Abstract
An important entraining signal for the endogenous circadian clock, independent of light, is food intake. The circadian and immune systems are linked; forced desynchrony of the circadian clock via nighttime light exposure or genetic ablation of core clock components impairs immune function. The timing of food intake affects various aspects of the circadian clock, but its effects on immune function are unknown. We tested the hypothesis that temporal desynchrony of food intake alters innate immune responses. Adult male Swiss Webster mice were provided with food during the night, the day, or ad libitum for 4 wk, followed by administration of LPS prior to the onset of either the active phase (zeitgeber time [ZT]12: Experiment 1) or the inactive phase (ZT0: Experiment 2). Three hours after LPS administration, blood was collected, and serum was tested for bacteria-killing capacity against Escherichia coli, as a functional assay of immune function. Additionally, cytokine expression was examined in the serum (protein), spleen, and hypothalamus (mRNA). Day-fed mice suppressed bacteria-killing capacity and serum cytokine responses to LPS during the active phase (ZT12). Night-fed mice increased bactericidal capacity, as well as serum and hypothalamic mRNA responses of certain proinflammatory cytokines during the active phase. Only day-fed mice enhanced serum cytokine responses when LPS challenge occurred during the inactive phase (ZT0); this did not result in enhanced bactericidal capacity. These data suggest that mistimed feeding has functional relevance for immune function and provide further evidence for the integration of the circadian, metabolic, and immune systems.
Introduction
The circadian and immune systems are fundamentally connected; most cells of the immune system display autonomous circadian oscillations in gene expression (1, 2), and core clock genes regulate several key immune transcription factors driving rhythmic expression of cytokines, cell proliferation, and immune receptor expression and function (2–7). Additionally, circadian clock proteins regulate inflammatory responses, immune cell trafficking, and phenotype (4, 5, 7–10). Therefore, optimal immune function is dependent on a functional circadian clock.
Light is the most potent entraining cue for the circadian clock, acting via signal transduction in the suprachiasmatic nuclei (SCN) of the hypothalamus. The SCN convey timing information to the immune system via autonomic and humoral pathways, “setting” peripheral clocks and governing daily variations in immune function (11). Disruption of the circadian clock markedly deregulates inflammatory responses. Specifically, chronic jet lag enhances LPS-induced increases in serum cytokine concentrations (12). Chronic jet lag and exposure to light at night enhance hypothalamic cytokine expression in response to LPS (12, 13). Shortening of the circadian period dampens serum cytokine concentrations in response to LPS (14). These changes in immune response occur independently from effects of sleep loss or stress (12, 14–17).
Timing of food intake can also act as an entraining signal to the circadian clock; timed feeding synchronizes rhythmic clock gene expression in the SCN of mice housed in constant darkness and restores rhythmicity in mice housed in constant light (18, 19). Hypocaloric feeding (∼50% daily food intake) can alter the phase of clock gene expression in the SCN, independent of the light/dark cycle (20, 21). Additionally, timed feeding regimens can phase shift clock gene expression in the liver, independent of SCN phase or light/dark cycle (22, 23). Metabolic homeostasis requires robust circadian gene expression in the liver. Indeed, liver-specific clock gene knockout and disruption, through a high-fat diet and feeding during the inactive phase, disrupt metabolic homeostasis (24–32).
Because of the association between circadian timing and metabolic homeostasis, several research groups have used time-restricted feeding paradigms to tease apart circadian contributions to metabolic diseases, such as diabetes and obesity (32, 33). Most individuals in the developed world live in a “24/7” society, with access to food at all times of the day; obesity and metabolic disease are frequently related to environmental circadian disruption (34, 35) and a chronic inflammatory state (36–38). Obese patients exhibit increases in proinflammatory cytokines, acute-phase proteins (APPs), and chemokines that are reduced with weight loss and positively correlate with comorbidities, such as insulin resistance (39–43). However, the interaction between the timing of food intake and immune responses remains unspecified.
