ORIGINAL INVESTIGATION Linkage and association analysis of candidate genes for TB and TNFa cytokine expression: evidence for association with IFNGR1, IL-10, and TNF receptor 1 genes Catherine M. Stein Æ Sarah Zalwango Æ Allan B. Chiunda Æ Christopher Millard Æ Dmitry V. Leontiev Æ Amanda L. Horvath Æ Kevin C. Cartier Æ Keith Chervenak Æ W. Henry Boom Æ Robert C. Elston Æ Roy D. Mugerwa Æ Christopher C. Whalen Æ Sudha K. Iyengar Received: 22 December 2006 / Accepted: 16 March 2007 / Published online: 13 April 2007 � Springer-Verlag 2007 Abstract Tuberculosis (TB) is a growing public health threat globally and several studies suggest a role of host genetic susceptibility in increased TB risk. As part of a household contact study in Kampala, Uganda, we have taken a unique approach to the study of genetic suscepti- bility to TB by developing an intermediate phenotype model for TB susceptibility, analyzing levels of tumor necrosis factor-a (TNFa) in response to culture filtrate as the phenotype. In the present study, we analyzed candidate genes related to TNFa regulation and found that interleukin (IL)-10, interferon-gamma receptor 1 (IFNGR1), and TNFa receptor 1 (TNFR1) genes were linked and associated to both TB and TNFa. We also show that these associations are with progression to active disease and not susceptibility to latent infection. This is the first report of an association between TB and TNFR1 in a human population and our findings for IL-10 and IFNGR1 replicate previous findings. By observing pleiotropic effects on both phenotypes, we show construct validity of our intermediate phenotype model, which enables the characterization of the role of these genetic polymorphisms on TB pathogenesis. This study further illustrates the utility of such a model for disentangling complex traits. Introduction Tuberculosis (TB) is a growing public health problem globally; approximately one-third of the world’s popula- tion is infected with the causal bacterium, Mycobacterium tuberculosis (Mtb), and the incidence of TB disease is increasing in the face of the HIV pandemic (Raviglione et al. 1995). TB is a complex trait for several reasons. Phenotype definition is not trivial because TB disease may be expressed with varying severity in a number of organ systems after a long and variable latency period. Both host and environmental factors affect the risk of infection by Mtb following exposure and progression to TB disease, and few studies have examined whether genetic suscep- tibility to these two stages in disease progression differ (Flores-Villanueva et al. 2005). Previous genetic studies C. C. Whalen and S. K. Iyengar contributed equally as senior authors of this work. C. M. Stein � A. B. Chiunda � C. Millard � D. V. Leontiev � A. L. Horvath � K. C. Cartier � R. C. Elston � S. K. Iyengar (&) Department of Epidemiology and Biostatistics, Case Western Reserve University, Wolstein Research Building Room 1303, 2103 Cornell Rd, Cleveland, OH 44106, USA e-mail: ski@cwru.edu C. C. Whalen Department of Epidemiology and Biostatistics, Case Western Reserve University, 10900 Euclid Avenue, Cleveland, OH 44106, USA C. M. Stein � S. Zalwango � A. B. Chiunda � K. Chervenak � W. H. Boom � R. D. Mugerwa � C. C. Whalen Tuberculosis Research Unit, Case Western Reserve University, Cleveland, OH, USA C. M. Stein � C. C. Whalen Center for Modern Epidemiology of Infectious Diseases, Case Western Reserve University, Cleveland, OH, USA S. Zalwango � R. D. Mugerwa Clinical Epidemiology Unit, Makerere University School of Medicine, Kampala, Uganda 123 Hum Genet (2007) 121:663–673 DOI 10.1007/s00439-007-0357-8 have also differed dramatically in TB diagnostic criteria and characterization of controls. Very few of these studies have been able to establish a link between candidate genes and TB pathogenesis. Furthermore, few studies have examined the linkage disequilibrium (LD) structure within these genes, and thus are unable to account for untyped polymorphisms that may influence TB risk. To address the complexities of TB, we have examined expression of tumor necrosis factor-a protein (TNFa) in response to mycobacterial antigens as an intermediate phenotype for TB disease. We chose TNFa as an inter- mediate phenotype because it is a central cytokine in TB pathogenesis that is involved in granuloma formation, in- duces symptoms including fever and weight loss (Barnes et al. 1990; Roach et al. 2002), and is important in the containment of latent Mtb infection (Keane et al. 2001). In our preliminary work, we have found that TNFa expression following antigen stimulation has a high heritability that is partially attributable to a complicated major gene effect and potentially influenced by gene-environment interaction (Stein et al. 2003, 2005). Based on our intermediate phenotype model, we iden- tified 12 candidate genes that may have roles in TNFa regulation and expression or have been associated with TB susceptibility in either human studies or mouse models (Table 1). The TNFa gene (TNFA) resides in the MHC class III region on chromosome 6 and is in high LD with MHC Class I and Class II genes, which code for the various HLA subtypes. TNFA knockout mice die from infection by Mtb (Flynn et al. 1995). TNFA promoter polymorphisms have been shown to be associated with TNFa levels (Pociot et al. 1993; Wilson et al. 1997). A number of HLA poly- morphisms have been associated with TB as well (Bellamy 2003). There are two receptors for TNFa, coded by TNFR1 (also referred to as TNFRSF1A) and TNFR2, and efficient binding of TNFa depends on the self-assembly of these two receptors. Both receptors have been shown to influence TNFa levels (Peschon et al. 1998), and TNFR1 and TNFR2 knockout mice succumb to Mtb infection and Bacille Calmette-Guérin (BCG) infection (Peschon et al. 1998; Piguet et al. 2002). Toll-like receptors (TLRs) also play a role in TB. Mtb bacilli contain distinct ligands that activate cells via TLR2 and TLR4 (Means et al. 1999), which results in TNFa production by macrophages (Underhill et al. 1999). TLR2 deficient mice produce less TNFa and lack functional granulomas (Bochud et al. 2003; Drennan et al. 2004). It has been shown that Mtb activates cells via TLR2 and TLR4 (Means et al. 1999) and that Mtb-induced TNFa production is differentially affected by a TLR4-specific antagonist (Means et al. 2001). The first TB susceptibility locus mapped in mice was Nramp1, which has since been renamed Slc11a1 (Blackwell et al. 2004; Skamene et al. 1998); the human ortholog of this gene (originally named NRAMP1, now named SLC11A1) has been studied in association with TB in humans in a number of populations (Bellamy 2003; Li et al. 2006). In mice, Slc11a1 has pleiotropic effects, including an influence on TNFa release by macrophages (Blackwell et al. 2004; Formica et al. 1994). Interleukin (IL)-10 is an inhibitor of Th-1 cell and macrophage functions, and drug and molecular studies suggest that