TNF-α promoter polymorphisms (G-238A and G-308A) are associated with susceptibility to Systemic Lupus Erythematosus (SLE) and P. falciparum malaria: a study in malaria endemic area

Tumor necrosis factor-α (TNF-α) is a proinflammatory cytokine associated with autoimmune and infectious diseases. Importance of TNF-α in P. falciparum malaria and systemic lupus erythematosus (SLE) have been demonstrated. However, association of functional promoter variants with SLE and malaria is lacking in malaria endemic population. A total of 204 female SLE patients and 224 age and sex matched healthy controls were enrolled in the study. Three hundred fourteen P. falciparum infected patients with different clinical phenotypes were included. TNF-α polymorphisms (G-238A & G-308A) were genotyped by PCR-RFLP. Plasma levels of TNF-α was quantified by ELISA. Heterozygous mutants and minor alleles of TNF-α (G-238A and G-308A) polymorphisms were significantly higher in SLE patients compared to healthy controls and associated with development of lupus nephritis. In addition, both promoter variants were associated with severe P. falciparum malaria. SLE patients demonstrated higher levels of plasma TNF-α compared to healthy controls. TNF-α (G-238A and G-308A) variants were associated with higher plasma TNF-α. In conclusion, TNF-α (G-238A & G-308A) variants are associated with higher plasma TNF-α levels in SLE patients residing in malaria endemic areas and could be a contributing factor in the development of SLE and susceptibility to severe P. falciparum malaria.

tnf-α (G-238A and G-308A) polymorphism are associated with SLE. As shown in Table 2, the prevalence of GA and minor allele ' A' for TNF-α (G-308A) polymorphism were significantly high in SLE patients compared to healthy controls (GA: P = 0.005, OR = 2.18; A: P = 0.002, OR = 1.99). Similarly, frequency of GA for TNF-α (G-238A) polymorphism were more frequent in SLE patients than healthy female controls (P = 0.008, OR = 2.02). Although minor allele for TNF-α (G-238A) polymorphism was more frequent in SLE patients compared to controls, the difference was not significant after Bonferroni correction (P = 0.032, OR = 1.69).

Distribution of TNF-α (G-238A and G-308A) polymorphism in patients with nephritis.
Since the study revealed a significant association between TNF-α promoter (G-238A and G-308A) polymorphisms and SLE, we analyzed the association of these polymorphisms with organ involvement. Lupus nephritis was the most important clinical phenotype observed and often associated with increased mortality in SLE. In our study, 41% of SLE patients had lupus nephritis. We categorized the patients into two broad groups: (1) Patients with lupus nephritis (LN+), and (2) patients without nephritis (LN−). As depicted in Table 3, GA genotype and minor allele (A) of TNF-α (G-238A) polymorphism was more frequent in patients with lupus nephritis (LN + ) compared to those patients without renal involvement (LN − ) (GA: P = 0.002, OR = 2.89;A: P < 0.001, OR = 2.92). However, distribution of TNF-α (G-308A) polymorphism was comparable among both groups.   (14) - Genotype-phenotype association of TNF-α (G-238A and G-308A) polymorphisms. Several studies have demonstrated a functional relevance of TNF-α promoter polymorphisms (G-238A and G-308A) with expression of TNF -α. We compared plasma levels of TNF-α among different genotypes of TNF-α (G-238A and G-308A). As shown in Fig. 2A,B, for both promoter polymorphism (G-238A and G-308A) the major genotype GG expressed significantly lower levels of plasma TNF-α compared to heterozygous mutant (GA)(P < 0.0001) and homozygous minor genotypes (AA) (G-238A: P = 0.005; G-308A: P = 0.002). Furthermore, we analyzed association of both promoter polymorphism with plasma levels of TNF-α in SLE patients and healthy controls independently (data not shown) and interestingly the observations remained consistent.
Association of TNF-α (G-238A and G-308A) polymorphisms with P. falciparum malaria. Association between TNF-α (G-238A and G-308A) polymorphisms and susceptibility to P. falciparum malaria have been well documented. In the present study, we enrolled 314 P. falciparum infected cases comprising of 103 uncomplicated cases and 211 severe malaria patients and genotyped for TNF-α (G-238A and G-308A) polymorphisms. As shown in Table 4, heterozygous genotype for TNF-α (G-238A) polymorphism and minor allele of G-308A polymorphism were more frequent in SM than UM (GA: P = 0.02, OR = 2.09; A: P = 0.02, OR = 2.05).
Severe malaria patients were further sub-categorized in to CM, MOD and NCSM and distributions of genotypes and allele were compared with UM cases. Results are shown in Table 4. Distributions of heterozygous genotype (GA), minor allele (A) were significantly higher in MOD compared to UM for both TNF-α promoter  Table 3. Prevalence of TNF-α (G-308A and G-238A) polymorphisms in lupus nephritis and non-nephritis SLE patients. Note: Data are no. (%) of participants unless otherwise specified. LN + = lupus nephritis; LN − = nonlupus nephritis; OR = odds ratio; 95% CI = 95% confidence interval. Patients with MOD displayed significantly higher levels of plasma TNF-α compared to NCSM (P = 0.004) and UM (P = 0.0002). In addition, a significant difference in mean plasma levels of TNF-α was also observed among CM and UM (P = 0.04) (Fig. 3B).
In silico analysis. We observed a significant correlation between TNF-α polymorphisms (G-238A and G-308A) and plasma levels of TNF-α. To validate the above findings, we analyzed functional relevance of these variants in silico. SNPs (rs1800629 and rs361525) were submitted to the FuncPred program and results obtained are shown in Supplementary Table 2. Both the SNPs were found to affect transcription factor binding site (TFBS). However, none of them affect miRNA binding site. SNP with ID rs1800629 was found to have a regulatory potential (RegPot) of 0.0401, which was also an indication of regulatory effects on binding and expression of gene targets.
RegulomeDB database has divided both the SNPs into two distinct categories (Category 1d and Category 4 as shown in Supplementary Table 3). rs1800629 showed RegulomeDB score of 1d and rs361525 which has minimal binding evidence (Category 4). The top ranked SNP rs1800629 had annotation for eQTL + TF binding + any motif + DNase peak and thus very likely to have regulatory functions.

