Genes and Immunity (2003) 4, 351–355 & 2003 Nature Publishing Group All rights reserved 1466-4879/03 $25.00 www.nature.com/gene

Genetic susceptibility to visceral in The Sudan: linkage and association with IL4 and IFNGR1

HS Mohamed1,2, ME Ibrahim1, EN Miller2, CS Peacock2, EAG Khalil1, HJ Cordell2, JMM Howson2, AM El Hassan1, REH Bereir1 and JM Blackwell2 1Institute of Endemic Diseases, University of Khartoum, Khartoum, Sudan; 2Cambridge Institute for Medical Research, Addenbrooke’s Hospital, Cambridge, UK

Longitudinal studies in Sudan show ethnic differences in and clinical phenotypes associated with donovani. Immunologically, bias in type 1 vs type 2 cytokine responses is important. To determine whether polymorphisms at IL4/IL9 or IFNGR1 contribute to susceptibility, we examined 59 multicase families of (VL) with/without post Kala-azar dermal leishmaniasis (PKDL). Multipoint nonparametric analysis (Allegro) linked IL4/IL9 to VL per se (P ¼ 0.002). disequilibrium testing with robust variance estimates confirmed association in the presence of linkage between VL per se and IL4 (P ¼ 0.008) but not IL9. Stepwise logistic regression analysis showed both IL4RP2 and IL4RP1 markers contributed significantly to the association, suggesting a common disease-associated haplotype. In contrast, IFNGR1 was linked (P ¼ 0.031) and associated (P ¼ 0.007) to PKDL but not VL or VL per se. Hence, polymorphism in a type 2 cytokine gene influences underlying susceptibility to VL, whereas IFNGR1 is specifically related to susceptibility to PKDL. Genes and Immunity (2003) 4, 351–355. doi:10.1038/sj.gene.6363977

Keywords: leishmaniasis; association; linkage; IL4; IFNGR1

Introduction Results Longitudinal studies in Sudan show marked differences Linkage and allelic association for VL per se with in incidence and clinical phenotypes, for example, IL4/IL9 clinical vs subclinical visceral leishmaniasis (VL), inci- Results of multipoint nonparametric analysis using the dence of post Kala-azar dermal leishmaniasis (PKDL), Spairs scoring function to look for linkage between IL4/IL9 associated with sensu strictu in and susceptibility to leishmaniasis are presented in different ethnic groups with shared environments.1 Table 1. The results provide evidence for linkage between Immunologically, balance between interferon-g (IFNg) all three markers and a putative disease-susceptibility secreting type 1 vs interleukin-4 (IL4) secreting type 2 T locus controlling susceptibility to VL with PKDL cells is important.2 More recent studies also suggest a (IL4RP2: P ¼ 0.008; IL4RP1: P ¼ 0.031; IL9: P ¼ 0.009) role for regulatory T cells that secrete both IFNg and and VL per se (IL4RP2: P ¼ 0.002; IL4RP1: P ¼ 0.013; IL9: IL10,3 with IL10 playing an important role in the P ¼ 0.034). These data provide suggestive evidence for a development of PKDL.4 Nonexposed individuals also gene(s) at 5q23.3–q33 controlling underlying suscept- show differences in patterns of IFNg and/or IL4 ibility to VL in this Sudanese population. production upon stimulation with L. donovani and L. To narrow down the region of interest that might carry major antigens,5 suggesting an inherent bias in response the susceptibility allele within this type 2 cytokine gene that could influence clinical outcome upon exposure to cluster, allelic association was determined using the . In this context, genes that regulate, or control transmission disequilibrium test (TDT). Table 2a shows response to, type 1 and type 2 cytokines are good the global w2 statistics for TDT with correction for candidate susceptibility loci. Here, we present data multiple affected sibs in a pedigree. The single markers showing that polymorphism at loci within the type 2 IL4RP2 and IL4RP1 show significant allelic association cytokine gene cluster on chromosome 5q23.3–q31 may with VL alone, PKDL and VL per se. IL4RP2 remains influence underlying susceptibility to VL in Sudan, while significant for allelic association in the presence of polymorphism at the gene IFNGR1 on chromosome linkage (ie robust test), suggesting that this marker lies 6q23–q24 encoding the type 1 receptor for IFNg may be close to the functional polymorphism. Using the ‘case– specifically related to susceptibility to PKDL. pseudo-control’ data set generated from the families and conditional logistic regression analysis (Table 2b) sig- nificant allelic associations with VL per se in the presence Correspondence: Professor JM Blackwell, Cambridge Institute for Medical Research, Wellcome Trust/MRC Building, Addenbrooke’s Hospital, Hills of linkage (robust tests) were observed for allele 2 at Road, Cambridge CB2 2XY, UK. IL4RP2 (P ¼ 0.0034) and allele 4 at IL4RP1 (P ¼ 0.0007) E-mail: [email protected] under a multiplicative model (ie two copies of an allele IL4 and IFNGR1 polymorphism and leishmaniasis H Mohamed et al 352 Table 1 Multipoint nonparametric linkage analysis using the Spairs function in Allegro to examine linkage with different disease phenotypes at the IL4/IL9 region

