Metabolic and functional impairment of CD8⁺ T cells from the lungs of influenza-infected obese mice

Abstract

Obesity is an independent risk factor for morbidity and mortality in response to influenza infection. However, the underlying mechanisms by which obesity impairs immunity are unclear. Herein, we investigated the effects of diet-induced obesity on pulmonary CD8⁺ T cell metabolism, cytokine production, and transcriptome as a potential mechanism of impairment during influenza virus infection in mice. Male C57BL/6J lean and obese mice were infected with sub-lethal mouse-adapted A/PR/8/34 influenza virus, generating a pulmonary anti-viral and inflammatory response. Extracellular metabolic flux analyses revealed pulmonary CD8⁺ T cells from obese mice, compared with lean controls, had suppressed oxidative and glycolytic metabolism at day 10 post-infection. Flow cytometry showed the impairment in pulmonary CD8⁺ T cell metabolism with obesity was independent of changes in glucose or fatty acid uptake, but concomitant with decreased CD8⁺GrB⁺IFNγ⁺ populations. Notably, the percent of pulmonary effector CD8⁺GrB⁺IFNγ⁺ T cells at day 10 post-infection correlated positively with total CD8⁺ basal extracellular acidification rate and basal oxygen consumption rate. Finally, next-generation RNA sequencing revealed complex and unique transcriptional regulation of sorted effector pulmonary CD8⁺CD44⁺ T cells from obese mice compared to lean mice following influenza infection. Collectively, the data suggest diet-induced obesity increases influenza virus pathogenesis, in part, through CD8⁺ T cell-mediated metabolic reprogramming and impaired effector CD8⁺ T cell function.

1 |. INTRODUCTION

Influenza virus causes significant global morbidity and mortality each year. High-risk groups for increased mortality from influenza include children under the age of five, the elderly, pregnant women, individuals with compromised lung or heart function, and individuals with obesity. Following infection with influenza virus, individuals with obesity have an increased risk of morbidity and mortality.^1–4 Importantly, as of 2014, the global burden of obesity surpassed underweight for the first time in recorded history, signifying an epidemic shift in the burden of increased weight on human health.⁵ This intersection between a chronic disease condition (obesity) and an acute infectious disease (influenza) represents a confluence of public health problems for the 21^st century. Currently, 10% of the global population, including 42% of the U.S. population, are classified as obese,^6,7 increasing the risk for influenza-associated morbidity and mortality.

The immune response to influenza infection and vaccination relies on innate and adaptive immune cell systems, both of which are impaired in the obese host.^8–17 These impairments are thought to be responsible for the observed increased risk to both seasonal and epidemic/pandemic influenza in adults with obesity.^2,18 In humans, obesity is associated with greater rates of hospitalization and severe influenza,^3,19 and a sustained duration of influenza viral shedding.²⁰ Our own studies have found influenza vaccination of adults with obesity, when compared with lean adults, results in a suboptimal vaccine response.¹³ When stimulated with influenza antigens, influenza-specific CD4⁺ and CD8⁺ T cells from vaccinated adults with obesity have reduced expression of activation markers and impaired function.¹² Alarmingly, despite a capacity to generate a “protective” level of influenza-specific antibodies, impaired T cell function in vaccinated obese adults is associated with twice the risk of influenza infection and influenza-like illness.²¹

Animal models of obesity mirror these findings in humans. Diet-induced obese (DIO) mice exhibit increased mortality to primary¹⁶ and secondary⁸ influenza infections, decreased influenza-specific memory T cell populations,⁹ and suppressed memory T cell function.⁸ Other infection models, such as lymphocytic choriomeningitis virus (LCMV), demonstrate DIO mice have greater mortality to secondary LCMV infection with increased CD8⁺ T cell infiltration in white adipose tissue causing increased viral pathogenesis.²² Consistent between these infection models is a greater inflammatory response among DIO mice leading to increased viral-induced pathology^16,22; albeit with decreased type II IFN-specific T cell responses upon influenza infection.⁸ Despite this understanding, the mechanism(s) by which obesity drives T cell dysfunction remains unclear.

Antigen inexperienced CD8⁺ T cells give rise to long-lived memory cells following effector resolution through a shift in transcription factor expression and dynamic changes in cellular metabolism.^23,24 These long-lived memory cells can be broadly defined as central and effector memory (Tcm and Tem) cells, with a resident memory (Trm) population residing within the previously infected tissue providing rapid, localized protection.^25,26 Naïve and memory T cells utilize oxidative metabolism of fatty acids and glucose-derived pyruvate to drive and support immune surveillance and survival.^27,28 Activated effector T cells display predominately glycolytic and glutamine oxidative metabolism fueling daughter cell generation and effector function in response to antigen.^29–31 Considering obesity is inherently a metabolic condition associated with systemic chronic inflammation,^15,32 we hypothesized obesity impairs the immune response to the influenza virus through diminished metabolic and functional responses via differential gene expression patterns. We infected lean and obese mice with a sublethal dose of mouse-adapted influenza virus (A/PR/8/34). Here, for the first time, we report obesity impairs the metabolic reprogramming of effector CD8⁺ T cells within the lungs during influenza infection.

