Rapidly Escalating Hepcidin and Associated Serum Iron Starvation Are Features of the Acute Response to Typhoid Infection in Humans

Abstract

Background: Iron is a key pathogenic determinant of many infectious diseases. Hepcidin, the hormone responsible for governing systemic iron homeostasis, is widely hypothesized to represent a key component of nutritional immunity through regulating the accessibility of iron to invading microorganisms during infection. However, the deployment of hepcidin in human bacterial infections remains poorly characterized. Typhoid fever is a globally significant, human-restricted bacterial infection, but understanding of its pathogenesis, especially during the critical early phases, likewise is poorly understood. Here, we investigate alterations in hepcidin and iron/inflammatory indices following experimental human typhoid challenge.

Methodology/principal findings: Fifty study participants were challenged with Salmonella enterica serovar Typhi and monitored for evidence of typhoid fever. Serum hepcidin, ferritin, serum iron parameters, C-reactive protein (CRP), and plasma IL-6 and TNF-alpha concentrations were measured during the 14 days following challenge. We found that hepcidin concentrations were markedly higher during acute typhoid infection than at baseline. Hepcidin elevations mirrored the kinetics of fever, and were accompanied by profound hypoferremia, increased CRP and ferritin, despite only modest elevations in IL-6 and TNF-alpha in some individuals. During inflammation, the extent of hepcidin upregulation associated with the degree of hypoferremia.

Conclusions/significance: We demonstrate that strong hepcidin upregulation and hypoferremia, coincident with fever and systemic inflammation, are hallmarks of the early innate response to acute typhoid infection. We hypothesize that hepcidin-mediated iron redistribution into macrophages may contribute to S. Typhi pathogenesis by increasing iron availability for macrophage-tropic bacteria, and that targeting macrophage iron retention may represent a strategy for limiting infections with macrophage-tropic pathogens such as S. Typhi.

Conflict of interest statement

The authors have declared that no competing interests exist.

Figures

Fig 1. Changes in hepcidin, iron and inflammatory indices between baseline–the day of typhoid challenge–and the day of typhoid diagnosis.

Serum samples were available from both day of typhoid challenge and day of typhoid diagnosis from 19 individuals from Study A (placebo arm of vaccine/typhoid challenge study). (A) temperature, (B) hepcidin, (C) serum iron, (D) transferrin saturation, (E) CRP, (F) ferritin and (G) hemoglobin were measured on the day of challenge and day of typhoid diagnosis. P-values represent results of paired t tests based on geometric means for hepcidin (baseline, 10.4 ng/mL [95% CI: 7.1–15.3]; diagnosis, 98.2 ng/mL [75.9–126.9]), ferritin (baseline, 46.3 μg/L [30.8–69.8]; diagnosis, 86.4μg/L [56.9–131.0]) and CRP (baseline: 1.46 mg/L [0.84–2.54]; diagnosis, 34.1 mg/L [24.2–48.2]), and arithmetic means for temperature (baseline, 36.3°C [36.1–36.5]; diagnosis, 37.6°C [37.3–37.9]), serum iron (baseline, 12.6 μmol/L [9.9–15.3]; diagnosis, 4.4 μmol/L [2.7–6.1]), transferrin saturation (baseline, 22.3% [17.2–27.4]; diagnosis, 7.4% [4.7–10.2]) and hemoglobin (baseline, 14.3 g/dL [13.6–15.1]; diagnosis 13.9 g/dL [13.1–14.8]).

Fig 2. Kinetics of perturbations in hepcidin, iron and inflammatory parameters in individuals diagnosed with typhoid infection following experimental Salmonella Typhi challenge.

(A) Temperatures, (B) serum hepcidin concentrations, (C) serum iron, (D) transferrin saturations, (E) hemoglobin, (F) CRP, and (G) ferritin concentrations were measured in 13 individuals from Study B who received typhoid diagnosis following challenge with Salmonella Typhi. Analyte values are plotted relative to the day of typhoid diagnosis, TD = day 0; since not all individuals were diagnosed on the same day post-challenge, baseline samples from the day of challenge (Chall) are considered together, as are data from the final day 14 visit (Final). (Left-hand panels) Data available from each individual for each day were plotted using box and whiskers, representing median values and interquartile ranges (IQR); whiskers represent the data point occurring furthest from the first or third quartile but still within 1.5*IQR of the quartile; outliers (further than 1.5*IQR from the quartile) are shown as isolated data points. Smoothed curves were also interpolated from the mean data for each day and overlaid on the plots. The Wald test was employed after fitting linear mixed effects models to test the null hypothesis that there is no difference between parameter values between days. Pairwise differences between baseline values on the day of typhoid challenge (Chall) and other days were examined by t-tests after accounting for subject-specific variability. Significant perturbations from baseline are indicated with asterisks (*p<0.05, **p<0.01, ***p<0.001). For ferritin, outliers at (a) 1014.07 μg/L and 1075.24 μg/L, (b) 2433.54 μg/L, and (c) 1008.87 μg/L are beyond the y-axis limits and not depicted on the figure, but are included in the analysis. (Right-hand panels) Smoothed interpolated curves as described above, but depicting 95% pointwise prediction intervals (thick error bar) and conservative simultaneous Bonferroni bounds (thin error bar) of the interpolated curves.

Fig 3. Opposite associations of hepcidin with serum iron and transferrin saturation in the presence and absence of inflammation.

Associations between (A) hepcidin and serum iron, (B) hepcidin and transferrin saturation (Tsat), (C) ferritin and serum iron, and (D) ferritin and Tsat, when acute inflammation (defined as CRP>5 mg/L, open circles) or when no inflammation (CRP5 mg/L: 69 observations, 36 clusters. Tsat analyses (B) and (D): CRP 5 mg/L: 69 observations, 36 clusters). Regression with clustered errors adjusts the confidence intervals of the regression coefficients to account for intra-cluster correlation, as is likely when multiple observations from the same individuals are included. Pearson correlation coefficients and p-values are stated.

Fig 4. Kinetics of perturbations in IL-6 and TNF-alpha in individuals diagnosed with typhoid infection following experimental Salmonella Typhi challenge.

(A) Plasma IL-6 and (B) TNF-alpha concentrations were measured in 13 individuals from Study B who received typhoid diagnosis following challenge with Salmonella Typhi. Analyte values are plotted relative to the day of typhoid diagnosis, TD = day 0; since not all individuals were diagnosed on the same day post-challenge, baseline samples from the day of challenge (Chall) are considered together, as are data from the final day 14 visit (Final). (Left-hand panels) Data available from each individual for each day were plotted using box and whiskers, representing median values and interquartile ranges (IQR); whiskers represent the data point occurring furthest from the first or third quartile but still within 1.5*IQR of the quartile; outliers (further than 1.5*IQR from the quartile) are shown as isolated data points. Smoothed curves were also interpolated from the mean data for each day and overlaid on the plots. The Wald test was employed after fitting linear mixed effects models to test the null hypothesis that there is no difference between parameter values between days. Pairwise differences between baseline values on the day of typhoid challenge (Chall) and other days were examined by t-tests after accounting for subject-specific variability. Significant perturbations from baseline are indicated with asterisks (***p<0.001). (Right-hand panels) Smoothed interpolated curves as described above, but depicting 95% pointwise prediction intervals (thick error bar) and conservative simultaneous Bonferroni bounds (thin error bar) of the interpolated curves.