Haenszel method) (Berlin included in the following analysis. Overall, there was no et al. 1989). Otherwise, the random effects model (DerSi- significant difference in the male/female ratio in either monian and Laird method) was adopted (DerSimonian and groups (P = 0.62), nor was there a significant difference in Kacker 2007). In addition, a funnel plot was drawn to age between groups (P = 0.26). In the case group, 254 assess possible publication bias, in which a symmetric subjects had fever (referring to a body temperatures above inverted funnel shape indicated a lack of publication bias 37.5 °C) and 369 subjects had cavities in chest radio- (Higgins and Thompson 2002). graphs. No control subjects displayed fever or lung The THESIAS program was used to estimate the hap- cavities. lotype frequency using the stochastic expectation–maxi- mization (EM) algorithm (Tregouet et al. 2004; Tregouet Single SNP association and Garelle 2007). The association between individual haplotypes and TB was tested using the Chi-squared test All genotyped SNPs were found to be in agreement with with 1 df to generate a haplotype-specific test. The global the Hardy–Weinberg equilibrium in the control group test for the association of each gene with TB was the (n - 1) (Supplementary Table 4). Among them, two SNPs were df test, here, n refers to the number of haplotypes tested for significantly associated with TB after Bonferroni correc- each gene. tion, including rs3809849 in MYBBP1A (Pcor = 0.0038) For simultaneous analysis of SNPs in the same gene, we and rs9061 in SP110 gene (Pcor = 0.019). Two more used the SNP-set Kernel Association Test (SKAT) to SNPs were found to be significantly associated with TB after evaluate the association between each gene and TB (Wu 10,000 permutations, including rs9905742 in MYBBP1A et al. 2010). This method uses a logistic machine model (Pcor = 0.049) and rs1135791 in SP110 gene (Pcor = that takes into account the joint effect of the SNPs in the 0.045) (Table 2). considered SNP-group. Individuals with missing genotypes The database for the meta-analysis included work from for a certain gene were not included in this gene-based five papers. Of these, four papers reported a possible analysis. association of the rs1135791 and rs3948464 SNPs with TB, The gene–gene interaction was tested using the logistic and three papers reported a possible association of rs9061 regression. The SNP with the lowest P value in a gene was with TB. Among these three SNPs, meta-analysis including selected as potential one to represent the gene for logistic the data from the current study confirmed that rs9061 was regression analysis. significantly associated with TB (P = 0.002, heterogeneity Single SNP tests were performed using Stata SE 10 P = 0.13) in East Asian populations (Supplementary (StataCorp LP, Texas, USA). Gene-based association tests Table 5). Moreover, no publication bias was found to exist

123 Hum Genet (2013) 132:265–273 269

Table 2 Allele frequencies of MYBBP1A, SP110 and RELA in patients with TB and control subjects in a Chinese Han population SNP Minor allele Minor allele frequency OR (95 % CI) P Pcor* Pcor** Controls TB

MYBBP1A rs12603519 A 0.287 0.265 0.90 (0.74–1.08) 0.25 1 0.97 rs3809849 C 0.158 0.223 1.53 (1.23–1.92) 0.0002 0.0038 0.0028 rs10852863 G 0.472 0.421 0.81 (0.69–0.97) 0.019 0.361 0.24 rs9907224 T 0.508 0.495 0.95 (0.80–1.13) 0.54 1 1.00 rs7211078 G 0.488 0.489 1.00 (0.84–1.18) 0.99 1 1.00 rs751670 C 0.418 0.401 0.93 (0.79–1.11) 0.44 1 1.00 rs9905742 A 0.035 0.064 1.87 (1.23–2.86) 0.003 0.057 0.049 rs1045845 G 0.453 0.419 0.87 (0.73–1.04) 0.12 1 0.81 SP110 rs11679983 A 0.039 0.030 0.79 (0.49–1.26) 0.32 1 0.97 rs11556887 T 0.071 0.105 1.55 (1.14–2.12) 0.006 0.114 0.08 rs9061 T 0.185 0.244 1.42 (1.15–1.76) 0.001 0.019 0.017 rs1365776 G 0.085 0.114 1.39 (1.04–1.86) 0.027 0.513 0.33 rs3948464 A 0.028 0.053 2.00 (1.24–3.22) 0.0036 0.0684 0.086 rs1135791 C 0.192 0.144 0.71 (0.57–0.89) 0.0028 0.0532 0.045 RELA rs7101916 T 0.362 0.358 0.98 (0.82–1.17) 0.84 1 1.00 rs11820062 T 0.391 0.405 1.06 (0.89–1.27) 0.49 1 1.00 rs2306365 A 0.382 0.368 0.94 (0.79–1.12) 0.50 1 1.00 rs11227247 G 0.382 0.368 0.94 (0.79–1.12) 0.48 1 1.00 rs7119750 T 0.382 0.368 0.94 (0.79–1.12) 0.54 1 1.00 *Pvalues were corrected based on Bonferroni method; ** P values were corrected based on 10,000 permutations

