IFNG and IFNGR1 gene polymorphisms and susceptibility to post kala-azar dermal leishmaniasis in Sudan

Abstract

Post kala-azar dermal leishmanaisis (PKDL) in Sudan is associated with elevated interferon-γ. To study interferon-γ pathways in PKDL, we genotyped 80 trios from the Masalit ethnic group for polymorphisms at −470 ins/delTT, -270T/C, -56T/C and +95T/C in IFNGR1, and at −179G/A and +874T/A in IFNG. No associations occurred at IFNG. Global association with haplotypes comprising all 4 markers at IFNGR1 (χ²_10df = 21.97, P=0.015) was observed, associated with a significant (χ²_1df=4.54, P=0.033) bias in transmission of the haplotype insTT T T T, and less (χ²_1df=5.59, P=0.018) than expected transmission of insTT C C C. When compared with data on malaria associations from The Gambia, the results suggest a complex pattern of haplotypic variation at the IFNGR1 promoter locus associated with different infectious disease in African populations that reflect the complex roles of IFN-γ in parasite killing versus inflammation and pathogenesis.

In Sudan more than 50% of cases successfully treated for visceral symptoms caused by Leishmania donovani go on to develop post kala-azar dermal leishmaniasis (PKDL).¹ Interferon- γ (IFN-γ) is a key T helper 1 and natural killer cell cytokine. Its activates macrophages to kill parasites,² exerting its effects through its receptor IFNGR1. IFNγ is present in PKDL lesions, and PKDL is associated with elevated antigen-specific IFNγ T cell responses.^3,4

IFNG and IFNGR1, located on chromosomes 12q15 and 6q23.3, are important candidate genes for PKDL. Rare mutations in the coding region of IFNGR1 are the cause of Mendelian inheritance of susceptibility to non-tuberculous mycobacteria in consanguineous families.^5-7 Compound heterozygosity for rare alleles at IFNGR1 was also associated with disseminated infection with non-tuberculous mycobacteria, and impaired cellular response to IFN-γ, in a child born to nonconsanguinous parents ⁸. In mapping this deficiency, a CA repeat in intron 5 of IFNGR1 was identified and designated FA1 ⁸. We found FA1 was linked and associated specifically with PKDL, and not visceral leishmaniasis (VL) per se, in the Masalit ethnic group in Sudan ⁹.

To understand natural population variation at IFNGR1, sequence analysis of the promoter locus between -1400 and +100 of the translational start site was undertaken in 140 Gambian chromosomes.¹⁰ This identified 3 T-to-C single nucleotide polymorphisms (SNPs) at +95, -56 and -270 nucleotides, and a TT deletion at -470/-471 nucleotides. Variant alleles at the -56 bp (T/C alleles at frequency 0.5) and +95 bp SNPs (C allele 0.49) were common, and heterozygotes for the -56 bp SNP were protected against cerebral malaria and death from cerebral malaria.¹⁰ Variant alleles at -270 (C allele 0.02) and -470 (delTT allele 0.11) were less frequent in the Gambia. Nevertheless, for the Mandinka ethnic group, the double deletion at -470 bp was protective against severe malaria. No associations were observed for any of these polymorphisms with pulmonary tuberculosis caused by Mycobacterium tuberculosis in the Gambia,¹¹ or at -56 T/C in Croatians.¹²

Two possible functional polymorphisms occur at IFNG. The +874 A/T SNP in intron 1 lies within a nuclear factor κB binding site, and the variant T allele is in complete linkage disequilibrium (LD) with a CA repeat allele associated with high IFN-γ production.¹³ A -179 G/A SNP in the proximal promoter also controls gene transcription.¹⁴

The aim of the present study was to determine whether further information on the association between PKDL and the IFN-γ pathway could be obtained by genotyping additional SNPs at IFNGR1, as well as the putative function-associated SNPs at IFNG. Samples were collected from two adjacent villages, Um Salala and El-Rugab, in the highly endemic area of eastern Sudan. All individuals were from the Masalit tribe, which is highly susceptible to VL and PKDL. Details of the site and diagnostic criteria for PKDL were described previously.⁹ Ethical approval for this study was from the Local Research Ethics Committee of the University of Khartoum. DNA from buccal swabs from 80 trios was genotyped for the -470 ins/del, -270 T/C, -56 T/C (rs2234711) and +95 (rs7749390) SNPs at IFNGR1, and the -179 G/A (rs2069709) and +874 (rs2430561) SNPs at IFNG. Tests using genetically independent family members confirmed that all four IFNGR1 polymorphisms, and the two IFNG gene SNPs, were in HWE (P>0.05). As determined for Gambian populations (0.49 and 0.56, respectively),¹⁰ the frequencies of the minor allele in IFNGR1 +95C and IFNGR1 −56C were high (0.45 and 0.46, respectively). Frequencies for minor alleles -270C and -470delTT were 0.05 and 0.03 in the Masalit, compared to 0.02 and 0.11 in Mandinka Gambians. Eleven haplotypes that were present in the Masalit (Table 1), 5 of which were not reported at frequencies >1% for the Mandinka.¹⁰ Of interest is the presence in the Masalit of two haplotypes, insTT T C T at 16% and insTT T T C at 16%, that were present at 3% and 2% in the Mandinka.¹⁰ Minor allele frequencies at IFNG -179G/A and +874A/T were 0.06 and 0.1 in the Masalit.

