Influence of Environmental Exposures on T Follicular Helper Cell Function and Implications on Immunity: a comparison of Bangladeshi and American Children

T follicular helper (Tfh) cells exhibit age-specific changes in functional marker expression in Bangladeshi children, at age 2

In both American and Bangladeshi children, the overall frequencies of CD4+ T cells and Tfh cells were comparable, suggesting that population-level differences in total T cell populations were minimal (Fig. 1A) and aligned with previous findings.¹² However, significant differences in the expression patterns of CD40L and CD57 emerged in Bangladeshi children by age two, pointing to functional divergences in Tfh cell maturation or responsiveness. At age two, Bangladeshi children exhibited a diminished ability to upregulate CD40L following PMA/Ionomycin stimulation, a phenotype not observed in their American counterparts (Fig. 1B). This lack of CD40L upregulation signified a potentially reduced functional capacity of Tfh cells in supporting B cell responses. Additionally, Bangladeshi children at age two showed a significant decrease in CD57 expression upon stimulation, in contrast to American children, who did not exhibit this pattern (Fig. 1C), suggesting an accelerated apoptotic process in these cells. Collectively, these findings indicated that at age two, Bangladeshi children exhibited altered Tfh cell functionality, resembling an exhaustion-like phenotype potentially driven by environmental factors. This distinct immune profile was not observed in American children and may have implications for understanding immune competency across different environmental contexts.

• Download figure
• Open in new tab

Figure 1. Comparison of CD4+ T cell and T follicular helper (Tfh) cell populations and function marker expression in American and Bangladeshi children.

(A) Percentage of CD4+ T cells among all T cells and Tfh cells (CXCR5+, PD-1+) among all CD4+ T cells in American (red) and Bangladeshi (black) children across difference ages. (B) Percentage of CD40L+ Tfh cells out of all Tfh cells, comparing untreated (blue) and PMA/Ionomycin-stimulated (orange) conditions. (C) Percentage of CD57+ Tfh cells out of all Tfh cells, comparing untreated (blue) and PMA/Ionomycin-stimulated (orange) conditions. (B-C) Bangladeshi samples were displayed in the top panel and American samples in the bottom panel. Statistical significance was calculated by multiple paired T-test and denoted by asterisks (*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001).

Environmental factors predict expression changes of functional markers on T follicular helper cells upon stimulation, in Bangladeshi children at age 2

The variability in CD40L and CD57 expression on Tfh cells in Bangladeshi children, at age two, underscored the potential influence of environmental factors on immune cell functionality. Significant fluctuations in these markers upon stimulation revealed how external exposures may have uniquely shaped Tfh cell responses in this population (Fig. 2). To identify specific environmental factors associated with these functional changes, we evaluated feature importances with Random Forest and applied Generalized Estimating Equations (GEE) to assess the relative importance of external exposures and risk factors, including nutrition, infections, and breastfeeding practices influenced Tfh cell response to stimulation. Functional response to PMA/Ionomycin stimulation was measured by the ratio of percent CD40L+ or CD57+ Tfh cells post-stimulation to pre-stimulation levels.

• Download figure
• Open in new tab

Figure 2: Variability in CD40L and CD57 expression on Tfh cells in response to stimulation in Bangladeshi children, at age two.

Spaghetti plots show log-transformed percentages of CD40L+ (left) and CD57+ (right) Tfh cells before (UT) and after (PMA) stimulation with PMA/Ionomycin in individual Bangladeshi children, at age 2.

Key predictors identified included days of only exclusive breastfeeding, total number of antibiotic treatments by age 2, total number diarrheal episodes by age 2, and malnutrition scores such as weight-for-age Z-score (WAZ) (Fig. 3A). Partial dependency plots (Fig. 3B) further highlighted inverse relationships between these environmental factors and Tfh cell marker expression. Children who had longer exclusive breastfeeding, fewer antibiotic treatments, fewer diarrheal episodes, and better WAZ scores were more likely to upregulate CD40L expression upon stimulation. Conversely, children with more frequent antibiotic use, more diarrheal episodes, poorer malnutrition scores, and reduced exclusive breastfeeding duration were associated with increased CD57 expression. These findings suggested that early environmental exposures may influence Tfh cell functional programming, towards either an activation-supportive phenotype (CD40L+) or an exhaustion-prone phenotype (CD57+), with potential implications for immune competency and disease susceptibility in later life.