for rs9061 (Fig. 1). The pooled OR associated with the rs1135791 SNP (OR = 0.44, 95 % CI = 0.03–0.62, T allele for TB subjects was 1.24 (95 % CI = 1.08–1.43). P = 5E-6).

Haplotype analysis Gene-based association

Haplotypes with a frequency above 0.03 are listed As a complementary tool for SNP-based and haplotype- in Table 3. Haplotype association analysis showed that based analysis, gene-based analysis showed that MYBBP1A MYBBP1A and SP110 were associated with TB (P-glo- and SP110 were associated with TB (P = 0.0037 and bal = 4.17E-5 and 1.88E-8, respectively), while the 0.012, respectively; the Dunn–Sidak corrected P values RELA gene was not (P-global = 0.73). For MYBBP1A, were 0.011 and 0.035, respectively, Table 4). with the CTTGCAGG haplotype as a reference, both GTCTTGGG and GTCTTGCG haplotypes had significant Gene–gene interaction test protective effects on TB (OR = 0.47, 95 % CI = 0.35– 0.64 and OR = 0.46, 95 % CI = 0.24–0.89, respectively). The gene–gene interaction between SP110 and MYBBP1A Among the haplotypes studied, no associated haplotypes identified was tested using logistic regression, and it was were significantly affected by each SNP. For SP110, three found that interactions were significant for overall analysis haplotypes, TGATCG (P = 0.038), CGACCG (P = (P = 1.0E-15; Table 5). 5.00E-6) and TGATTG (P = 2.59E-4), were found to be significantly associated with TB. Among the haplotypes Protein 3D structure analysis studied, the associated haplotype TGATCG was signifi- cantly affected by the rs9061 SNP (OR = 1.65, 95 % Since no experimental structural data whatsoever are CI = 1.03–2.65, P = 0.038) and CGACCG by the available for SP110, we have to resort to structural

123 270 Hum Genet (2013) 132:265–273

Fig. 1 Meta-analysis of SP110 rs9061 C allele versus T allele with four studies. a The heterogeneity test and pooled OR with fixed model. b The funnel plot of four studies, X-axis: OR of rs9061 T allele, Y-axis: standard error of log OR