Location (cM/Mb) LOD Zlr P-value (one-tailed) Information content Marker

VL with PKDL 0 1.263 2.412 0.008 0.708 IL4RP2 0.001 0.761 1.873 0.031 0.706 IL4RP1 3.850 1.179 2.331 0.009 0.654 IL9

VL per se 0 1.840 2.911 0.002 0.697 IL4RP2 0.001 1.081 2.232 0.013 0.675 IL4RP1 3.850 0.721 1.822 0.034 0.590 IL9

Table 2 Allelic association testing: (a) Results of transmission disequilibrium tests (TDT), and (b) uses a ‘case–pseudo-control’ data set and conditional logistic regression to look at the association of alleles at each marker (rows 1–2) with VL per se

(a) Results of TDT

TDT statistics VL alone VL with PKDL VL per se

IL4RP2 IL4RP1 IL9 IL-4RP2 IL4RP1 IL9 IL4RP2 IL4RP1 IL9

w2 5.04 18.95 6.14 4.24 8.65 2.13 8.44 25.69 3.49 df 1 8 4 1 3 1 1 10 4 P 0.025 0.015 0.189 0.039 0.034 0.144 0.004 0.004 0.479 w2 robust test 4.84 7.3828 6.71 2.57 6.55 1.68 6.94 11.33 3.31 df 1 8 4 1 3 1 1 10 4 P corrected 0.028 0.496 0.151 0.109 0.088 0.1944 0.008 0.3323 0.5064

(b) case–pseudo-control data set Null model Alternative model Test statistic Robust test statistic w2dfP w2dfP

F IL4RP2(a2) 8.54 1 0.0035 8.60 1 0.0034 F IL4RP1(a4) 15.94 1 0.0001 11.54 1 0.0007 IL4RP1(a4) IL4RP1(a4)+IL4RP2(a2) 7.52 1 0.0061 8.02 1 0.0046 IL4RP2(a2) IL4RP2(a2)+IL4RP1(a4) 8.21 1 0.0042 5.37 1 0.0205 IL4RP1(a4) IL4RP1(a4)+IL4RP2(a2) 10.85 2 0.0044 8.60 2 0.0136 IL4RP2(a2) IL4RP2(a2)+IL4RP1(a4) 8.84 2 0.0144 6.64 2 0.0361

Likelihood ratio tests and robust Wald w2 tests (rows 3–6) of main effects in a stepwise logistic regression procedure were used to determine the relative contributions of disease alleles at IL4RP2 (a2 = allele 2) and IL4RP1 (a4=allele 4) to the observed associations. 1 df tests report allelic contributions; 2 df tests report genotype contributions. All test statistics are shown with/without robust variance estimates to control for nonindependence of marker genotypes in pedigrees.

