MicroRNA-139 Expression Is Dispensable for the Generation of Influenza-Specific CD8⁺ T Cell Responses

MicroRNAs (miRNAs/miRs) are the most well-studied group of noncoding RNAs (1). miRNAs are small (∼22 nucleotides) single-stranded noncoding RNAs that modulate gene expression by directly binding to target mRNA in a sequence-specific manner (2), resulting in the inhibition of mRNA translation or RNA instability and thereby promoting the degradation of target mRNAs. We and others have previously demonstrated that modulating the expression of a single miRNA in immune cells can have a profound impact on the immune response to infection and tumors (3–7). It has also become clear that miRNA-directed gene regulation is heavily influenced by cell-extrinsic and -intrinsic factors such as the local inflammatory environment or the activation state of a given cell type (8, 9). It is therefore of great interest to explore miRNA expression and regulation of miRNAs’ specific targets in the context of different infection and tumor systems because the miRNAs involved and their effects may differ in these conditions.

A crucial component of the immune response to infection is the development of effector CD8⁺ CTLs, because these cells are ultimately responsible for the clearance of intracellular infections and the control of tumors (10, 11). Influenza virus infection of an immunocompetent host, both in humans and in mice, results in an acute viral infection localized in the lungs that is efficiently cleared by CD8⁺ T cells followed by the establishment of long-lived influenza-specific memory CD8⁺ T cells (12). Over the course of this response, CD8⁺ T cells undergo significant changes in gene expression (13) that are, in part, mediated by miRNAs. miRNAs are of critical importance to overall T cell development as well as the development of effective responses to pathogens. The importance of miRNAs in T cell development was first shown when Dicer, an enzyme critical for the maturation of miRNAs (14), was deleted in T cells (15). Previously, the role of specific miRNAs had been studied by our group (3, 5) and others. However, large-scale changes in miRNA expression during the development of CTL responses to influenza virus infection remain unknown. We hypothesized that in vivo differentiated influenza virus–specific CD8⁺ T cells would demonstrate miRNA expression patterns unique to the activation state of the cell. To address this, we performed miRNA microarray analysis on naive or in vivo generated effector influenza virus–specific CD8⁺ T cells to determine the relative miRNA expression profiles during acute viral infection in a murine model. We determined the miRNA profile of an in vivo CTL response and identified 120 mature miRNAs with significant changes in expression that accompany effector CTL differentiation.

One miRNA that was greatly downregulated in effector CTLs was miR-139. miR-139 has previously been identified as a tumor suppressor, such as in myeloid progenitors, where miR-139 regulates the expression of EIF4G2 and TSPAN3, and during interstrand cross-link (ICL)-induced stress, where miR-139-3p was suggested to repress the RNA binding protein HuR (16–19). In addition, miR-139 is downregulated in chronic myeloid leukemia, where miR-139-5p serves as a regulator of proliferation and Brg1, an ATP-dependent helicase with a known role in BCR-ABL transformation in chronic myeloid leukemia (20). Decreased expression of miR-139-5p has also been associated with several other types of cancer, including breast cancer (21) and non–small cell lung cancer (22). However, a role for miR-139 in immune cell responses remains poorly understood. One study by Trifari et al. was the first to identify a role for miR-139 (particularly miR-139-3p) in CD8⁺ T cell differentiation and function by assessing the impact of miR-139 overexpression in wild-type (WT) or Dicer-deficient (and therefore lacking all mature miRNAs) CD8⁺ T cells (23). Trifari et al. specifically noted that miR-139 regulated the transcription factor Eomesodermin (Eomes), which is known to be upregulated in effector T cells and to a greater extent in memory and exhausted CD8⁺ T cells (24, 25). Our miRNA profiling identified that miR-139 was significantly downregulated in effector CTL and upon in vitro activation. We therefore sought to test the direct contribution of miR-139 in effector and memory anti-influenza CTL responses.