IL-10 activity leads to decreased TNFa pro- duction (Balcewicz-Sablinska et al. 1999; Goldman et al. Table 1 Microsatellite markers analyzed by candidate gene Candidate gene Gene (Mb location from pter) Markers Mb location (Mb) Tumor necrosis factor-a and human leukocyte antigen TNFA/HLA (31.6) D6S306 28.0 D6S265 30.1 D6S273 31.8 D6S439 35.2 TNFB 44.2 D6S1701 47.7 Tumor necrosis factor-a receptor type-1 TNFR1 (6.3) D12S221 4.9 D12S825 5.9 D12S1625 7.0 Tumor necrosis factor-a receptor type-2 TNFR2 (11.9) D1S244 10.3 D1S489 11.7 D1S228 13.3 Interleukin-10 IL-10 (204.0) D1S256 202.0 D1S177 204.0 D1S2692 205.1 Natural resistance-associated macrophage protein 1 SLC11A1 (219.4) D2S301 218.1 D2S2179 219.0 D2S2359 221.1 Interleukin-12 sub-unit A IL12A (161.0) D3S1553 160.6 D3S3580 160.9 D3S3708 163.4 D3S1607 158.3 Interleukin-12 sub-unit B IL12B (158.7) D5S487 155.6 D5S1971 158.5 D5S2047 160.5 Interferon-c receptor 1 IFNGR1 (137.5) D6S270 134.6 D6S1587 138.4 D6S1675 140.0 Interferon-c receptor 2 IFNGR2 (33.7) D21S223 32.2 D21S2039 33.6 D21S65 35.0 Toll-Like receptor 2 TLR2 (155.2) D4S233 154.6 D4S3049 155.3 D4S2976 156.4 Toll-like receptor 4 TLR4 (115.8) D9S154 114.7 D9S1864 115.8 D9S275 116.9 664 Hum Genet (2007) 121:663–673 123 1996). IL-12, which is composed of two subunits coded for by IL12A and IL12B, prompts the release of TNFa by T cells (van Crevel et al. 2002). Finally, macrophages may have decreased function due to deficient receptors. Inter- feron-c (IFNc), a key cytokine in the response to Mtb, is released by T-cells; its receptors, IFNGR1 and IFNGR2, are on macrophages. In this way, IFNc may influence TNFa production (Knight and Kwiatkowski 1999). IL-10, IL-12B, and IFNGR1 have been studied in human popu- lations, but with mixed results (Bellamy 2003). In this analysis, we took a two-stage approach to the analysis of these candidate genes. First, we conducted a positional candidate linkage analysis in each of these can- didate genes. After detecting nominally significant linkage, we followed-up those genes of interest by conducting a family-based association analysis. In the association stage, we analyzed single nucleotide polymorphisms (SNPs) that characterized the LD structure of the genes of interest. This study provides additional validity to our intermediate phe- notype model by demonstrating pleiotropic effects of TB susceptibility genes. Additionally, this analysis illustrates the utility of the intermediate phenotype approach to dem- onstrate construct validity and elucidate the impact of risk alleles on the pathogenesis of a complex disease. Materials and methods Subject recruitment and assessment The families presented in this analysis are part of an ongoing household contact study (Guwattude et al. 2003) in Kawempe Division of Kampala, Uganda. Index cases and household contacts were enrolled between April 2002 and July 2003 [Phase II of the study (Stein et al. 2005)]. Index case patients were diagnosed with TB disease if their sputum contained Mtb on Ziehl–Nielsen stain or myco- bacterial culture. These index cases were referred to the study through the Uganda National Tuberculosis and Leprosy Programme, public and private clinics in Kaw- empe Division, or by self-referral after community sensi- tization efforts or word of mouth from friends currently enrolled in the study. To be included in the study, both the index case and household members provided informed consent. The institutional review boards at University Hospitals of Cleveland and the Uganda Council for Science and Technology approved the study. Upon enrollment, the index case and all household members underwent a clinical examination, including a health and symptom survey, tuberculin skin test using purified protein derivative and the Mantoux method (PPD, 5 TU, Tubersol; Connaught Laboratories, Limited, Toronto, Canada), and HIV testing using traditional ELISA methods (Cambridge BioScience). Individuals who had suspected TB also had chest radiography, sputum micros- copy and cultures performed. Patients with active TB dis- ease were given the recommended course of therapy (Blumberg et al. 2003). In addition, household members who did not have active disease but demonstrated latent Mtb infection by the tuberculin skin test were given iso- niazid preventive therapy. In this paper, ‘‘TB’’ refers to active pulmonary disease, and ‘‘Mtb infection’’ refers to latent infection that has been contained and has not pro- gressed to active disease. TNFa protein production in re- sponse to Mtb culture filtrate, the intermediate phenotype, was measured using the whole blood cytokine assay. Blood cells were stimulated with Mtb culture filtrate obtained from Colorado State University. After 18 h of incubation, TNFa concentrations in the supernatants were measured using ELISA (Endogen) (Stein et al. 2005). Molecular methods DNA was extracted from buffy coats using a salting pro- cedure (Miller et al. 1988) and consequently quantified using PicoGreen (Invitrogen) and standardized to 10 ng/ lL. Quality of DNA was validated using the amelogenin marker. DNA was arrayed in a 96-well format, and PCR was performed in a MJ Tetrad thermocycler. Three microsatellite markers were typed per candidate gene region, with the exception of the HLA/TNFA region, which was covered with 6 microsatellite markers (Table 1) that were chosen to have high heterozygosity (>70%) and mapped within 2 Mb of the gene of interest. Genetic map locations of these markers were based on the most recent map (Kong et al. 2004); when markers were not included on this map, their locations were obtained by linear inter- polation on physical map locations on the UCSC Genome Browser (http://www.genome.ucsc.edu/). The markers were multiplexed and run on an ABI 3700 capillary ma- chine (Applied Biosystems). Internal controls were in- cluded on each gel, consisting of two CEPH controls and four replicate samples. The ABI ROX 500 standard (present in every lane) was used to estimate size of alleles. Familial relationships were verified or reclassified using all microsatellite data based on likelihood methods imple- mented in RELPAIR (Epstein et al. 2000); a total of 20 relationships were reclassified. Inconsistencies in the seg- regation of the genotypes within families were examined using MARKERINFO (S.A.G.E. 2006). Individuals dem- onstrating Mendelian inconsistencies at multiple markers that could not be resolved by retyping or reclassifying familial relationship were treated as missing for the pur- pose of this analysis. In total, 5% of the microsatellite data were treated as missing, resulting in a loss of 6% of the sibpairs. Marker allele frequencies conformed to Hardy– Hum Genet (2007) 121:663–673 665 123 Weinberg expectations. We estimated the allele frequen- cies for each genetic marker by simple gene counting (disregarding relationships), using FREQ (S.A.G.E. 2006). To follow-up candidate genes that demonstrated nomi- nal