To investigate interactions among these systems, we used a time-restricted feeding protocol allowing adult male Swiss Webster mice access to food during the day, during the night, or ad libitum (AdLib) for 4 wk before administering an endotoxin challenge at the onset of the active phase (zeitgeber time [ZT]12: Experiment 1) or the onset of the inactive phase (ZT0: Experiment 2). We hypothesized that time-restricted feeding alters immune responses to an endotoxin challenge. If true, then we predicted that day-restricted feeding would impair bacteria-killing capacity and cytokine responses to LPS compared with night-fed or AdLib-fed mice.
Materials and Methods
Animals
Sixty adult male (>8 wk old) Swiss Webster mice (Charles River Laboratories, Kingston, NY) were used for each experiment; a separate set of 36 mice was used for the nighttime bacteria-killing assay. Mice were housed individually and allowed to recover for 1 wk after arriving at our facility to entrain to a 12:12 light/dark cycle (09:00–21:00 EST) and recover from shipping. Food (7912; Teklad; Harlan) and filtered tap water were provided AdLib. The following week, all mice were habituated to twice-daily cage switching (32). Mice were then randomly assigned to one of three groups: day fed, night fed, or AdLib fed. Day-fed mice were allowed access to chow during the 12-h light phase, after which they were transferred to a second cage that contained only water (no food). After the 12-h restriction period, mice were placed back into their food-containing cage. This was done to prevent “crumbling” or “hoarding” behavior, which may allow animals to eat outside of the restricted time frame. Night-fed animals were similarly allowed food only during the 12-h dark phase and were transferred to their second cage with water at the beginning of the light phase. AdLib animals were allowed access to food at all times but also experienced twice-daily cage switching to control for any possible stress of this manipulation. All procedures and experiments were approved by the Ohio State University Institutional Animal Care and Use Committee.
Endotoxin administration and tissue collection
After 4 wk of timed food restriction, mice received an i.p. injection of 0.5 mg/kg LPS (serotype 0111:B4; Sigma-Aldrich, St. Louis, MO) in sterile saline or saline alone 1 h prior to the onset of the dark phase (ZT11–12: Experiment 1) or the light phase (ZT23–0: Experiment 2). Three hours later (ZT15 and ZT3), mice were lightly anesthetized with isoflurane and rapidly decapitated, and blood and tissues were collected. Blood was centrifuged at 4°C, and serum was removed, aliquoted, and stored at −80°C until bacteria-killing and multiplex assays. Spleens, adrenal glands, and brains were also collected. Spleens and adrenal glands were weighed and immediately flash frozen on dry ice. Brains were placed into RNAlater reagent (QIAGEN) on ice, and hypothalami were later dissected, flash frozen, and stored at −80°C until quantitative PCR analysis.
Bacteria killing assay
The bacteria-killing assay is an ex vivo assessment of innate immunity mediated by complement proteins and natural Abs. Samples were immediately centrifuged at 4°C for 25 min at 4000 × g, and serum aliquots were stored at −80°C until assayed. Serum samples were diluted 1:20 in CO2-independent media (Life Technologies, Carlsbad, CA) under a laminar flow hood. A standard number of CFU Escherichia coli (E-POWER 0483E7; Thermo Fisher Scientific) was added to each serum sample at a ratio of 1:10. Serum–bacteria mixtures were incubated for 30 min at 37°C and plated in duplicate onto tryptic-soy agar plates (Teknova, Hollister, CA) using sterile technique. Two plates were spread with diluted bacteria alone as positive controls, and two were spread with media alone as negative controls. All plates were incubated at 37°C overnight, and total CFU were quantified by an experimenter who was blinded to group assignments. Total CFU were averaged across the duplicates for each animal and then compared with the average of the positive control plates to calculate the percentage of bacteria killed. Neither negative control plates contained CFU.
Serum multiplex assay
To examine serum cytokine protein profiles, a 10-plex cytokine (V-PLEX Proinflammatory Panel 1; Meso Scale Diagnostics) panel was performed, according to the manufacturer’s instructions. This kit measures protein levels of IFN-γ, IL-10, IL-12p70, IL-1β, IL-2, IL-4, IL-5, IL-6, KC/GRO (CXCL1), and TNF-α.