Our study reports the miRNA changes that CTLs undergo as they differentiate from naive CD8⁺ T cells into effector CTLs in an in vivo viral infection. We specifically examine miR-139, which previously has been reported to affect effector CTL differentiation and is significantly downregulated upon TCR stimulation. Using miR-139^−/− mice and an overexpression system, we demonstrate here that in vivo miR-139 expression does not affect effector or memory CTL development during influenza virus infection. Using two models of influenza virus infection (PR8 and Wilson Smith Neurotropic/33 expressing OVA_(257-264) [WSN-OVA]), we evaluated the ability of miR-139 to regulate CTL responses. By all parameters assessed, neither miR-139 deficiency nor miR-139 overexpression altered CTL development or responses to influenza virus infection. However, in a head-to-head competition format during influenza virus infection, miR-139–sufficient CTLs outcompeted miR-139–deficient CTLs. To assess if miR-139 expression was dispensable for the generation of CTL responses to a nonviral infection, we used Listeria monocytogenes (LM) as a model of intracellular bacterial infection. In these studies, we observed that miR-139 deficiency led to decreased effector CD8⁺ T cell responses, but Eomes expression was unaffected, similarly to our studies in influenza virus infection. We also evaluated if lack of miR-139 affected the development of exhausted CD8⁺ T cells using an in vitro model system of T cell exhaustion driven by TCR stimulation (26). In this system, we found no evidence for miR-139 deficiency affecting the development of T cell exhaustion. We therefore conclude that, despite significant downregulation upon activation of CD8⁺ T cells, miR-139 expression is dispensable for the in vivo generation of influenza-specific effector and memory Ag-specific CD8⁺ T cell responses.

C57BL/6 Tg(TcraTcrb)1100Mjb/J (OT-I) mice were backcrossed with B6.SJL-Ptprca Pepcb/BoyJ (CD45.1⁺) mice (both from The Jackson Laboratory) to generate OT-I CD45.1⁺ mice on the C57BL6/J background. miR-139–deficient (miR-139^−/−) mice were generated on the C57BL/6 background using CRISPR/Cas9 nickase mRNA and four single-guide RNAs (sgRNAs) containing an sgRNA 5′ plus strand: 5′-GTCAGTACAGTGGGAGTGCC-3′; sgRNA 5′ minus strand: 5′-GGTGATAGAGAGTGGGAAGG-3′; sgRNA 3′ plus strand: 5′-GTGAGTGAAAGGCACGTCTC-3′; sgRNA 3′ minus strand: 5′-GGTGTAAAGACCGGGATGTA-3′. This was done to introduce four DNA nicks flanking miR-139 encoding sequences in C57BL/6J zygotes before transplant into a BCBA (F1: C57BL/6J × CBA) female (Janvier) as published elsewhere (Supplemental Fig. 4F–H) (27). F1 heterozygote offspring were crossed until complete homozygote knockouts were confirmed using PCR and sequence analysis of multiple tissues. miR-139^−/− mice were backcrossed with OT-I CD45.1 mice until complete miR-139 deficiency was confirmed using PCR analysis, and OT-I CD45.1 expression was confirmed using PCR and flow cytometry analysis. C57BL/6J mice, OT-I CD45.1 mice, miR-139^−/−, and miR-139^−/− OT-I mice (both on the C57BL/6J background) were kept in a barrier facility (certified by the Association for Assessment and Accreditation of Laboratory Animal Care) at Drexel University College of Medicine or in a barrier facility at Erasmus University Medical Center. This study was carried out in accordance with the recommendations of the institutional animal care and use committee or the Instantie voor Dierenwelzijn. The protocols were approved by the institutional animal care and use committee or the Instantie voor Dierenwelzijn. For influenza infections, female mice at 8–12 wk old were anesthetized with 2.5% isoflurane gas and were infected intranasally with 1.6 × 10⁴ PFU influenza virus strain A/WSN-OVA (a gift from D. Topham, University of Rochester Medical Center) or 1.3 × 10⁵ PFU influenza virus strain A/Puerto Rico/8/1934 (PR8; H1N1; a kind gift from W. Gerhard, Wistar Institute). Virus infectious doses were titrated such that WT mice lost ∼20% of initial body weight by day 10 of infection before recovering. For LM-OVA (a gift from Ananda Goldrath) infections, female mice 8–12 wk old were infected i.v. with 3.5 × 10⁴ CFU LM-OVA grown in brain/heart infusion media containing 5 μg/ml erythromycin.

