Glucocorticoid-mediated Suppression of Effector Programming Assists the Memory Transition of Virus-specific CD8⁺ T Cells

The cells of the immune system perform three major functions, namely infection clearance, establishing homeostasis, and immune surveillance. However, unchecked immune activation could invariably result in adverse reactions that can range from limited tissue damage to several life-threatening immunopathological conditions (1). Many cellular products such as galectins, the derivatives of the arachidonic acid pathway such as lipoxins, and omega-3 polyunsaturated fatty acid derivatives such as resolvins and protectins reduce the extent of proinflammatory damage (2–4). Additionally, infection-induced stress can stimulate the hypothalamic-pituitary-adrenal axis to produce glucocorticoids (GCs), and a systemic rise in GC levels is evident upon viral infection (5, 6). GCs have anti-inflammatory functions and their adrenal synthesis coincides with increased production of proinflammatory cytokines such as IFN-γ, TNF, and IL-6 by innate immune cells (7). A lack of adrenal GC synthesis results in rapid clearance of infection but enhances immunopathology and mortality of the infected animals due to uncontrolled T cell responses (8, 9). Therefore, the role of GCs during infection is yet to be adequately established. Ever since the discovery of their immunosuppressive activity, GCs have been routinely used to mitigate inflammatory and autoimmune pathologies. Although the use of GCs in the management of COVID-19–mediated immunopathology remains debatable (https://www.covid19treatmentguidelines.nih.gov/management/clinical-management-of-adults/nonhospitalized-adults–therapeutic-management/), initial success in the treatment of a subset of COVID-19 patients (10) has reinvigorated interest in harnessing its utility in managing virus-induced immunopathology.

Dexamethasone, along with other synthetic analogs as well as the naturally occurring corticosteroids (cortisol in humans and corticosterone in mice), acts via binding to the cytosolic GC receptor (GR; protein: NR3C1, gene: Nr3c1). Upon ligand binding, the GR translocates to the nucleus to modulate a plethora of cellular processes. GCs can suppress Ag presentation, inhibit the production of proinflammatory cytokines, or cause apoptosis of lymphocytes (11). We previously reported that effector cells were refractory to dexamethasone-mediated killing while naive and memory cells showed enhanced susceptibility to apoptosis (12). Recent studies have also highlighted the role of GCs on the differentiation and functionality of the CD8⁺ T cells (13–15). We undertook this study to elucidate the effects of GC exposure on the differentiation of responding CD8⁺ T cells during a localized viral infection. The findings of this study attest to the role of GC signaling in the differentiation of virus-specific CD8⁺ T cell memory.

C57BL/6 (B6; strain 000664), B6 OT1 (C57BL/6-Tg(TcraTcrb)1100Mjb/J; strain 003831), B6 Rag1^−/− (B6.129S7-Rag1^tm1Mom/J; strain 002216), and B6 CD45.1 (B6.SJL-Ptprc^a Pepc^b/BoyJ; stock 002014) were procured from The Jackson Laboratory (Bar Harbor, ME). All animals were housed in individual polysulfone cages with aspen bedding in the Small Animal Facility for Experimentation (SAFE) of the Indian Institute of Science Education and Research (IISER), Mohali. The animals were fed LabDiet from the United States and maintained under a 12-h light/12-h dark cycle. The animal experiments were performed strictly per the protocol approved by the Institutional Animal Ethics Committee (IAEC), IISER, Mohali, constituted under the aegis of the Committee for the Purpose of Control and Supervision of Experiments on Animals (CPCSEA).

γ-Herpesvirus (MHV68-SIINFEKL) and influenza A virus (WSN-SIINFEKL) were used for the in vivo experiments (16–18). MHV68-SIINFEKL was propagated, harvested, and titrated using Vero cells and stored as aliquots at −80°C until further use. WSN-SIINFEKL was propagated and titrated using Madin–Darby canine kidney cells. WSN-SIINFEKL was used for intranasal (i.n.) infections at the indicated dose. MHV68-SIINFEKL was used for infecting i.p. or inducing heterologous challenge i.n. in mice previously infected with WSN-SIINFEKL at the indicated dose.

For the adoptive transfer experiments, OT1 × RAG1^−/− mice were sacrificed and their lymph nodes and spleen were harvested. The CD8⁺ T cells were MACS purified from single-cell suspensions of the pooled lymph nodes and splenocytes. The number of cells indicated in the schematics was then transferred into CD45.1⁺ congenic mice. The donor and recipient animals were sex-matched and of the same age group (6–8 wk). The frequency of the donor cells in circulation was checked just before infection with WSN-SIINFEKL via the i.n. route. The animals were then given recurrent doses of dexamethasone (10 mg/kg body weight) from 4 to 6 days postinfection (dpi) or mifepristone (40 mg/kg body weight) from 1 to 5 dpi or the diluent. The donor OT1 cells were tracked in circulation. A heterologous challenge was given i.n. with MHV68-SIINFEKL. The animals were sacrificed at indicated time points following reinfection from each of the groups, and the collected organs were subjected to cellular analysis. Similar experiments were performed to measure the optimal dose response of dexamethasone and mifepristone administration (data not shown).

In vitro stimulation of OT1 cells with anti-CD3ε and anti-CD28 was performed with Abs for 16 h followed by their exposure to 1 µM dexamethasone or the diluent for 1 h. The cells were washed five times and transferred in separate groups of animals and their survival and retention were measured at 1 and 6 d after transfer. The transferred cells were then recalled by infecting recipients i.n. with MHV68-SIINFEKL (2 × 10⁶ PFU) 30 d after transfer. The cells were analyzed at the indicated time points by flow cytometry.