bioinformatics tools, such as homology modeling, to per- of the genetic associations reported by previous candidate form further and in-depth investigation. For the most pre- gene studies implies that advanced technologies have not cise modeling, the 3D structure of the human SP110 SAND been sufficiently used to study TB. Furthermore, genetic domain was constructed with the known 3D structure of effects on TB susceptibility are more complicated com- 1h5pA. Results showed that the structure containing the pared with non-communicable diseases because of the Met residue at the rs1135791 variant site was composed of complex effects of factors involved in MTB exposure, more a-helices and less b-sheets than the structure con- infection, and TB activation (Newport and Finan 2011). taining the Thr residue at this variant site (Supplementary Thus, new research strategies with low costs are necessary Fig. 1). to clarify the genes involved in the development of TB and its underlying molecular mechanisms in humans. To our knowledge, this is the first study demonstrating Discussion an association between TB and SP110-associated protein partners in humans, which have previously been identified The association between pulmonary tuberculosis and in mice (Cai et al. 2010). Mice have proven to be valuable SP110 variants has been investigated by several research- for aiding in understanding pathogenesis and dissecting ers, with conflicting results arising from their reports (Babb complex genetic backgrounds of various human traits (Apt et al. 2007; Cong et al. 2010; Liang et al. 2011; Png et al. and Kramnik 2009). In fact, many findings regarding the 2012; Szeszko et al. 2007). Use of genome-wide associa- response to TB in murine models have been subsequently tion studies (GWAS) is currently the favored approach to confirmed in humans (North and Jung 2004). SP110 has identifying variants involved in common diseases. Recent been found to limit MTB growth and induce apoptosis of data from two GWAS conducted in Gambian and Ghanaian infected cells (Pan et al. 2005). MYBBP1A has been subjects were combined to identify 17 SNPs with P \ 10-5 identified as a binding protein of SP110 in mice, which (Thye et al. 2011). Interestingly, none of these 17 loci helps SP110 induce apoptosis in macrophages (Cai et al. harbored any genes with known function that may partic- 2010). MYBBP1A has also been found to interact directly ipate in host resistance to MTB infection or TB activation. with RELA and repress NF-jB-dependent gene transcrip- Among these 17 loci, only one single SNP in a gene-poor- tion, including that of apoptosis factors (Owen et al. 2007). region of chromosome 18q11 which conferred a rather Therefore, MYBBP1A and RELA, along with SP110, may modest OR of 1.19 was replicated in samples from Ghana. have important roles in the pathogenesis of TB and may be The fact that GWAS studies cannot replicate the majority functional candidates for susceptibility to TB in humans.

123 Hum Genet (2013) 132:265–273 271

Table 3 Distribution of haplotypes (frequency [0.03) in TB patients and control subjects Haplotype TB (freq) con (freq) OR (95 % CI)* P value

MYBBP1A rs1045845-rs9905742-rs751670-rs7211078-rs9907224-rs10852863-rs3809849-rs12603519 CTTGCAGG 0.292 0.269 Ref GTCTTGGG 0.108 0.19 0.47 (0.35–0.64) 2.00E-06 GTCTTGGA 0.045 0.067 0.63 (0.39–1.03) 0.065 CTTGCACA 0.02 0.039 0.48 (0.22–1.02) 0.056 CTTTTGGG 0.032 0.038 0.68 (0.42–1.09) 0.109 GTCTTGCG 0.02 0.035 0.46 (0.24–0.89) 0.02 CTTTCAGG 0.02 0.032 0.61 (0.37–1.02) 0.06 Global test 1398 846 v2 = 29.86 with 6 df, P = 4.17E-5 SP110 rs1135791-rs3948464-rs1365776-rs9061-rs11556887-rs11679983 TGACCG 0.636 0.687 Ref CGACCG 0.028 0.073 0.44 (0.30–0.62) 5.00E-06 TGATCG 0.043 0.029 1.65 (1.03–2.65) 0.038 CGATCG 0.036 0.03 1.01 (0.58–1.78) 0.95 CGGTCG 0.039 0.047 0.78 (0.49–1.27) 0.33 TGATTG 0.052 0.018 3.52 (1.79–6.92) 2.59E-04 Global test 1404 850 v2 = 44.45 with 5 df, P = 1.88E-8 RELA rs7119750-rs11227247-rs2306365-rs11820062-rs7101916 CTGTC 0.392 0.387 Ref TGACT 0.317 0.329 0.94 (0.76–1.15) 0.536 CTGCC 0.202 0.198 0.78 (0.79–1.27) 0.984 TGACC 0.037 0.049 0.76 (0.50–1.21) 0.265 CTGCT 0.0319 0.0316 0.98 (0.57–1.69) 0.955 Global test 1404 850 v2 = 2.05 with 4 df, P = 0.73 df degree of freedom * Adjusted for age and gender, P values were calculated by logistic regress