evidence for linkage to both traits (at the a = 0.05 le- vel), we genotyped SNPs within these genes (Table 2). Freely-available databases were used to select SNPs, including HapMap Phases I and II with Haploview soft- ware (Altshuler et al. 2005; Barrett et al. 2005), dbSNP (http://www.ncbi.nlm.nih.gov/SNP/index.html), and SNP- Browser by Applied Biosystems. SNP assays were avail- able as Assays-on-Demand or Assays-by-Design (Applied Biosystems). Haplotype-tagging SNPs were selected using an r2 threshold of 0.90, restricted to those with a minor allele frequency (MAF) of at least 10% in African or African–American populations and that have been vali- dated by at least two sources. If a haplotype-tagging SNP did not meet the MAF criterion, a nearby SNP within the same LD bin (within 5 kb) was identified to replace it. SNPs were genotyped on the TaqMan platform, and data cleaning was conducted as described above, resulting in 2.95% of genotypes being deleted due to Mendelian inconsistencies or low quality scores. SNP genotypes were tested for Hardy–Weinberg proportions using PEDSTATS (Wigginton and Abecasis 2005) in a random sample of unrelated individuals, separately in individuals affected and unaffected for TB. Statistical analysis Because of the multifactorial nature of TB and TNFa, we used two complementary model-free linkage analysis strategies to analyze the microsatellite data. Both single- point and multipoint analyses were conducted; for brevity, only multipoint analysis results are shown. For each rela- tive pair, the proportion of alleles shared identical by descent (IBD) was estimated using the GENIBD program (S.A.G.E. 2006). First, both the quantitative trait TNFa and the binary trait TB (presence of active disease), were analyzed using Haseman–Elston regression (Haseman and Elston 1972), in which a measure of sibling trait similarity is regressed on the proportion of alleles shared IBD by that sibpair. There is a variety of ways to parameterize the phenotypic similarity between sibs to account for the non- independence of siblings within a sibship; we used the W4 option in SIBPAL (S.A.G.E. 2006), as it is asymptotically the most powerful (Shete et al. 2003). Recently, the Haseman–Elston method has been extended to include both full-sibs as well as half-sib pairs (S.A.G.E. 2006). Sec- ondly, the conditional logistic model (Goddard et al. 2001; Olson 1999) was used to further analyze the binary trait TB, as is implemented in LODPAL (S.A.G.E. 2006). This affected sibpair (ASP) analysis model compares the ob- served IBD sharing to Mendelian expectation under the null hypothesis of no linkage. Unaffected individuals can be incorporated in the analysis by including an indicator variable covariate for discordant pair status; when the model uses this covariate, the LOD score is asymptotically distributed as 1 2 v2 1 þ 1 2 v2 2 (Goddard et al. 2001). To adjust for the possible confounding effects, HIV serostatus was used as a covariate in the Haseman–Elston models, and conditional logistic regression models were run on the subset of HIV concordantly seronegative relative pairs. Since this was a focused candidate gene linkage study, we did not apply genome-wide significance thresholds (Lander and Kruglyak 1995; Witte et al. 1996), but instead fol- lowed-up loci significant at a = 0.05 with SNP association analysis. After finding nominal evidence for linkage to IL-10, TNFR1, and IFNGR1, association analysis of SNPs within these genes was conducted. Again, both TB and TNFa were examined as phenotypes of interest. In these data, the traditional transmission-disequilibrium test (TDT) is not appropriate because most of our affected individuals are parents, not offspring. Instead, we used the family-based test of association between a marker and continuous phe- notype developed by George and Elston (1987), which allows for familial correlations by simultaneously esti- mating residual and multifactorial (polygenic, familial, and marital) variance components (Elston et al. 1992). This model has since been extended to allow for binary traits by incorporating a logit link function (Gray-McGuire 2004). This regression framework has the flexibility to allow families of any size or structure, and is implemented in ASSOC (S.A.G.E. 2006), which tests the significance of covariate coefficients using both the likelihood ratio test and the Wald test. Within this regression model, alleles are coded according to three to three possible modes of inheritance (recessive, additive, and dominant). Incorpo- Table 2 Single nucleotide polymorphisms (SNPs) analyzed by can- didate gene Candidate gene SNP name Minor allele Minor allele frequency (MAF) IFNGR1 rs4896243 C 0.245 rs1327474 A 0.018 rs2234711 C 0.405 IL-10 rs1518111 A 0.406 rs1554286 T 0.415 rs1800872 A 0.418 TNFR1 rs4149623 T 0.392 rs4149639 G 0.208 rs4149622 A 0.398 rs4149578 A 0.282 666 Hum Genet (2007) 121:663–673 123 ration of covariates other than locus-specific effects was done in a stepwise fashion; since the effect sizes for the SNPs were consistent across models, only the models with both a marker and significant covariates are presented. Covariates included HIV status and a composite environ- mental variable that was derived to depict components of shared environment and modeled nutritional status and shared environment—frequency of contact with the index TB case, clinical characteristics of the index case, poor ventilation within the home and poverty (Stein et al. 2005). Because of the possible confounding effects of HIV infection on the association between these phenotypes and loci, we also evaluated the effects of interaction between a locus and HIV within the association models. To examine whether the associations between SNPs and TB disease actually reflected susceptibility to latent infection, we constructed a variable to contrast these two stages of disease. For this analysis, we considered latent Mtb infection to be evidenced by a TST reading of 10 mm or greater but no clinical or microbiological evidence of TB disease (CDC 2003). The variable was coded as 1 = active TB disease and 0 = latent Mtb infection; indi- viduals without TB disease or latent Mtb infection were treated as missing. Haplotypes of SNPs within genes were estimated using the expectation-maximization algorithm as implemented in DECIPHER (S.A.G.E. 2006). When phase could not be resolved with 100% certainty, the phased genotypes with the highest posterior probability was selected. The most likely haplotypes were coded as indicator variables in the data; only haplotypes with greater than 10% sample fre- quency were considered. Haplotypes were individually included as covariates within the ASSOC models to ascertain the locus-specific effects. Finally, to evaluate whether a SNP accounted for the observed linkage effects, we included the SNPs as cova- riates within the Haseman–Elston regression, coding the genotypes according to the