Abs used for measuring the expression of different molecules were procured from eBioscience and BioLegend. The Abs used were against the following: CD8α-PerCP-Cy5.5 (clone 53-6.7; stock concentration: 200 µg/ml, working concentration: 0.50 µg/ml), CD45.2-FITC/-Allophycocyanin/-PE (clone 104; stock concentration: 500 µg/ml, working concentration: 1.66 µg/ml), CD44-allophycocyanin (clone IM7; stock concentration: 200 µg/ml, working concentration: 0.40 µg/ml), CD127-FITC/-Allophycocyanin (clone A7R34; stock concentration: 200 µg/ml, working concentration: 1.00 µg/ml), KLRG1-FITC/-PE/-AF488 (clone 2F1; stock concentration: 500 µg/ml, working concentration: 1.25 µg/ml), IFN-γ/-Allophycocyanin/-PE (clone XMG1.2; stock concentration: 200 µg/ml, working concentration: 1.00 µg/ml), TNF-α-allophycocyanin (clone TN3-19; stock concentration: 200 µg/ml, working concentration: 1.00 µg/ml), granzyme B–PE (clone NGZB; stock concentration: 200 µg/ml, working concentration: 1.00 µg/ml), Ki67-FITC (clone SolA15; stock concentration: 500 µg/ml, working concentration: 1.00 µg/ml), Bcl2-FITC (clone 10C4; stock concentration: 100 µg/ml, working concentration: 0.50 µg/ml). The MHC class I tetramers used in this study were synthesized in-house and the monomers were refolded with SIINFEKL (OVA_257–264) as described earlier (19). All Abs were diluted in FACS buffer (PBS with 2% FBS). Dexamethasone, mifepristone, hematoxylin, and eosin Y were purchased from Sigma-Aldrich. 2,2,2-Tribromoethanol (Avertin) was purchased from TCI Chemicals. A MojoSort CD8⁺ T cell purification kit was purchased from BioLegend. CFSE, intracellular fixation-permeabilization buffer, and streptavidin-PE were purchased from Thermo Fisher Scientific. Monensin, brefeldin A, purified anti-CD3ε (17A2), and anti-CD28 (37.51) were purchased from eBioscience. The primary Abs for Western blot analysis were against STAT3, phospho-STAT3 (Y705), STAT5β, and phospho-STAT5β (Y694). The PhosphoPair STAT3 (Tyr⁷⁰⁵) Ab Set was purchased from BioLegend (catalog no. 699952), and the primary Abs against STAT5β (catalog no. 71-2500) and phospho-STAT5β (Y694) (ST5P-4A9, catalog no. 33-6000) were purchased from Invitrogen. GAPDH loading control Ab was purchased from Invitrogen (GA1R, catalog no. MA5-15738).

For blood staining, blood samples were collected in EDTA (0.7 mg/100 µl of blood). Then, 25 µl of blood was aliquoted directly in a centrifuge tube and 5 µl of the Ab mix was added. The cells were then incubated at 4°C for 45 min in the dark. Thereafter, 300 µl of 1× RBC lysis solution (155 mM NH₄Cl, 12 mM NaHCO₃, 0.1 mM EDTA [pH 7.3]) was added to lyse RBCs for 5 min at room temperature, after which 300 µl of PBS was added, and the cells were acquired and analyzed by flow cytometry using BD C6 Accuri, BD FACSCalibur, or BD FACSAria Fusion.

The mice were euthanized and subsequently perfused with 15–20 ml of PBS via the left ventricle and then the right ventricle to remove residual circulating cells from organs by incising the inferior vena cava traversing through the diaphragm via the abdominal cavity. Different lymphoid and nonlymphoid organs were collected from the mice. Single-cell suspensions were then prepared. In brief, the organs were placed in a cell strainer (70-µm cutoff) with 2 ml of cold complete RPMI 1640 and gently crushed using the soft end of a 2.5-ml syringe plunger. The suspension was then passed through the strainer and collected in a 15-ml centrifuge tube. The cells were then washed twice with cold PBS by centrifugation at 1200 rpm for 5 min at 4°C. The cells were finally resuspended in complete RPMI 1640 for further cellular analysis. The nonlymphoid organs were prepared by first digesting with type IV collagenase for 1 h at 37°C. Single-cell suspensions were then made as described above. The cells were stained using the indicated fluorescent Abs at 4°C for 30 min. For intracellular staining, cells were first surface stained and then treated with fixation buffer (eBioscience) for 20 min at 4°C. This was followed by permeabilization using an intracellular permeabilization buffer from eBioscience. The stained cells were washed three times with cold PBS and acquired using BD Accuri C6 or BD FACSAria Fusion. FlowJo v10 software was used for the analysis of the acquired data.

The caudal right lobe of the lungs from each group of mice was collected and fixed overnight at 4°C in 4% paraformaldehyde prepared in 1× PBS. Tissues were then dehydrated at room temperature in a 5–20% gradient of sucrose prepared in 1× PBS at 4°C. The tissues were embedded in an OCT compound and the tissue blocks were used for cutting tissue sections of 5 μm on a Leica Cryotome. The sections were stained by H&E. The dried sections were imaged using a microscope from Leica (DMi8). The images were analyzed using ImageJ software.

The mice were sacrificed at the indicated time postinfection and the lung tissues were collected from PBS-perfused mice infected with MHV68-SIINFEKL. The organs were weighed and homogenates of lung tissues were prepared to determine the viral burden on Vero cells.

Double-stranded oligonucleotides for short hairpin RNA (shRNA) against Nr3c1 were cloned into pLKO.1-GFP between AgeI and EcoRI restriction enzyme sites. The sequences for the shRNA are 5′-CCGGACCGGTCAGGCTGGCTTTATTAAATTCAAGAGATTTAATAAAGCCAGCCTGGAATTCC-3′ (top strand) and 5′-GGAATTCCAGGCTGGCTTTATTAAATCTCTTGAATTTAATAAAGCCAGCCTGACCGGTCCGG-3′ (bottom strand). Vector-specific primers were used to screen positive colonies by PCR. The recombinant lentivirus was made by cotransfection with pLKO.1GFP-nr3c1-shRNA, pCMVR8.74 (packaging vector), pMD2.G (envelope vector), and Tat and Rev plasmids in HEK293T cells using polyethyleneimine (PEI, stock concentration: 1 mg/ml). For transfection of a 100-mm dish with 70–80% confluency, the concentrations of the aforementioned were as follows: pLKO.1GFP-nr3c1-shRNA (10 µg), pCMVR8.74 (9 µg), pMD2.G (6 µg), Tat (6 µg), and Rev plasmid (6 µg). These were mixed in 1 ml of serum-free DMEM along with one-third the volume of PEI (in microliters). Mixing of all of the components was done by vortexing the mixture for 30 s followed by incubation at room temperature for 15 min. This mixture was used for transfection of HEK293T cells. The medium was replaced with complete DMEM after 6 h to remove extra PEI. At 72 h after transfection, the culture supernatants were collected and concentrated using 20% polyethylene glycol–NaCl. The concentrated virus was then used for transduction of OT1 cells. A similar process was used for generating retroviruses containing the scrambled shRNA sequences. The sequences for the scrambled shRNA were as follows: 5′-CCGGACCGGTAGTATGGTTCAGATCGAGTTCAAGAGACTCGATCTGAACCATACTGAATTCC-3′ (top strand) and 5′-GGAATTCAGTATGGTTCAGATCGAGTCTCTTGAACTCGATCTGAACCATACTACCGGTCCGG-3′ (bottom strand).