2 |. MATERIALS AND METHODS

2.1 |. Mice and diets

C57BL/6J Control (Stock No. 380056, “lean”) and C57BL/6J DIO (Stock No. 380050, “obese”) 18-week-old, male mice were obtained from The Jackson Laboratory (Bar Harbor, Maine, USA) and allowed one week of acclimation. Mice were group-housed (4 per cage), maintained at ambient temperature, and given ad libitum access to food and water. Forty-eight lean mice were placed on a purified low-fat control diet (D12450B, Research Diets, New Brunswick, NJ, USA) and forty-eight DIO mice were maintained on 60% kcal high-fat diet (HFD) (Research Diets, D12492). All procedures were performed in accordance with protocols approved by the Institutional Animal Care and Use Committee (IACUC) at the University of North Carolina at Chapel Hill.

2.2 |. Influenza infection

Influenza infection followed previously described protocols.³³ Briefly, mice were lightly anesthetized with isoflurane and infected intranasally with 50μL of sterile PBS containing 0.05 hemagglutination units (HAU) of A/PR/8/34 (ATCC® VR-95™, Manassas, VA, USA), an H1N1 influenza virus. Mice were weighed daily and euthanized at day 0 (uninfected), 10 and 24-post infection.

2.3 |. Flow cytometry and cell sorting

Lungs were lavaged and cells were collected from the bronchoalveolar lavage fluid (BALF) and prepared for flow cytometry analysis as previously described.^11,33 The following anti-bodies were used: Alexa Fluor 700 Rat anti-mouse CD3 (17A2, BioLegend, San Diego, CA, USA), APC-Cy7 anti-mouse CD19 (6D5, BioLegend, Sand Diego, CA, USA), Pacific Blue Rat anti-mouse CD4 (GK1.5, BioLegend), Alexa Flour 594 Rat anti-mouse CD8a (53–6.7, BioLegend), FITC Rat anti-mouse CD62L (MEL-14, BioLegend), APC Rat anti-mouse Granzyme B (NGZB, eBioscience, San Diego, CA, USA), PE Rat anti-mouse Interferon gamma (XMG1.2, eBioscience), PE Rat anti-mouse CD62L (MEL-14, BioLegend), APC Rat anti-mouse CD44 (IM7, BioLegend), APC-Cy7 Armenian Hamster CD69 (H1.2F3, BioLegend), APC Rat anti-mouse CD11a (M17/4, BioLegend), PE Armenian Hamster anti-mouse CD103 (2E7, BioLegend), APC-Cy7 Rat anti-mouse CD8a (53–6.7, BioLegend), PE Rat anti-mouse TNFα (MP6-XT22, BioLegend), Pacific Blue Rat anti-mouse CD44 (IM7, BioLegend), Zombie Aqua Fixable Viability dye (BioLegend), UltraComp eBeads compensation beads (ThermoFisher, Waltham, MA, USA), and purified Rat anti-mouse CD16/CD62 FcBlock (2.4G2, BD Bioscience). To determine fatty acid uptake, single cells isolated from the lung were incubated with 5μM BODIPY FL C₁₆ (ThermoFisher) for 30 min at 37°C prior to extracellular staining.

To determine glucose uptake, single cells isolated from the lung were incubated with 100μM 2NBDG (Cayman, Ann Arbor, Michigan, USA) for 30mins at 37°C prior to extracellular staining. To determine mitochondrial membrane potential and reactive oxygen species, lung isolated single cells were incubated with 15 nM Mitotracker Deep Red (ThermoFisher) and 100nM Mitotracker Red CMXRos (ThermoFisher) according to the method of Clutton et al., respectively.³⁴ All samples were acquired on an Attune NxT flow cytometer (ThermoFisher), and data were analyzed using FlowJo v10 (Treestar). For cell sorting, lung isolated single cells from day 0 (uninfected) and day 10 post-PR8 infected mice were stained for viable CD3⁺CD8⁺CD44⁺ using the following antibodies: Alexa Fluor 700 Rat anti-mouse CD3 (17A2, BioLegend, San Diego, CA, USA), Alex Fluor 594 Rat anti-mouse CD8a (53–6.7, BioLegend), APC Rat anti-mouse CD44 (IM7, BioLegend), and Zombie Aqua Fixable Viability dye (BioLegend). Cells were sorted using the Becton Dickson FACSAria II with FACSDiva 8.0.1 software. Cells were sorted directly into ice-cold Trizol LS (ThermoFisher) and frozen immediately on dry ice.

2.4 |. Extracellular metabolic flux analysis

CD8⁺ T cells were isolated from the single cell suspension of mouse BALF and digested lungs at day 10 post PR8 infection using magnetic bead negative selection (Stemcell, Vancouver, Canada) in Easy-Sep buffer (PBS + 2% FBS + 1 mM EDTA). Isolated cells were counted using the Bio-Rad TC20 with trypan blue exclusion for viability. XFe96 cell culture microplates were treated with Cell-Tak™ (Corning, Corning, NY, USA) in 0.1 M sodium bicarbonate to allow for cell adherence. CD8⁺ T cells were plated in non-buffered RPMI-1640 with freshly added 10 mM glucose, 2 mM glutamine, and 1 mM pyruvate at 300,000 cells per well. Extracellular acidification (ECAR) and oxygen consumption rates (OCR) were determined using the Seahorse XFe96 Flux analyzer (Agilent, Santa Clara, CA, USA) at 37°C as previously described.³⁵ OCR and ECAR were normalized to cell number.