most significant genetic model. Results This analysis included 398 individuals, comprising 66 pedigrees, 232 full sibling pairs, and 157 half sibling pairs. Of these individuals, 15.0% had active TB disease, 16.4% were HIV seropositive, and 74.3% had latent Mtb infection based on the tuberculin skin test. After elimination of individuals with Mendelian inconsistencies, there were 213 full sibling pairs and 120 half sibling pairs available for linkage analysis; all of these individuals had complete data for TB and TNFa. We analyzed linkage between our two phenotypes, TNFa and TB, and microsatellite markers that map to our candidate genes of interest (Table 1). Using the Haseman– Elston regression model and adjusting for HIV status, we found that TNFR1 and IFNGR1 were linked to both TNFa and TB (Table 3). The most significant marker for TNFR1 was D12S1625 for both TNFa (p = 0.0184) and TB (p = 9 · 10–7), and the most significant marker for IFNGR1 was D6S270 for both traits (TNFa p = 0.006 and Table 3 Summary of significant linkage results NS not significant at a = 0.05 Gene Marker Haseman–Elston p-values Conditional logistic p-values, TB TNFa TB All sibs HIV negative only TNFR1 D12S221 0.0202 2.30 · 10–6 NS NS D12S374 0.0401 2.06 · 10–5 NS NS D12S1625 0.0184 9.00 · 10–7 NS NS IL-10 D1S236 0.0321 NS 0.029 NS D1S177 0.0788 NS NS NS D1S2692 0.0023 NS 0.021 NS IFNGR1 D6S270 0.006 2.82 · 10–5 0.004 0.009 D6S1587 0.061 2.15 · 10–4 0.003 0.002 D6S1675 0.049 2.84 · 10–4 0.012 0.045 TNFR2 D1S244 NS 0.0053 NS NS D1S489 NS 0.0034 0.015 NS D1S228 NS 5.80 · 10–4 0.008 NS TLR4 D9S154 NS 1.60 · 10–4 0.012 NS D9S1864 NS 3.70 · 10–6 0.030 NS D9S1675 NS 0.0023 NS NS TLR2 D4S233 NS NS 0.016 NS D4S3049 NS NS 0.014 NS D4S2976 NS NS 0.017 NS Hum Genet (2007) 121:663–673 667 123 TB p = 2.82 · 10–5). In addition, the TNFR2 and TLR4 genes also demonstrated significant linkage to TB only, and IL-10 demonstrated significant linkage to TNFa. The conditional logistic analysis results confirmed many of the Haseman–Elston analysis results. When including all relative pairs, significant linkage between TB disease and microsatellite markers at the a = 0.05 threshold was at- tained for IL-10, IFNGR1, TNFR2, TLR4, and TLR2 (Ta- ble 3). When the analysis of the TB phenotype was repeated for HIV negative pairs only, the statistical sig- nificance diminished for all genes except IFNGR1 which remained significant. The decline in significance may be attributed to a substantial drop in sample size (entire sample included 71 relative pairs, compared to 47 con- cordantly HIV negative pairs, for this analysis). Because IFNGR1, IL-10, and TNFR1 were linked to both TNFa and TB status, we genotyped tagSNPs in these genes (Table 2) and analyzed them using family-based association analysis. All of these SNPs were in Hardy– Weinberg equilibrium in cases and controls (p > 0.10, data not shown). Using the model that allowed for multifactorial correlations, we evaluated association between both traits and all the SNPs, where genotype coding schemes were used for recessive, dominant, and additive models; the referent allele for each gene is noted (Tables 4, 5). In addition, we included as covariates HIV status and an environmental variable that we developed in a path anal- ysis (Stein et al. 2005). In our analysis of TNFa as the phenotype, we found association with SNPs in all three genes (Table 4), with several highly significant p-values, especially in the IL-10 and TNFR1 genes. These results were significant even after applying a conservative Bon- ferroni correction (a* = 0.05/10 = 0.005). When analyzing TB as the trait, we found association with TNFR1 and IFNGR1, but not IL-10 (Table 5). Though less statistically significant than the results for TNFa, we still observed multiple SNPs within both genes demonstrating signifi- cance. All but one of these SNPs were significant using the Table 5 Results of association analysis of TB as the phenotype Gene Marker Referent allele Best fitting model b coef Genotype p-value* HIV · locus interaction§ IFNGR1 rs4896243 C Additive 0.467 0.002 NS rs1327474 G Dominant –1.312 0.243 NS rs2234711 T Dominant 0.677 0.002 0.067 IL-10 rs1518111 A Dominant –0.163 0.376 NS rs1554286 C Recessive 0.199 0.276 NS rs1800872 A Dominant –0.196 0.270 NS TNFR1 rs4149623 T Recessive –0.784 0.003 NS rs4149639 A Recessive 0.210 0.196 0.019 rs4149622 A Recessive –0.794 0.002 NS rs4149578 G Additive 0.293 0.030 NS * Larger p-value reported, though LRT always similar in magnitude to the Wald test § Non-significant (NS) p-values >0.10 Table 4 Results of association analysis of TNFa as the phenotype Gene Marker Referent allele Best fitting model b coef Genotype p-value* HIV · locus interaction§ IFNGR1 rs4896243 C Dominant –0.700 2.89 · 10–11 0.004 rs1327474 G Dominant 0.044 0.814 NS rs2234711 T Additive –0.307 8.978 · 10–7 NS IL-10 rs1518111 A Dominant –0.774 1.54 · 10–13 NS rs1554286 C Recessive 0.779 6.31 · 10–14 NS rs1800872 A Dominant –0.770 1.50 · 10–13 NS TNFR1 rs4149623 T Dominant –0.467 1.13 · 10–5 1 · 10–7 rs4149639 A Recessive –0.435 4.15 · 10–6 0.070 rs4149622 A Dominant –0.443 3.44 · 10–5 1 · 10–7 rs4149578 G Dominant 0.410 0.010 NS * Larger p-value reported, though LRT always similar in magnitude to the Wald test § Non-significant (NS) p-values >0.10 668 Hum Genet (2007) 121:663–673 123 above Bonferroni correction. Based on the regression coefficients from the association models for IFNGR1, the allele associated with increased risk for TB (through either an additive or dominant model) was associated with in- creased TNFa expression. The same appears to be true for TNFR1, though the results are less consistent. When including the SNPs within the Haseman–Elston model, we found that the most significant SNP (per gene) accounted for the whole linkage signal. The exception was the linkage analysis of TNFa and IFNGR1, when both rs2234711 and rs4896243 were needed to account for the linkage signal (data not shown). To explore whether the genetic associations seen with the TB phenotype depict risk for active disease or latent Mtb infection, we conducted an analysis where we con- trasted TB patients with individuals having latent Mtb infection but no evidence of disease. The associations seen with IFNGR1 and TNFR1 remained significant (same SNPs, p < 0.01, data not shown), implying that these loci influence progression from Mtb infection to active disease. When we examined the effect of interaction between HIV status and the SNPs on both phenotypes, we found that TNFR1 is the only gene that showed consistent evi- dence for a locus · HIV interaction on both phenotypes; three out of four markers had significant (at a = 0.10) HIV · SNP interactions when TNFa was the phenotype (Table 4), and one out of four markers had a significant interaction when TB was the phenotype (Table 5). It is likely that we did not have the power to detect interactions for TB as the trait because there were only 