After 9 years of prospective observation by household visits every other day, 220 of the original 289 children remained enrolled in the cohort. Of these children, 185 represented nuclear families and were included in this genetic analysis. Ninety percent of the children had been infected at least once with E. histolytica
in 9 years. Forty percent of the children were either malnourished or stunted at baseline (ref. 20
, Supplemental Table 1, and Supplemental Figure 1; supplemental material available online with this article; doi:
We genotyped 15 SNPs in leptin (LEP) and 77 SNPs in the leptin receptor (LEPR); 1 SNP showed significant deviation (P < 0.01) from Hardy-Weinberg equilibrium and was excluded. Five SNPs in LEPR and 5 SNPs in LEP were either monomorphic or had a minor allele frequency (MAF) of less than 5% in this population and were not considered in further analyses. A test of association was performed between the remaining SNPs and E. histolytica infection, and 6 SNPs of LEPR had an empiric P value of less than 0.05 (Supplemental Figure 2 and Supplemental Table 2). All markers showing significant associations were in LEPR, located between introns 5 and 19. A single haplotype block (haplotype block 7) including 12 SNPs spanned intron 2 through intron 6, of which rs4655537 in intron 6 was significant for the association both as a single SNP and as part of this haplotype (Supplemental Figure 2A).
To determine whether this region harbored a potential causal variant, known coding SNPs within this significant haplotype block or near individual significant SNPs were identified. Five nonsynonymous SNPs were genotyped: rs1137100 (K109R, exon 4), rs1137101 (Q223R, exon 6), rs13306526 (I503V, exon 11), rs8179183 (K656N, exon 15), and rs34499590 (T699M, exon 15). Two SNPs, rs13306526 and rs34499590, were both monomorphic and could not be analyzed, which is consistent with HapMap data for individuals of European ancestry (CEU). No significant association was seen for rs1137100, K109R, with risk of amebiasis (odds ratio [OR] = 0.98; P
= 0.674). However, a strong association was identified with coding SNP rs1137101, Q223R, which appeared to have a dominant effect. Individuals segregating for the G allele, which translates to arginine at position 223 of LEPR
, were nearly 4 times more likely to have an E. histolytica
infection than those homozygous for the ancestral A allele (glutamine; Q223) (OR = 3.91, P
= 0.007, 95% CI 1.44, 10.6, empirical P
= 0.009, population attributable risk = 69%). This association persisted after adjusting for age, sex, and malnutrition at baseline (OR = 3.98, P
= 0.009, 95% CI 1.4, 11.3). In addition, we adjusted for HLA class II alleles DQB1*0601 and DQB1*1501, which were previously shown to be associated with infection in this same cohort (21
), and the strength of the association remained (OR = 4.19, P
= 0.01, 95% CI 1.39, 12.6). When we stratified the outcome by number of E. histolytica
infections, the risk attributable to 223R was pronounced among those with greater than 3 infections (Table ). Since this Q223R LEPR
allele had been associated with adiposity in some but not all human studies (reviewed in ref. 22
), we compared the baseline malnutrition and stunting of children with different genotypes at this locus and saw no association (Supplemental Table 3). We concluded the effect of the Q223R leptin receptor polymorphism on amebiasis was not confounded by the nutritional status at baseline.
Association of leptin receptor Q223R polymorphism with susceptibility to amebiasis
To determine whether this genetic variant altered not only the risk of having an E. histolytica infection, but also the time to an infection, we performed a Kaplan-Meier analysis (Figure ). The median time from study entry to an infection for individuals homozygous for the glutamine allele (QQ) was 2.56 years, compared with 1.51 years for those with only 1 glutamine (QR) allele and 1.01 years for individuals with no glutamine allele (RR) (log rank P = 0.002). The estimated hazard ratio for individuals segregating for at least one copy of the arginine allele (QR or RR) was 1.82 (P = 0.002, 95% CI 1.1.24–2.67). To insure these effects were not confounded by nutritional status, we compared the time to infection stratified on malnutrition status at baseline. The median time to infection for those malnourished at baseline was 1.23 years compared with 1.61 years for those not malnourished. This difference was not statistically significant (log rank P = 0.17, RH = 1.25, P = 0.17, 95% CI 0.91–1.72). We concluded that the 223Q allele of LEPR was associated with protection from E. histolytica infection and a delay in time to E. histolytica infection.
Kaplan-Meier analysis of time to E. histolytica infection for the RR, QR, and QQ variants.
The importance of LEPR was also seen in an ALA cohort in Bangladesh. ALA results from extension of E. histolytica infection from the intestine. Thus, it was reasonable to expect the Q223R LEPR polymorphism might also affect susceptibility to liver abscess if it predisposed children to E. histolytica infection. Because ALA occurs predominantly in men, we used 86 adult male cases and 161 adult male blood donor controls. ALA was nearly 2 times more likely in individuals carrying RR compared with those who were QR or QQ (OR = 1.9, P = 0.021, 95% CI 1.10–3.27) (Table ). Interestingly, the association with liver abscess was restricted to those segregating for 2 arginines (RR), whereas susceptibility to E. histolytica infection in the child cohort was conveyed by even a single copy of the arginine allele (QR or RR).