Quantitative real-time PCR
RNA was extracted using TRIzol Reagent (Invitrogen), according to the manufacturer’s instructions. RNA quality and quantity were determined using a spectrophotometer (NanoDrop), and cDNA was synthesized using M-MLV reverse transcriptase. Forty nanograms of cDNA per reaction were used in subsequent PCRs. TaqMan Fast Advanced Master Mix (Life Technologies) containing AmpliTaq Fast DNA Polymerase was used in a 20-μl duplex reaction with one of the primer/probe pairs listed below and a primer-limited primer/probe for the endogenous control eukaryotic 18s rRNA. Gene expression was assayed for tnf-α (Mm00443260_g1), il-6 (Mm00446190_m1), and il-1β (Mm00434228_m1). The two-step real-time PCR cycling conditions used were 95°C for 20 s, 40 cycles of 95°C for 3 s, and then 60°C for 30 s. Relative gene expression was calculated using the Pfaffl method (44). Data are expressed as fold change from saline-injected animals on the same feeding regimen.
Statistics
Differences in body mass were analyzed using one-way ANOVA. Two way ANOVA, assessing the effects of time of feeding, injection, and interactions, was conducted on other somatic measures, bacteria killing capacity, nighttime serum IL-10, IL-12, IL-1β, IL-6, and TNF-α, splenic tnf-α expression, as well as daytime serum IFN-γ, IL-1β, IL-5, and TNF-α. If the data did not meet the assumptions of normality or equal variance, then nonparametric tests (Mann–Whitney and Kruskal–Wallis tests) were conducted, followed by a Dunn post hoc test. Data that fell into this latter category were nighttime serum IFN-γ, IL-2, IL-4, IL-5, and CXCL1; splenic il-1β and il-6; hypothalamic tnf-α, il-1β, and il-6; and daytime serum IL-10, IL-12, IL-2, IL-4, IL-6, and CXCL1; and daytime splenic and hypothalamic gene expression. Samples were excluded by Z-score analysis (>2 SD) and gene expression analysis if 18S CT value was >15. Mean differences were considered statistically significant at p ≤ 0.05 for all analyses. Statistical analyses were conducted using SPSS Statistics V22.0 (IBM, Armonk, NY) and visualized using Prism 7 (GraphPad Software, La Jolla, CA).
Results
Experiment 1: nighttime LPS
Somatic measures.
Neither body nor adrenal mass was affected by 4 wk of timed food restriction at ZT15 (p > 0.05, Fig. 1A, 1B, respectively). Spleen mass was increased in mice that received LPS (F1,53 = 8.48, p < 0.01, Fig. 1C).
Night-fed mice had enhanced, whereas day-fed mice eliminated, bactericidal capacity in response to a nighttime LPS challenge. Graphs depict data from nighttime (A–D) and daytime (E–H) LPS administration in timed-fed mice. (A and E) Body mass after 4 wk of time-restricted feeding. Adrenal (B and F) and spleen (C and G) mass 3 h post LPS injection (D and H). n = 5–10 mice per feeding group per injection; error bars represent SEM. Bars sharing the same letter are not statistically significant in comparison to one another. *p < 0.05 saline versus LPS.
Night-fed mice had enhanced, whereas day-fed mice eliminated, bactericidal capacity in response to a nighttime LPS challenge. Graphs depict data from nighttime (A–D) and daytime (E–H) LPS administration in timed-fed mice. (A and E) Body mass after 4 wk of time-restricted feeding. Adrenal (B and F) and spleen (C and G) mass 3 h post LPS injection (D and H). n = 5–10 mice per feeding group per injection; error bars represent SEM. Bars sharing the same letter are not statistically significant in comparison to one another. *p < 0.05 saline versus LPS.
Bacteria killing.
Night-fed mice injected with LPS killed more bacteria than all other groups in a serum bactericidal assay (F2,28 = 7.96, p < 0.05; Tukey test, p < 0.05, Fig. 1D). Day-fed mice did not display a bactericidal response to LPS (Tukey test, p > 0.05).
Serum.