Table 1. Estimated haplotype probabilities at the IFNGR1 promoter.

Haplotype identifier	Haplotype	Estimated haplotype probabilities
		Masalit, Sudan	Mandinka, Gambia
1	insTT T T T	0.33	0.47
2	insTT T C T	0.16	0.03
3	insTT T C C	0.23	0.37
4	insTT T T C	0.16	0.02
5	delTT T C C	0.004	0.08
6	insTT C C C	0.03	0.02
7	insTT C T T	0.025	0
8	insTT C T C	0.01	0
9	insTT C C T	0.02	0
10	delTT T T C	0.02	0
11	delTT T C T	0.011	0

DNA from buccal swabs from 80 trios was amplified using Multiple Displacement Amplification (Molecular Staging, USA; now supplied by Qiagen). The -470 ins/del, -270 and +95 (rs7749390) SNPs at IFNGR1, and the +874 SNP (rs2430561) at IFNG, were genotyped by Amplification Refractory Mutational System PCR using primers and probes as described.¹⁰,¹³ Restriction fragment length polymorphisms were used to genotype the IFNGR1 -56 T/C SNP (rs2234711) and the IFNG -179 G/A SNP (rs2069709). PCR primers were: IFNG -179 forward 5′-ATC AAT GTG CTT TGT GAA TGA A-3′ and reverse 5′-CCG AGA GAA TTA AGC CAA AGA -3′ ; IFNGR1 -56 forward 5′ - TCA GGC TCC AAG ACA ACC AGG -3′ and reverse 5′-CCA CGG AGC CCC AGT CTC G-3′. PCR products digested with Ava11 (IFNG −179) and Bts1 (IFNGR1 −56) were visualised on 1.5% agarose gels. Tests for deviation from Hardy Weinberg equilibrium (HWE) for genotype data, and LD (D’ values range 0 for no LD to 1 for full LD) between markers, were performed within STATA v8.0 (Electronic Database Information 1) using the GenAssoc package available from (Electronic Database Information 2). Haplotype probabilities were estimated on the basis of maximum likelihood calculations performed within the TRANSMIT program (Electronic Database Information 3), using data from the Masalit families with PKDL from Sudan, and as previously reported¹⁰ from Mandinka families with severe malaria from Gambia. Only haplotypes with an estimated probability of >1% were reported for the Mandinka. Markers are shown in chromosomal order (i.e. IFNGR1 -470, -270, -56 and +95).

Single marker transmission disequilibrium tests carried out using TRANSMIT (Table 2) showed a significant bias in transmission of alleles at IFNGR1 -270T/C (χ²_1df=6.98, P=0.0087) from heterozygous parents to PKDL offspring, but not for other IFNGR1 or IFNG (data not shown) markers. Conditional logistic regression showed that PKDL was associated with the common T allele at IFNGR1-270 (odds ratio 3.67; 95% CI 1.02-13.45; P=0.045), while the minor C allele was associated with protection (odds ratio 0.27; 95% CI 0.08-0.98; P=0.046). Considering the complex haplotypic structure observed for the 4 promoter locus polymorphisms in African populations, we also looked for haplotype associations using TRANSMIT (Table 2). Haplotype analysis using all 4 markers showed a significant (P=0.015) global χ² (21.97) on 10df, associated with a significant (χ²_1df=4.54, P=0.033) bias in transmission of haplotype insTT T T T but not in transmission of the haplotypes insTT T C C or insTT T C T. This suggests involvement of a functional variant in PKDL susceptibility that is in LD with the -56 bp SNP (underlined), even though single marker analysis did not show a significant association with PKDL disease. The global haplotype association for all 4 markers was also associated with significantly (χ²_1df=5.59, P=0.033) less than expected transmission of haplotype insTT C C T that is at low frequency (2%) in the Masalit. Significant associations (Table 2) for 3 marker (e.g. insTT C C; insTT C T) and 2 marker (e.g. C C and C T for -270/-56) haplotypes were also associated with less than expected transmission of haplotypes carrying the -270 C allele (underlined), i.e. independently of variation at the -56 bp SNP. The involvement of more than one etiological variant across the IFNGR1 promoter locus in determining susceptibility to, or protection against, PKDL in the Masalit population is consistent with data showing that the -56 bp SNP is in LD with the -470 bp SNP (D’=0.73) but not the -270 bp SNP (D’=0.13) or the +95 bp SNP (D’=0.22), and that the -270 bp SNP is not in LD with the +95 bp SNP (D’=0.26).

Table 2. Haplotype association testing using TRANSMIT.