To transduce OT1 cells, MACS-purified OT1 cells were activated in vitro with anti-CD3 and anti-CD28 Abs for 48 h. These cells were then spin-transduced at 800 × g for 90 min at 37°C with freshly collected lentivirus. The cells were then kept in a humidified CO₂ incubator for 4 d. Total RNA was isolated from the nr3c1-shRNA and scrambled shRNA-transduced cells, respectively, by the TRIzol method. The isolated RNA was then converted to cDNA using a first-strand cDNA synthesis kit (Verso cDNA synthesis kit, Thermo Fischer Scientific) according to the manufacturer’s protocol. Quantitative real-time PCR (qPCR) was then performed using a 2× DyNAmo ColorFlash SYBR Green qPCR kit from Thermo Fisher Scientific (F416L). The reaction was carried out using the QuantStudio real-time PCR system (Thermo Fisher Scientific). The expression of the HPRT gene served as the endogenous control, and the relative expression of the various genes was assessed. The reaction conditions used were as follows: initial denaturation (95°C for 7 min), denaturation (95°C for 10 s), annealing, and extension (60°C for 30 s) for 40 cycles. Subsequently, melt curve analysis was performed. The primers used and the product size, respectively, for the different genes were as follows: Nr3c1 (forward, 5′-AGTGATTGCCGCAGTGAAAT-3′, reverse, 5′-GCCATGAGAAACATCCATGA-3′) = 105 bp; Tbx21 (forward, 5′-CAATGTGACCCAGATGATCG-3′, reverse, 5′-GCGTTCTGGTAGGCAGTCAC-3′) = 168 bp; Ifng (forward, 5′-TGAATGTCCAACGCAAAGCA-3′, reverse, 5′-CTGGGATGCTCTTCGACCTC-3′) = 122; Il7r (forward, 5′-CAGCAAGGGGTGAAAGCAAC-3′, reverse, 5′-CTCGCTCCAGAAGCCTTTGA-3′) = 149; Itgae (forward, 5′-CCACAGGACGAAGATCACTGT-3′, reverse, 5′-CCCTCCTTGTGCTCTCCAAG-3′) = 169.

CD8⁺CD44⁺ cells were sorted at 8 dpi from the spleens of MHV68-SIINFEKL–infected mice that had received either three doses of dexamethasone from 4 to 6 dpi or the diluent. Splenocytes of six mice per group were pooled into a single sample for sorting of the cells. RNA was isolated from the sorted cells by the TRIzol method. RNA sequencing (RNA-seq) was performed by Eurofins Genomics India. Briefly, the qualities and quantities of the RNA samples were checked on NanoDrop followed by Agilent TapeStation using high-sensitivity RNA ScreenTape. The RNA-seq paired-end sequencing libraries were prepared from the quality control–passed RNA samples using the Illumina TruSeq stranded mRNA sample prep kit as per the manufacturer’s protocol. Briefly, mRNA was enriched from the total RNA using poly-T–attached magnetic beads. Thereafter, enzymatic fragmentation was performed. First-strand cDNA conversion was achieved using SuperScript II and actinomycin D followed by second-strand synthesis. The double-stranded cDNA was then purified using AMPure XP beads, which was then followed by poly(A)-tailing and adapter ligation. A total of 15 PCR cycles were performed for enrichment. The PCR-enriched libraries were then purified using AMPure XP beads and analyzed on a 4200 TapeStation system (Agilent Technologies) using high-sensitivity D1000 screen tape as per the manufacturer’s instructions. After obtaining the Qubit concentration for the libraries and the mean peak sizes from the Agilent TapeStation profile, the PE Illumina libraries were loaded onto NextSeq 500 for cluster generation and sequencing. The kit reagents were used in the binding of samples to complementary adapter oligonucleotides on paired-end flow cells. The adapters were designed to allow selective cleavage of the forward strands after resynthesis of the reverse strand during sequencing. The copied reverse strand was used to sequence from the opposite end of the fragment. The data have been uploaded to the Gene Expression Omnibus under the accession number GSE254273 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE254273).

Raw fastq files were assessed for quality using FASTQC, and the 5′ ends were trimmed accordingly using an in-house Python script. The trimmed fastq files were subsequently mapped to the reference genome assembly mm10 of Mus musculus using Hi-SAT2. The output Bam files were processed using SAMtools and the processed bam files were used to count the number of reads that map to each gene in the mm10 assembly using HT-seq count. Genes with raw counts of ≤10 were eliminated from the analysis before normalization and calculation of FPKM (fragments per kilobase of transcript per million mapped reads) values. The raw read counts were then converted to FPKM + 1 values using a self-written Python script. Log₂ fold change for all of the genes was calculated with respect to sham and used for gene set enrichment analysis using GSEApy (20).

The lymphocytic choriomeningitis virus–specific CD8⁺ T cell single-cell RNA-seq data are publicly available at GSE131535 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE131535). Scanpy and other single-cell RNA-seq modules as implemented in Python were used for the analysis. Briefly, genes that were expressed in less than three cells and cells that expressed <200 genes were removed. In addition, cells that had a higher abundance of mitochondrial genes were also removed. Raw counts were then normalized and log-transformed before dimensionality reduction using principal component analysis. Subsequently, the neighbors graph was calculated and cells were visualized in a two-dimensional space using uniform manifold approximation and projection. A Leiden algorithm was used for clustering cells based on gene expression. All cells were scored for memory or effector signature using the “tl.score_genes” function as provided in the scanpy module, based on previously described input gene lists (21). Transcription factor activities were inferred using a univariate linear model as implemented within the decoupler. For differentially expressed gene analysis between cluster 3 and cluster 5, pseudobulk replicates were generated and processed using PyDESeq2. A PAGA graph was calculated and root cell defined to generate diffusion pseudotime for each cluster. A cell rank pseudotime kernel was used to compute a cell–cell transition matrix, which was then visualized using a stream plot. Matplotlib and Seaborn were used throughout for plotting.

OT1 splenocytes were pulsed with SIINFEKL for 24 h. The OT1 cells were then isolated by MACS. The enriched cells were then divided into two groups. One group was treated with 1 µM dexamethasone while the other group was sham treated. At 6 h posttreatment, the cells were washed three times with cold PBS and the lysate was prepared by using a lysis buffer (20 mM HEPES, 0.2 mM EDTA, 1.5 mM MgCl₂, 100 mM KCl, 20% [v/v] glycerol, 0.02% [v/v] Nonidet P-40 [pH 7.5]). The lysate was size fractionated in 10% denaturing polyacrylamide gel before transferring onto an Immun-Blot polyvinylidene difluoride membrane (Bio-Rad, 162-0177), followed by probing with the specific primary Abs and HRP- or alkaline phosphatase–conjugated secondary Abs using Clarity Western ECL (Bio-Rad, 170-5061) or femtoLUCENT PLUS-AP (G-Biosciences, 786-10AP) substrates, respectively. When necessary, the blots were stripped using a mild acidic stripping buffer (for 1 l of buffer, 2.8 g of glycine, 10 g of SDS [pH 2.2]).

