Four-week-old outbred female CD-1 mice (17∼22 g, n = 20) were purchased from Charles River Laboratories. Mice were individually caged and had free access to diet and water. All experimental procedures involving animals were approved by the Animal Care and Use Committee of Tufts University and conducted according to the NIH Guidelines for the Care and Use of Laboratory Animals (approval number H2013-135). Mice were randomly assigned to two groups: Low fat (LF, 10% calories from fat, Research Diets, #D12450J) diet (n = 6) and high fat (HF, 45% calories from fat, Research Diets, #D12451) diet (n = 14) group. The compositions of the two purified diets are provided in Supplementary Table 1. During the feeding period, body weight was recorded weekly. After 10-week of feeding with their respective diets, body weight increased in all mice but the weight gain in HF group varied greatly. Based on the average body weight of the obese mice, mice in HF group were further divided into HF lean (HF-L, BW < 34.6 g) and HF obese group (HF-O, BW > 34.6 g) (Figure 1A).

Next, daily food intake was recorded for three consecutive days. Total food intake (g/day), relative food intake (g/g of body weight per day), and total caloric intake (kcal/day), and relative caloric intake (kcal/g of body weight per day) were determined. Mice were fed their respective diets for another 15-week, at which time they were immunized twice with influenza vaccine and infected with influenza virus as detailed below.

After a total of 25-week of feeding, mice were immunized with an influenza A/Puerto Rico/8/34 vaccine as previously reported (22). Briefly, the mice were intraperitoneally injected with 200 μL of the influenza vaccine, which was prepared by mixing the hemagglutinin protein of influenza A strain A/Puerto Rico/8/1934 (H1N1) (Sino Biological) and aluminum adjuvant (Thermo Fisher Scientific). The immunization was repeated 3-week later (booster). Blood was collected 4-week after the booster for antibody titer analysis. Five week after booster injection, under anesthesia, mice were infected intranasally with 50 hemagglutinin units of influenza A/Puerto Rico/8/34 virus diluted in 0.05 mL PBS (kindly provided by Dr. Melinda Beck of UNC Gillings School of Global Public Health). Five unvaccinated CD-1 mice fed LF diet were infected with influenza virus as the control group to validate success of infection. Body weight was recorded daily for one month as indicator of severity of infection, and then the mice were euthanized with CO₂ asphyxiation followed by exsanguination.

Blood was collected after influenza vaccination boost and 30 days post-infection at animal sacrifice, respectively, and the serum was separated and stored in −80°C for determining serum antibody titer. The serum antibody titer was assessed according to published methods (23, 24), with minor modification. Briefly, serum was incubated with receptor destroying enzyme (RDE) (Denka Seiken, Tokyo, Japan) overnight at 37°C to remove non-specific inhibitors of hemagglutination (HA), The remaining RDE enzyme was heat inactivated at 56°C for 40 min, followed by adding PBS for a final 1:10 dilution of the serum. Twenty-five μL RDE-treated serum was twofold serially diluted in v-bottom microtiter plates, and then, 25μL of influenza virus (4 HAU) was added to each well followed by incubation at room temperature for 40 min. Next, 50 μL 1% Turkey red blood cells (TRBC) (Lampire Biologicals, Pipersville) in PBS was added to each well, and the plates were mixed by agitation, covered, and the RBCs were allowed to settle for 45 min at room temperature. HAI titer was determined by the reciprocal dilution of the last row which contained non-agglutinated RBC. Positive and negative serum controls were included for each plate.

Splenic lymphocytes were isolated for assessment of ex vivo cytokine production as previously reported (25). Levels of selected T cell cytokines in the cell culture supernatant of the cultured splenic lymphocytes stimulated by LPS, T cell mitogen concanavalin A (Con A), or anti-CD3/CD28 T cell receptor antibodies were determined by ELISA. Briefly, splenic lymphocytes were cultured in the presence of LPS (1.0 μg/mL) for 24 h for inflammatory interleukin (IL)-1β, IL-6, TNF-α production or in the presence of Con A (1.5 μg/mL) or anti-CD3 (5 μg/mL)/anti-CD28 (2 μg/mL) for 48 h for IFN-γ, IL-2, IL-4, and IL-10 production. Cell-free supernatants were collected at the end of incubation and analyzed using ELISA kits (R&D Systems, Minneapolis, MN) for all the cytokines according to the manufacturer’s instructions.

All results were tested for normality and homogeneity of variance. A two-tailed t test was used to compare two group means, while the one-way analysis of variance (ANOVA) and Tukey’s post hoc tests were performed to compare three group means. Correlation coefficients were calculated by using the non-parametric Spearman correlation. Statistical analysis was performed using IBM SPSS Statistics (version 25) software. Results are means ± SD. Significance was set at p < 0.05.

