REL and BHLHE40 Variants Are Associated with IL-12 and IL-10 Responses and Tuberculosis Risk

Tuberculosis (TB) is a leading cause of death from infection worldwide. The current bacillus Calmette–Guérin (BCG) vaccine remains the only approved vaccine against TB despite its partial and variable effects across populations (1). Vaccine efforts are hampered by a lack of understanding of the immune correlates of protection (2). Understanding the factors required to induce effective, long-lasting immunity to infections may provide tools to improve TB vaccines.

Twin, Mendelian, linkage, genome-wide association and candidate gene studies suggest that genetic factors influence susceptibility to TB (3, 4). Multiple clinical TB phenotypes show a high degree of heritability, including host susceptibility to pulmonary TB (PTB) (5–10), TB meningitis (11, 12), and latent TB infection (13–17). Common genetic variation influences the cellular innate immune response to Mycobacterium tuberculosis. Multiple studies demonstrate the impact of genetic variation on innate immune cellular distribution and cytokine responses (18–21). Quantitative trait loci of gene expression demonstrate immune cell–specific effects (20, 22). Recent advances permit the evaluation of innate immune cytokine responses from rare cell populations (23, 24). Variants that influence functional responses in immune cells of interest represent attractive secondary traits that can be correlated with TB susceptibility, and these correlations may provide insight into genetic mechanisms of disease susceptibility (25).

Dendritic cells (DCs) are essential for host defense against mycobacteria (7, 15, 26), and IL-10 and IL-12, which are produced in large quantities by DCs, are particularly important cytokines that shape immune responses to M. tuberculosis. Genetic variation in the IL-10 gene has been modestly associated with TB susceptibility (27–29). However, rare inborn mutations in IL-12 signaling genes lead to widely disseminated mycobacterial disease (30, 31).

Recently, several novel mediators of IL-12 and IL-10 signaling were identified. After pathogen receptor recognition, the transcription factor CNBP and its binding partner c-REL translocate to the nucleus and induce IL12B (32, 33). The transcription factors BHLHE40, with CNBP, control IL10 transcription from both myeloid and lymphoid cells (32–34). The roles of these genes and their genetic variants in human regulation of DC and T cell responses are unknown. We hypothesized that variants that influences DC phenotypes would lead to multiple downstream effects, including influence on BCG-specific T cell responses and susceptibility to TB disease. In this study, we investigated the impact of common human genetic variation in the CNBP, REL, and BHLHE40 gene regions on LPS and mycobacteria-induced cytokine responses in DCs, BCG-specific T cell responses, and TB susceptibility.

Approval for human study protocols was obtained from the Institutional Review Boards at local sites in Vietnam, South Africa, and the University of Washington School of Medicine. Genomic DNA was purified from blood samples using genomic DNA isolation kits (Qiagen). DNA concentrations were confirmed using NanoDrop. Seattle cohort study subjects were local volunteers self-described as healthy without a history of recurrent of serious infections. Fifty-two percent of individuals were female, and 48% were male. The ethnic composition of this study group was 69% white, 19% Asian, 2% black or African American, and 2% Latinx. The average age of study participants was 39, with an interquartile range of 29–46 at the time of their enrollment. In this study, we tested immune responses on a subset of 35 participants (discovery) and a second set of 48 participants (validation).

For genetic studies in Vietnam, approval for human study protocols was obtained from the Human Review Boards at the University of Washington School of Medicine, the Hospital for Tropical Diseases (Ho Chi Minh City, Vietnam), Pham Ngoc Thach Hospital for Tuberculosis and Lung Diseases (Ho Chi Minh City, Vietnam), Health Services of Ho Chi Minh City, Hung Vuong Hospital (Ho Chi Minh City, Vietnam), and the Oxford Tropical Research Ethics Committee. For South African pediatric TB cohorts, the study was conducted according to the U.S. Department of Health and Human Services and Good Clinical Practice guidelines. This study included written informed consent from the parent or legal guardian of the study participant and protocol approval by the University of Cape Town Research Ethics Committee and the University of Washington Human Subjects Review Board.