2.5 |. Ultra-low RNA-sequencing

Ultra-low RNA-sequencing was performed using a previously defined method.³⁶ Live⁺CD3⁺CD8⁺CD44⁺ T cells were sorted from BALF and digested lung single cell suspensions using the FACSAria II into Trizol LS (Invitrogen) on ice. Samples were shipped frozen on dry ice to Genewiz (NJ, USA) for RNA isolation using poly-A enrichment for full length transcripts. NA sequencing libraries were prepared using the NEBNext Ultra RNA Library Prep Kit for Illumnia using manufacturer’s instructions. The sequencing libraries were multiplexed and clustered on two lanes of a flowcell. After clustering the flowcell was loaded onto the Illumina HiSeq instrument (4000 or equivalent). The samples were sequenced using a 2×150bp Paired End configuration. Fastq file quality was checked using FastQC with low sequence read depths excluded following ENCODE guidelines. All fastq sequence files were aligned to the Mus musculus UCSC MM10 reference genome using Bowtie2,³⁷ and the resulting SAM file was converted to a BAM file for further downstream analysis using the subread featureCounts package.³⁸ We included a stringency parameter (-B) to ensure the mapping of both paired ends when assigning a count to a specific gene. We then used DESeq2 for normalization and statistical analysis of the gene counts file output from the featureCounts program. DESeq2 is a conservative software that utilizes a negative binomial distribution with a generalized linear model to adjust for false positive significance values that result from running statistical tests on tens of thousands of genes.³⁹ The DESeq2 analysis generated Benjamini-Hochberg (BH) adjusted p-values, log2 fold changes, base mean expression values, and normalized counts for each gene, as detailed in the DESeq2 manual.³⁹ All subsequent statistical analyses only utilized BH-adjusted P-values below 0.1, or a false discovery rate (FDR) of 10%.

2.6 |. Statistical analysis

Seahorse data were analyzed using custom-built R version 4.0.3 and Python version 2.7 scripts, linked here (https://github.com/abrar-alshaer/seahorse_analysis). We filtered outliers outside of the Q1/Q3±1.5*IQR range for technical or biological replicates. Parameters of the Mitochondrial Stress Test were calculated in our Python program as detailed in the Agilent Seahorse user guide. For statistical analyses, normality was tested using the Shapiro-Wilks test, followed by testing for equal variance using the Bartlett test. If the data satisfied the assumption of normality, we utilized a parametric Student T-test; otherwise, we utilized the non-parametric Wilcoxon rank-sum test. Finally, all data were adjusted for multiple comparisons using the Benjamini-Hochberg correction for false discovery rate of 10% FDR. Details describing the statistical analysis and sample size of each experiment can be found in the figure legends. All statistical analysis was performed using GraphPad Prism 8 for Mac OS X, version 7.0c (GraphPad Software, Inc., La Jolla, CA) or R version 4.0.3. All data were determined as significant by adjusted-P < 0.10.

2.7 |. Data availability

RNA-seq data are available in the GEO database (GSE167112). Additional information and materials will be made available upon request.

3 |. RESULTS

3.1 |. Primary influenza infection elicits a robust pulmonary inflammatory response

Male C57BL/6J lean and obese mice were infected with a sublethal dose of A/PR/8/36 influenza virus (Figure 1A). Both lean and obese mice lost ∼20% of their body weight by day 9 and 10 post-infection, with no difference in weight loss between lean and obese mice (Figure 1A and B). Lung and BALF cell numbers in uninfected (Day 0), during infection (Day 10), and following immune resolution (Day 24) did not vary between lean and obese mice (Figure 1C). As expected, both lean and obese mice exhibited a dramatic increase in infiltrating immune cells at day 10 (Figure 1C) relative to baselines (7.26 fold increase in lean; 7.50 fold increase in obese). Infection resolution resulted in significant decreases in cell numbers from day 10 peaks (62.6% decrease in lean, 65.0% decrease in obese), albeit with continued significantly higher cell numbers than baseline (4.46 fold greater in lean; 4.54 fold greater in obese). Similar increases in BALF protein levels (a measure of lung epithelial damage caused by infection) occurred in both lean and obese mice at day 10 post-infection compared to baseline (20.3-fold increase in lean; 50.3-fold increase in obese) (Figure 1D). Resolution of infection resulted in a significant decrease in BALF protein in both lean and obese mice at day 24 post-infection; however, obese mice had ∼70% lower BALF protein compared to lean mice (Figure 1D).

FIGURE 1. Sublethal influenza infection induces a robust primary immune response in the lungs of lean and obese mice.