Resampling analysis.
As the samples size investigated in the present study was smaller, we performed a resampling analysis and data are shown in Supplementary Table 4. TNF-α (G-308A and G-238A) variants and minor alleles were more frequent in SLE cases and lupus nephritis cases suggesting an important role of TNF-α variants with predisposition to SLE and clinical manifestations.

Discussion
TNF-α is an important cytokine in the pathogenesis as well as control of P. falciparum infection 16 . Therefore, higher levels observed in malaria infection is a protective phenomenon but very high levels can contribute to severity and mortality. The role of TNF-α in SLE is still conjectural but there are studies implicating it as a contributory factor in the pathogenesis based on experimental and associational studies 6,55 . It is important to understand the link between TNF-α and SLE in patients residing in malarial endemic areas. In the present study, we observed elevated plasma levels of TNF-α in SLE patients. Furthermore, TNF-α promoter polymorphisms (G-238A and G-308A) were significantly associated with higher plasma levels. These observations provide evidence of a possible role for TNF-α in the pathogenesis of SLE but the precise mechanism(s) is not yet known 6,55 . TNF-α is a pleotropic cytokine and acts at multiple levels 1 . In genetically susceptible SLE individuals, malaria might be a trigger for increased production of TNF-α, besides other cytokines, triggering a cascade of events contributing to the development of SLE 10,56-58 . www.nature.com/scientificreports www.nature.com/scientificreports/ P. falciparum malaria is predominantly endemic in the Eastern and Northeastern parts of India 59 . But it remains endemic in most parts of the country. Prevalence of TNF-α promoter polymorphisms (G-238A and G-308A) have been reported in Indian population. Distributions of TNF-α (G-238A and G-308A) genotypes were comparable with previous reports from different parts of the country 60 ,-South-West 61 , North 62 and North-West 63 regions. In most of the earlier reports 10,60-63 distribution of TNF-α promoter variants were compatible with HWE in healthy controls. In the present study distribution of TNF-α (G-308A) genotypes deviated from HWE in healthy females (P = 0.0007). The geographical area of Odisha is highly endemic for P. falciparum malaria which contributes to high mortality due to malaria in the country 64 . Deviation of genotype distribution has been attributed to several factors, and selection pressure remains one of the important causes 65 . The studied population is endemic to various infectious diseases other than malaria and could be the reason for increased selection pressure on host genome 66 . Interestingly, two independent studies from South 67 and North India 68 have reported higher prevalence of heterozygotes (GA) compared to homozygous (GG or AA). They have also shown a deviation of TNF-α G-308A genotypes from HWE. These abnormalities in observations could be due to genotyping methods (ARMS PCR/sequence specific primer PCR) which could give spurious results.
Role of TNF-α promoter variants in SLE have been widely investigated. A recent meta-analysis including 41 published studies worldwide, showed association of minor allele (A) and AA genotype of TNF-α (G-308A) polymorphism with susceptibility to development of SLE 11 . In the present analysis, we observed higher prevalance of heterozygous (GA) and minor allele (A) in SLE patients compared to healthy females, suggesting a possible    www.nature.com/scientificreports www.nature.com/scientificreports/ role of TNF-α (G-308A) polymorphism in susceptible to SLE. Similar observations have been reported in SLE patients from different geographical areas such as Brazil, Colombo, Mexico, North America, Spain, Taiwan 11 and South India 10 . However, contradictory results have also been reported in Portugese, Thai, Chinese, Italian, African American, Japanese and Argentenian populations 11 . These discrepancies have been attributed to ethnicity of subjects enrolled for case-control studies and further supported by ethnicity related meta-analysis which revealed significant link between allele ' A' with predisposition to SLE in Europeans, Asians and South and North Americans but surprisingly not in African population 11 . In the present study, TNF-α (G-238A) heterozygous and minor allele (A) were also associated with susceptibility to SLE and it corroborated with other observations. Although the exact mechanism related to TNF-α polymorphism and SLE is yet to be understood, results of the present study and previous reports across the world indicates a strong association of TNF-α promoter variants, higher expression of TNF-α m-RNA and elevated levels of plasma TNF-α in SLE patients from malaria endemic regions.