Haplotype groups	IFNGR1 polymorphisms
	-470ins/del	-270 T/C	-56 T/C	+95 T/C
4 markers	χ2 = 21.97; df=10; p=0.015
3 markers	χ2 = 8.34; df=5; p=0.14
3 markers		χ2 = 16.86; df=7; p=0.018
2 markers	χ2 = 8.30; df=2; p=0.016
2 markers		χ2 = 6.62; df=3; p=0.085
2 markers			χ2 = 10.58; df=3; p=0.014
1 marker	χ2 = 0.94; df=1; p=0.33	χ2 = 6.57; df=1; p=0.010	χ2 = 0.01; df=1; p=0.91	χ2 = 0.27; df=1; p=0.60

Family based transmission disequilibrium tests for association were performed using TRANSMIT (available at Electronic Database Information 3) which allows for missing parental data and the analysis of extended haplotypes. Global χ² statistics ar shown for associations between single markers or all possible IFNGR1 haplotypes for the 4 polymorphisms -470, -270, -56 and +95 and PKDL in the Masalit ethnic group in Sudan.

IFN-γ provides first line defense against infectious diseases by activating macrophages to kill intracellular pathogens. IFN-γ is also part of an inflammatory response that, if exaggerated, can be associated with adverse pathology. In VL, IFN-γ production is low at diagnosis and increases with treatment and clinical cure from visceral symptoms in all patients.⁴ However, peripheral blood mononuclear cells from patients that go on to develop PKDL show elevated proliferative and IFN-γ responses to Leishmania antigen compared to patients that do not, with no differences in IL-4, IL-5 or IL-10.⁴ Our study shows that this elevated IFN-γ response in PKDL patients does not appear to be associated with polymorphisms at the IFNG gene, despite studies showing high IFN-γ associated with the +874 bp T allele and low IFN-γ with the A/A genotype.¹³ Failure to detect associations with PKDL could be due to low frequency for variant alleles at IFNG SNPs in the Masalit.

PKDL lesions contain parasites and antigen, and an inflammatory infiltrate comprising macrophages, lymphocytes, and plasma cells.³ IL-10 is the most prominent cytokine in lesions, but IFN-γ is always present and IL-4 occurs in most lesions.³ One reason for failure of IFN-γ to activate macrophages to kill Leishmania in lesions could be inhibitory IL-10. Another, possibly related to the action of IL-10, could be failure to adequately express IFNGR1 due to regulatory polymorphisms in the gene. Previously⁹ we demonstrated association between different alleles at the non functional FA1 IFNGR1 microsatellite⁸ and susceptibility to, or protection from, PKDL. Using a larger sample of PKDL trios we genotyped the four promoter locus polymorphisms at IFNGR1, some or all of which may regulate IFNGR1 expression. These were originally identified in Gambians,¹⁰ where the rare variant allele -470 delTT was associated with all severe malaria in the Mandinka ethnic group, while heterozygosity at the -56 T/C SNP was associated with protection from cerebral malaria. In our study we found no direct association between PKDL and either -470 bp or -56 bp polymorphisms. However, markedly different frequencies of certain haplotypes across the four promoter locus polymorphisms were observed when comparing the Masalit from Sudan with the Mandinka from Gambia. In particular, we found two haplotypes, insTT T T C and insTT T C T that were rare (2% and 3%, respectively) in the Mandinka were more common (both 16%) in the Masalit. Whereas the common haplotype insTT T T T was significantly associated with PKDL, insTT T T C T and insTT T C C showed no bias in transmission to PKDL patients, suggesting that a functional variant in LD with the -56 bp SNP (at position 3 in these haplotype) might contribute to susceptibility to PKDL even though we found no direct association with the -56 bp SNP itself. In addition, we found that the rare -270 C allele, and haplotypes that contained this variant, were associated with protection from PKDL independently of variation at the -56 bp SNP. This suggests that a complex pattern of haplotypic variants exists in different African ethnic groups that may relate to differing exposures and susceptibilities to infectious disease. In this respect it is interesting that the Masalit have a larger range of haplotypes involving the -270 C allele, that are protective for PKDL, compared to the Mandinka.

At present there is little data to determine whether these promoter locus variants are functional in determining IFNGR1 expression. Although experimental studies¹⁵ show differences in luciferase reporter gene activity associated with variants at -56 T/C, with lower activity for the C allele, our studies suggest that this SNP itself does not itself contribute to PKDL but could be in LD with a functional SNP on different haplotypes. Malaria is endemic in the region of Sudan where the present work on PKDL took place, but transmission is highly seasonal and the incidence of severe malaria is low. In the Masalit, the -56 bp C and T alleles are both common, which could relate to heterozygous advantage against cerebral malaria shown for this SNP¹⁰. However, Jüliger and coworkers¹⁵ also failed to find a direct association between this SNP and either mild or severe malaria in Gabonese children, suggesting that the earlier association with protection from cerebral malaria¹⁰ might also be due to haplotypic variation between Gambians and Gabonese, and a functional variant in LD with this SNP rather than -56 T/C itself. Further experimental analysis will be important in elucidating the role of regulatory polymorphisms at IFNGR1 in determining susceptibility to these important clinical manifestations of protozoan parasitic diseases throughout Africa.