South African study participants were enrolled at the South African Tuberculosis Vaccine Initiative field site in Worcester, South Africa, near Cape Town, as part of a larger study on BCG vaccination with 11,680 infants (35, 36). This area has one of the highest rates of TB incidence in the world, with an incidence of 3% among children under 3 y of age in the study population (35, 36). A nested genetics case control study was performed with identification of cases and controls during a 2-y prospective observation period after vaccination at birth. Household controls were children without TB disease during the 2-y follow period living with an individual with active TB disease. Community controls had no history of TB disease in the first 2 y of life. Community-wide passive surveillance systems identified patients with TB disease and children with symptoms suggestive of TB disease. The criteria for detection of TB cases have been described previously (37). All infants who had symptoms compatible with TB disease or who had contact with an adult with TB disease were admitted to a dedicated research ward for clinical examination, chest radiography, tuberculin skin testing, two early-morning gastric aspirations, and two sputum inductions for M. tuberculosis smear and culture. Participants were defined as having “definite TB” when they had a positive M. tuberculosis culture, a positive smear, or a positive M. tuberculosis PCR from one of their samples. Participants with a chest radiograph compatible with or suggestive of TB combined with one or more additional laboratory or clinical features consistent with TB (smear negative, cough >2 wk, purified protein derivative skin test ≥15 mm, failure to thrive, and recent weight loss) were defined as having “probable TB.” Individuals without radiography consistent with TB who were diagnosed with TB by the treating physician and had two or more clinical features suggestive of TB were defined as “possible TB.” All others were described as “not TB.” All infants admitted to the research ward were also tested for HIV infection, and positive tests resulted in exclusion from the study. The following were additional exclusion criteria at 10 wk of age: mother known to be infected with HIV; BCG not received by infant within 24 h of birth; significant perinatal complications in the infant; any acute or chronic disease in the infant at the time of enrollment; clinically apparent anemia in the infant; and household contact with any person with TB disease or any person who was coughing.

Whole blood from South African infants was collected as previously described (37, 38). Infants were vaccinated with BCG on the day of birth, and at 10 wk of age, heparinized blood was collected from BCG-vaccinated infants. None of the infants in this study had active TB at the time of their 10-wk blood draw or during 2 y of follow-up observation. Flow cytometry was performed on a total of 181 infants, divided into discovery (n = 89) and validation (n = 92) cohorts.

Study subjects from the Vietnam cohort were described previously and are briefly summarized here (12). Subjects with tuberculous meningitis (TBM) were recruited from two centers in Ho Chi Minh City, Vietnam: Pham Ngoc Thach Hospital for Tuberculosis and the Hospital for Tropical Diseases. Subjects with PTB were recruited from a network of district TB control units within Ho Chi Minh City that provide directly observed therapy to TB patients. In addition, PTB subjects were recruited from the Pham Ngoc Thach hospital from 2006 through 2008. Vietnamese population controls are otherwise healthy adults with primary angle closure glaucoma who have been previously described (39). All case and control participants were unrelated, and >95% were of the Vietnamese Kinh ethnicity. Previous genetic studies of this population indicate minimal population substructure (12). Written informed consent was obtained from patients or their relatives when the patient could not provide consent (i.e., was unconscious). Individuals in the TBM group were defined as follows. Individuals at least 15 y old, admitted to these centers with clinical meningitis (defined as nuchal rigidity and abnormal cerebrospinal fluid parameters), a negative HIV test result, and a positive Ziehl–Neelsen stain for acid-fast bacilli or M. tuberculosis culture, or both, from cerebrospinal fluid (“definite TBM”) were recruited for genetics studies from 2001 to 2008. In addition to definite TBM, the cohort included subjects with probable TBM, defined as clinical meningitis plus at least one of the following: chest radiograph consistent with active TB, acid-fast bacilli found in any specimen other than cerebrospinal fluid, and clinical evidence of other extrapulmonary TB. The PTB group was defined as follows: participants were outpatients who were at least 18 y old, had no previous history of treatment for TB, no evidence of miliary or extrapulmonary TB, chest radiograph results consistent with non-miliary PTB, negative HIV test results, and a sputum smear positive for acid-fast bacilli or M. tuberculosis cultured from sputum.