Day-fed mice had decreased serum TNFα, IL-12p70, IL-10, IL-2, IL-5, CXCL1, and IL-6 concentrations in response to LPS relative to night-fed and AdLib-fed mice (Tukey test, p < 0.05, Fig. 2A–G). Night-fed mice had increased serum IL-6 and IFN-γ concentrations in response to LPS relative to day- and AdLib-fed mice (Tukey test, p < 0.05, Fig. 2G, 2H). The factorial main effects of LPS and timed feeding, as well as interactions, are presented in Table I.
Serum cytokines following a nighttime (ZT12) endotoxin challenge. Four weeks of time-restricted feeding blunted serum TNF-α (A), IL-12p70 (B), IL-10 (C), IL-2 (D), IL-5 (E), CXCL1 (F), and IL-6 (G) responses to LPS in day-fed mice. Night-fed mice had increased IL-6 (G) and IFN-γ (H) responses to LPS. n = 6–10 mice per feeding group. Error bars represent SEM for parametric data and 95% confidence intervals for nonparametric data. Bars sharing the same letter are not statistically significant in comparison to one another.
Serum cytokines following a nighttime (ZT12) endotoxin challenge. Four weeks of time-restricted feeding blunted serum TNF-α (A), IL-12p70 (B), IL-10 (C), IL-2 (D), IL-5 (E), CXCL1 (F), and IL-6 (G) responses to LPS in day-fed mice. Night-fed mice had increased IL-6 (G) and IFN-γ (H) responses to LPS. n = 6–10 mice per feeding group. Error bars represent SEM for parametric data and 95% confidence intervals for nonparametric data. Bars sharing the same letter are not statistically significant in comparison to one another.
Cytokine . | LPS . | Timed Feeding . | Interaction . |
---|---|---|---|
TNF-α | p < 0.001, U = 0.00 | p = 0.41, χ2 = 1.79 | p < 0.01, χ2 = 39.49 |
IL-1β | p < 0.001, U = 0.00 | p = 0.58, χ2 = 1.11 | p < 0.01, χ2 = 37.64 |
IL-6 | p < 0.001, U = 7.00 | p = 0.18, χ2 = 3.47 | p < 0.01, χ2 = 38.22 |
IFN-γ | p < 0.001, U = 0.00 | p = 0.166, χ2 = 3.59 | p < 0.01, χ2 = 37.68 |
IL-12p70 | p < 0.001, U = 16.00 | p = 0.31, χ2 = 2.34 | p < 0.01, χ2 = 31.63 |
IL-2 | p < 0.001, U = 96.00 | p = 0.19, χ2 = 3.32 | p < 0.01, χ2 = 21.77 |
IL-4 | p = 0.09, U = 118.00 | p = 0.39, χ2 = 1.88 | p = 0.28, χ2 = 6.32 |
IL-5 | p < 0.05, U = 150.00 | p < 0.01, χ2 = 32.32 | p < 0.01, χ2 = 41.94 |
IL-10 | p < 0.001, U = 0.00 | p = 0.45, χ2 = 1.61 | p < 0.01, χ2 = 38.56 |
CXCL1 | p < 0.001, U = 90.00 | p < 0.05, χ2 = 7.25 | p < 0.001, χ2 = 26.36 |
Cytokine . | LPS . | Timed Feeding . | Interaction . |
---|---|---|---|
TNF-α | p < 0.001, U = 0.00 | p = 0.41, χ2 = 1.79 | p < 0.01, χ2 = 39.49 |
IL-1β | p < 0.001, U = 0.00 | p = 0.58, χ2 = 1.11 | p < 0.01, χ2 = 37.64 |
IL-6 | p < 0.001, U = 7.00 | p = 0.18, χ2 = 3.47 | p < 0.01, χ2 = 38.22 |
IFN-γ | p < 0.001, U = 0.00 | p = 0.166, χ2 = 3.59 | p < 0.01, χ2 = 37.68 |
IL-12p70 | p < 0.001, U = 16.00 | p = 0.31, χ2 = 2.34 | p < 0.01, χ2 = 31.63 |
IL-2 | p < 0.001, U = 96.00 | p = 0.19, χ2 = 3.32 | p < 0.01, χ2 = 21.77 |
IL-4 | p = 0.09, U = 118.00 | p = 0.39, χ2 = 1.88 | p = 0.28, χ2 = 6.32 |
IL-5 | p < 0.05, U = 150.00 | p < 0.01, χ2 = 32.32 | p < 0.01, χ2 = 41.94 |
IL-10 | p < 0.001, U = 0.00 | p = 0.45, χ2 = 1.61 | p < 0.01, χ2 = 38.56 |
CXCL1 | p < 0.001, U = 90.00 | p < 0.05, χ2 = 7.25 | p < 0.001, χ2 = 26.36 |
Main effects and interactions of results from a cytokine multiplex. For parametric tests, p and F values are reported. For nonparametric tests, p and U values are reported for the Mann–Whitney U test, and p and χ2 values are presented for the Kruskal–Wallis test. Bold type indicates statistical significance.