(A) Male, 18-week-old C57BL/6J mice were fed control (Lean, n = 36) or high fat diet (Obese, n = 36). After one-week acclimation, mice were weighed for four consecutive days before infection with 0.05 HAU/50μL PR8 influenza virus. Mice were maintained on lean and high fat diets for 24 days after infection. Body weights were measured daily. (B) Percent weight loss was calculated relative to original body weight at time of infection. (C) Lungs were harvested, digested, and homogenized in single cell suspension at each time point. Total cell number of combined bronchoalveolar lavage fluid and digested lungs from lean and obese mice at day 0, day 10, and day 24 post-influenza infection. (D) Bronchoalveolar lavage fluid (BALF) protein levels were determined at day 0, day 10, and day 24. (E) Frequency of CD8⁺ T cells relative to total cells from BALF+Lungs. (F) Cell numbers of CD8⁺ T cells from BALF+Lungs. Data represent mean ± SD (A-F) where each dot represents data obtained from one mouse. Student’s t test or the Wilcoxon rank sum test were used to compare groups. Multiple comparisons were corrected for a false discovery rate of 0.10 using the Benjamini-Hochberg correction, where significance was defined as *adj.P < 0.05, **adj.P < 0.05, ***adj.P < 0.001, ****adj.P < 0.0001. ns indicates not significant

Influenza infection resulted in significant changes in the frequency and number of CD8⁺ T lymphocytes within the total BALF and lungs. Using flow cytometry to identify CD8⁺ T cell populations (Supplemental Figure 1), no significant differences were found between lean and obese mice at day 0, day 10, or day 24 for percent (Figure 1E) or number (Figure 1F) of CD8⁺ T cells within the lungs. However, influenza infection resulted in a significant increase in the percent of CD8⁺ T cells within the lungs at day 10 post PR8 infection (Figure 1E; 3.09-fold increase in lean, 3.83-fold increase in obese) as well as the number of total CD8⁺ T cells at day 10 (Figure 1F; 24.1-fold increase in lean, 29.1-fold increase in obese) relative to day 0. Infection resolution at day 24 occurred with significant reduction in both percent (Figure 1E) and number (Figure 1F) of pulmonary CD8⁺ T cells relative to day 10. Similar to previous reports,³³ both the percent and number of pulmonary CD8⁺ T cells in lean and obese mice were significantly greater at day 24 post-infection compared to day 0 (percent, 1.6-fold greater in lean, 1.5-fold greater in obese; number, 3.8-fold greater in lean, 3.74-fold greater in obese).

CD8⁺ T cell memory subsets also dramatically changed during the course of infection (Figure 2A-G). Figure 2A depicts the gating strategy used for analyses. Central memory CD8⁺ T cells (CM; CD8⁺CD44⁺CD62L⁺) were significantly greater in terms of percent (Figure 2B) and number (Figure 2E) in both uninfected day 0 lean and obese mice compared with naïve and effector memory cells. Naïve CD8⁺ T cells (CD8⁺CD44^–CD62L^–) comprised a third of the total CD8⁺ T cells. At day 10 post-infection, over 85% of all pulmonary CD8⁺ T cells were effector memory (EM; CD8⁺CD44⁺CD62L^–) CD8⁺ T cells (Figure 2C) and had approximately 9-fold greater numbers compared to CM and ∼40-fold greater numbers compared to naive (Figure 2F).

FIGURE 2. Influenza infection results in the increase in CD8⁺CD44⁺ T cells within the lungs of lean and obese mice.

Lungs were harvested, digested, and homogenized into single cell suspension from Lean (n = 12–18) and Obese (n = 12–18) mice, and 1×10⁶ cells were stained for (A) naïve (CD44^–CD62L⁺), central memory (CM; CD44⁺CD62L⁺), and effector memory (EM; CD44⁺CD62L^–) CD8⁺ T cells by flow cytometry at day 0, day 10, and day 24. (B) Percent of naïve, CM, and EM CD8⁺ T cells relative to total CD8⁺ T cells from uninfected day 0 mice. (C) Percent of naïve, CM, and EM CD8⁺ T cells relative to total CD8⁺ T cells from day 10 PR8 infected mice. (D) Percent of naïve, CM, and EM CD8⁺ T cells relative to total CD8⁺ T cells from day 24 PR8 infected mice. (E) Number of naïve, CM, and EM CD8⁺ T cells from uninfected day 0 mice. (F) Number of naïve, CM, and EM CD8⁺ T cells from day 10 PR8 infected mice. (G) Number of naïve, CM, and EM CD8⁺ T cells from day 24 PR8 infected mice. Data represent mean ± SD, with each dot representing one mouse. Student’s t test or Wilcoxon rank sum test was used to compare groups. Multiple comparisons were corrected for a false discovery rate of 0.10 using the Benjamini-Hochberg correction, where significance was defined as **adj.P < 0.01, ****adj.P < 0.0001. ns indicated not significant.

Following infection resolution at day 24, the percent (Figure 2D) and number (Figure 2G) of EM CD8⁺ T cells remained elevated compared to naïve and CM CD8⁺ T cells in both lean and obese mice. However, compared to lean mice, obese mice had 42% lower number of naïve CD8⁺ T cells (Figure 2G). Together, these data indicate sublethal PR8 influenza infection elicits an acute primary infection in both lean and obese mice during infection response at day 10, with greater than 95% of CD8⁺ T cells gaining expression of CD44⁺ to facilitate homing to the non-lymphoid pulmonary tissue.