RPMI 1640 medium and l-glutamate were obtained from Life Technologies. Ultrapure LPS (TLR4 ligand) isolated from Salmonella minnesota R595 was obtained from List Biological Labs. Whole-cell lysate from M. tuberculosis strain H37Rv was obtained as part of the National Institutes of Health/National Institute of Allergy and Infectious Diseases contract no. HHSN266200400091C, titled “Tuberculosis vaccine testing and research materials” (Colorado State University, Fort Collins, CO). Lyophilized live BCG (20 × 10⁶ CFU/ml) was obtained from Statens Serum Institute (Copenhagen, Denmark).

For whole-blood stimulation assays, the BCG Russia strain (gift of David Sherman) was grown to an OD value of 1.0 at 600 nm in 7H9 broth medium supplemented with 0.2% glycerol and 10% OADC (oleic acid, albumin, dextrose, catalase) enrichment (complete medium for bacterial culture) and harvested by centrifugation at 4000 rpm for 25 min. The bacterial pellet was resuspended and adjusted to an OD value of 1.0 at 600 nm in RPMI 1640 with 10% BSA supplementation. These aliquots were frozen at −80°C until use.

LPS (10 ng/ml), BCG (10⁶ CFU/ml), and M. tuberculosis whole-cell lysate (TBWCL; 50 µg/ml) were added to 500 µl of whole blood. Media-stimulated cells were included as a negative control. Brefeldin A (Sigma-Aldrich) was added at a concentration of 500 μg/ml (50-fold the desired final concentration of 10 ng/ml). Additionally, protein transport inhibitor containing monensin (BD Biosciences) was added according to the manufacturer’s protocol to all wells 4 h prior to the completion of the experiment. Primary monocytes were isolated from human subjects as described previously (12). Briefly, peripheral blood was obtained and PBMCs were collected via Ficoll gradient separation and cryopreserved. Subsequently, CD14⁺ cells were purified using a human monocyte negative selection kit (Miltenyi Biotec). Monocytes were >95% pure for CD14⁺ using this method.

The following Abs (clones and source) were used in these experiments: PE-Texas Red anti-CD3 (UCHT1, Beckman Coulter), allophycocyanin anti-CD11c (S-HCL-3, BD Biosciences), V500 anti-CD14 (M5E2, BD Biosciences), BV650 anti-CD16 (3G8, BioLegend), biotin anti-CD66a/c/e (ASL-3, BioLegend), PE-Cy7 anti-CD123 (6H6, BioLegend), SB600 anti–HLA-DR (LN3, eBioscience), FITC anti–IL-10 (BT-10, eBioscience), EF450 anti–IL-12 (C8.6, eBioscience), BV786 streptavidin (BD Biosciences), and near-IR Avid fixable Live/Dead stain (Invitrogen). The concentrations of all Abs were titrated prior to use.

Infant T cell assays were collected as described previously (37, 38). Infants were vaccinated with BCG on the day of birth, and at 10 wk of age, heparinized blood was collected from BCG-vaccinated infants and 1 ml was incubated ex vivo with 1.26 × 10⁶ CFU of BCG (Danish strain 1331). None of the infants in this study had active TB at the time of their blood draw at 10 wk of age or during 2 y of follow-up observation. Whole blood was incubated for 12 h with either media control, BCG, or staphylococcal enterotoxin B (SEB), then fixed and frozen in liquid nitrogen. Flow cytometry was performed on a total of 181 infants, divided into discovery (n = 89) and validation (n = 92) cohorts.

Frozen samples were thawed in a 37°C water bath and spun, and then pellets were resuspended in 200 μl of Perm/Wash solution (BD Biosciences) and incubated at room temperature for 10 min. After one wash in Perm/Wash, cells were stained for 30 min at room temperature. After two further washes with Perm/Wash, cells were immediately analyzed on an LSR II flow cytometer (BD Biosciences). Positivity thresholds were determined by gating on fluorescence minus one control samples followed by selective testing of thresholds control samples.