Spleen.
Night-fed mice had elevated splenic gene expression of tnf-α 3 h post-LPS relative to other timed feeding groups (F2,21 = 3.70, p < 0.05, Fig. 3A). No statistical differences were observed in splenic il-1β or il-6 expression (p > 0.05, Fig. 3B, 3C). Timed feeding did not affect expression of tnf-α, il-1β, or il-6 in the spleens of saline-treated mice, regardless of the feeding group (p > 0.05).
Splenic and hypothalamic proinflammatory gene expression following a nighttime (ZT12) endotoxin challenge. Night-fed mice had increased splenic tnf-α in response to LPS (A), but it had no effect on il-1β or il-6 expression (B and C). Night-fed mice had increased hypothalamic tnf-α, il-1β, and il-6 expression (D–F, respectively). n = 7–10 mice per feeding group; error bars represent SEM for parametric data and 95% confidence intervals for nonparametric data. Bars sharing the same letter are not statistically significant in comparison to one another.
Splenic and hypothalamic proinflammatory gene expression following a nighttime (ZT12) endotoxin challenge. Night-fed mice had increased splenic tnf-α in response to LPS (A), but it had no effect on il-1β or il-6 expression (B and C). Night-fed mice had increased hypothalamic tnf-α, il-1β, and il-6 expression (D–F, respectively). n = 7–10 mice per feeding group; error bars represent SEM for parametric data and 95% confidence intervals for nonparametric data. Bars sharing the same letter are not statistically significant in comparison to one another.
Hypothalamus.
Night-fed mice had elevated hypothalamic tnf-α, il-1β, and il-6 gene expression 3 h post-LPS relative to all other feeding groups (F2,23 = 4.07, F2,24 = 5.62, F2,24 = 4.18, p < 0.05, Fig. 3D–F, respectively). Baseline gene expression of tnf-α, il-1β, and il-6 was not altered by timed feeding in the hypothalami of saline-treated mice (p > 0.05).
Experiment 2: daytime LPS
Somatic measures.
Body mass was also equivalent among groups after 4 wk of timed food restriction when assessed at ZT3 (p > 0.05, Fig. 1E), although adrenal mass was altered by timed feeding at this time point (F2,53 = 3.32, p < 0.05, Fig. 1F). LPS increased adrenal mass in day-fed mice relative to AdLib-fed mice (p < 0.05, Fig. 1F). Spleen mass was not altered by LPS or timed feeding in daytime-injected animals (p > 0.05, Fig. 1G).
Bacteria killing.
Neither LPS nor timed feeding altered serum bactericidal capacity when injected at ZT0 (p > 0.05, Fig. 1H).
Serum.
Compared with night-fed and AdLib-fed mice, day-fed mice had increased concentrations of IL-6 and CXCL1 in response to LPS (Tukey test, p < 0.05, Fig. 4A, 4B). Day-fed mice also had increased IFN-γ relative to night-fed mice (Dunn test, p < 0.05, Fig. 4C). The factorial main effects of LPS and timed feeding, as well as interactions, are presented in Table II.
Serum cytokines following a daytime (ZT0) endotoxin challenge. Daytime feeding blunted serum IL-6 (A), CXCL1 (B), and IFN-γ (C) responses to LPS. n = 7–9 mice per feeding group. Error bars represent SEM for parametric data and 95% confidence intervals for nonparametric data. Bars sharing the same letter are not statistically significant in comparison to one another.