• Download figure
• Open in new tab

Figure 3: Environmental predictors of CD40L and CD57 expression in Tfh cells, at age 2, in Bangladeshi children.

The response to PMA/Ionomycin stimulation was quantified by the percentage of CD40L+ or CD57+ Tfh cells after stimulation divided by their percentage before stimulation. Response to stimulation was then adjusted for a log scale distribution. (A) Variable importance plots generated by Random Forest model, ranking predictors of CD40L and CD57 response to stimulation (CD40L on the top left and CD57 on the bottom left). Variables are ranked by importance, with significant predictors determined by GEE model and marked by asterisks. (B) Partial dependency plots depict the relationship between key environmental factors (e.g., antibiotics, diarrheal episodes, WHZ, exclusive breastfeeding) and Tfh cell response, with CD40L response on the left y-axis (pink) and CD57 response on the right y-axis (blue). Statistical significance was calculated by GEE model and denoted by asterisks (*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001).

Functional T follicular helper cell responses are correlated with ability to mount an antibody response after vaccination or infection

To evaluate the relationship between Tfh cell functionality and downstream immune responses, we examined antibody concentration following vaccination or Cryptosporidium infection and correlated the increase of CD40L and CD57 expression on Tfh cells post-stimulation. Cryptosporidium, an endemic parasitic infection in Bangladesh, presented a relevant model for assessing pathogen-specific immune responses in this population.

Correlational analysis revealed that increased expression of CD40L and CD57 on Tfh cells in response to PMA/Ionomycin stimulation was associated with higher antibody titers against certain vaccine antigens, specifically Type 2 and Type 3 Poliovirus (Fig. 4A). These findings implied that the ability of Tfh cells to upregulate activation and senescence markers may have been most necessary to support robust antibody responses following Polio vaccination specifically. Additionally, we observed a significant positive correlation between high concentrations of IgA antibodies against the Cryptosporidium sporozoite protein Cp17 and CD40L expression. This indicated that Tfh cells expressing higher levels of CD40L may have played a role in promoting protective mucosal immunity against Cryptosporidium. Conversely, IgA levels against Cp17 were negatively correlated with CD57 expression, which suggested that Tfh cells with lower CD57 expression may have been more functionally active in mediating responses to parasitic infections (Fig. 4B).

• Download figure
• Open in new tab

Figure 4: Correlation of Tfh cell functional response with antibody responses to vaccination and Cryptos-poridium infection.

(A) Heatmap displaying correlation coefficients between Tfh cell CD40L and CD57 response to stimulation and antibody titers against vaccine antigens. (B) Heatmap displaying correlation of CD40L and CD57 response to stimulation with IgA and IgG responses against Cryptosporidium sporozoite proteins Cp17 and Cp23.

Data Acquisition and Analysis

This study utilized data from two cohorts: PROVIDE of Bangladeshi children and STORK of U.S. children.

• PROVIDE Study: Based in Mirpur, Dhaka, Bangladesh, this cohort consisted of children from lower socioeconomic areas with high microbial exposure. Children were enrolled within seven days of birth and monitored biweekly. At each visit, questionnaires collected data on diarrheal illnesses, antibiotic use, breastfeeding practices (exclusive and non-exclusive), and nutritional status. Whole blood samples were collected annually up to age four. At study enrollment, mothers completed a lifestyle questionnaire. Vaccinations, including Polio (Types 1, 2, and 3), Haemophilus influenzae type B (Hib), Tetanus, Diphtheria, and acellular Pertussis, were administered before age one, while the measles vaccine was administered at 65 weeks of age.¹⁷

• STORK Study: Based in the San Francisco Bay Area, California, this cohort included children from areas with low microbial exposure. Pregnant mothers were recruited before 36 weeks of gestation and interviewed weekly post-birth. Whole blood samples were collected annually up to age three. Blood samples collected within 15 weeks of a child’s birthday were grouped together for analysis. For example, samples collected between 37–67 weeks of age were categorized as year 1.¹⁸