3.2 |. Obesity impairs CD8⁺ T cell metabolism in the lungs of influenza-infected mice independent of changes in glucose or fatty acid uptake

CD8⁺ T cells provide critical defense against influenza infection in the lungs. To establish if obesity alters the metabolic profile of these cells in influenza-infected lungs, we isolated pulmonary CD8⁺ T cells from lean and obese mice at day 10-post infection. Oxidative and glycolytic metabolism were determined using an Agilent Seahorse extracellular flux analyzer by measuring oxygen consumption rate (OCR; a proxy for mitochondrial metabolism), and extracellular acidification rate (ECAR; a proxy for lactate production and glycolysis). Real-time measurements of OCR (Figure 3A) and ECAR (Figure 3B) for pulmonary CD8⁺ T cells from influenza-infected lean and obese mice show distinct metabolic differences at day 10-post infection. Specifically, compared to pulmonary CD8⁺ T cells from influenza-infected lean mice, pulmonary CD8⁺ T cells from obese mice had 30% lower basal OCR (Figure 3C) and 24.9% lower basal ECAR (Figure 3D).

FIGURE 3. Obesity impairs CD8⁺ T cell metabolism in the lungs of influenza-infected mice.

Lungs were harvested, digested, and homogenized into single cell suspension. Total CD8⁺ T cells from day 10 influenza infected mice were isolated using negative selection magnetic bead separation. Cells were plated at 300,000 cells per well, and extracellular flux analysis was performed using the Agilent Seahorse XF_e96 flux analyzer. (A) Oxygen consumption rates (OCR) and (B) Extracellular Acidification Rates (ECAR) were measured in response to the Mitochondrial Stress Test for CD8⁺ T cells from Lean (n = 11) and Obese (n = 12) mice. (C) Basal OCR and (D) Basal ECAR were determined as the last point before oligomycin injection. (E) Maximal respiration was measured as peak OCR following injection of 1.5 μM FCCP. (F) Spare Respiratory Capacity was calculated as the difference in Max OCR minus Basal OCR. Student’s t-test or Wilcoxon rank sum test was used to compare groups. Multiple comparisons were corrected for a false discovery rate of 0.10 using the Benjamini-Hochberg correction, where significance was defined as *adj.P < 0.05, **adj.P < 0.01

Maximal mitochondrial respiration, measured as the highest OCR following administration of the mitochondrial uncoupling agent carbonyl cyanide 4-(trifluoromethoxy) phenylhydrazone (FCCP), was 47.4% lower in pulmonary CD8⁺ T cells from obese mice compared to lean (Figure 3E). Similarly, compared to lean mice, CD8⁺ T cells from obese mice had 61.6% lower spare respiratory capacity (SRC; Figure 3F), a measure of the potential mitochondrial metabolic response to energy demand.

Subsequently, we investigated if the impairment in CD8⁺ T cell metabolism of obese mice was driven by dysregulated glucose or fatty acid uptake. There were no differences in pulmonary CD8⁺ T cell glucose uptake (Supplemental Figure 2A, B) or expression of glucose transporter 1 (GLUT1; Supplemental Figure 2C, D). Day 10 pulmonary CD8⁺ T cells from lean and obese mice also had no difference in fatty acid uptake (BODIPY; Supplemental Figure 2E, F), nor was there a difference in mitochondrial membrane potential (Supplemental Figure 2G), as measured by the top 90% of cells incorporating Mitrotracker Deep Red; or mitochondrial reactive oxygen species determined by the 90^th percentile of CMXRos (Supplemental Figure 2H). While the ratio of basal OCR to basal ECAR showed no difference between lean and obese mice (Supplemental Figure 2I), the comparison of basal OCR versus basal ECAR before and after injection of oligomycin, an ATP-synthase inhibitor, showed impaired glycolytic potential of pulmonary CD8⁺ T cells from obese mice compared to lean (Supplemental Figure 2J), signifying an inability of CD8⁺ T cells from influenza-infected obese mice to use glycolysis over mitochondrial respiration.

3.3 |. Obesity reduces pulmonary CD8⁺ T effector cells and cytokine production in influenza-infected mice

T cell metabolism critically drives T cell function.^31,40 Thus, we examined whether obesity altered key cytokine expression of effector CD8⁺ T cells during influenza infection (Figure 4A). Similar to previous findings,^8,11,16,33 compared to lean mice, obese mice had no significant difference in percent or number of granzyme B (GrB) single positive pulmonary CD8⁺ T cells (Figure 4B). Relative to lean controls, obese mice had 1.34-fold lower percent of CD8⁺IFNγ⁺ T cells, but no difference in the number of CD8⁺IFNγ⁺ T cells (Figure 4C). Obese mice had significantly lower percent and number (Figure 4D) of GrB and IFNγ double-positive CD8⁺ T cells at day 10 post-infection compared to lean mice. This ∼39% reduction in pulmonary effector CD8⁺ T cells occurred concomitant with reduced CD8⁺ T cell glycolytic and oxidative metabolism (Figure 3), despite no difference in nutrient uptake or mitochondrial potential (Supplemental Figure 2). However, there was no difference in median fluorescent intensity (MFI) of GrB or IFN-γ among single positive or double-positive pulmonary effector CD8⁺ T cells at day 10 post-infection (Supplemental Figure 3A-D). Following influenza clearance at day 24-post infection, there were no differences in percent or number of pulmonary effector CD8⁺GrB⁺IFNγ⁺, CD8⁺GrB⁺, or CD8⁺IFNγ⁺ T cells (data not shown).

FIGURE 4.