Serum cytokines following a daytime (ZT0) endotoxin challenge. Daytime feeding blunted serum IL-6 (A), CXCL1 (B), and IFN-γ (C) responses to LPS. n = 7–9 mice per feeding group. Error bars represent SEM for parametric data and 95% confidence intervals for nonparametric data. Bars sharing the same letter are not statistically significant in comparison to one another.
Cytokine . | LPS . | Timed Feeding . | Interaction . |
---|---|---|---|
TNF-α | p < 0.001, U = 26.00 | p = 0.695, χ2 = 4.39 | p < 0.001, χ2 = 1.19 |
IL-1β | p < 0.001, U = 25.00 | p = 0.78, χ2 = 0.50 | p < 0.001, χ2 = 33.68 |
IL-6 | p < 0.001, U = 27.00 | p = 0.088, χ2 = 4.87 | p < 0.001, χ2 = 38.87 |
IFN-γ | p < 0.001, U = 25.00 | p = 0.07, χ2 = 5.31 | p < 0.001, χ2 = 39.02 |
IL-12p70 | p < 0.001, U = 17.00 | p = 0.98, χ2 = 0.61 | p < 0.001, χ2 = 37.34 |
IL-2 | p < 0.001, U = 79.00 | p = 0.78, χ2 = 0.50 | p < 0.001, χ2 = 24.40 |
IL-4 | p < 0.001, U = 53.00 | p = 0.83, χ2 = 0.36 | p < 0.001, χ2 = 29.17 |
IL-5 | p < 0.001, F = 44.82 | p = 0.053, F = 3.14 | p = 0.053, F = 3.14 |
IL-10 | p < 0.001, U = 24.00 | p = 0.66, χ2 = 0.83 | p < 0.001, χ2 = 33.73 |
CXCL1 | p < 0.001, F = 395.13 | p < 0.05, F = 4.48 | p = 0.052, F = 3.16 |
Cytokine . | LPS . | Timed Feeding . | Interaction . |
---|---|---|---|
TNF-α | p < 0.001, U = 26.00 | p = 0.695, χ2 = 4.39 | p < 0.001, χ2 = 1.19 |
IL-1β | p < 0.001, U = 25.00 | p = 0.78, χ2 = 0.50 | p < 0.001, χ2 = 33.68 |
IL-6 | p < 0.001, U = 27.00 | p = 0.088, χ2 = 4.87 | p < 0.001, χ2 = 38.87 |
IFN-γ | p < 0.001, U = 25.00 | p = 0.07, χ2 = 5.31 | p < 0.001, χ2 = 39.02 |
IL-12p70 | p < 0.001, U = 17.00 | p = 0.98, χ2 = 0.61 | p < 0.001, χ2 = 37.34 |
IL-2 | p < 0.001, U = 79.00 | p = 0.78, χ2 = 0.50 | p < 0.001, χ2 = 24.40 |
IL-4 | p < 0.001, U = 53.00 | p = 0.83, χ2 = 0.36 | p < 0.001, χ2 = 29.17 |
IL-5 | p < 0.001, F = 44.82 | p = 0.053, F = 3.14 | p = 0.053, F = 3.14 |
IL-10 | p < 0.001, U = 24.00 | p = 0.66, χ2 = 0.83 | p < 0.001, χ2 = 33.73 |
CXCL1 | p < 0.001, F = 395.13 | p < 0.05, F = 4.48 | p = 0.052, F = 3.16 |
Main effects and interactions of results from a cytokine multiplex. For parametric tests, p and F values are reported. For nonparametric tests, p and U values are reported for the Mann–Whitney U test, and p and χ2 values are presented for the Kruskal–Wallis test. Bold type indicates statistical significance.
Spleen.
Timed feeding did not alter splenic cytokine response to LPS (p > 0.05, Fig. 5A–C). Timed feeding did not affect expression of tnf-α, il-1β, or il-6 in the spleen of saline-treated mice, regardless of feeding group (p > 0.05).