We analysed the possible association of TNF-α polymorphisms with clinical manifestations of SLE, namely lupus nephritis which is one of the major clinical phenotypes linked to SLE mortality. We observed that heterozygous (GA) and minor allele ' A' of TNF-α (G-238A) polymorphism were significantly associated with lupus nephritis. These observations have been corroborated by a recent study on Chinese SLE patients 11 . However it contradicts an observation from South Indian population 10 and a recent meta-analysis 11 . Furthermore, patients with lupus nephritis had higher levels of plasma TNF-α than those without nephritis.
Functional relevance of TNF-α promoter polymorphisms (G-308A and G-238A) have not been widely investigated. Minor allele for TNF-α (G-308A) polymorphism has been observed to enhance the binding of transcription factors and is associated with increase in mRNA production compared to major allele (G) 69 . In vitro stimulation of peripheral blood mononuclear cells (PBMC) derived from heterozygous subjects (GA) with lipopolysaccharide, displayed higher TNF-α than those of wild type individuals (GG) 70 . Furthermore, elevated plasma TNF-α has been associated with mutants for TNF-α (G-308A) polymorphism 10 . In the present study, we observed higher plasma levels of TNF-α in GA and AA genotypes compared to GG, corroborating earlier observations. Interestingly, other TNF-α promoter polymorphism (G-238A) also revealed similar results: mutants (GA and AA) were associated with higher plasma TNF-α than wild type (GG), which corroborates with an earlier report 71 . Furthermore, we performed in silico analysis which revealed regulatory effect in binding of transcription factors and enhanced expression of TNF-α gene. Results of the present investigation and earlier reports demonstrate signifcant regulatory role of promoter polymorphisms.
Investigations on possible link between malaria and SLE are limited and contradictory. Epidemiological data have shown lower prevalence of autoimmune diseases in areas where malaria incidence is high 72 . However, in an earlier observation, we have demonstrated protection against severe malaria and malarial death in complement receptor 1 variants and concluded a possible reason for higher prevalence of CR1 mutants in malaria endemic areas 14 . We had also observed that CR-1 mutants are susceptible to development of SLE and lupus nephritis since they expressed lower surface CR1 which affects clearance of apoptotic debris 73 . Furthermore, similar association of FcγRIIb variant (codon 232) with susceptibility to SLE and protection against P. falciparum malaria has been reported 74 . Lower parasitaemia and minimal clinical severity has been reported in FcγRIIb deficient mice when infected with non-lethal murine plasmodium strain indicating protective nature of the truncated or deficient FcγRIIb against malaria 74 . This observation has been further supported by higher prevalence of FcγRIIb codon 232 mutant in African and Asian population when compared to other populations across the world where malaria is endemic. The results of the present study and earlier reports collectively demonstrate that certain genotypes are beneficial in protecting humans against P. falciparum malaria and are highly prevalent in malaria endemic areas. Unfortunately, subjects genetically susceptible to SLE and residing in malarial endemic areas have a greater chance to develop SLE compared to those residing in non-endemic areas.
In conclusion, elevated plasma TNF-α is observed in SLE patients and associated with clinical severity. Furthermore, promoter variants of TNF-α gene, associated with higher TNF-α expression, were more prevalent in SLE patients. TNF-α is essential for clearance of malarial parasites 49 and people residing in malarial endemic areas often produce optimal levels of TNF-α 19 which could be helpful in combating the infection. It could also be one of the contributory factors for inducing SLE in genetically susceptible individuals. Further studies from other malarial endemic areas in the world are important to validate our findings.