contribute twice as much as one copy). No dominance showed that both IL4RP1 allele 4 and IL4RP2 allele 2 effects were observed. For IL4RP2, the odds ratio of contribute significantly to the association, suggesting a disease associated with allele 2 relative to allele 1 was common disease-associated functional variant in linkage 1.68 (P ¼ 0.003; 95% CI 1.19–2.39). The genotype relative disequilibrium (LD) with a 2–4 haplotype for IL4RP2– risk for heterozygous 1/2 relative to 1/1 was 2.57 IL4RP1. Overall, the results are consistent with a gene in (P ¼ 0.008; 95% CI 1.27–5.18) and for homozygous 2/2 the close vicinity of the IL4 locus controlling underlying relative to 1/1 was 3.59 (P ¼ 0.003; 95% CI 1.56–8.27). For susceptibility to VL. IL4RP1, the odds ratio of disease associated with allele 4 relative to all other alleles was 2.5 (P ¼ 0.001; 95% CI Linkage and allelic association for PKDL with IFNGR1 1.47–4.24). The genotype relative risk for heterozygous Results of single-point nonparametric analysis using the

other/4 relative to other/other was not significant, but Spairs scoring function to look for linkage between for homozygous 4/4 relative to other/other was 5.72 IFNGR1 and susceptibility to leishmaniasis are shown (P ¼ 0.001; 95% CI 2.11–15.46). No significant allelic or in Table 3. Weak evidence for linkage was observed genotype associations were observed for IL9 using robust for PKDL (P ¼ 0.031), but not for VL per se. Similarly, variance estimates. Results of the likelihood ratio tests of evidence for allelic association between IFNGR1 and main effects in a stepwise logistic regression procedure6 susceptibility to disease was found only for the PKDL