Genotyping was performed with Illumina MEGA^EX chip for the Seattle cohorts and Illumina OmniExpress in Vietnam. Imputation in Vietnam was performed as described previously (40). Selected genotyping was also performed using a Fluidigm genotyping 96 × 96 array. We selected a region 10 kb upstream and downstream of genes of interest using a minor allele frequency cutoff of 5%. Single-nucleotide polymorphisms (SNPs) were excluded when they demonstrated Hardy–Weinberg equilibrium (HWE) p < 0.001. In the Seattle cohort, genotypes were determined by Illumina MEGA^EX array annotated to the human genome (hg19) with annotatr basic genes (v1.14.0) in R (v4.0.2) (41). The linkage disequilibrium (LD) Pearson r² was calculated between SNPs with two or more genotypes and within 10 kb of a gene using the genetics package (v1.3.8.1.2). Haplotype tags were selected by the SNPinfo Web server with an LD threshold of 0.8, maximum distance of 250 kb between SNPs, and a minimum of two SNPs tagged using each SNP, using a European population for the Seattle cohort and an East Asian population for the Vietnam cohort (42).

Genotyping in the South Africa cohort was performed on haplotype-tagging SNPs in the REL and BHLHE40 gene region using the Fluidigm genotyping 96 × 96 array using the SNP selection approach detailed above. Haplotype tags were selected by the SNPinfo Web server using the African population (42).

SNPs of interest were identified using a haplotype tagging approach from whole-genome data in the Seattle cohort. The association between SNPs and development of TBM was compared by χ² analysis between cases and controls. Statistical significance of cytokine responses were initially determined using a linear model. In case control studies, an allelic model was used as a screening approach. For SNPs of interest, a dominant genotypic model (major allele homozygotes are compared with a composite of heterozygotes and minor allele homozygotes) and recessive genotypic model (minor allele homozygotes are compared with a composite of heterozygotes and major allele homozygotes) were also evaluated to further define the genetic model. Haplotype blocks and tagging SNPs were identified in CNBP, REL, and BHLHE40 using the SNP info Web server (https://manticore.niehs.nih.gov/cgi-bin/snpinfo/snptag.cgi) selecting for genotype data “dbSNP,” and the European population (43). We used 10-kb flanking regions, including any upstream genomic regions that may participate in gene regulation. We used an LD threshold of 0.8, minor allele frequency cutoff of 0.05, and a minimum of two SNPs tagged by each SNP. Measures of correlation (r²) and allelic linkage (D^′) between SNPs of interest were calculated using the R package genetics for the Seattle population. The values of r² and D′ range from 0 to 1, with lower values indicating allelic independence and higher values suggesting that alleles are in complete LD (e.g., that the alleles are fully coinherited).

In our primary analysis, we examined whether polymorphism genotype frequencies were associated with cytokine production using a simple generalized linear model. For genetic analysis, background-corrected values were used to estimate cytokine induction by inflammatory stimuli. We modeled effect size for our primary analyses by calculating the R² correlation coefficient. For each result with p < 0.05 in this model, we added a second p value that adjusts for ethnicity using a linear regression with ethnic origin as a covariate. For secondary analyses, SNPs were investigated for associations under additional genetic models (dominant, recessive, and additive). In the recessive model, carriers of allele 1 (00 and 01 genotypes) were compared with homozygous subjects for allele 2 (11 genotype). In the dominant model, carriers of allele 2 (01 and 11 genotypes) were compared with homozygous subjects for allele 1 (00 genotype). For genetic studies of continuous variables, simple linear models were used, modeling the additive contribution of genotypes, and effect sizes were estimated using the slope of the linear regression model (model r²; generalized linear model command, Stata 14.1). Analyses adjusted for ethnicity are linear regressions evaluating the relationship between genotype and cytokine, with ethnicity as a covariate (regress command, Stata 14.1).