61 HIV negative TB cases and 48 HIV positive TB cases. There was one SNP showing a significant HIV · SNP interaction for IFNGR1. These results are weaker because the other SNPs within IFNGR1 did not show a significant interaction (p > 0.10) and the specific SNP involved in the interaction differed across phenotypes. To further examine the inter- action effects, we redid the analysis of TNFa, stratifying on HIV status (Table 6). We did not conduct a stratified analysis for the TB trait because of the sample size limi- tations noted above. For TNFa as the trait of interest, we found the SNP-phenotype associations remained significant within strata (p < 0.10). Furthermore, we found the SNPs had stronger negative associations with TNFa levels in the HIV negative subgroup than in the HIV positive subgroup. Next, we evaluated associations between haplotypes in these genes with both phenotypes. A haplotype in IL-10 (T-A-A) was associated with protection against TB (b = –0.958, p = 0.002, data not shown), even though the SNPs analyzed individually did not demonstrate associa- tion. This suggests that another polymorphism within IL-10 may convey TB risk, though these particular SNPs do not. This haplotype, consisting of all three minor alleles, increases risk for TB, and the minor alleles also appear to be associated with decreased TNFa levels (Table 4). Discussion Several studies have suggested that risk for developing TB disease is influenced by host genetic factors. However, many of the previous studies have been hindered by inconsistent phenotype definitions that ignore the complex nature of TB. In this study conducted in Kampala, Uganda, a country endemic for TB, we took a unique approach to evaluate candidate genes for TB by using two phenotypes. Pulmonary TB, the disease phenotype, was evaluated and diagnosed in a rigorous and consistent way throughout the study. Our study is distinctive in that the circumstances of exposure are known and accounted for in the analysis as a covariate. Furthermore, because we assess latent Mtb infection prospectively, our study has the ability to dis- tinguish whether genetic loci predispose to latent Mtb infection or active disease. The other phenotype is an intermediate phenotype based on known components of the host immune response to Mtb. By examining linkage and association for both phenotypes in the same study sample, we were able to show construct validity and internal con- sistency in a single study. We have developed expression levels of TNFa in re- sponse to Mtb culture filtrate as an intermediate phenotype (Stein et al. 2003, 2005) and, in the present study, we have examined linkage and association of candidate genes hypothesized to regulate TNFa expression. The interme- diate phenotype model is useful because such traits are thought to be more powerful than binary traits because of their quantitative nature (Duggirala et al. 1997) and be- cause they are more closely tied to the level of gene expression (Rice et al. 2001; Risch 2000). Table 6 Association analysis stratified by HIV status with TNFa as the phenotype Gene Marker b in HIV+ p-value in HIV+ b in HIV– p-value in HIV– IFNGR1 rs4896243 –0.431 0.074 –1.270 4.44 · 10–16 TNFR1 rs4149623 –0.255 0.011 –1.451 4.44 · 10–16 rs4149639 –0.398 3.32 · 10–5 –0.507 1.00 · 10–5 rs4149622 –0.223 0.016 –1.482 1.00 · 10–7 Hum Genet (2007) 121:663–673 669 123 By observing linkage and association between markers and both traits, we can delineate the role of these genes and TB pathogenesis. Our analyses for TNFR1 and IFNGR1 suggest that the overexpression of a proinflammatory cytokine, TNFa, is associated with disease. These results are consistent with previous studies demonstrating a posi- tive correlation between TB severity and TNFa levels, and a corresponding decrease in TNFa levels after treatment (Bekker et al. 2000; Ribeiro-Rodrigues et al. 2002). Fur- thermore, our results suggest that HIV modifies the effect of TNFR1 on TNFa expression. These results illustrate that the impact of TNFa release on TB risk clearly depends on the specific genes involved, implying that the function of these genes must be considered in the immunologic anal- yses. Certainly, these association results do not imply causality, but they do generate new hypotheses that warrant follow-up in functional studies. In addition, it is possible that the relationship between TNFa levels and TB depends on the stage of disease. At this time we do not have suf- ficient data to test this hypothesis, but it will be the focus of future research. To our knowledge, we are the first to find an association between TNFR1 and TNFa regulation and TB risk in a human population. It has been suggested that the soluble portion of TNFa receptors may block TNFa activity, increasing susceptibility to TB (Keane et al. 2001). Moreover, a role for TNFR1 in TB susceptibility and TNFa protein production has been demonstrated through knock- out mouse models (Peschon et al. 1998; Piguet et al. 2002). Despite this association with disease and cytokine expres- sion, none of the tagSNPs we genotyped appear to have known functional implications. Since this is the first report finding an association between TB risk and the human TNFR1 gene, further studies are needed to replicate the finding. The first report of an association with IFNGR1 was from a Maltese pedigree; this association was observed with non-tuberculous mycobacterial infection (Newport et al. 1996). Association between a SNP in IFNGR1 and disease caused by specifically M. tuberculosis was recently re- ported in a Gambian population (Cooke et al. 2006), though previous analyses of microsatellites within IFNGR1 were inconsistent (Awomoyi et al. 2004; Fraser et al. 2003). Our analysis independently confirms that of Cooke and colleagues who found the CC genotype at the –56 SNP (rs2234711) to be associated with increased risk for TB disease. In addition to replicating this association, we make the novel observation that this SNP is associated with TNFa levels, and that two additional SNPs within this gene are associated with both traits. Since IFNc has a role in TNFa production (Knight and Kwiatkowski 1999), defi- ciencies in IFNGR1 may also affect TNFa activity, ulti- mately influencing TB susceptibility. In this analysis, a haplotype in IL-10 showed protection against TB disease. The relationship between IL-10 SNPs and TNFa is less clear; further molecular experiments are needed to clarify the relationship between the IL-10 gene and TNFa regulation. IL-10 is an inhibitor of Th-1 cell and macrophage functions (Janeway et al. 2001), and drug and molecular studies suggest that IL-10 activity leads to de- creased TNFa production (Balcewicz-Sablinska et al. 1999; Goldman et al. 1996). There are several case-control studies examining the association between the IL-10 gene and TB risk, but their results are inconsistent. Though a promotor SNP, –1082, has been associated with