Table 4 Gene-based association analysis Table 5 Gene–gene interactions between MYBBP1A and SP110

Gene Number of SNPs P Pcor* MYBBP1A SP110 Controls (%) TB (%) Overall

MYBBP1A 8 0.0037 0.011 OR (95 % CI)* P SP110 6 0.012 0.035 GG CC 263 (61.88) 294 (41.7) 1.00 (Ref) RELA 5 0.68 1.00 GG CT/TT 38 (8.94) 135 (19.15) 3.5 (2.33–5.27) 3.66E-09 Total 19 GC/CC CC 14 (3.29) 103 (14.61) 2.69 (2.00–3.61) 1.74E-12 GC/CC CT/TT 110 (25.88) 173 (24.54) 1.14 (1.02–1.26) 0.012 * Dunn–Sidak-corrrected P values (3 tests) Test for interaction 1.00E-15

* Adjusted for age and gender In the present study, we used multiple analyses to identify an association between SP110/MYBBP1A and Chinese Han population (Cong et al. 2010; Liang et al. pulmonary TB in the Chinese Han population. Two SNPs 2011). Cong et al. observed no association of either alleles in SP110 (rs9061 and rs1135791) and two SNPs in or genotypes at rs9061 with TB. However, this may be MYBBP1A (rs3809849 and rs9905742) were found to be explained by the fact that only 100 cases and 100 controls associated with TB after correction using 10,000 permu- were used in this study, which may reduce the power to tation testing for multiple testing. However, using more detect association. Although Liang et al. did not observe an conservative Bonferroni correction, only rs9061 and association between alleles and TB, there was a significant rs3809849 were found to be associated with TB. Meta- association with genotype at rs9061 (P = 6.63 9 10-5). analysis of the SNPs in the SP110 gene also supported that Our genotype analysis supported this finding. rs9061 was associated with TB. During meta-analysis, it In addition, the SP110 rs1135791 variant is located was noted that two association studies performed in the within the functional SAND domain that is involved in