Lower percent of CD8⁺IFNγ⁺GrB⁺ effector T cells in the lungs of influenza-infected obese mice correlates with impaired pulmonary CD8⁺ T cell metabolism. Lungs were harvested, digested, and homogenized into single-cell suspension and 1×10⁶ cells were stained for intracellular cytokines in effector T cells by flow cytometry. (A) Representative scatter plot of IFN-γ and granzyme B (GrB) CD8⁺ T cells at day 10 PR8 infection from Lean (n = 10–12) and Obese (n = 10–12) mice. (B) Percent (left) and cell number (right) of CD8⁺GrB⁺ T cells. (C) Percent (left) and cell number (right) of CD8⁺IFNγ⁺ T cells. (D) Percent (left) and cell number (right) of CD8⁺GrB⁺IFNγ⁺ T cells. (E) Representative scatter plot of TNF-α in CD8⁺ T cells from day 10 PR8 infected Lean (n = 6) and Obese (n = 6) mice. (F) Percent (left) and cell number (right) of CD8⁺TNFα+ T cells. (G) Pearson correlation of basal ECAR (mpH/min) by the percent of CD8⁺GrB⁺IFNγ⁺ T cells of total CD8⁺ T cells. (H) Pearson correlation of basal OCR (pmoles/min) by the percent of CD8⁺GrB⁺IFNγ⁺ T cells of total CD8⁺ T cells. Each dot represents data obtained from one mouse with mean ± SD. Student’s t-test or Wilcoxon rank sum test was used to compare groups. Multiple comparisons were corrected for a false discovery rate of 0.10 using the Benjamini-Hochberg correction, where significance was defined as *adj.P < 0.05, **adj.P < 0.01, ***adj.P < 0.001. ns indicates not significant

Pulmonary CD8⁺ T cells making TNFα also represent a critical early effector population,⁴¹ although recent data suggests loss of TNFα production by CD8⁺ T cells during influenza A virus infection propagates viral-specific CD8⁺ T cells.⁴² Thus, pulmonary single cells were stained for TNFα at day 10 post-PR8 infection in lean and obese mice (Figure 4E). No difference in percent or number of CD8⁺TNFα+ T cells were observed between lean and obese mice (Figure 4F). Of note, the percent of pulmonary effector CD8⁺GrB⁺IFNγ⁺ T cells correlated positively with both total CD8⁺ basal ECAR (Figure 4G) and basal OCR (Figure 4H). Together, these data suggest impaired pulmonary CD8⁺ effector T cell metabolism reduces the presence of effector CD8⁺GrB⁺IFNγ⁺ T cells during influenza infection. This impairment in effector CD8⁺ T cell function is specific for type II interferon and is not observed in CD8⁺GrB⁺ or CD8⁺TNFα⁺ pulmonary T cells.

3.4 |. Obesity modulates the transcriptome of pulmonary CD8⁺ T cells during influenza virus infection

To discern potential mechanisms for the functional and metabolic dysregulation of obesity in anti-viral CD8⁺ T cells, we utilized next generation RNA sequencing to investigate the transcriptome of pulmonary CD8⁺CD44⁺ T cells in response to influenza PR8 infected mice. Approximately 1.1 − 2 × 10⁴ (day 0 uninfected) and 1 − 3 × 10⁵ (day 10 PR8 infected) viable CD3⁺CD8⁺CD44⁺ T cells were sorted from lean (n = 3) and obese (n = 2–3) mice. Using Ultra-Low RNAseq library preparation, we profiled the transcriptome to determine the differential expression of gene transcripts in uninfected and day 10 PR8 infected mice. Normalized gene sets were compared between day 0 and day 10 mice within each group and significant genes defined by an FDR < 0.10 were combined to identify uniquely expressed and shared gene transcripts. We identified 1314 exclusive up-regulated and 446 exclusive down-regulated genes among lean CD8⁺CD44⁺ T cells, with 1,066 exclusive up-regulated and 829 exclusive down-regulated genes among obese CD8⁺CD44⁺ T cells from PR8 infected mice (Figure 5A). Between both lean and obese mice, there were 1,925 significantly shared up-regulated genes and 1,088 significantly shared down-regulated genes.

FIGURE 5.

Pulmonary CD8⁺CD44⁺ T cells from obese mice display differential gene expression. Lungs were harvested, digested, and homogenized into single cell suspension. Cells were stained for viable⁺CD3⁺CD8⁺CD44⁺ T cells and sorted using the FACSAria II for Ultra-Low RNA sequencing from uninfected Lean (n = 3), uninfected Obese (n = 2), day 10 PR8 infected Lean (n = 3), and day 10 PR8 infected Obese (n = 3) mice. DESeq2 was used to generate normalized gene counts between uninfected and PR8 infected viable⁺CD3⁺CD8⁺CD44⁺ T cells. (A) Significant (BH adj. P value < 0.10) shared and unique up-regulated and down-regulated gene transcripts between Lean and Obese mice. (B) PCA clustering of 1925 shared up-regulated and 1038 shared down-regulated significant genes. (C) KEGG pathways of relevant up-regulated genes identified in Day 10 Lean and Obese mice are listed on the y-axis and plotted by fold enrichment on the x-axis, with BH adj.P-value annotated by color and the number of genes identified in that KEGG pathway (count) portrayed by circle size. (D) KEGG pathways of relevant down-regulated genes identified in Day 10 Lean and Obese mice are listed on the y-axis and plotted by fold enrichment on the x-axis, with BH adj.P-value annotated by color and the number of genes identified in that KEGG pathway (count) portrayed by circle size. (E) Expression of shared significant genes from (A-D) in day 10 Lean and Obese mice. Red dots represent gene transcripts with >2-fold expression in day 10 obese mice compared to lean, while blue dots represent gene transcripts with >2-fold expression in day 10 lean mice compared to obese. (F) Heatmap of z-score normalized median expression of select genes for inhibitory receptors, effector molecule, cell surface receptors, and transcription factors that were significantly different between uninfected and day 10 PR8 infected Lean and Obese mice. Bold gene transcripts are >2-fold higher as shown in (E)