protection against TB in studies in Malawi (Fitness et al. 2004), Cambodia (Delgado et al. 2002), and Sicily (Scola et al. 2003), nonsignificant results were found in studies in Spain (López-Maderuelo et al. 2003), Hong Kong (Tso et al. 2005), Korea (Shin et al. 2005), and the Gambia (Bellamy et al. 1998). We did not genotype the –1082 SNP (rs1800896) because it was not a tagSNP. However, rs1800872 is also a promoter SNP (–592) and is in strong LD with –1082 (Kurreeman et al. 2004). Interestingly, though the Korean study (Shin et al. 2005) did not find association between TB and –1082, it did detect association between TB and –592, so our report provides replication of that findings. Apart from the disease phenotype, we found linkage and association between IL-10 and TNFa levels. Since we used a haplotype-tagging approach to select SNPs and observed association between TB and a haplotype in IL-10, but not with the individual SNPs, it is possible that we are detecting association to a polymorphism in LD with our SNPs, such as –1082. Fine mapping is necessary to clarify this hypothesis. One of the most studied genes with regard to TB risk is SLC11A1, which has been associated with TB across a number of populations, as shown by a recent meta-analysis (Li et al. 2006). We did not detect linkage to this gene and consequently, we did not pursue association analysis. A study conducted in a Gambian population also did not find association between SLC11A1 and TNFa response to lipopolysaccaharide (LPS) (Awomoyi et al. 2002). On the one hand, it is possible that the association between SLC11A1 does not exist in this Ugandan population; on the other, SLC11A1 may have a small effect on TB risk, and small effects are less likely to be detected by linkage analysis (Risch and Merikangas 1996). The linkage analysis results have other noteworthy findings. First, TNFR2, TLR4, and TLR2 were linked to TB disease but not to TNFa levels. These findings suggest that these genes are related to TB susceptibility but not via TNFa regulation. Second, the Haseman–Elston and con- ditional logistic methods provided slightly different results for the binary trait—TNFR1, IL-10, TNFR2 and TLR2 were only detected by one method. The explanation for this 670 Hum Genet (2007) 121:663–673 123 apparent inconsistency is that these two models are based on different, complementary models. The Haseman–Elston model contrasts concordant and discordant pairs, but when extrapolated, allows for impossible values for IBD sharing less than 0 and greater than 1. The conditional logistic model as implemented in LODPAL compares observed allele sharing IBD to that expected under Mendelian seg- regation alone. This model avoids the aforementioned problem of IBD sharing less 0 and greater than 1, but relies heavily on the assumption of Mendelian segregation without any selective forces acting to distort it. Sometimes one model may be better than another, but the underlying biologic reasons for this phenomenon are not understood. However, it is important to emphasize that these analyses were based on only 109 individuals with TB, so this small sample size is a limitation of our study. In this first molecular genetic study of TB susceptibility conducted from Uganda, we took a unique approach to the characterization of a complex trait. In doing so, we found a new association between TB and TNFR1 in a human population and replicated previous findings for IFNGR1 and IL-10 from different ethnic populations. These findings are consistent with the immunologic model for TB patho- genesis that Th-1-like cytokines such as IFNc and TNFa create a microenvironment within granuloma which con- tains growth of Mtb and that is down-regulated by IL-10. Validation of these findings in other populations is war- ranted as well as finer mapping of these candidate genes, especially since previous studies have focused on a few polymorphisms per gene, and it is possible that there are polymorphisms in LD with those studied that actually convey risk. Acknowledgments This study would not be possible without the support of the Ugandan National Tuberculosis and Leprosy Control Programme, the generous participation of the Ugandan patients and families, and the medical officers, project coordinators, laboratory personnel and home health visitors who helped collect the epidemi- ological data. This work is supported in part by the National Institutes of Health, National Center for Research Resources (NCRR) Multi- disciplinary Clinical Research Career Development Programs Grant (8K12RR023264), Tuberculosis Research Unit (grant N01-AI95383 from the NIAID), National Institute of General Medical Sciences (GM-28356), a developmental grant from the STERIS Corporation, and the Gene Expression and Genotyping Facility of the Compre- hensive Cancer Center of Case Western Reserve University and University Hospitals of Cleveland (P30 CA43703). Some of the re- sults of this paper were obtained by using the program package S.A.G.E., which is supported by a U.S. Public Health Service Resource Grant (RR03655) from the NCRR. References Altshuler D, Brooks LD, Chakravarti A, Collins FS, Daly MJ, Donnelly P (2005) A haplotype map of the human genome. Nature 437:1299–1320 Awomoyi A, Marchant A, Howson J, McAdam K, Blackwell J, Newport M (2002) Interleukin-10, Polymorphism in SLC11A1 (formerly NRAMP1), and susceptibility to tuberculosis. J Infect Dis 186:1808–1814 Awomoyi AA, Nejentsev S, Richardson A, Hull J, Koch O, Podinovskaia M, Todd JA, McAdam KPWJ, Blackwell JM, Kwiatkowski D, Newport MJ (2004) No association between interferon-c receptor-1 gene polymorphism and pulmonary tuberculosis in a Gambian population sample. Thorax 59:291– 294 Balcewicz-Sablinska M, Gan H, Remold H (1999) Interleukin 10 produced by macrophages inoculated with Mycobacterium avium attenuates mycobacteria-induced apoptosis by reduction of TNF-a activity. J Infect Dis 180:1230–1237 Barnes P, Fong S, Brennan P, Twomey P, Mazumder A, Modlin R (1990) Local production of tumor necrosis factor and IFN- gamma in tuberculosis pleuritis. J Immunol 145:149–154 Barrett JC, Fry B, Maller J, Daly MJ (2005) Haploview: analysis and visualization of LD and haplotype maps. Bioinformatics 21:263– 265 Bekker LG, Haslett P, Maartens G, Steyn L, Kaplan G (2000) Thalidomide-induced antigen-specific immune stimulation in patients with human immunodeficiency virus type 1 and tuberculosis. J Infect Dis 181:954–965 Bellamy R (2003) Susceptibility to mycobacterial infections: the importance of host genetics. Genes Immun 4:4–11 Bellamy R, Ruwende C, Corrah T, McAdam K, Whittle H, Hill A (1998) Assessment of the interleukin 1 gene cluster and other candidate gene polymorphisms in host susceptibility to tubercu- losis. Tuberc Lung Dis 79:83–89 Blackwell J, Barton C, White J, Searle S, Baker A, Williams H, Shaw M (2004) Genomic organizaton and sequence of the human NRAMP gene: identification and mapping of