Within the Vietnam genome-wide association studies, no correction for sex or ethnicity was provided. This is due to prior data demonstrating that sex is not a confounding factor in genetic studies (44). For Vietnam genome-wide association studies, all participants were unrelated, and all were of the Kinh Vietnamese ethnicity. In addition, we previously genotyped a panel of ancestry informative markers and found no evidence of population admixture between case and control participants (45). Furthermore, principal-component analysis confirmed the genetic homogeneity of this population (46). Therefore, we did not perform regression with ethnicity as a covariate.

Graphs were created with Prism version 8.0 (GraphPad Software). Results are reported without correction for multiple comparisons due to the heterogeneous sources of data (mixture of cellular and clinical), including varied availability of validation datasets for some datasets.

To evaluate genetic regulation of IL-10 and IL-12 production from healthy human donors, we used flow cytometry to measure the proportion of peripheral blood MHC class II⁺CD11c⁺ DCs producing IL-10 and IL-12 after stimulation of whole blood with LPS or TBWCL (Fig. 1A). Initially, we chose LPS as a stimulation because it is a well-characterized ligand that potently induces IL-12 and IL-10 from innate immune cells both in vitro and in vivo. We also used TBWCL due to its breadth of mycobacterial ligands and relevance for studying TB pathogenesis as well as its induction of IL-10 and IL-12. We demonstrate the kinetics of IL-12 and IL-10 responses in Supplemental Fig. 1A. LPS (10 ng/ml) and TBWCL (50 µg/ml) both strongly induced IL-12 (p < 0.0001 for both tests, Fig. 1B) and IL-10 (p < 0.0001 for both tests, Fig. 1C) from DCs 24 h after stimulation. We also measured cytokine responses to LPS (10 ng/ml) and live BCG (10⁶ CFU/ml) 6 h after stimulation (Fig. 1D). We chose BCG (10⁶ CFU/ml) because it was a live bacillus relevant to M. tuberculosis pathogenesis and due to its capacity to induce T cell responses in humans, nonhuman primates, and mice that are relevant to M. tuberculosis control. We found that LPS and BCG induced IL-12 6 h after stimulation in CD11c⁺ DCs (p < 0.001 for LPS, p = 0.01 for BCG, Fig. 1D). However, we did not detect IL-10 above background levels from DCs at this time point (data not shown).

We next examined whether candidate gene variants were associated with LPS- or TBWCL-induced IL-12 in DCs. We interrogated 5 haplotype-tagging SNPs from CNBP, 6 from REL, and 20 from BHLHE40 in a local cohort of healthy volunteers (the complete SNP list is presented in Supplemental Fig. 1B–D). REL SNP rs842634 was associated with increased IL-12 after TBWCL and LPS stimulation (Fig. 2A, p = 0.037, generalized linear model, ethnicity-adjusted p = 0.017, r² = 0.10; (Fig. 2B, p = 0.044, adjusted p = 0.042, r² = 0.09). A second haplotype-tagging SNP, REL SNP rs842618, was associated with increased IL-12 after TBWCL and LPS stimulation (Fig. 2C, p = 0.013, adjusted p = 0.03, r² = 0.13; (Fig. 2D, p = 0.04, adjusted p = 0.04, r² = 0.09). CNBP SNP rs11709852 was associated with increased IL-12 production after TBWCL stimulation, but not LPS stimulation (Fig. 2E, p = 0.003, adjusted p = 0.24; (Fig. 2F, p = 0.48). However, after adjustment for ethnicity, this association did not meet criteria for statistical significance. No SNPs from BHLHE40 were associated with IL-12 (Supplemental Dataset 1).

We evaluated the association of our candidate SNPs with an earlier time point with different healthy donors. We examined whole blood incubated with BCG (10⁶ CFU/ml) or LPS (10 ng/ml) for 6 h, followed by measurement of cytokine responses, as described above. REL SNP rs842634 was associated with increased IL-12 after BCG infection (Fig. 3A; p = 0.046; ethnicity-adjusted p = 0.040, r² = 0.11) and LPS stimulation (Fig. 3B; p = 0.024; ethnicity-adjusted p = 0.014, r² = 0.15), and REL SNP rs842618 was associated with increased IL-12 after BCG (Fig. 3C, p = 0.042; ethnicity-adjusted p = 0.014, r² = 0.10) and LPS (Fig. 3D, p = 0.002; ethnicity-adjusted p = 0.004, r² = 0.26).