Genes and Immunity IL4 and IFNGR1 polymorphism and leishmaniasis H Mohamed et al 353 Table 3 Single-point nonparametric using the Spairs function in polymorphisms to fine map the disease causing gene/ Allegro to examine linkage between IFNGR1 and VL with PKDL, or mutations across this region of 5q23.3 in this Sudanese VL per se, phenotypes population. The corollary to genes regulating type 2 cytokines as Phenotype LOD Zlr Zlr one-tailed Information candidate susceptibility genes in leishmaniasis are genes P-value content causing a defect in the type 1 cytokine pathway. An interesting candidate for interference with type 1 effector VL with PKDL 0.753 1.862 0.031 0.785 activity is polymorphism in the IFNGR1 gene encoding VL per se 0.548 1.588 0.056 0.576 the type 1 IFNg receptor. Studies in man have identified mutations in the gene IFNGR1 in Mendelian inheritance of susceptibility to disseminated atypical mycobacterial phenotype (global TDT w2 ¼ 15.84, 5 df, P ¼ 0.007). infection.13 Results presented here suggest that poly- Robust TDT showed allelic association between allele morphism at IFNGR1 is associated with susceptibility to 11 (w2 ¼ 4.12, 1 df, P ¼ 0.042) of IFNGR1 and PKDL. PKDL but not to underlying VL. Again, further work is Conditional logistic regression analysis indicated that required to verify this result and to identify the disease- alleles 10 (odds ratio ¼ 3.0; P ¼ 0.033; 95% CI 1.09–8.25) associated mutation/polymorphism in this population. and 11 (odds ratio ¼ 2.18; P ¼ 0.032; 95% CI 1.07–4.45) Recent studies suggest that regulatory T cells expressing were associated with disease, while allele 12 (odds IFNg and IL10 are expanded in individuals from Sudan ratio ¼ 0.15; P ¼ 0.014; 95% CI 0.035–0.682) was asso- with a history of drug-cured VL.3 Furthermore, that IL10 ciated with protection. plays an important role in the development of PKDL,4 consistent with studies in mice showing a role of IL10 in the persistence of L. major in the after healing.14 Our Discussion results suggest that failure of PKDL patients to respond adequately to IFNg at this post-treatment phase of In this study, we have shown suggestive evidence for infection may result in bias towards the down regulatory linkage and significant allelic association between mar- role played by IL10, leading to the recrudescence of kers within the genes encoding type 2 cytokines IL4 and parasites in remote sites in the skin. IL9 and underlying susceptibility to VL in Sudan. In Overall, our study has demonstrated important differ- man, immunological studies suggest that the balance ences in genetic susceptibility to VL and PKDL that may in type 1 vs type 2 cytokine responses is important in shed further light on the immunoregulatory pathways L. donovani infection.2,7,8 In relation to clinical VL, IL4 was that are important in determining these different detected in all Brazilian patient sera14 and its level manifestations of disease in the Sudanese population. increased 13-fold in Indian Kala-azar patients when compared with healthy controls.15 The role of IL9 in human leishmaniasis is not well documented. During Study population and methods L. major infection in the mouse, no functionally active IL9 is detected in sera of either susceptible BALB/c or Ascertainment and sampling of families resistant C57BL/6 mice, but IL9 mRNA is expressed at The study was carried out on the Nilosaharan-speaking higher levels in BALB/c lymph nodes and spleens and Masalit population who migrated from western Sudan in IL-9 protein produced at higher level after restimulation the early-mid 1980s to occupy villages along the Rahad of in vitro.9 In mice, the region on murine River in the heart of the endemic area in eastern Sudan. chromosome 11 that has conserved synteny with human Epidemiological and demographic details relating to the chromosome 5q23.3–q33 has been specifically shown study site are described in detail elsewhere.1 Long- to be involved in controlling later phases of cutaneous itudinal data from the study area collected between 1991 lesion development and visceralisation of L. major in and 1993 showed that 56% of treated VL patients that susceptible BALB/c mice.10 Extensive visceralisation and were followed up for 6 months developed PKDL.1 This death following L. major infection in susceptible BALB/c high rate of PKDL has been confirmed in more recent mice make it a better model of clinical visceral studies from the same study site, and from other study leishmaniasis than L. donovani infection in the mouse. sites within eastern Sudan (reviewed in Zijlstra and El- Interestingly, while the human IL9 gene maps to 5q31– Hassa15). Furthermore, PKDL in Sudan usually occurs q32, the mouse gene lies on chromosome 13 outside the within 0–6 months after treatment for VL, with a smaller region of conserved synteny between human 5q and percentage (8%16) occurring between 7 and 13 months mouse 11.11 Hence, if the same gene is responsible for after treatment. This is in stark contrast to where susceptibility to leishmaniasis in mouse and man, IL9 PKDL occurs 2–7 years after VL.15 Multicase families ought not to be the candidate susceptibility gene in the with VL with/without PKDL were ascertained from present study. In fact, allelic association data suggest that epidemiological and medical records of the Institute the IL4 locus itself may be the susceptibility gene in the of Endemic Diseases. Diagnosis was made on the basis Sudanese population. However, other candidate genes of clinical, parasitological and serological criteria as lie within this region, including genes encoding IL5 and described.1,15 All VL cases were monitored for a mini- IRF1 B130–190 kb proximal to IL4 and IL13 B360 kb mum of 2 years after treatment. Hence, cases classified as distal to IL4. Studies in mice using single and double VL alone were unlikely to have proceeded to the PKDL knockouts for IL4 and IL13, and transgenic reintroduc- clinical phenotype. Clinical subgroups are referred to as tion of IL13, indicate an independent role for IL13 in VL alone, VL with PKDL ( ¼ PKDL), or ‘VL alone plus susceptibility to L. major infection.12 Further studies are VL with PKDL,’ which we refer to as VL per se, that is, in progress using a large number of single nucleotide looking for a gene controlling underlying susceptibility