a promotor region polymorphism. Mol Med 1:194–205 Blumberg H, Burman W, Chaisson R, Daley C, Etkind S, Friedman L, Fujiwara P, Grzemska M, Hopewell P, Iseman M, Jasmer R, Koppaka V, Menzies R, O’Brien R, Reves R, Reichman L, Simone R, Starke J, Vernon A, American Thoracic Society, Centers for Disease Control, Prevention, Infectious Diseases Society (2003) American Thoracic Society/Centers for Disease Control and Prevention/Infectious Diseases Society of America: treatment of tuberculosis. Am J Respir Crit Care Med 167:603– 662 Bochud P-Y, Hawn T, Aderem A (2003) Cutting edge: a toll-like receptor 2 polymorphism that is associated with lepromatous leprosy is unable to mediate mycobacterial signaling. J Immunol 170:3451–3454 Centers for Disease Control and Prevention (2003) Treatment of tuberculosis, American Thoracic Society, CDC, and Infectious Diseases Soceity of America. MMWR 52:1–77 Cooke GS, Campbell SJ, Sillah J, Gustafson P, Bah B, Sirugo G, Bennett S, McAdam KP, Sow O, Lienhardt C, Hill AV (2006) Polymorphism within the interferon gamma/receptor complex is associated with pulmonary tuberculosis. Am J Respir Crit Care Med 174:339–343 Delgado J, Baena A, Thim S, Goldfeld A (2002) Ethnic-specific genetic associations with pulmonary tuberculosis. J Infect Dis 186:1463–1468 Drennan M, Nicolle D, Quesniaux V, Jacobs M, Allie N, Mpagi J, Frémond C, Wagner H, Kirschning C, Ryffel B (2004) Toll-like receptor 2-deficient mice succumb to Mycobacterium tubercu- losis infection. Am J Pathol 164:49–57 Duggirala R, Williams J, Williams-Blangero S, Blangero J (1997) A variance components approach to dichotomous trait linkage using a threshold model. Genet Epidemiol 14:987–992 Hum Genet (2007) 121:663–673 671 123 Elston RC, George VT, Severtson F (1992) The Elston–Stewart algorithm for continuous genotypes and environmental factors. Hum Hered 42:16–27 Epstein M, Duren W, Boehnke M (2000) Improved inference of relationship for pairs of individuals. Am J Hum Genet 67:1219– 1231 Fitness J, Floyd S, Warndorff D, Sichali L, Malema S, Crampin A, Fine P, Hill A (2004) Large-scale candidate gene study of tuberculosis susceptibility in the Karonga district of Nothern Malawi. Am J Trop Med Hyg 71:341–349 Flores-Villanueva PO, Ruiz-Morales JA, Song CH, Flores LM, Jo EK, Montano M, Barnes PF, Selman M, Granados J (2005) A functional promoter polymorphism in monocyte chemoattractant protein-1 is associated with increased susceptibility to pulmo- nary tuberculosis. J Exp Med 202:1649–1658 Flynn J, Goldstein M, Chan J, Triebold K, Pfeffer K, Lowenstein C, Schreiber R, Mak T, Bloom B (1995) Tumor necrosis factor- alpha is required in the protective immune response against Mycobacterium tuberculosis in mice. Immunity 2:561–572 Formica S, Roach T, Blackwell J (1994) Interaction with extracellular matrix proteins influences Lsh/Ity/Bcg (candidate Nramp) gene regulation of macrophage priming/activation for tumor necrosis factor-a and nitrite release. Immunology 82:42–50 Fraser D, Bulat-Kardum L, Knezevic J, Babarovic P, Matakovic- Mileunic N, Dellacasagrande J, Matanic D, Pavelic J, Beg-Zec Z, Dembic Z (2003) Interferon-c receptor-1 gene polymorphism in tuberculosis patients from Croatia. Scand J Immunol 57:480– 484 George V, Elston RC (1987) Testing the association between polymorphic markers and quantitative traits in pedigrees. Genet Epidemiol 4:193–202 Goddard K, Witte J, Suarez B, Catalona W, Olson J (2001) Model- free linkage analysis with covariates confirms linkage of prostate cancer to chromosomes 1 and 4. Am J Hum Genet 68:1197–1206 Goldman M, Marchant A, Schandené L (1996) Endogenous interleu- kin-10 in inflammatory disorders: Regulatory roles and pharma- cological modulation. Ann NY Acad Sci 796:282–293 Gray-McGuire C (2004) Assessment of a variance component method for binary phenotype data: model misspecification and effects of ascertainment. PhD Dissertation, Case Western Reserve Univer- sity Guwattude D, Nakakeeto M, Jones-Lopez E, Maganda A, Chiunda A, Mugerwa R, Ellner J, Bukenya G, Whalen C (2003) Tubercu- losis in household contacts of infectious cases in Kampala, Uganda. Am J Epidemiol 158:887–898 Haseman J, Elston R (1972) The investigation of linkage between a quantitative trait and a marker locus. Behav Genet 2:3–19 Janeway C, Travers P, Walport M, Shlomchik M (2001) Immuno- biology, 5th edn. Garland Publishing, New York Keane J, Gershon S, Wise R, Mirabile-Levens E, Kaszinca J, Schwietermann W, Siegel J, Braun M (2001) Tuberculosis associated with inflixamab, a tumor necrosis factor a-neutraliz- ing agent. N Engl J Med 345:1098–1104 Knight J, Kwiatkowski D (1999) Inherited variability of tumor necrosis factor production and susceptibility to infectious disease. Proc Assoc Am Phys 111:290–298 Kong X, Murphy K, Raj T, He C, White P, Matise T (2004) A combined linkage-physical map of the human genome. Am J Hum Genet 75:1143–1148 Kurreeman FA, Schonkeren JJ, Heijmans BT, Toes RE, Huizinga TW (2004) Transcription of the IL10 gene reveals allele-specific regulation at the mRNA level. Hum Mol Genet 13:1755–1762 Lander E, Kruglyak L (1995) Genetic dissection of complex traits: guidelines for interpreting and reporting linkage results. Nat Genet 11:241–247 Li HT, Zhang TT, Zhou YQ, Huang QH, Huang J (2006) SLC11A1 (formerly NRAMP1) gene polymorphisms and tuberculosis susceptibility: a meta-analysis. Int J Tuberc Lung Dis 10:3–12 López-Maderuelo D, Arnalich F, Serantes R, González A, Codeoceo R, Madero R, Vazquez J, Montiel C (2003) Interferon-c and interleukin-10 gene polymorphisms in pulmonary tuberculosis. Am J Respir Crit Care Med 167:970–975 Means T, Wang S, Lien E, Yoshimura A, Golenbock D, Fenton M (1999) Human Toll-like receptors mediate ceulluar activation by Mycobacterium tuberculosis. J Immunol 163:3920–3927 Means T, Jones B, Schromm A, Shurtleff B, Smith J, Keane J, Golenbock D, Vogel S, Fenton M (2001) Differential effects of a toll-like receptor antagonist on Mycobacterium tuberculosis- induced macrophage responses. J Immunol 166:4074–4082 Miller S, Dykes D, Polesky H (1988) A simple salting out procedure for extracting DNA from human nucleated cells. Nucleic Acids Res 16:1215 Newport MJ, Huxley CM, Huston S, Hawrylowicz CM, Oostra BA, Williamson R, Levin M (1996) A mutation in the interferon-c- receptor gene and susceptbility to mycobacterial infection. New Engl J Med 335:1941–1949 Olson J (1999) A general conditional-logistic model for affected- relative-pair linkage studies. Am J Hum Genet 65:1760–1769 Peschon JJ, Torrance DS, Stocking KL, Glaccum MB, Otten C, Willis CR, Charrier K, Morrissey PJ, Ware CB, Mohler KM (1998) TNF receptor-deficient mice reveal divergent roles for p55 and p75 in several models of inflammation. J Immunol 160:943–952 Piguet PF, Da Kan C, Vesin C (2002) Rolle of tumor necrosis factor receptor 2 (TNFR2) in cerebral malaria in mice. Lab Invest 82:1155–1166 Pociot F, Briant L, Jongeneel CV, Mölvig J, Worsaae H, Abbal