The MTT assay was performed to ascertain mitochondrial function. OT1 splenocytes pulsed with SIINFEKL for 24 h were MACS purified to obtain CD8⁺ T cells. The cells were then treated with varying doses of dexamethasone for 6 h, after which they were washed five times and seeded at 2 × 10⁵ cells per well in a gelatin-coated flat-bottom 96-well optical plate. To these cells, MTT was added to a final concentration of 500 µg/ml by diluting a stock solution of 10 mg/ml (prepared in serum-free RPMI 1640). The cells were then incubated with MTT in the CO₂ incubator for 4 h. Following incubation, the cells were washed by centrifugation and the blue MTT formazan product was extracted by adding 100 µl of 100% DMSO with intermittent agitation (by pipetting) at an interval of 10 min for a total of 30 min. After the formazan extraction, the absorbance of the formazan product was spectrophotometrically determined by absorbance measurements at 570 nm. The mitochondrial function in the cells of each group was determined after normalizing with the cells in the sham-treated group using the following formula: % Mitochondrial function (relative to sham treatment) = (OD₅₇₀ dexamethasone-treated cells/OD₅₇₀ sham-treated cells) × 100.

The data presented in the figures are representative of two individual replicates. GraphPad Prism 9 was used for statistical analysis. A Student t test was used to determine the statistical significance between the two groups. For comparing more than one group, one-way ANOVA or two-way ANOVA was used. The degree of significance is denoted as follows: *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001.

We first analyzed the publicly available single-cell RNA-seq dataset of lymphocytic choriomeningitis virus–specific CD8⁺ T cells isolated during the acute phase (7 dpi) of a resolving viral infection (22). The correlation between GR (Nr3c1) activity and previously established effector and memory cell signatures was assessed (21). Whereas the GR activity positively correlated with the signatures of effector cells (r = 0.15, p < 0.0001), its negative correlation was observed with the signature of memory cells (r = −0.08, p < 0.002) (Fig. 1A). A similar correlation analysis between Id2 activity, a known marker of effector cells (23), and the signatures of effector and memory cells was also performed. As expected, Id2 activity negatively correlated with the signatures of memory cells (r = −0.14, p < 0.0001) but showed a positive correlation with those of the effector cells (r = 0.15, p < 0.0001) (Fig. 1A).

Next, we used the expression of certain key molecules such as Tcf7 (encoding TCF1), Tbx21 (encoding T-bet), Eomes, Il7r, Klrg1, Cx3cr1, and Sell (encoding CD62L or L-selectin), which are associated with different states of the effector cells, to gain insights into the heterogeneity of the cells within the dataset (Fig. 1B, 1C). The expression of CD27 within each cluster established their activated status (Fig. 1C). However, the expression of lineage-defining markers such as Tcf7, Id3, Id2, Eomes, and Il7r demonstrated heterogeneity in each cluster, which indicated that each cluster presented itself as a microstate within the pool of the responding CD8⁺ T cells. Of note, cluster 5 exhibited a phenotype similar to that of memory precursors. Accordingly, this cluster exhibited higher expression levels of Il7r, Id3, Eomes, and Tcf7 and lower levels of Cx3cr1, Klrg1, and Id2 (Fig. 1C). Next, we performed a trajectory analysis to ascertain the hierarchical emergence of each population (Fig. 1D, 1E). Cluster 7 was defined as the root population due to lower Cx3cr1 and Cd27 expression, indicating a lower level of effector differentiation as compared with the other clusters (Fig. 1C). Accordingly, cluster 7 bifurcated into clusters 8 and 9, which merged at cluster 4, which subsequently bifurcated via cluster 6 into cluster 1 and cluster 2. Whereas cluster 2 showed an association with both cluster 0 and 3, cluster 5 showed a very weak association with cluster 0. However, a comparatively stronger association of the same from cluster 3 was evident (Fig. 1D, 1E). This was in agreement with the previous observation that effector cells dedifferentiate to form memory cells (24). Clusters 3 and 5, therefore, presented themselves as the most differentiated states within the pool of the responding CD8⁺ T cells. Whereas cluster 5 was marked by higher expression of memory precursor-associated markers (Il7r, Id3, Eomes, and Tcf7), cluster 3 was marked by the expression of effector molecules such as Cx3cr1, Klrg1, and Id2 (Fig. 1C). Therefore, cluster 3 is more likely to harbor the short-lived effectors, whereas cluster 5 contains the memory precursors. We next scored each cluster for the activity of the GR (Fig. 1F). Interestingly, cluster 5 exhibited the lowest mean Nr3c1 activity, whereas cluster 3 exhibited the highest activity. Indeed, cluster 7 also exhibited lower Nr3c1 activity, but given that this cluster was the root population and expressed higher levels of the receptor (Fig. 1C), the cells contained might have induced low or no GC signaling. This was contrary to cluster 3, which exhibited the higher activity, but lower receptor expression, potentially occurring due to the negative feedback loop (Fig. 1C, 1F). The difference in Nr3c1 activity between cluster 5 and cluster 3 was statistically significant (p < 0.0001, Mann–Whitney U test, Fig. 1F). We, therefore, performed a differential gene expression analysis between cluster 3 and cluster 5 (Fig. 1G). Interestingly, cluster 5, which scored the lowest for Nr3c1 activity, exhibited a higher expression of genes associated with memory CD8⁺ T cell differentiation (Bcl2, Il7r, Tcf7, Id3, Eomes) whereas cluster 3 expressed genes synonymous with the terminal effector cells (Gzmb, Id2, Stat1, Ifng, Klrg1, Nkg7, Gzma, Havcr2). A similar trend was observed when clusters 5 and 7 were compared; however, the differences were not as profound as seen between clusters 5 and 3 (data not shown). Therefore, Nr3c1 activity negatively correlated with memory T cell signature and the memory precursor cells had the lowest Nr3c1 activity. Effector cells downregulate the GR and its downregulation in memory precursor effector cells (MPECs) could drive the memory transition possibly by inducing basal GR signaling (12).