Splenic and hypothalamic proinflammatory gene expression following a daytime (ZT0) endotoxin challenge. Time-restricted feeding did not alter splenic (A–C) or hypothalamic (D–F) tnf-α, il-1β, or il-6 expression in response to LPS. n = 6–10 mice per feeding group; error bars represent SEM for parametric data and 95% confidence intervals for nonparametric data. Bars sharing the same letter are not statistically significant in comparison to one another.
Splenic and hypothalamic proinflammatory gene expression following a daytime (ZT0) endotoxin challenge. Time-restricted feeding did not alter splenic (A–C) or hypothalamic (D–F) tnf-α, il-1β, or il-6 expression in response to LPS. n = 6–10 mice per feeding group; error bars represent SEM for parametric data and 95% confidence intervals for nonparametric data. Bars sharing the same letter are not statistically significant in comparison to one another.
Hypothalamus.
Timed feeding did not alter hypothalamic cytokine response to LPS (p > 0.05, Fig. 5D–F). Baseline expression of tnf-α, il-1β, or il-6 in the hypothalami of saline-treated mice was not altered by timed feeding (p > 0.05).
Discussion
The reciprocal relationship between the immune and circadian systems has been thoroughly established (7, 45). Circadian disruption, through alterations in the lighting environment, alters innate immune responses (12–14). However, the circadian system can also be entrained by nonphotic factors. The liver and, in some cases, the SCN can be entrained by time-restricted feeding (20–23). With the increase in industrialization during the last century, most individuals in the developed world live in a 24/7 society with access to food at all times of the day. Mistimed food consumption is associated with altered circadian rhythms in the liver and disruptions in metabolic homeostasis (30–32). However, the effects of timed feeding on physiological systems other than metabolism have not been investigated. We hypothesized that feeding during the inactive phase would alter functional and cytokine responses to a bacterial endotoxin challenge.
Daytime-restricted feeding impaired LPS-primed bacterial killing during the active and inactive phase (ZT12 and ZT0) compared with saline-treated mice (Fig. 1D, 1H). In contrast, night-fed and AdLib-fed mice showed significant enhancement of bacteria-killing capacity when challenged prior to the active, but not the inactive, phase (Fig. 1D, 1H). Blood bactericidal capacity has been used in humans and animal models to test constitutive innate immunity and predict susceptibility to bacterial infections (46–48). Serum bactericidal capacity is dependent on the production of APPs and complement (49). The acute-phase response (APR) is a rapid complex nonspecific innate immune response mediated by a large set of APPs (50). Killing of E. coli (0483E7) is complement dependent (47), and decreased bactericidal capacity in day-fed animals may suggest that circadian disruption induced deficits in signals driving APP and complement synthesis (49, 51, 52).
IL-6 has long been identified as a major regulator of the APR (53). Diurnal variation in IL-6 and TNF-α responses to endotoxin are driven by macrophage circadian rhythms, with maximal responses during the late light/early dark phase (2). Day-fed mice eliminated serum IL-6 and TNF-α responses at ZT12 and enhanced IL-6 responses at ZT0 (Figs. 2G, 2C, 4A). Circadian variation in IL-6 and TNF-α responses to LPS is driven by expression of the core clock gene REV-ERBα in macrophages (9). REV-ERBα couples circadian rhythms and fed/fasted states to maintain metabolic homeostasis. Indeed, daytime-restricted feeding inverses the phase of several clock genes, including rev-erbα (nr1d1) in the liver (54). REV-ERBα may act as a functional integrator of metabolic state and immune function driving altered circadian phase in immune cells, resulting in deregulation of innate immune responses.
In addition to direct regulation of cytokine expression, clock genes alter the expression of key transcription factors regulating immune responses, including Stat3 and NF-κB (55, 56). STAT3 mediates the IL-6–driven transcriptional regulation of APPs (57). Conversely, STAT3 inhibition is necessary for induction of CXCL1, to recruit and promote survival of neutrophils (58). Deletion of STAT3 results in abrogated serum IL-6, TNF-α, IL-1β, and IL-10 responses to LPS (57). With the exception of IL-1β, these responses are similar to the diminished responses to LPS in day-fed mice when administered LPS at the onset of the active phase (Fig. 2A, 2C, 2G). Overall, this suggests that day-fed mice may have impaired IL-6 release or downstream signaling through the JAK–STAT pathway.