123 272 Hum Genet (2013) 132:265–273 transcriptional regulation through DNA binding (Bottom- this study, especially to Prof. Krzysztof Trzcin´ski and Dr. Daniel M. ley et al. 2001). It was found to drive haplotypes associa- Czajkowsky for their warm-hearted help and critical comments on the manuscript. This work was supported by the Chinese National Sci- tion. Homology modeling of the 3D structure of the SP110 ence Foundation (No. 31101015), the National Infectious Disease SAND domain showed that substituting the hydrophobic Prevention and Cure Special Project (No. 2008ZX10003-012) and the Met amino acid to hydrophilic Thr at the rs1135791 variant Scientific Research Foundation for the Returned Overseas Chinese site significantly altered the 3D structure of the SAND Scholars, State Education Ministry, China. domain, which indicated that this mutant might affect the Conflict of interest All authors have no conflicts to report. ability of SP110 to function by binding with DNA and regulating gene transcriptions. Both haplotype-based and gene-based analysis confirmed that both MYBBP1A and SP110 were associated with TB. The logistic regression analysis results suggested inter- References actions potentially exist between SP110 and MYBBP1A. A Apt A, Kramnik I (2009) Man and mouse TB: contradictions and combination of the CC genotype of rs9061 and GC/CC solutions. Tuberculosis 89:195–198 genotype of rs3809849 was found to be markedly and Babb C, Keet EH, van Helden PD, Hoal EG (2007) SP110 significantly associated with TB (P = 1.74E-12), sug- polymorphisms are not associated with pulmonary tuberculosis gesting that it might be useful as a predictive marker for in a South African population. Hum Genet 121:521–522 Bellamy R (2003) Susceptibility to mycobacterial infections: the identifying the Chinese Han population at high risk of importance of host genetics. Genes Immun 4:4–11 developing TB. Berlin JA, Laird NM, Sacks HS, Chalmers TC (1989) A comparison Compared with previous association studies examining of statistical methods for combining event rates from clinical SP110, the minor allele frequency of SNP rs9061 in the trials. Stat Med 8:141–151 Bloch DB, Nakajima A, Gulick T, Chiche JD, Orth D, de la Monte SM, Chinese Han population was lower than that in the West Bloch KD (2000) Sp110 localizes to the PML-Sp100 nuclear Africa population (91.5 vs. 95.5 %) (Tosh et al. 2006). body and may function as a nuclear hormone receptor transcrip- Furthermore, in our study the rs9061 genotype was sig- tional coactivator. Molecular Cellular Biology 20:6138–6146 nificantly associated with TB; however, in the Gambian Bonizzi G, Karin M (2004) The two NF-kappaB activation pathways and their role in innate and adaptive immunity. Trends Immunol population no association of this SNP with TB was 25:280–288 observed. These results confirm differences in the fre- Bottomley MJ, Collard MW, Huggenvik JI, Liu Z, Gibson TJ, Sattler quencies of some genetic variations among different ethnic M (2001) The SAND domain structure defines a novel DNA- groups and geographic regions. At the same time, the use binding fold in transcriptional regulation. Nat Struct Biol 8:626–633 of gene-based analysis as a complementary tool is sug- Cai L, Zhang D, Shi Y, Oaks MK, Chen L, Jin J, Qian J, Du F, Li N, gested to replicate the present study results in different Han X, Ren D (2005) Association of the CTLA-4 gene with ethnic background. This is due to three main reasons: rheumatoid arthritis in Chinese Han population. Eur J Hum firstly, a lot of single SNPs have only mild effects; sec- Genetics 13:823–828 Cai L, Pan H, Trzcinski K, Thompson CM, Wu Q, Kramnik I (2010) ondly, the heterogeneity that alleles at different loci cause MYBBP1A: a new Ipr10s binding protein in mice. Mol Biol Rep diseases in different populations reduces the chances of the 37:3863–3868 association of a single marker being replicate; thirdly, a Chan ED, Morris KR, Belisle JT, Hill P, Remigio LK, Brennan PJ, gene consists of a group of SNPs acting in concert to Riches DW (2001) Induction of inducible nitric oxide synthase- NO* by lipoarabinomannan of Mycobacterium tuberculosis is perform specific biological tasks. Thus, replication of mediated by MEK1-ERK, MKK7-JNK, and NF-kappaB signal- associations found at the gene level or even at the pathway ing pathways. Infect Immun 69:2001–2010 level is more likely to occur compared to replication at the Cong J, Li G, Zhou D, Tao Y, Xiong Y (2010) Study on relation SNP level. between Sp110 gene polymorphism and tuberculosis genetic susceptibility of Chongqing Han people. J Hyg Res 39:540–544 In conclusion, MYBBP1A and SP110 may be potentially Cooke GS, Hill AV (2001) Genetics of susceptibility to human gene candidates for determining the risk of developing infectious disease. Nat Rev Genet 2:967–977 pulmonary TB, and a combination of SP110 and DerSimonian R, Kacker R (2007) Random-effects model for meta- MYBBP1A polymorphisms may serve as a novel predictive analysis of clinical trials: an update. Contemp Clin Trials 28:105–114 marker for the risk of developing TB in the Chinese Han Emily M, Mailund T, Hein J, Schauser L, Schierup MH (2009) Using population. Further functional studies in humans are biological networks to search for interacting loci in genome- required to validate the predictive value of this novel wide association studies. Eur J Hum Genet 17:1231–1240 biomarker and to elucidate the effects of these two genes Higgins JP, Thompson SG (2002) Quantifying heterogeneity in a meta-analysis. Stat Med 21:1539–1558 on the development and outcome of TB. Liang L, Zhao YL, Yue J, Liu JF, Han M, Wang H, Xiao H (2011) Association of SP110 gene polymorphisms with susceptibility to Acknowledgments We thank three anonymous reviewers for their tuberculosis in a Chinese population. Infect Genet Evol useful comments. And we are grateful to all subjects participating in 11:934–939