We next tested whether a lower GR activity within the memory precursors could make them less susceptible to GC-mediated apoptosis, as such an effect could promote memory response (12). Splenocytes from OVA TCR transgenic (OT1) mice were pulsed with SIINFEKL peptide for 24 h. The activated OT1 cells were then treated with varying doses of dexamethasone and analyzed for the expression of CD127 (IL-7Rα, a marker of memory precursor cells [25]) as well as annexin V (Fig. 2A). Lower frequency of CD127⁺ as compared with the CD127⁻ cells were annexin V⁺ at even at higher doses of dexamethasone, suggesting a refractoriness of CD127⁺ cells to the killing effects of GCs (Fig. 2B). Furthermore, the CD127⁺annexin V⁻ cells were more abundant in the groups treated with intermediate doses (1–10 µM) of dexamethasone than those treated with the high or low doses (Fig. 2C). Could a modulation of GC signaling in vivo in the acute phase of a virus infection alter the representation of short-lived effector cells (SLECs; CD127⁻KLRG1⁺) and MPECs (CD127⁺KLRG1⁻) within the effector CD8⁺ T cells? To this end, we tracked the fate of adoptively transferred OT1 cells in influenza A virus (WSN-SIINFEKL)–infected animals that were additionally administered with either dexamethasone or mifepristone, an antagonist of the GR (Fig. 2D). At 7 dpi, the ratio of the absolute number of MPECs and SLECs in each group was plotted. A clear skewing toward MPECs was observed in the draining mediastinal lymph nodes (medLNs), lungs, and spleens of the dexamethasone-treated group, whereas mifepristone did not show any significant differences (Fig. 2E). Therefore, the administration of dexamethasone skewed the virus-specific CD8⁺ T cell pool toward a memory precursor phenotype.

Next, we longitudinally tracked the fate of adoptively transferred OT1 cells in the circulation of sham-, mifepristone-, and dexamethasone-treated WSN-SIINFEKL–infected animals. As compared with the mice treated with mifepristone or the diluent, dexamethasone-treated mice showed an initial reduction in the frequency of OT1 cells in their circulation, but a burst was evident at 12 dpi (Fig. 3A). Accordingly, the donor OT1 cells represented more than ∼40% of the total CD8⁺ T cells in circulation whereas such cells were less than ∼10% in the sham group. The levels remained surprisingly high (∼10% among CD8⁺ T cells) at 90 dpi in the circulation of the animals receiving a transient dexamethasone therapy whereas their frequencies were ∼1% in the other groups (Fig. 3A). Upon normalizing the frequencies to the peak of the response for each group, a slower contraction within the dexamethasone-treated animals was observed (Fig. 3B). Conversely, a steeper contraction of OT1 cells in the mifepristone-treated group was also evident. Upon heterologous challenge with MHV68-SIINFEKL at 90 dpi, the dexamethasone-treated group showed higher frequencies of the recalled OT1 cells until 7 dpi.

To better image the response at the site of infection, we compared the absolute counts of the donor cells within the lungs and draining medLNs at 4 and 7 dpi of both the primary and secondary infection. Whereas a higher count of donor cells was evident within the lungs of the mifepristone-treated animals at 7 dpi of the primary response, a significantly lower count was observed at 7 dpi of the secondary response. The increased counts in the mifepristone-treated group during the acute phase could be due to a sparing of the naive clones from GC-induced apoptosis as proposed earlier (12). However, a lower count during the secondary response was indicative of the hindered memory transition of the effector cells within the mifepristone-treated group, which is further supported by the observed steeper contraction (Fig. 3B). Conversely, whereas the dexamethasone-treated group exhibited lower counts during the primary response, a significantly higher count was observed during the secondary response in the lungs (Fig. 3D). Therefore, GC signaling could have influenced the memory transition of the virus-specific CD8⁺ T cells.

The long-term persistence of memory cells in lymphoid and nonlymphoid tissues of dexamethasone-treated animals was also evident (Supplemental Fig. 2). Accordingly, donor OT1 cells were more abundant even after 7 mo of primary infection in different secondary lymphoid organs as well as lung tissues of dexamethasone-treated mice in comparison with the control animals (Supplemental Fig. 2B, 2C). The cognate peptide stimulated cells that produced IFN-γ (single positive) or IFN-γ and TNF-α (double positive) were higher in lung tissues and medLNs of the dexamethasone-treated animals as compared with the control animals (Supplemental Fig. 2D–F). Surprisingly, very few double-positive cells were present in animals from the sham and mifepristone groups, suggesting a GC-mediated differentiation of multipotent Ag-specific CD8⁺ T cells. The elevated counts of Ag-specific CD8⁺ T cells in the dexamethasone group could be due to an enhanced turnover. We therefore measured the expression level of Ki-67, a molecule associated with cell cycle progression (26). Although the counts of Ki-67⁺ OT1 cells were comparable in spleen and medLNs of different groups, such cells were more abundant in lung tissues of the dexamethasone group when compared with control animals (Supplemental Fig. 2G, 2H). These results suggested that GC signaling enhanced the response of memory cells at tissue sites by promoting their turnover. Taken together, the data showed that a transient dexamethasone therapy during the expansion phase of a localized infection served to elevate the pool of resting memory CD8⁺ T cells, whereas a blockade of endogenous GC signaling via pharmacological antagonism attenuated the memory transition of the virus-specific CD8⁺ T cells.

A systemic administration of dexamethasone could indirectly influence the differentiation of CD8⁺ T cells because NR3C1 is ubiquitously expressed by all nucleated cells (11). Therefore, it was important to delineate the contribution of CD8⁺ T cell–intrinsic changes or the altered environment in the GC-mediated differentiation of CD8⁺ T cell memory. We sorted the virus-specific cells from WSN-SIINFEKL–infected dexamethasone-, mifepristone-, and sham-treated animals at 7 dpi (Supplemental Fig. 4A) and transferred these cells into naive, unmanipulated mice (Fig. 4A). Tracking of these cells in the circulation of the recipients revealed a nonsignificant increase in the frequency of the tetramer-positive cells in the circulation of dexamethasone-treated cell recipients 2 wk after transfer. Upon i.n. challenge with MHV68-SIINFEKL, a rapid increase in virus-specific cells was observed in the circulation of the dexamethasone-treated cell recipients, with the tetramer-positive cells ∼30-fold more in the circulation at 3 d after rechallenge (Fig. 4B). A higher count of the dexamethasone-treated cells was also found in the lung tissues, whereas these cells were barely detectable at other sites (Fig. 4C, 4D). Additionally, the cells collected from the lungs of dexamethasone-treated recipients divided more efficiently following an in vitro peptide pulse as analyzed by a CFSE dilution assay (Fig. 4E, 4F). These results suggested a dexamethasone-induced programming within the CD8⁺ T cells during their initial antigenic exposure that led to efficient recall, preferential homing, and residence in nonlymphoid organs such as lungs. Both IFN-γ as well as IFN-γ and TNF-α producers were more abundant in the animals that had received dexamethasone-treated cells in comparison with the control cells (Fig. 4G, 4H). A significantly higher count of granzyme B–producing dexamethasone-treated cells was also observed in the lungs as compared with the other groups (Fig. 4I, 4J). This approach, however, did not discriminate between the endogenous or the dexamethasone-exposed donor cells expanded following infection. We therefore performed adoptive transfer experiments to track the fate of donor cells into congenic CD45.1⁺ recipients. Similar results were obtained when MHV68-SIINFEKL infection-expanded CD45.2⁺OT1 cells exposed to dexamethasone, mifepristone, or sham treatment were tracked following their adoptive transfer in CD45.1⁺ congenic mice by homologous infection 4 mo later (Fig. 4K–M). The dexamethasone-treated cell recipients exhibited better IFN-γ and TNF-α production (Fig. 4N, 4O) and, consequently, these animals had lower viral load in the lung tissues (Fig. 4P).