Day-fed mice also had decreased concentrations of IL-12p70, IL-2, and IL-5 in response to LPS during the active phase relative to their AdLib- and night-fed counterparts (Fig. 2B, 2D, 2E), suggesting deficits in immune cell communication, specifically in the differentiation and expansion of Th1 cells (59–61). IL-12p70 and IL-2 also contribute to the APR by increasing production of opsonizing and complement-fixing IgG subclasses (62, 63). These data further support diminished immune responses in day-fed mice in response to LPS.
We further investigated splenic and hypothalamic cytokine responses to the endotoxin challenge. Night-fed mice had increased splenic tnf-α relative to day-fed counterparts when challenged prior to the active phase (Fig. 3A). Disruption of circadian clock function in peripheral organs through alterations to the light/dark cycle requires communication between the SCN and peripheral oscillators. Time-restricted feeding is not sufficient to alter SCN phase without caloric restriction (20, 21). Phase shifts of circadian clocks in the liver (22, 23), kidney, heart, pancreas (54), and muscle (64) have been reported in response to time-restricted feeding. Within the immune system, the effect of restricted feeding has only been assessed on platelet-producing cells; daytime-restricted feeding phase shifts clock genes in megakaryocytes, altering gene expression of transcription factors involved in megakaryopoiesis (65, 66). These data suggest that timed feeding can modulate immunity, independent from descending input from the brain.
Activation of the innate immune system triggers physiological responses, such as the APR, as well as behavioral responses. Sickness responses are a host of behavioral adaptations that develop over the course of an infection to aid in survival and are triggered by the production of proinflammatory cytokines TNF-α, IL-1β, and IL-6 in the hypothalamus. Specifically, IL-1β is the primary trigger of fever and anorexic and anhedonic behavior, as well as activation of the hypothalamic–pituitary–adrenal axis (67–70). Night-fed mice had increased hypothalamic tnf-α, il-1β, and il-6 expression when challenged prior to the inactive phase (Fig. 3D–F). Combined with increased serum IL-6 responses to LPS, these data suggest that night-fed mice enhance central and peripheral signaling in response to an LPS challenge during the inactive phase.
Immune function can be modulated by a variety of factors, including adiposity (71, 72). Restricting food intake to the inactive phase, the daytime in nocturnal rodents, has been shown to increase body mass (17). In this study, we did not observe increased body mass in response to daytime feeding (Fig. 1A, 1E). Previous studies used a 16:8 light/dark cycle, where only the food hoppers, but not the cages, were switched twice daily, and mice were maintained on timed food restriction for 8 wk (17). In our study, mice were maintained on a 12:12 light/dark cycle, and cages were switched twice daily in accordance with Hatori et al. (32), for a total of 4 wk of timed food restriction. The differential response to daytime feeding in body mass may be a result of differences in methodology.
Together, our data provide evidence that nonphotic circadian disruption, through mistimed feeding, impairs immune function. Decreases in serum bactericidal capacity in day-fed animals are accompanied by deficits in proinflammatory cytokine concentrations in serum and production in the spleen. Conversely, night-restricted feeding is comparable to, and, in some cases, enhances, cytokine production and bacteria killing relative to AdLib-fed mice. Future studies should address whether timed feeding disrupts clock gene expression in immune organs. Around-the-clock feeding is commonplace in modern society, and its effects on metabolism have been well documented (73, 74). These data expand upon the effects of mistimed feeding and suggest that mistimed food intake contributes to deficits in immune defenses.
Acknowledgements
We thank the undergraduate research assistants for aid with daily cage switching during this experiment, including Tial KaiKai Tinkai, Curtis Stegman, Reuben Don, Adam Weiss, Evan Thomas, and Anna Suresh. We further thank Jamie Tussing and The Ohio State University Laboratory Animal Resource staff for providing excellent care of the animals in these studies.
Footnotes
This work was supported by National Science Foundation Grant IOS-11-18792. Y.M.C. was funded by National Institutes of Health/National Institute of Environmental Health Sciences Grant F31-ES026890-02.
References
Disclosures
The authors have no financial conflicts of interest.