Genes and Immunity IL4 and IFNGR1 polymorphism and leishmaniasis H Mohamed et al 354 to VL. Ethical approval for this study was obtained from type associations, and relative risk estimates, were the Ethical Committee of the University of Khartoum. obtained by creating a ‘case–pseudo-control’ study Buccal swab DNA was prepared from 312 individuals where the ‘cases’ comprise the genotypes of the affected from 59 extended multicase VL per se families (67 nuclear offspring, and the ‘controls’ are the one to three other families; 177 affected individuals) with one to six affected genotypes (depending on whether phase is known or offspring per nuclear family. Nuclear families with one inferred) which the affected offspring might have case were always part of an extended multicase pedigree. received from the parents.6 The relative risks were For the clinical subtypes VL alone and PKDL, there were estimated using conditional logistic regression analysis, 53 nuclear families (123 affected individuals) and 26 again employing robust variance estimates to control for nuclear families (54 affected individuals), respectively. pedigree clustering and a Wald w2 test to indicate overall The numbers of nuclear families for the different significance for allelic or genotype associations. A subtypes do not add up to the total of 67 for VL per se stepwise logistic-regression procedure6 was used to since some families were multicase for both phenotypes. evaluate the relative importance of variants at the different sites within IL4. Likelihood ratio tests or Wald Genotyping and analysis of markers w2 tests (robust test) were used to compare models in Two polymorphisms17 were typed for IL4: a variable which the main effects for both loci are modeled with number tandem repeat (VNTR) in intron 2 designated one in which the main effects at the primary locus only IL4RP2 and a microsatellite in intron 3 designated are included. TDT tests allowed for grouping of alleles IL4RP1. Microsatellites in intron 4 of IL918 and intron 6 with expected transmission frequencies of o5%. Condi- of IFNGR119 were also typed. For the VNTR, PCR tional logistic regression analyses allowed for group- products were run on 1% agarose gels. Microsatellites ing of alleles/genotypes with frequencies of o5%. All were typed using FAM-, TAMRA- or HEX-labeled statistical packages employed were developed by David forward primers with PCR products analyzed by Clayton at the Cambridge Institute for Medical Research electrophoresis on 6% polacrylamide gels using an and are available at http://www-gene.cimr.cam.ac.uk/ automated sequencer (model ABI377, Applied Biosys- clayton/software/ tems) and the computer software genescan and genoty- per (Applied Biosystems). IL4RP1 had 17 (14 45%) alleles, IL9 10 (745%) alleles and IFNGR1 11 (945%) alleles, respectively, in this population. According to Acknowledgements the December 2001 freeze of the draft human genome We thank the people of Sudan who contributed to this (http://genome.ucsc.edu/), IL4 lies at position 134.2 Mb study. The work was supported by grants from the on 5q23.3, IL9 lies 3.85 Mb distal to IL4 at position 138 Mb Wellcome Trust and the UNDP/World Bank/WHO on 5q31. Using all unrelated individuals in the sample to 0 Special Programme for Research and Training in Tropical calculate the D statistic for LD, LD between IL4RP2 and Diseases (TDR). IL4RP1 was 0.454, between IL4RP2 and IL9 was 0.185 and between IL4RP1 and IL9 was 0.373. Values of D0 between 0.35 and 0.5 are considered to indicate weak LD. All markers were in Hardy–Weinberg equilibrium. References

Linkage analyses 1 Zijlstra EE, El Hassan AM, Ismael A, Ghalib HW. Endemic Nonparametric linkage analysis was performed in kala-azar in Eastern Sudan, a longitudinal study on the 20 incidence of clinical and subclinical infection and post-kala- Allegro using the Spairs scoring function with 0.5 azar dermal leishmaniasis. Am J Trop Med Hyg 1994; 51: weighting to take account of differences in family size. 826–836. Allegro reports allele sharing LOD scores and maximum 2 Kemp M, Kurtzhals JAL, Bendtzen K et al. Leishmania Z-scores for the likelihood ratio (Zlr). One-sided P-values donovani-reactive Th1- and Th2-like T-cell clones from in- associated with LOD and Zlr scores are used throughout. dividuals who have recovered from visceral leishmaniasis. For multipoint analysis, genetic distances between Infect Immun 1993; 61: 1069–1073. markers were entered in cM calculated on the basis of 3 Kemp K, Kemp M, Kharazmi A et al. Leishmania-specific T the physical map distance (1 Mb ¼ 1 cM). The fraction cells expressing interferon-gamma (IFN-gamma) and IL-10 of the total inheritance information extracted by the upon activation are expanded in individuals cured of visceral available marker data is indicated by the ‘information leishmaniasis. Clin Exp Immunol 1999; 116: 500–504. 4 Kharazmi A, Kemp K, Ismail A et al. T-cell response in human content’. Simulations performed in Allegro using a leishmaniasis. Immunol Lett 1999; 65: 105–108. standard set of six microsatellite markers (7–10 alleles; 5 Kurtzhals JA, Kemp M, Poulsen LK, Hansen MB, Kharazmi A, 0.73 heterozygosity) demonstrated that the 59 families Theander TG. Interleukin-4 and interferon-gamma production had 99% power to detect linkage up to a critical value by Leishmania stimulated peripheral blood mononuclear cells equivalent to an allele sharing LOD score of 3.00 from nonexposed individuals. Scand J Immunol 1995; 41: (P ¼ 0.0001) for PKDL and 100% power to detect 343–349. LOD ¼ 3.95 (P ¼ 0.00001) for the combined phenotype 6 Cordell HJ, Clayton DG. A unified stepwise regression of VL per se. procedure for evaluating the relative effects of polymorphisms within a gene using case/control or family data: application to HLA in type 1 diabetes. Am J Hum Genet 2002; 70: 124–141. Allelic association testing 7 Zwingenberger K, Harms G, Pedrosa C, Omena S, Sandkamp Family-based allelic association tests were performed B, Neifer S. Determinants of the immune response in visceral 21 using the TDT correcting for clustering at the pedigree leishmaniasis: evidence for predominance of endogenous level and nonindependence between sibs using a robust interleukin 4 over interferon-gamma production. Clin Immunol sandwich estimator for the variance. Allelic and geno- Immunopathol 1990; 57: 242–249.