For all study samples, peripheral blood mononuclear cells (PBMCs) were isolated using conventional methods as previously described.¹² Cryopreserved PBMCs from both studies were thawed, stimulated non-specifically with PMA/Ionomycin for six hours, and stained with metal-conjugated antibodies. Data acquisition was performed on a CyTOF2 mass cytometry instrument following established protocols. Detailed methodology for CyTOF immunophenotyping and analysis has been documented in detail,¹² and the corresponding data files are publicly accessible through the Flow Repository database (Experiment ID: FR-FCM-ZYV8).

Data Processing

CD4+ T cells were gated manually from the CD3+ T cell population following the exclusion of any dead cells, B cells, monocytes, and natural killer cells. Samples with fewer than a count of 400 CD4+ T cells were excluded from further analysis. Additionally, samples lacking matched untreated and stimulated groups were removed, ensuring that only paired untreated and stimulated samples were included in the final analysis. T follicular helper (Tfh) cells were gated from the CD4+ T cell population and defined as CXCR5+ PD-1+, CD4+ T cells. From the total Tfh cell population, CD40L+ and CD57+ Tfh cells were further gated out of the total Tfh cell population. The proportion of CD40L+ or CD57+ Tfh cells was calculated by dividing the count of CD40L+ or CD57+ Tfh cells by the total Tfh cell count within each sample, separately for untreated and stimulated groups. The Tfh cell response to stimulation was defined as the ratio of marker expression in the stimulated group to the untreated group. This ratio was calculated as the proportion of CD40L+ or CD57+ Tfh cells in the stimulated sample divided by the corresponding proportion in the untreated sample, providing a quantitative measure of Tfh functional response to stimulation.

Prediction Modeling and Statistics

Random forest feature importance evaluation was used to identify key environmental factors influencing Tfh cell functionality. The variables of interest included environmental factors such as exclusive and total breastfeeding duration, number of antibiotic treatments, nutritional metrics (WAZ, HAZ, and WHZ scores), number of diarrheal episodes, CRP levels, annual family income, and parental education levels. The outcome variables were the log-transformed ratios of CD40L+ and CD57+ Tfh cells as a ratio of stimulated to unstimulated conditions. Models were trained using 10-fold cross-validation with tree counts of 100, 300, 500 and 700, and a minimum node sizes ranging from 1 to 9. Through the grid search, the optimal tuning parameters are selected to fit the Random Forest modeling. Variable importance was evaluated based on the mean decrease in the Gini index, which quantifies the predictive contribution of each variable.

Generalized Estimating Equation (GEE) models were subsequently employed to model associations between the top five predictors identified by Random Forest analysis. The log-transformed ratio of stimulated to unstimulated cell counts for CD40L+ and CD57+ Tfh cells (as a proportion of all Tfh cells) served as the response variables. GEEs accounted for the repeated measures and clustering inherent in the data, providing robust estimates of significant associations between environmental factors and Tfh cell functional outcomes. Statistical analyses were performed using appropriate software with significance thresholds set at p < 0.05.

Antibody Quantification

Antibody concentrations post-vaccination for measles, Bordetella pertussis toxin, filamentous hemagglutinin, pertactin, diphtheria toxoid, tetanus, and Haemophilus influenzae type B antigens were quantified using standardized assays, reported as IU/mL, EU/mL, or ng/mL. IgA and IgG concentrations against Cryptosporidium sporozoite antigens Cp17 and Cp23 were measured using enzyme-linked immunosorbent assays (ELISA).

For ELISA, microtiter plates were coated with antigen solution and incubated overnight at 4°C. Plates were then washed, and a blocking buffer was applied to minimize nonspecific binding. Plasma samples were added to the wells and incubated, followed by another blocking buffer application. Subsequently, HRP-conjugated anti-human secondary antibodies were added to detect bound primary antibodies. TMB (3,3′,5,5′-tetramethylbenzidine) substrate solution was then added for color development, and the reaction was stopped with a stop solution. Absorbance was measured at 450 nm to quantify antibody levels. All measurements were performed in triplicate to ensure reproducibility.