The data above highlighted a T cell–intrinsic role of GC signaling in the optimal differentiation of memory. However, the enhanced response observed in the dexamethasone-treated group could be due to the transfer of cells biased toward an MPEC phenotype (as shown in Fig. 2E), and not necessarily by the functional enhancement of memory cells. To address this issue, we FACS sorted MPECs (CD127⁺KLRG1⁻, Fig. 5A) from sham-, mifepristone-, and dexamethasone-treated mice for adoptively transferring into the unmanipulated naive CD45.1⁺ mice. The recipients were then infected i.n. with 2 × 10⁵ PFU of MHV68-SIINFEKL and analyzed at 10 dpi. Whereas higher counts of the donor cells were observed within the draining medLNs of the recipients of mifepristone- and dexamethasone-treated cell, the lungs of the recipients of mifepristone-treated cells had lower counts of the donor cells (Fig. 5B). This showed a poor recall of the mifepristone-treated MPECs. Upon assessing the functionality of the donor cells by an intracellular cytokine staining (ICCS) assay, better functionality of the dexamethasone-treated cells was evident. Accordingly, a higher proportion of both IFN-γ⁺ single-positive and IFN-γ⁺TNF-α⁺ double-positive cells was observed in the lungs as well as medLNs of the dexamethasone cell recipients (Fig. 5C, 5D). These results implied that a higher magnitude of cells survived in the dexamethasone-treated group than the mifepristone group, thus showing an apparent enhancement of memory responses.

Next, to clearly assess the T cell–intrinsic role of GR signaling in memory transition, we genetically ablated the GR by shRNA-mediated knockdown and then phenotypically characterized these cells. A schematic of the experiment is shown in Fig. 6A. The knockdown of NR3C1 was evident at the mRNA and protein levels (Fig. 6B). The cells lacking NR3C1 expression had a significant reduction in the mRNA expression of Tbx21 (T-bet), Ifng (IFN-γ), Itgae (CD103), and Il7r (CD127), which suggested the role of GC signaling in modulating the response of the virus-specific CD8⁺ T cells (Fig. 6B). Control and GR-deficient cells were then adoptively transferred into naive, unmanipulated recipients. A lower expression of CD127 by the GR-depleted cells was evident at the time of transfer (Fig. 6C, gray histogram [Nr3c1_shRNA]: median fluorescence intensity = 995 versus histogram with no fill [scrambled]: median fluorescence intensity = 1477). Whereas equal frequencies of tetramer-positive cells were observed in the circulation at 1 d posttransfer, the animals receiving the control cells showed a significantly higher frequency of tetramer-positive cells at 7 d posttransfer (Fig. 6D). At 14 d posttransfer, the recipients were infected i.n. with WSN-SIINFEKL to monitor the expansion of the virus-specific cells in the circulation. The recipients of Nr3c1_shRNA cells showed significantly lower counts of the Ag-specific cells in the draining medLNs and lungs as compared with those receiving the control (scrambled) cells following infection (Fig. 6E, Supplemental Fig. 3A). SIINFEKL peptide–stimulated cells that produced IFN-γ were significantly lower in both the draining medLNs and lungs of animals that received Nr3c1-depleted OT1 cells as compared with those receiving the control cells (Fig. 6F, 6G). Endogenous WT cells could contribute to the overall pool of K^b-SIINFEKL-tetramer positive cells, we therefore performed such experiments by adoptively transferring control or the Nr3c1 knock-down OT1 cells (CD45.2⁺) into congenic CD45.1⁺ mice. We observed an efficient recall response of the control donor cells in recipients as compared with those of the Nr3c1 knockdown cells (Fig. 6H, Supplemental Fig. 3B). Therefore, signaling via Nr3c1 is crucial in the differentiation of CD8⁺ T cells to become long-lived memory cells.

We next compared the proliferative potential of the scrambled and Nr3c1-knockdown OT1 cells in a lymphopenic environment, as this would help assess their ability to transition into memory cells by undergoing homeostatic turnover (27, 28). To this end, 10³ scrambled or Nr3c1-shRNA transduced OT1 cells were transferred into Rag1^−/− mice. After 30 d of transfer, the GFP⁺ cells were analyzed (Fig. 6I). A lower count of Nr3c1-depleted cells was recorded in PBMCs, pooled lymph nodes (cervical, mandibular, brachial, axillary, inguinal, mesenteric, mediastinal, popliteal), and the splenic tissues of the recipients (Fig. 6J, Supplemental Fig. 3B). Therefore, a lack of Nr3c1 expression reduced the survival and homeostatic expansion of the CD8⁺ T cells in a lymphopenic environment, and in so doing reduced the persisting memory cells. Given that the receptor was knocked down in the activated CD8⁺ T cells, it can be assumed that GR signaling intrinsic to CD8⁺ T cells seems to be important in the memory transition of the effectors.

Taken together, the data demonstrated that GR signaling within the responding CD8⁺ T cells plays a key role in their transition to generate memory.

To gain mechanistic insights into the GC-mediated changes within the activated CD8⁺ T cells following virus infection, we performed RNA-seq analysis of polyclonal CD8⁺CD44⁺ cells sorted from sham- and dexamethasone-treated MHV68-infected animals (Fig. 7A). A pathway enrichment analysis showed that the dexamethasone-treated cells exhibited a negative enrichment of the effector cell signature (normalized enrichment score [NES] = −2.565, p < 0.0001, false discovery rate [FDR] < 0.0001) and of the genes involved in fatty acid oxidation (NES = −1.413, p = 7.595e−02, FDR = 1.093e−01), oxidative phosphorylation (NES = −2.419, p < 0.0001, FDR < 0.0001), and the citric acid cycle (NES = −1.561, p = 1.961e−02, FDR = 1.655e−01) (Fig. 7B). Furthermore, a positive enrichment of the memory signature (NES = 1.342, p = 2.515e−02, FDR = 2.464e−02) and of the genes involved in cytokine–cytokine receptor signaling (NES = 1.808, p < 0.0001, FDR = 1.22e−01) and JAK-STAT signaling (NES = 1.633, p = 1.236e−02, FDR = 2.578e−01) was observed (Fig. 7B). Therefore, dexamethasone exposure of the effector T cells perturbed multiple pathways encompassing cellular metabolism as well as the cytokine responsiveness.