M, Thomsen M, Nerup J, Cambon-Thomsen A (1993) Association of tumor necrosis factor (TNF) and class II major histocompat- ibility complex alleles with the secretion of TNF-a and TNF-b by human mononuclear cells: a possible link to insulin-depen- dent diabetes mellitus. Eur J Immunol 23:224–231 Raviglione M, Snider D, Kochi A (1995) Global epidemiology of tuberculosis: morbidity and mortality of a worldwide epidemic. JAMA 273:220–226 Ribeiro-Rodrigues R, Resende CT, Johnson JL, Ribeiro F, Palaci M, Sa RT, Maciel EL, Pereira Lima FE, Dettoni V, Toossi Z, Boom WH, Dietze R, Ellner JJ, Hirsch CS (2002) Sputum cytokine levels in patients with pulmonary tuberculosis as early markers of mycobacterial clearance. Clin Diagn Lab Immunol 9:818–823 Rice JP, Saccone NL, Rasmussen E (2001) Definition of the phenotype. Adv Genet 42:69–76 Risch N (2000) Searching for genetic determinants in the new millenium. Nature 405:847–856 Risch N, Merikangas K (1996) The future of genetic studies of complex human diseases. Science 273:1516–1517 Roach DR, Bean AGD, Demangel C, France MP, Briscoe H, Britton WJ (2002) TNF regulates chemokine induction essential for cell recruitment, granuloma formation, and clearance of mycobacte- rial infection. J Immunol 168:4620–4627 S.A.G.E. (2006) Statistical analysis for genetic epidemiology. Case Western Reserve University, Cleveland, OH Scola L, Crivello A, Marino V, Gioia V, Serauto A, Candore G, Colonna-Romano G, Caruso C, Lio D (2003) IL-10 and TNF- alpha polymorphisms in a sample of Sicilian patients affected by tuberculosis: implication for ageing and life span expectancy. Mech Ageing Dev 124:569–572 Shete S, Jacobs KB, Elston RC (2003) Adding further power to the Haseman and Elston method for detecting linkage in larger sibships: weighting sums and differences. Hum Hered 55:79–85 672 Hum Genet (2007) 121:663–673 123 Shin HD, Park BL, Kim YH, Cheong HS, Lee IH, Park SK (2005) Common interleukin 10 polymorphism associated with de- creased risk of tuberculosis. Exp Mol Med 37:128–132 Skamene E, Schurr E, Gros P (1998) Infection genomics: Nramp1 as a major determinant of natural resistance to intracellular infec- tions. Annu Rev Med 49:275–287 Stein CM, Guwattude D, Nakakeeto M, Peters P, Elston RC, Tiwari HK, Whalen CC (2003) Heritability analysis of cytokines as intermediate phenotypes of tuberculosis. J Infect Dis 187:1679– 1685 Stein CM, Nshuti L, Chiunda AB, Boom WH, Elston RC, Mugerwa RD, Iyengar SK, Whalen CC (2005) Evidence for a major gene influence on tumor necrosis factor-alpha expression in tubercu- losis: path and segregation analysis. Hum Hered 60:109–118 Tso H, Ip W, Chong W, Tam C, Chiang A, Lau Y (2005) Association of interferon gamma and interleukin 10 genes with tuberculosis in Hong Kong Chinese. Genes Immun 6:358–363 Underhill DM, Ozinsky A, Smith KD, Aderem A (1999) Toll-like receptor-2 mediates mycobacteria-induced proinflammatory sig- naling in macrophages. Proc Natl Acad Sci 96:14459–14463 van Crevel R, Ottenhoff THM, van der Meek JWM (2002) Innate immunity to mycobacterium tuberculosis. Clin Microbiol Rev 15:294–309 Wigginton JE, Abecasis GR (2005) PEDSTATS: descriptive statis- tics, graphics and quality assessment for gene mapping data. Bioinformatics 21:3445–3447 Wilson AG, Symons JA, McDowel TL, McDevitt HO, Duff GW (1997) Effects of a polymorphism in the human tumor necrosis factor a promotor on transcriptional activation. Proc Natl Acad Sci 94:3195–3199 Witte JS, Elston RC, Schork NJ (1996) Genetic dissection of complex traits. Nat Genet 12:355–356 Hum Genet (2007) 121:663–673 673 123 Linkage and association analysis of candidate genes for TB �and TNF&agr; cytokine expression: evidence for association �with IFNGR1, IL-10, and TNF receptor 1 genes Abstract Introduction Materials and methods Subject recruitment and assessment Molecular methods Statistical analysis Results Discussion Acknowledgments References << /ASCII85EncodePages false /AllowTransparency false /AutoPositionEPSFiles true /AutoRotatePages /None /Binding /Left /CalGrayProfile (None) /CalRGBProfile (sRGB IEC61966-2.1) /CalCMYKProfile (ISO Coated) /sRGBProfile (sRGB IEC61966-2.1) /CannotEmbedFontPolicy /Error /CompatibilityLevel 1.3 /CompressObjects /Off /CompressPages true /ConvertImagesToIndexed true /PassThroughJPEGImages true /CreateJDFFile false /CreateJobTicket false /DefaultRenderingIntent /Perceptual /DetectBlends true /ColorConversionStrategy /sRGB /DoThumbnails true /EmbedAllFonts true /EmbedJobOptions true /DSCReportingLevel 0 /SyntheticBoldness 1.00 /EmitDSCWarnings false /EndPage -1 /ImageMemory 524288 /LockDistillerParams true /MaxSubsetPct 100 /Optimize true /OPM 1 /ParseDSCComments true /ParseDSCCommentsForDocInfo true /PreserveCopyPage true /PreserveEPSInfo true /PreserveHalftoneInfo false /PreserveOPIComments false /PreserveOverprintSettings true /StartPage 1 /SubsetFonts false /TransferFunctionInfo /Apply /UCRandBGInfo /Preserve /UsePrologue false /ColorSettingsFile () /AlwaysEmbed [ true ] /NeverEmbed [ true ] /AntiAliasColorImages false /DownsampleColorImages true /ColorImageDownsampleType /Bicubic /ColorImageResolution 150 /ColorImageDepth -1 /ColorImageDownsampleThreshold 1.50000 /EncodeColorImages true /ColorImageFilter /DCTEncode /AutoFilterColorImages false /ColorImageAutoFilterStrategy /JPEG /ColorACSImageDict << /QFactor 0.76 /HSamples [2 1 1 2] /VSamples [2 1 1 2] >> /ColorImageDict << /QFactor 0.76 /HSamples [2 1 1 2] /VSamples [2 1 1 2] >> /JPEG2000ColorACSImageDict << /TileWidth 256 /TileHeight 256 /Quality 30 >> /JPEG2000ColorImageDict << /TileWidth 256 /TileHeight 256 /Quality 30 >> /AntiAliasGrayImages false /DownsampleGrayImages true /GrayImageDownsampleType /Bicubic /GrayImageResolution 150 /GrayImageDepth -1 /GrayImageDownsampleThreshold 1.50000 /EncodeGrayImages true /GrayImageFilter /DCTEncode /AutoFilterGrayImages true /GrayImageAutoFilterStrategy /JPEG /GrayACSImageDict << /QFactor 0.76 /HSamples [2 1 1 2] /VSamples [2 1 1 2] >> /GrayImageDict << /QFactor 0.15 /HSamples [1 1 1 1] /VSamples [1 1 1 1] >> /JPEG2000GrayACSImageDict << /TileWidth 256 /TileHeight 256 /Quality 30 >> /JPEG2000GrayImageDict << /TileWidth 256 /TileHeight 256 /Quality 30 >> /AntiAliasMonoImages false /DownsampleMonoImages true /MonoImageDownsampleType /Bicubic /MonoImageResolution 600 /MonoImageDepth -1 /MonoImageDownsampleThreshold 1.50000 /EncodeMonoImages true /MonoImageFilter /CCITTFaxEncode /MonoImageDict << /K -1 >> /AllowPSXObjects false /PDFX1aCheck false /PDFX3Check false /PDFXCompliantPDFOnly false /PDFXNoTrimBoxError true /PDFXTrimBoxToMediaBoxOffset [ 0.00000 0.00000 0.00000 0.00000 ] /PDFXSetBleedBoxToMediaBox true /PDFXBleedBoxToTrimBoxOffset [ 0.00000 0.00000 0.00000 0.00000 ] /PDFXOutputIntentProfile (None) /PDFXOutputCondition () /PDFXRegistryName (http://www.color.org?) /PDFXTrapped /False /Description << /DEU /ENU >> >> setdistillerparams << /HWResolution [2400 2400] /PageSize [2834.646 2834.646] >> setpagedevice