Next, we evaluated for associations between genetic variants in CNBP, REL, and BHLHE40 with IL-10 production from DCs. BHLHE40 SNP rs4496464 was associated with increased IL-10 production after TBWCL stimulation (Fig. 4A; p = 0.005; ethnicity-adjusted p = 0.005, r² = 0.15). In contrast, rs4496464 was not associated with IL-10 after LPS stimulation (Fig. 4B, p = 0.18, r² = 0.04). Although rs79248174 was associated with IL-10 after TBWCL stimulation, this was due to a single data point, and therefore this SNP was not included for further analysis (Supplemental Dataset 1). BHLHE40 SNP rs4496464 was not associated with IL-12 expression after stimulation with either TBWCL or LPS (Fig. 4C, 4D). No REL SNPs were associated with IL-10 expression after LPS stimulation (data not shown).

We evaluated whether rs4496464 genotypes were associated with BHLHE40 mRNA expression in peripheral blood monocyte-derived macrophages (MDMs) from healthy donors. The uncommon G allele of rs4496464 was associated with increased BHLHE40 in unstimulated monocytes using a dominant model of inheritance (Fig. 5A; p = 0.026, A/A versus [G/A + G/G], Mann–Whitney U test). No other BHLHE40 SNPs were associated with expression. The T allele of REL variant rs842618 was associated with increased mRNA expression in monocytes (Fig. 5B; p = 0.0015, r² = 0.20). However, REL SNP rs842634 was not associated with mRNA expression in monocytes (Fig. 5C).

To validate our association between rs496464 and IL-10 expression in DCs, we measured IL-10 secreted from MDMs stimulated with either LPS (50 ng/ml) or TBWCL (25 µg/ml) overnight (Fig. 6A, n = 26). The rs4496464 G allele was associated with increased IL-10 after LPS stimulation (Fig. 6B, p = 0.01, generalized linear model). SNP rs4496464 was also associated with increased IL-10 after TBWCL (Fig. 6C, p = 0.005, generalized linear model). SNP rs4496464 was not associated with TNF secretion after either LPS (Fig. 6D) or TBWCL stimulation (Fig. 6E), which suggests that variation in BHLHE40 is associated with IL-10 production specifically, over proinflammatory cytokine responses.

Our data suggest that rs842634 and rs842618 are associated with increased IL-12 in DCs and rs4496464 is associated with increased IL-10 production from peripheral blood monocytes and DCs in our local population. We hypothesized that these polymorphisms were associated with susceptibility to TB due to their influence on these critical immune phenotypes. Within a large genome-wide association study comparing Vietnamese individuals with adult PTB (n = 1598) or TBM (n = 407) with control subjects (n = 1139), we evaluated whether SNPs in REL and BHLHE40 were associated with adult PTB or TBM and in LD with our SNPs of interest (Supplemental Fig. 2). REL rs842634 was not associated with TBM, but the minor T allele of REL SNP rs842618 was associated with increased TBM susceptibility (p = 0.03; odds ratio [OR] 1.27, allelic model; Table I, Supplemental Dataset 2). These data best fit a dominant model (Table I, p = 0.035, OR 1.32, 95% confidence interval 1.02–1.73). No BHLHE40 SNPs were associated with TBM. We did not identify any associations between SNPs in REL, or BHLHE40 SNPs with PTB (Supplemental Dataset 2). We performed a haplotype analysis comparing the contribution of rs842634 and rs842618 with the TB meningitis phenotype. SNPs rs842634 and rs842618 are interrelated, and the presence of both minor alleles strengthens the association between these genotypes and TBM susceptibility (p = 0.003 describing an interaction between these two SNPs). Taken together, these data suggest that a causal REL SNP linked to rs842618 is associated with both increased IL-12 production, increased REL mRNA, and increased adult TBM susceptibility, but not morbidity or mortality.