Activated T cells are metabolically active (29–31), and this might result in the accumulation of reactive oxygen species (ROS), which contributes to cellular death in the contraction phase (32). Therefore, dexamethasone-mediated suppression of metabolic processes could reduce cellular ROS levels and spare the effector cells from apoptosis to facilitate their memory transition. To test this hypothesis, we first assessed mitochondrial functionality via an MTT assay, as the readout of the MTT assay directly depends on mitochondrial succinate dehydrogenase activity. To this end, splenocytes from OT1 mice were stimulated with SIINFEKL peptide for 24 h and MACS purified. The purified CD8⁺ T cells were then incubated with varying doses of dexamethasone for 6 h, after which their mitochondrial activity was assessed. Although the cells treated with 0.1 and 1 µM dexamethasone did not exhibit any drop in mitochondrial functionality relative to the sham-treated cells, a significant drop was observed for the 10 µM (20% decrease) and 100 µM (50% decrease) treated groups (Fig. 7C). We next assayed the accumulation of cellular ROS in control or the dexamethasone-treated cells by measuring the mean intensity of DCFDA fluorescence by fluorescence microscopy. Dexamethasone at 1 as well as a 100 µM dose led to a significant reduction in ROS levels (Fig. 7D, 7E). The higher dose could have induced toxicity, but the lower doses of dexamethasone reduced the cellular ROS accumulation without affecting viability. This could result in better survival and memory transition of the activated CD8⁺ T cells.

Higher cellular ROS levels are known to downregulate the expression of prosurvival proteins such as Bcl2 (reviewed in Ref. 27). We therefore analyzed the expression of Bcl2 in dexamethasone-treated cells since recovery of Bcl2 expression is an important marker of memory transition (25). To this end, SIINFEKL-pulsed OT1 cells were MACS purified and exposed to 1 or 100 µM dexamethasone in vitro for 6 h. DMSO-treated cells were taken as a control. Upon completion of treatment, the cells were washed five times with PBS and then cultured for 2 more days after treatment in fresh medium. The expression of Bcl2 was analyzed by flow cytometry at 0 h (before treatment or just after activation), 6 h (just after treatment), 24 h (at 1 d posttreatment) and 48 h (at 2 d posttreatment) (Fig. 7F). Although a slight but significant increase in Bcl2 expression was observed just after dexamethasone exposure, an ∼2-fold increase was observed in the cells treated with 1 µM dexamethasone after 48 h (Fig. 7F). Therefore, dexamethasone-exposed CD8⁺ T cells accumulate lower levels of ROS and upregulate the antiapoptotic marker Bcl2, which could result in their preferential survival to become memory cells.

In addition to gene enrichment for metabolic pathways, dexamethasone-exposed CD8⁺ T cells showed significant enrichment scores for cytokine receptors as well as and JAK-STAT signaling (Csf3r, Ccr1, Csf1r, Il1b, Tnfsf8, Ccr6, Tnfrsf14, Csf2rb, Cxcl10, Csf2ra, Tnfrsf4, Ccl15, Il6st, Tnfrsf21, Il7r, Ifngr2, Tnfsf4, Acvr1b, Cxcr5, Clcf1, Il6r, Il10, Il2ra, Ccr7, Tnfrsf25, Lta, Ccr9, Il12rb2, Tgfbr1, Il4r, Tnfsf9). The upregulation of IL-7 and IL-6 receptors was of particular interest given their synergistic effects in CD8⁺ T cell differentiation (25, 33). Therefore, we analyzed the phosphorylation of STAT5 and STAT3 because of their role in cell survival and turnover (Fig. 7G). To this end, OT1 splenocytes were stimulated with SIINFEKL peptide for 24 h, after which the cells were either dexamethasone or sham treated for 6 h. This was followed by the MACS purification of the CD8⁺ OT1 cells, which were then used for lysate preparation and Western blot analysis. The dexamethasone-treated cells had significantly higher levels of both phosphorylated STAT3 and STAT5β (Fig. 7G). Interestingly, smaller-sized products of STAT5β were also observed that could be the cleaved products of STAT5β, which might be crucial in the differentiation process of memory CD8⁺ T cells. Therefore, increased phosphorylation of both STAT5β and STAT3 was observed within the dexamethasone-treated cells. Given that these cells were not treated with any specific cytokine, both autocrine or paracrine signals via multiple cytokine receptors could be involved in the induction of STAT3 and STAT5 phosphorylation. Although IL-6 family members, IL-10 family members, and IL-21, IL-27, G-CSF, and type I IFNs can induce STAT3 phosphorylation, the γ-chain cytokines possess the ability to induce STAT5 phosphorylation (34). Indeed, the dexamethasone-treated cells were also observed to have an enrichment of receptors involved in the same (Fig. 7B). The exact cytokines involved would have to be delineated by targeted studies to elucidate the direct or indirect effect that GCs have on JAK-STAT induction.

Therefore, prosurvival STAT signaling within the activated CD8⁺ T cells upon dexamethasone treatment programmed these cells to generate memory cells.

Finally, we stimulated OT1 cells in vitro with anti-CD3 and anti-CD28 Abs followed by a brief exposure to 1 µM dexamethasone or the diluent to directly assay memory transition upon in vitro dexamethasone exposure. The dose was shown to enhance Bcl2 recovery in vitro (Fig. 7F). A detailed schematic of the experiment is shown in Fig. 8A. That the recipients had an equal number of transferred cells at 1 d posttransfer was shown by K^b-SIINFEKL-tetramer staining. The cells were then tracked over time. At 6 d posttransfer, K^b-SIINFEKL-tetramer⁺ cells in the recipients of dexamethasone-treated cells were 2-fold higher as compared with those receiving control cells (Fig. 8B, 8C). The recipients were infected with MHV68-SIINFEKL i.n. 30 d posttransfer to recall the persisting cells. We observed significantly higher frequencies and numbers of K^b-SIINFEKL-tetramer⁺ CD8⁺ T cells in bronchoalveolar lavage and spleens of infected mice at 8 dpi in the animals receiving dexamethasone-treated cells than those receiving the control cells (Fig. 8D–G). Significantly higher frequencies of the peptide-stimulated cells produced IFN-γ in the recipients of dexamethasone-treated cells than in the control cell recipients (Fig. 8H, 8I). The virus titers were ∼16-fold lower in the lung tissues of the animals receiving dexamethasone-treated cells as compared with those injected with control cells (Fig. 8J). The extent of tissue disruption in the lungs and elevated cellular infiltrates was less in the recipients of dexamethasone-treated cells as compared with those infused with the control cells (Fig. 8K). More alveolar spaces were also visible in animals receiving dexamethasone-treated cells as compared with the control cells (Fig. 8K). The data, therefore, suggested a better protection rendered to the virus infected host by in vitro dexamethasone-treated cells during a recall response, suggesting a better memory transition of the dexamethasone-treated cells.