Materials and Methods
Subjects. Gender wise analysis has been recommended in numerous earlier reports of genetic association studies 73,75,76 . SLE is a chronic inflammatory autoimmune disorder and mostly prevalent in females 77 . In the present study, we enrolled 428 female subjects (224 healthy controls and 204 SLE patients) to investigate possible association of TNF-α polymorphism in SLE. Patients of SLE were diagnosed based on the revised American College of Rheumatology (ACR) classification criteria 78 and analyzed based on various clinical manifestations (Table 1). In addition, we enrolled 314 P. falciparum infected patients who reported to or were admitted to Department of Medicine, SCB Medical College, Cuttack, Odisha. Clinical categorization of falciparum infected patients were performed as described earlier 14,15,73,79 . Healthy females, age matched and residing in the same geographical areas, with no prior history of autoimmune disorders were enrolled as controls (HC). About 5 ml blood was collected from each participant. Plasma was separated and stored at −80 degrees centigrade for later use. The study was approved by the Institutional Human Ethics Committee of Central University of Jharkhand, India and S.C.B. Medical College Cuttack, Odisha, India. Informed written consent was obtained from each patient. The study was conducted in accordance with methods approved by IHECs. www.nature.com/scientificreports www.nature.com/scientificreports/ polymerase chain reaction followed by restriction fragment length polymorphism method as described earlier (Galbraith et al. 1998).
tnf-α quantification. The plasma TNF-α levels in SLE patients, healthy controls and P. falciparum infected cases were quantified by enzyme linked immunosorbent assay (ELISA) kit (eBiosciences) according to manufacturer's instructions.
Non-coding SNP functional analysis. In order to recognize the effect of SNPs in non-coding regions, tools predicting probable functional effect of SNPs at transcription factor binding sites (TFBS), Intron/exon border consensus sequences (splice sites), Exonic splicing enhancers (ESEs), and miRNA binding were utilized. SNPinfo (FuncPred) and RegulomeDB offer a pool of functional information using series of tools. The SNPs functionality was defined by SNPinfo (FuncPred) (https://snpinfo.niehs.nih.gov/snpinfo/snpfunc.php) web server 80 , which helps in selecting SNPs for genetic association studies. For the current study, two SNPs (rs1800629 and rs361525) were uploaded for batch analysis with the default settings. The output information was a list of SNPs with possible functional effect.
To supplement SNP ranking, SNPs were further analyzed by RegulomeDB (http://regulomedb.org/) 81 . RegulomeDB categorizes variants into six categories ranging from 1 to 6, where category 1 variants are 'likely to affect binding and linked to expression of a gene target' , category 2 variants are 'likely to affect binding' , Category 3 variants are 'less likely to affect binding' , and Category 4, 5 and 6 variants have 'minimal binding evidence' . RegulomeDB also allocates a score of 7 for variants with no annotation data available. dbSNP rsIDs were utilized as input for the current study.
Statistical analysis. Genotype and allele distribution among different clinical categories was compared by Fisher's exact test. P value less than 0.02 was taken as significant (Bonferroni correction for two SNPs 0.05/2 = 0.02). The mean plasma levels of TNF-α in SLE patients and healthy controls was compared by student's t test and analysis of variance (ANOVA) was employed for study difference in plasma TNF-α in different clinical categories of P. falciparum malaria. The association of TNF-α (G-238A and G-308A) genotypes with plasma TNF-α levels were analyzed by unpaired 't' test or ANOVA followed by an appropriate post-test. Graph Pad Prism 5.01 software was used for these statistical analyses. Haplotype analysis was performed by SNAP Stats online tool. Resampling analysis was performed by bootstrap method in Microsoft excel sheet attached as supplementary file-1.

Data Availability
The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request.