Using the 3,013 significantly expressed genes shared among both lean and obese mice, we performed PCA dimension reduction (Figure 5B). Uninfected lean and obese mice clustered together with no separation between groups. Meanwhile, day 10 PR8 infected lean and obese mice had a significant separation from uninfected mice, albeit only slight separation between lean and obese PR8 infected mice based on the relative expression of the 3,013 significantly shared gene transcripts. We utilized the DAVID Functional Annotation Tool version 6.8 to identify the KEGG pathway clusters shared among the gene targets in day 10 up-regulated (Figure 5C) and down-regulated (Figure 5D) gene transcripts. Up-regulated genes included those for pathways such as PI3K-Akt signaling, cytokine and chemokine interactions, and other cellular processes in addition to other important cellular functions (Supplementary Table 1). Down-regulated genes shared among lean and obese PR8 infected mice included TNF signaling, T cell receptor signaling pathways, and pyrimidine and glutathione metabolism (Supplementary Table 2). KEGG analysis of uniquely expressed genes upregulated and down-regulated in PR8 infected lean and obese mice (Supplemental Figure 4) reveal the complex regulation of signaling cascades and metabolic enzymes responsible for the transcriptomic response to influenza A virus.

In order to identify the differences in gene expression from pulmonary CD8⁺CD44⁺ T cells, we compared the log10 expression from day 10 PR8 infected lean and day 10 PR8 infected obese mice (Figure 5E). We identified 69 significantly expressed genes with >2-fold higher expression in obese mice and 110 significantly expressed genes with >2-fold higher expression in lean mice. Among these genes were the exhaustion markers Pdcd1 and Tigit in obese mice, along with the cell surface receptor Csf1. Among lean mice, genes included cell surface receptors Ccr7 and Cxcr5, and the transcription factor Id3. Additionally, the metabolic genes, Apoe (4.27-fold greater in day 10 obese) and Odc1 (2.5-fold greater in day 10 lean), showed differential expression in PR8 infected mice. Genes related to lipid inflammation resolution pathways were also modified including LTA4H exclusively downregulated in lean mice while Ltb4r1 and Alox15 were exclusively upregulated in obese mice (Fig. 5E). Finally, we assessed the expression of critical genes important in inhibitory responses, effector molecules, cell surface receptors, and transcription factors (Figure 5F). We identified 29 gene transcripts with significant differential expression in pulmonary CD8⁺CD44⁺ T cells from lean and obese mice. Together, these data provide the first evidence of the complex and unique transcriptional regulation of effector CD8⁺CD44⁺ T cells driven by obesity following influenza infection.

4 |. DISCUSSION

Influenza remains a pertinent public health threat, causing approximately 3 to 5 million cases of severe illness and up to 500,000 deaths each year globally.⁴³ Within the U.S., seasonal influenza-related illness accounts for 9.2 to 35.6 million cases and anywhere from 140,000 to 710,000 hospitalizations.⁴⁴ Following the 2009 H1N1 swine flu pandemic, obesity was recognized as an independent risk factor for increased influenza-related morbidity and mortality.⁴⁵ Since identification as a high-risk population, several studies have described how obesity affects immunity to influenza virus,^{12,15,20,46–48} however the mechanism(s) responsible remain largely unknown. Alarmingly, the novel beta coronavirus, Severe Acute Respiratory Syndrome-Coronavirus 2 (SARS-CoV-2), which is responsible for the COVID-19 pandemic, is also associated with greater severity, increased rates of hospitalization, and increased mortality in adults with obesity.⁶ Thus, understanding how a chronic non-communicable disease like obesity increases the risk for infectious diseases like influenza virus and SARS-CoV-2 are of significant public health importance.

Herein, we demonstrate for the first time, suppressed oxidative and glycolytic metabolism of effector CD8⁺ T cells in the lungs of influenza-infected obese mice. Notably, pulmonary CD8⁺ T cells exhibited an inability to utilize glucose metabolism when forced using the ATP synthase inhibitor, oligomycin. This suggests obesity impairs CD8⁺ T cell metabolic plasticity. Surprisingly, these impairments in pulmonary CD8⁺ T cell metabolism were not caused by decreased glucose or fatty acid uptake, impaired mitochondrial membrane potential, nor accumulation of reactive oxygen species within the mitochondria. However, we cannot rule out other aspects of mitochondrial function including, but not limited to, an effect of obesity on mitochondrial structure-function.⁴⁹ For example, we have previously shown in other model systems that obesity impairs cardiac mitochondrial inner membrane cardiolipin composition and thereby mitochondrial function. Thus, studies focused on CD8⁺ T cell mitochondrial cardiolipin content and composition may be an avenue for future research.⁵⁰