Genes and Immunity IL4 and IFNGR1 polymorphism and leishmaniasis H Mohamed et al 355 8 Sundar S, Reed SG, Sharma S, Mehrotra A, Murray HW. 15 Zijlstra EE, El-Hassan AM. Leishmaniasis in Sudan. 4. Post Circulating T helper 1 (Th1) cell- and Th2 cell-associated kala-azar dermal leishmaniasis. Trans R Soc Trop Med Hyg cytokines in Indian patients with visceral leishmaniasis. Am J 2001; 95: S1/59–S1/76. Trop Med Hyg 1997; 56: 522–525. 16 Musa AM, Khalil EAG, Rahemm MA et al. Natural history of 9 Gessner A, Blum H, Rollinghoff M. Differential regulation of Sudanese post Kala-azar demal leishmaniasis: clinical im- IL-9-expression after infection with Leishmania major in suscep- munological and prognostic features. Trans R Soc Trop Med tible and resistant mice. Immunobiology 1993; 189: 419–435. Hyg 2003; in press. 10 Roberts M, Mock BA, Blackwell JM. Mapping of genes 17 Mout R, Willemze R, Landegent JE. Repeat polymorphisms controlling Leishmania major infection in CXS recombinant in the interleukin-4 gene (IL4). Nucleic Acids Res 1991; 19: inbred mice. Eur J Immunogenet 1993; 20: 349–362. 3763. 11 Mock BA, Krall M, Kozak CA et al. IL9 maps to mouse 18 Polymeropoulos MH, Xiao H, Rath DS, Merril CR. Dinucleo- chromosome 13 and human chromosome 5. Immunogenetics tide repeat polymorphism at the human interleukin 9 gene. 1990; 31: 265–270. Nucleic Acids Res 1991; 19: 688. 12 Matthews DJ, Emson CL, McKenzie GJ, Jolin HE, Blackwell 19 Altare F, Jouanguy E, Lamhamedi-Cherradi S et al. A causative JM, McKenzie AN. IL-13 is a susceptibility factor for relationship between mutant IFNgR1 alleles and impaired Leishmania major infection. J Immunol 2000; 164: 1458–1462. cellular response to IFNgamma in a compound heterozygous 13 Newport MJ, Huxley CM, Huston S et al. A mutation in the child. Am J Hum Genet 1998; 62: 723–726. interferon-g-receptor gene and susceptibility to mycobacterial 20 Gudbjartsson DF, Jonasson K, Frigge ML, Kong A. Allegro, a infection. N Engl J Med 1996; 26: 1941–1949. new computer program for multipoint linkage analysis. Nat 14 Belkaid Y, Hoffmann KF, Mendez S et al. The role of Genet 2000; 25: 12–13. interleukin (IL)-10 in the persistence of Leishmania major in 21 Spielman RS, McGinnis RE, Ewens WJ. Transmission test for the skin after healing and the therapeutic potential of anti- linkage disequilibrium: the insulin gene region and insulin- IL-10 receptor antibody for sterile cure. J Exp Med 2001; 194: dependent diabetes mellitus (IDDM). Am J Hum Genet 1993; 1497–1506. 52: 506–516.

Genes and Immunity