After 10-week of feeding the respective diets, mice fed the LF diet weighed 30.1 ± 3.6 g (n = 5, one mice fed the LF diet died of unknown cause), whereas those fed the HF diet exhibited high variation in weight gain (34.6 ± 8.2 g, n = 14) (Supplementary Figure 1A). Thus, based on the average body weight of the CD1 mice fed HF diet, we further divided the HF-fed mice into two sub-groups: HF lean (HF-L, BW < 34.6 g, n = 8) and HF obese group (HF-O, BW > 34.6 g, n = 6). The average body weight of HF-L group was similar to that of LF mice, while HF-O mice weighed significantly more than LF or HF-L mice (both at p < 0.001) (Figure 1A and Supplementary Figure 1B). Similarly, total food intake (g/day) (Figure 1C) and total calorie intake (kcal/day) (Figure 1E) in HF-O group were significantly higher than those in LF mice and HF-L mice. However, after normalizing to body weight, relative food intake (g/g BW) (Figure 1D) and relative caloric intake (Kcal/g BW) (Figure 1F) of HF-O mice were significantly lower than LF mice and HF-L mice, suggesting that HF-O mice might have less energy expenditure and/or higher energy absorption efficiency. In addition, both HF-L mice and HF-O mice had significantly higher total fat intake (mg/day) (Figure 1G) and relative fat intake (mg/g of BW) (Figure 1H) than LF mice. However, when corrected by body weight, HF-O mice had significantly lower relative fat intake than HF-L mice (Figure 1H). Taken together, these observations suggest that the observed differences in weight between HF-O and HF-L is independent of fat intake.

After the mice were fed their respective diets for another 15-week, the body weight of HF-O mice was still significantly higher than that of LF and HF-L mice while there were no significant differences in body weight between LF and HF-L mice (Figure 1B).

After total of 25-week of feeding, mice were vaccinated with influenza virus and boosted 3-week later (28-week of feeding). Five week after vaccination (33-week after feeding), mice in LF, HF-L and HF-O groups were infected with influenza virus, and a group (n = 5) of unvaccinated CD-1 mice fed the LF diet was also infected in parallel to serve as the positive control for infection. The unvaccinated control mice had significantly higher weight loss (a primary symptom of influenza infection in rodent models) than the vaccinated mice after infection. The weight loss started on day 6 post viral infection and continued to day 12 post-infection (peak days), which confirmed the effectiveness of both infection and protection by the vaccine (Supplementary Figure 2). It is worth noting that, although the unvaccinated mice gradually recovered to their initial weight by day 12 post-infection, the mean weight gain of unvaccinated mice before termination of infection study was generally lower than that of vaccinated mice.

Of the three vaccinated groups, HF-O mice had significantly more weight loss after influenza virus challenge compared to LF and HF-L mice, and there was no significant difference in body weight loss between LF and HF-L mice (Figure 2). Further, Spearman correlation analysis demonstrated significant associations between initial (pre-infection) body weight and infection-induced absolute body weight loss (rho = −0.6321, p = 0.01) and percentage weight loss (rho = −0.6143, p = 0.01), indicating that the mice with larger initial body weight had more weight loss after infection. These results suggest that obesity rather than high fat content of the diet contributes to more severe infection outcome following vaccination.

In order to elucidate mechanism(s) of more weight loss in HF-O mice, we examined antibody titer levels and ex vivo T cell cytokine production by splenic lymphocytes. There was no significant difference in antibody titer within the group and among the groups (Figure 3). HF-O mice had significantly higher IFN-γ production in both Con A and anti-CD3/CD28-stimulated splenic lymphocytes compared to LF mice. There was no significant difference in IFN-γ production from Con A or anti-CD3/CD28-stimulated splenic lymphocytes between LF mice and HF-L mice (Figures 4A,B). Further IFN-γ production tended to be higher in HFO mice compared to HF-L mice (p = 0.0774 for Con A and p = 0.0662 for anti-CD3/CD28 stimulation). In addition, we found that CD3/CD28-stimulated IFN-γ production from splenic lymphocytes was positively associated with final body weight (rho = 0.6536, p = 0.01), negatively correlated with absolute body weight loss (rho = -0.5571, p = 0.03), and percent body weight loss (rho = −0.5179, p = 0.05). These results suggest that obese mice may produce more IFN-γ, which, in turn, might contribute to an inflammatory state and higher body weight loss during influenza infection.

Excepting that HF-O mice tended to have higher TNFα production (215.5 ± 58.9 pg/mL) from Con A-stimulated splenic lymphocytes (p = 0.09) compared to LF mice (199.7 ± 42.1 pg/mL), no significant difference was found in ex vivo production of other cytokines (IL2, IL-4, and TNFα) among the three groups (Supplementary Figure 3).