We next evaluated whether variants in REL or BHLHE40 were associated with pediatric TB in South Africa (Supplemental Fig. 2) (47). SNP rs4496464 was not associated with pediatric TB. However, in this gene, BHLHE40 SNP rs11130215 was associated with decreased risk for pediatric TB in an allelic model (Supplemental Dataset 2, Table II; p = 0.001), which best fit a dominant model of inheritance (p = 3.3 × 10⁻⁴, OR 0.5 [0.33–0.75]). Rs11130215 was in low LD with rs4496464 in the South African cohort (r² = 0.10, D′ = 0.30). To adjust for ethnic heterogeneity, we genotyped a panel of 95 ancestry informational markers and performed principal components analysis, as described previously (37). The association between rs11130215 and pediatric TB remained statistically significant after adjustment for gender and the top five principal components of the tested ancestry informational markers (Table II, p = 0.01, OR 0.24–0.83). No REL SNPs were associated with pediatric TB. Taken together, these data suggest that a BHLHE40 polymorphism (rs11130215) with low linkage to rs4496464 is associated with decreased risk for pediatric TB.

We next examined whether rs842618, rs842634, rs496464, and rs11130215 were associated with adaptive immune responses, as a possible mechanism of TB susceptibility from DC regulation of T cell responses. We tested this hypothesis in a cohort of South African infants that were vaccinated with BCG at birth and whose BCG-specific CD4⁺ IL-2, TNF, and IFN-γ⁺ T cell responses were measured at 10 wk of age by flow cytometry (36, 37) (Fig. 7A). Overall media (Fig. 7B), BCG-induced responses (Fig. 7C), and SEB-induced responses (Fig. 7D) are shown. We evaluated the association between genetic variation in our SNPs of interest: rs842634, rs842618, rs4496464, and rs11130215, with the frequency of BCG-induced IL-2, TNF, and IFN-γ in CD4⁺ T cells. SNPs rs842618 and rs842634 were monoallelic in the South African cohort and not analyzed further. SNP rs4496464 was associated with a trend toward increased IL-2⁺CD4⁺ T cell frequency, but this did not achieve statistical significance (Fig. 7E, p = 0.15, generalized linear model). This SNP was not associated with TNF or IFN-γ frequency in CD4⁺ T cells (Fig. 7F, 7G). The G allele of BHLHE40 rs11130215 was associated with increased frequency of BCG-specific IL-2⁺CD4⁺ cells (Fig. 7H, p = 0.015, generalized linear model), but not TNF or IFN-γ (Fig. 7I, 7J). In a second cohort of South African infants, rs11130215 was associated with a trend toward increased IL-2 expression that did not achieve statistical significance (Fig. 7J, p = 0.06, generalized linear model). However, when these data were combined, we found that this SNP was associated with increased IL-2 from CD4⁺ T cells (Fig. 7L, p = 0.006, generalized linear model). Taken together, these data suggest that a BHLHE40 SNP rs11130215 G allele is associated with increased IL-2⁺CD4⁺ T cell frequency and decreased risk for pediatric TB in a genetic cohort of South African infants.

IL-12 and IL-10 are both essential for an effective host response to TB, and overexpression of either cytokine can similarly lead to adverse outcomes. In this study, we found that variation in REL SNP rs842618 and BHLHE40 SNP rs4496464 are associated with secretion of IL-12 and IL-10, respectively, from peripheral blood DCs using a flow cytometry–based assay. To our knowledge, this assay has not been used previously to evaluate the genetics of DC immune responses (20, 48). REL SNP rs842618 was associated with increased expression of IL-12, increased REL mRNA expression, and increased susceptibility to TBM. SNPs in BHLHE40 were associated with increased IL-10 and pediatric TB. These data represent the most comprehensive evaluation of the human genetic loci associated with IL-10 and IL-12 production in TB pathogenesis.