123 Hum Genet (2013) 132:265–273 273

Neale BM, Sham PC (2004) The future of association studies: gene- Warndorff DK, Sichali L, Malema S, Crampin AC, Ngwira B, based analysis and replication. Am J Hum Genet 75:353–362 Teo YY, Small K, Rockett K, Kwiatkowski D, Fine PE, Hill PC, Newport MJ, Finan C (2011) Genome-wide association studies and Newport M, Lienhardt C, Adegbola RA, Corrah T, Ziegler A, susceptibility to infectious diseases. Brief Funct Genomics Morris AP, Meyer CG, Horstmann RD, Hill AV (2011) Genome- 10:98–107 wide association analyses identifies a susceptibility locus for North RJ, Jung YJ (2004) Immunity to tuberculosis. Annu Rev tuberculosis on chromosome 18q11.2. Nat Genet 42:739–741 Immunol 22:599–623 Tosh K, Campbell SJ, Fielding K, Sillah J, Bah B, Gustafson P, Owen HR, Elser M, Cheung E, Gersbach M, Kraus WL, Hottiger MO Manneh K, Lisse I, Sirugo G, Bennett S, Aaby P, McAdam K, (2007) MYBBP1a is a novel repressor of NF-kappa B. J Mol Bah-Sow O, Lienhardt C, Kramnik I, Hill AVS (2006) Variants Biol 366:725–736 in the SP110 gene are associated with genetic susceptibility to Pahl HL (1999) Activators and target genes of Rel/NF-kappaB tuberculosis in West Africa. Proc Natl Acad Sci USA transcription factors. Oncogene 18:6853–6866 103:10364–10368 Pan H, Yan BS, Rojas M, Shebzukhov YV, Zhou H, Kobzik L, Tregouet DA, Garelle V (2007) A new JAVA interface implemen- Higgins DE, Daly MJ, Bloom BR, Kramnik I (2005) Ipr1 gene tation of THESIAS: testing haplotype effects in association mediates innate immunity to tuberculosis. Nature 434:767–772 studies. Bioinformatics 23:1038–1039 Png E, Alisjahbana B, Sahiratmadja E, Marzuki S, Nelwan R, Adnan Tregouet DA, Escolano S, Tiret L, Mallet A, Golmard JL (2004) A I, van de Vosse E, Hibberd M, van Crevel R, Ottenhoff TH, new algorithm for haplotype-based association analysis: the Seielstad M (2012) Polymorphisms in SP110 are not associated Stochastic-EM algorithm. Ann Hum Genet 68:165–177 with pulmonary tuberculosis in Indonesians. Infect Genetics WHO (2011) Global tuberculosis control: WHO report 2011. World Evol Health Organization, Avenue Purcell S, Cherny SS, Sham PC (2003) Genetic Power Calculator: Wu MC, Kraft P, Epstein MP, Taylor DM, Chanock SJ, Hunter DJ, design of linkage and association genetic mapping studies of Lin X (2010) Powerful SNP-set analysis for case-control complex traits. Bioinformatics 19:149–150 genome-wide association studies. Am J Hum Genet 86:929–942 Szeszko JS, Healy B, Stevens H, Balabanova Y, Drobniewski F, Todd Yu W, Gwinn M, Clyne M, Yesupriya A, Khoury MJ (2008) A JA, Nejentsev S (2007) Resequencing and association analysis of navigator for human genome epidemiology. Nat Genet the SP110 gene in adult pulmonary tuberculosis. Hum Genet 40:124–125 121:155–160 Zhang Y (2009) I-TASSER: fully automated protein structure Thye T, Vannberg FO, Wong SH, Owusu-Dabo E, Osei I, Gyapong J, prediction in CASP8. Proteins 77(Suppl 9):100–113 Sirugo G, Sisay-Joof F, Enimil A, Chinbuah MA, Floyd S,

123