GCs are induced following infection and modulate the functionality of the immunocytes (6, 11, 35, 36). The GR not only transduces signaling in the responding cells but also acts as a transcription factor that can modulate cellular programming (37). Given that GCs are regularly used to alleviate infection-induced immunopathology, understanding their roles in the fate determination of T cells is crucial. We had previously reported a differential responsiveness of the CD8⁺ T cells to GC-mediated apoptosis, with the effector cells being resistant while the quiescent memory and naive counterparts showed enhanced susceptibility (12). The current study showed a dexamethasone-induced skewing of the stimulated CD8⁺ T cells toward an MPEC phenotype (CD127^hiKLRG1^lo). Interestingly, we show that a controlled GR signaling within the effector cells reduces their propensity to undergo GC-mediated apoptosis that could facilitate memory responses, but a basal GR response in such cells is critical for optimal memory differentiation. Consequently, pharmacological blockade using mifepristone and genetic ablation via shRNA-mediated knockdown of the GR within the activated CD8⁺ T cells compromised memory formation and maintenance in vivo. We, therefore, propose that a “goldilocks zone” of GC signaling may be critical for the differentiation of memory CD8⁺ T cells, wherein high amounts of signaling result in apoptosis, and no signaling attenuates memory transition.

Two models of CD8⁺ T cell differentiation majorly explain the origin of memory CD8⁺ T cells. Whereas one category posits that the memory cells originate from naive cells via an early bifurcation into memory precursors and terminal effectors (38–40), another proposes a sequential differentiation from naive to effectors to memory (24, 41, 42). Even though distinct from each other, both of these models proposed a rigid framework for memory differentiation wherein an expected trajectory or predetermined fate pattern had to be followed. Recent work from Abadie et al. (43), however, has unified these two models by showing that the decision to differentiate into a memory cell can be made at multiple stages after Ag encounter, thus giving the CD8⁺ T cells the ability to scale the magnitude of memory generation based on the infection. Therefore, the “plasticity” of the CD8⁺ T cells in terms of fate determination might be exploited to augment memory T cell differentiation throughout the acute phase of response. Indeed, previous studies have shown the enhancement of memory differentiation via the suppression of the inflammatory programming within the responding CD8⁺ T cells (16, 44–46). We observed an inhibition of the effector and the promotion of memory signatures within the dexamethasone-treated cells (Fig. 7B). Additionally, a negative enrichment of genes involved in cellular metabolism was also observed. Activated CD8⁺ T cells accumulate ROS due to enhanced metabolism, and this induces cell death (32). Accumulation of ROS is a key marker of terminally differentiated cells that have lost their stemness (47–50). Therefore, dexamethasone-induced suppression of metabolism could ensure the survival of these cells, thus promoting memory differentiation. Accordingly, a lower accumulation of cellular ROS and an enhanced expression of Bcl2 by the dexamethasone-treated cells were evident. Most importantly, these effects were evident only at a low dose of dexamethasone whereas a high dose resulted in mitochondrial damage. Therefore, a controlled GC signaling in the effector CD8⁺ T cells served to fine tune memory differentiation program. Given that T cell exhaustion is caused by an inability to inhibit inflammation due to pathogen persistence, it would also be interesting to investigate whether a controlled induction of GC signaling would be able to rescue T cell exhaustion via T cell–intrinsic reprogramming. Our study could be a primer for future studies investigating the same.

Although acute stress experienced at the time of immune induction has been shown to enhance innate and adaptive immune responses, chronic stress has been associated with a suppression of immune function (51). Additionally, chronic exposure to GCs within the tumor microenvironment has been shown to induce the exhaustion of tumor-infiltrating CD8⁺ T cells (14). Conversely, the results presented in this study have highlighted the role that controlled exposure to GCs plays in the differentiation of memory CD8⁺ T cells in the context of a viral infection. Therefore, a tight regulation of GC signaling within the CD8⁺ T cells is apparent. Acute stress induces a targeting of leukocytes to the skin to augment skin-associated immunity, whereas such a redistribution is inhibited during a chronic stress response (52). We reported a GC-mediated targeting of the virus-specific CD8⁺ T cells to the site of infection, which was in part controlled by the upregulation of the migratory receptor, CXCR3 (12). Dhabhar et al. (15, 53–55) proposed that “psychophysiological stress response is nature’s fundamental survival mechanism that could be therapeutically harnessed to augment immune function during vaccination, wound healing or infection.” Induction of a controlled GC signaling might therefore be exploited to enhance the vaccine-induced memory CD8⁺ T cell responses. However, we propose this with caution because the levels of inflammation might vary between infection and immunization, and therefore a dose optimization would be imperative. Even though unexplored, the potential of such an approach calls for the investigation of dexamethasone as an adjuvant to assist the generation of virus-specific CD8⁺ T cell memory in response to vaccination.

Even though our study shows a role of GC signaling in memory CD8⁺ T cell differentiation, the findings raise certain questions that we believe could be addressed in the future. The first is regarding the exact mechanism in which GC signaling brings about the metabolic changes that facilitate memory differentiation. At this point, we can only speculate. Because the downregulation of many genes involved in metabolism was observed, direct regulation of gene transcription via the GR might be involved. Indeed, chromatin immunoprecipitation sequencing experiments on mouse C2C12 myotubes have shed light on the potential target genes of the GR, with multiple targets being involved in the inhibition of mTOR and PI3K-Akt signaling (56). This is a probable explanation for the phenotype we observed because mTOR inhibition via rapamycin treatment has previously been implicated in memory T cell differentiation (57). These studies would have to be replicated using CD8⁺ T cells to provide some explanation along these lines. We cannot rule out the nongenomic effects of the GR. Of note, a recent study has highlighted the nongenomic mechanism through which the GR induces anti-inflammatory signaling within macrophages (58). By directly interacting with the pyruvate dehydrogenase complex, the GR directed an increased activity of the TCA cycle, rewiring mitochondrial metabolism to produce the anti-inflammatory metabolite itaconate. Therefore, it is possible that direct modulation of various metabolic enzymes and subsequent manipulation of the concentration of metabolites within the effector CD8⁺ T cells could also be a probable mechanism through which GCs prevent ROS accumulation and modulate memory differentiation. The modulation of the concentration of metabolites could also be a probable explanation of the pleiotropic effects that are usually associated with GCs. Metabolic flux analyses on GC-treated CD8⁺ T cells could help reveal the same. These points, therefore, present themselves as exciting avenues for investigation and could help shed light on the dynamic modulation of CD8⁺ T cell differentiation by glucocorticoids.

The authors have no financial conflicts of interest.

The help provided by Prateek of the flow cytometry facility at IISER, Mohali is appreciated.