Our findings are consistent with the notion that CD8⁺ T cell metabolism is critical for effector function. Foundational work demonstrating the connection between T cell transcription, metabolism, and function establishes the critical link between cellular glycolysis, glutaminolysis, and fatty acid oxidation to support T cell proliferation, differentiation, function, and survival.⁵¹ Mature T cells respond to their environment through a coordinated transcriptional and metabolic network.⁵² These processes are tied together, where the metabolic needs of the cell vary in response to changes in the environment and to activation and stress signals. Resting naïve and memory CD8⁺ T cells maintain a quiescent metabolic phenotype predominately of fatty acid and glucose-derived pyruvate oxidative phosphorylation.³¹ Manipulation of fatty acid uptake and oxidation alters memory T cell formation and survival,^53,54 whereas a reduction of mitochondrial respiratory capacity can reduce memory CD8⁺ T cell formation.⁵⁵ Despite the heterogeneous nature of memory CD8⁺ T cells, which have differing transcriptional and epigenetic signatures for T_CM, T_EM, and T_RM cells,⁵⁶ the metabolic phenotype between these cells remains heavily reliant on oxidative phosphorylation to support immune surveillance and survival.^24,55,57

Our RNAseq data suggest two potential mechanisms. First, obesity may drive pulmonary CD8⁺ T cell exhaustion limiting effector function in the context of influenza infection. Second, obesity may result in sustained activation defined by greater expression of Pdcd1 and Tigit. Within the past two decades, establishment of functional exhaustion among antigen-experienced CD8⁺ T cells has been recognized within chronic infections^58,59 and cancer.⁶⁰ Numerous reports demonstrate memory precursor and early effector cells begin to express markers of exhaustion, developing terminal differentiation and impaired multi-potency, polyclonal function, and memory pool propagation.^58,59,61,62 However, much of this work in chronic infection models has not been investigated in the context of an acute pulmonary infection like influenza virus. To our knowledge, no study has investigated the presence of a chronic condition such as obesity on pulmonary CD8⁺ T cell metabolism and function, emphasizing the need to understand how a metabolic condition like obesity might influence the cellular metabolism and functional response of pathogen-specific CD8⁺ T cell populations.

Identification of differential gene expression in effector CD8⁺CD44⁺ T cells from lean and obese mice revealed several key metabolism-related pathways altered during influenza infection. For example, effector CD8⁺ T cells from obese mice expressed significantly greater Apoe, which encodes the apolipoprotein E and is involved in cholesterol and fatty acid metabolism. Previously, depletion of Apoe in APCs has been shown to enhance proliferation and IFNγ production of T lymphocytes.⁶³ Meanwhile, PR8 infected lean mice significantly expressed Odc1, the rate-limiting step in polyamine biosynthesis and the enzyme responsible for the catalysis of ornithine to putrescine.⁶⁴ KEGG analysis revealed down-regulation of purine metabolism in lean PR8 infected mice, while up-regulation of purine metabolism among PR8 infected obese mice. The changes in metabolic and transcriptional outcomes in response to obesity occurred concomitant with reduced effector CD8⁺ T cells making IFN-γ and GrB in obese mice at day 10 post PR8 infection. The metabolic impairment in pulmonary CD8⁺ T cells was positively correlated with the pulmonary frequency of CD8⁺IFNγ⁺GrB⁺ T cells, which have previously been shown to be critical in influenza protection and memory formation.^8,11

Together, these findings suggest that diet-induced obesity increases influenza virus pathogenesis through T cell-mediated metabolic reprogramming, inducing terminal differentiation and exhaustion, resulting in impaired pulmonary CD8⁺ T cell function. This obesity-induced metabolic reprogramming of CD8⁺ T cells results in suppressed T cell function and, as demonstrated in previous influenza infection models with obese mice, reduced memory cell fitness in response to secondary influenza infection.⁹ As obesity rates continue to rise, these results highlight the urgent need for targeted studies to identify the impact of obesity on pulmonary CD8⁺ T cell differentiation and function. Further understanding is required to improve vaccine and therapeutic strategies to reduce the burden of influenza in obese populations.

Abbreviations:

2NBDG

2-(N-(7-Nitrobenz-2-oxa-1,3-diazol-4-yl)Amino)-2-Deoxyglucose

A/PR/8/34

Influenza A Puerto Rico 8 34 H1N1 Virus

ATP

adenosine triphosphate

BALF

bronchoalveolar lavage fluid

BODIPY

boron-dipyrromethene (4,4-difluoro-4-bora-3a,4a-diaza-s-indacene)

CD

cluster of differentiation

CMXROS

MitoTracker CMX Ros

DIO

diet induced obese

ECAR

extracellular acidification rate

EDTA

ethylenediaminetetraacetic acid

FCCP

carbonyl cyanide 4-(trifluoromethoxy) phenylhydrazone

GAPDH

glyceraldehyde 3-phosphate dehydrogenase

GrB

granzyme B

H1N1

hemagglutinin 1 neuraminidase 1

HAU

hemagglutination units

HFD

high fat diet

IACUC

Institutional Animal Care and Use Committee

IQR

interquartile range

LDHA

lactate dehydrogenase A

MFI

median fluorescent intensity

MTDR

MitoTracker Deep Red

OCR

oxygen consumption rate

T_CM

central memory T cells

T_EM

effector memory T cells

T_RM

residential memory T cells