The allele frequencies for rs1137101 were different among children who never had an E. histolytica infection (0.31) and those with ALA (Figure ). Controls from the ALA study had frequencies of arginine (G nucleotide, R amino acid) that were not significantly different from those observed in the overall child cohort (0.52) or in the HapMap phase II Gujarati Indians from Houston, Texas (GHI) sample (0.52). These controls, as expected, reflect the general population allele frequencies, indicating that this common polymorphism is not unique to this study population. However, frequency of the R allele was substantially increased in liver abscess cases (0.62). These data suggest there is a dose effect with 1 (or more) arginines increasing susceptibility to common infection acquired during childhood and 2 arginines increasing susceptibility to ALA among adults. Figure B also outlines the allele frequencies for each of the populations worldwide. Differences in allele frequencies for rs1137101 are observed with the most striking differences between Indo-European countries and East Asian countries where the allele is nearly fixed. Thus, the Q223R LEPR polymorphism appears to be associated with susceptibility to amebiasis in 2 independent studies, and differences in allele frequencies may result in differences in susceptibility to amebiasis and ALA via this pathway.
The allele frequencies for the rs1137101 allele.
We assessed the effect of this SNP on serum concentrations of leptin and soluble leptin receptor. The analysis was stratified by gender and BMI at the time serum was collected. Leptin receptor concentrations were decreased for both boys and girls with the RR genotype (Supplemental Figure 3). This effect was most pronounced in children with BMI less than the median (BMI < 16). The logarithm of serum leptin receptor concentrations was statistically significant for the 3 genotypes (ANOVA, P
= 0.0009) and remained significant when stratified by sex (boys, P
= 0.02; girls, P
= 0.04) or BMI category (low BMI < 16, P
= 0.002; high BMI > 16, P
= 0.04). No trend in circulating leptin concentrations by genotype was detected (ANOVA, P
= 0.065; Supplemental Figure 3), although differences by sex (P
< 0.00001) and BMI levels (P
< 0.00001) without regard to genotype were present, as expected due to known effects of sex and adiposity on circulating leptin concentrations. Soluble leptin receptor levels have been shown to play a role in levels of total leptin and free leptin, although the relationship is complex (23
). These data indicate that differences in leptin receptor expression rather than circulating leptin levels were associated with the SNP.
Recent studies have shown that both adipose tissue and circulating peripheral blood levels of LEPR
are highly heritable in expression-QTL studies (25
). We evaluated mRNA levels for LEPR
in resting PBMCs, using quantitative RT-PCR (qRT-PCR) for both the long and short leptin receptor isoforms. Individuals segregating for 2 copies of the arginine allele (RR) had a significant decrease in expression of the long isoform (LEPR isoform 1 as measured by Lepr_vb_2 primer) compared with those segregating for least 1 glutamine (QR, P
= 0.015 or QQ, P
= 0.027), respectively (Supplemental Figure 4).
To determine whether the Q223R polymorphism specified the altered phenotype, we challenged 129P3/J mice carrying the QQ, QR, and RR alleles with E. histolytica. Mice with at least 1 copy of arginine (R) were more susceptible to infection, consistent with the human genotype-phenotype relationship (Figure A). In addition, the infected RR and QR mice exhibited significantly more severe epithelial ulceration compared with QQ mice that were successfully infected (Figure , B and C). These data demonstrate that a single copy of the 223R allele (QR) significantly increases susceptibility to amebiasis and disease severity in animals that are otherwise genetically identical.
The 223R allele of the leptin receptor increases susceptibility to intestinal amebic infection and mucosal destruction in the murine model of intestinal amebiasis.
It has been demonstrated that E. histolytica
kills host cells by activating caspase-3, which leads to host cell apoptosis and allows amebic invasion (26
). Expression of the leptin receptor is known to protect both immune cells and epithelial cells from apoptotic stimuli (6
). We hypothesized that the R223 allele of the leptin receptor was less effective than the Q223 allele in protecting the intestinal epithelium from apoptosis induced by E. histolytica
. To test this, cecal sections from E. histolytica
–challenged QQ, RR, and QR mice were stained for activated caspase-3. The RR and QR mice displayed significantly more active caspase-3 than the QQ mice in intestinal epithelial cells (IECs) (Figure , A and B) and mononuclear cells (Figure C), suggesting a greater level of apoptosis induced by E. histolytica
in mice with the R allele.
Increased caspase-3 activity and decreased expression of phosphorylated Akt and antiapoptotic genes in the ceca of RR and QR mice.
PI3K/Akt signaling is implicated in leptin receptor–mediated antiapoptotic and proproliferative effects by upregulating the expression of survival factors (i.e., Bcl-2, cyclin D3, etc.) (6
). We examined phosphorylated Akt expression in the cecum from infected and uninfected QQ, RR, and QR mice and observed a significant decrease in Akt phosphorylation in RR and QR mice relative to QQ littermates regardless of their infectious status. Representative sections showed minimal phospho-Akt staining in the infected RR cecum with mucosal ulcerations and invading trophozoites and uninfected RR cecum with mucosal hyperplasia (Figure G). In contrast, pAkt was abundantly present in most of the QQ animals throughout the crypts and the villi, as well as within the lymphoid compartment (Figure H). RR and QR mice also had decreased expression of downstream survival factors Bcl-2 and cyclin D3 compared with their QQ littermates (Figure , D–F). These data support a role for the PI3K/Akt pathway in controlling amebic-induced IEC apoptosis and suggest that the PI3K/Akt pathway is impaired by the Q223R polymorphism.