REL SNP rs842618 was associated with IL-12 expression in peripheral blood DC and monocytes, REL mRNA expression, and TBM risk in a Vietnamese cohort. These studies provide evidence that common variation in an upstream regulator of IL-12 and innate immune activation are associated with TB outcomes. Study of the literature and publicly available datasets demonstrates strong ancillary evidence for its functional significance. Whole-genome studies demonstrate that variation in rs842618 is associated with REL mRNA expression in lymphoblastoid cells immortalized rom European donors (49) and in monocytes overall (50). As a possible mechanism of action, rs842618 alters two transcription factor motifs, RREB-1 and Foxp1 (51). These data provide evidence from the literature supporting our translational observations that this variant is associated with TBM in Vietnam. Identification of genetic factors that modulate DC proinflammatory cytokines provides insight into the optimal balance of cytokines to control M. tuberculosis in adults.

Variants in the BHLEH40 gene region were associated with immune outcomes and TB susceptibility in a South African and European population. To our knowledge, these data represent the first association of variants within this gene region with innate immune activation. Prior genetic studies have provided evidence of association with developmental and metabolic phenotypes, including cleft palate (52), diabetes mellitus (53), and late-onset Alzheimer’s disease (54, 55). The relationship between these phenotypes and TB remains unexplored. BHLHE40 SNP rs4496464 was associated with IL-10 production from DCs and macrophages after TBWCL stimulation, but not with IL-10 from LPS-stimulated macrophages only. These data suggest that IL-10 production might be preserved via an alternate mechanism in the setting of TLR4 stimulation. Rs11103215 was also associated with increased BCG-specific IL-2⁺CD4⁺ T cells with stable frequencies of TNF⁺ and IFN-γ⁺CD4⁺ T cells in South African infants and decreased risk for developing pediatric TB. Our observations are consistent with a model whereby modest increases in BHLHE40 are associated with increased IL-10 in macrophages, expanded IL-2⁺CD4⁺ T cell responses, and protection from TB. Notably, these data support findings from the mouse model, where BHLHE40 deficiency was associated with early M. tuberculosis death due to an excessive neutrophil-dominant inflammatory response (34). Study of the factors that influence IL-10 expression may provide insight into a suite of macrophage or T cell changes that may provide insight into TB susceptibility and control.

This study has several limitations. We do not yet have evidence of functional SNPs that directly regulate gene function. Future fine-mapping studies with in vitro mechanistic assays are required to determine the specific alleles that regulate cellular function and clinical outcomes together. Some of these observations do not achieve statistical significance after adjustments for multiple comparisons with associations with clinical outcomes. Although this limitation is true for the clinical findings, the evidence supporting a genetic regulatory role of human cellular IL-12/IL-10 responses was robust and provided support for the possible clinical associations. Given this, we used a threshold of p < 0.05 as a measure of statistical significance, without the conservative Bonferroni correction. Further studies are needed in additional cohorts, particularly after discovery of the causal SNP that regulates cytokine production. Third, case-control studies of TB outcomes may have misclassification of controls, as we examined population controls in studies in our Vietnamese cohort. However, classification errors that arise from such control populations likely lead to reduction in the statistical power of these studies.

To our knowledge, this study represents the most comprehensive analysis to date of genetic regulation of DC IL-12 and IL-10 production by common polymorphisms and their association with TB outcomes. Although further studies are required, overlapping genetic studies of immune outcomes and TB clinical susceptibility may lead to important breakthroughs in TB vaccine design and immune drug development.

We thank the individuals and families who participated in the study. We also thank the immunology and clinical teams at the hospitals in Ho Chi Minh City, Vietnam and Worcester, South Africa for obtaining informed consent and collecting and processing samples from study participants. We thank Chetan Seshadri from the University of Washington Center for Emerging and Reemerging Infectious Diseases for technical discussions and troubleshooting flow cytometry. We acknowledge the support of the Center for Emerging and Reemerging Infectious Disease Flow Cytometry Facility at the University of Washington for use of the BD LSRFortessa flow cytometer.

The authors have no financial conflicts of interest.