Naive (CD44⁻CD62L⁺) and day 9 after influenza virus infection antiviral donor (CD45.1⁺) OT-I CD8⁺ T cells were sorted from the spleens or lungs of naive or infected mice, respectively, and frozen at −80°C in TRI Reagent (Molecular Research Center, Cincinnati, OH). Total RNA was isolated from sorted cells, and miRNA gene expression was assessed using an Affymetrix murine miRNA 4.0 gene chip array at the Thomas Jefferson University Cancer Genomics Laboratory (Thomas Jefferson University, Philadelphia, PA). Microarray probe intensity data (CEL) files were normalized and analyzed using the Affymetrix Expression Console (Thermo Fisher Scientific); miRNA annotation is from miRBASE version 21. Expression data were exported to GraphPad Prism 7 (GraphPad Software, La Jolla, CA). To assess miRNA expression in naive and effector cells, expression was counted if the average detection p value was p ≤ 0.05. To calculate fold change of miRNA expression in effector cells, naive expression values were averaged, and individual effector expression value fold change was calculated against the naive average; overall effector fold change was calculated by averaging the individual effector fold change. Fold change p value was calculated using GraphPad Prism 7’s “multiple T tests” feature, analyzing each miRNA independently and using the recommended two-stage step-up method of the Benjamini, Krieger, and Yekutieli false discovery rate (FDR) method with an FDR of 5%. miRNAs were considered significantly upregulated or downregulated if effector cells had a fold change ≥2 or less than or equal to −2 compared with naive cells, a p value ≤0.05, and a q value ≤0.05. The microarray data presented in this article have been submitted to the Gene Expression Omnibus (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc5GSE183146) under accession number GSE183146.

Significantly upregulated and downregulated miRNAs in effector CD8⁺ T cells and their fold change values, miRNAs detected in naive CD8⁺ T cells only, and effector CD8⁺ T cells only were uploaded for Ingenuity Pathway Analysis (Qiagen). Ingenuity Pathway Analysis miRNA Target Filter analysis paired with mRNA expression data from day 10 influenza virus infection effector OT-I CD8⁺ T cells (28) identified canonical pathways predicted to be affected by the upregulated and downregulated miRNAs. Targets were limited to those expressed in day 10 CTLs and targets predicted by pairing miRNA upregulation with mRNA downregulation or conversely miRNA downregulation with mRNA upregulation. In the figures, blue indicates downregulated genes, and red indicates upregulated genes in effector versus naive OT-I CD8⁺ T cells.

Spleens from uninfected female CD45.1⁺ OT-I or CD45.1⁺ miR-139^−/− OT-I transgenic mice were processed into a single-cell suspension by passing cells through a 40-µm cell strainer (BD Falcon; BD Biosciences) into RPMI 1640 containing 5% FBS + 1% penicillin/streptomycin and 1% l-glutamine. RBCs were lysed using ammonium-chloride-potassium solution. Cells were counted using trypan blue staining and a hemocytometer, and the frequency of CD8⁺ T cells was determined by flow cytometry. OT-I splenocytes were resuspended at a concentration of 10⁵ CD8⁺ T cells per milliliter in sterile 0.9% NaCl. Equal numbers (10⁴) of WT or miR-139^−/− CD45.1⁺ OT-I CD8⁺ T cells were transferred i.v. into female CD45.2⁺ WT C57BL/6J recipient mice through tail vein injection of 100 µl. For competition experiments, naive (CD44⁻CD62L⁺) OT-I CD8⁺ T cells were isolated from (CD45.1⁺/CD90.1⁺) WT OT-I mice and (CD45.1⁻/CD90.1⁺) miR-139^−/− OT-I mice using magnetic bead negative selection. WT and miR-139^−/− OT-I cells were mixed 1:1 and confirmed via flow cytometry prior to transfer into CD45.2⁺CD90.2⁺ C57BL/6 mice. At the time of transfer, all WT recipient host mice were between 8 and 12 wk of age. All donor OT-I mice were between 8 and 12 wk of age. Three hours later, the recipient mice were anesthetized using 2.5% isoflurane gas and infected intranasally with influenza virus WSN-OVA. Only female donor and recipient mice were used for these experiments.

The murine miR-139–encoding region was cloned into the murine stem cell virus–internal ribosome entry site–Thy1.1 vector (provided by P. Marrack, University of Colorado). A scrambled control insert producing no functional miRNA was similarly cloned into the murine stem cell virus–internal ribosome entry site–Thy1.1 vector. Retroviruses were produced in the Platinum-E cell line (Cell Biolabs, San Diego, CA). Retroviral transduction of primary OT-I CD8⁺ T cells was completed as previously described (3). Splenic CD8⁺ T cells were isolated by negative selection with magnetic beads (EasySep; STEMCELL Technologies) from uninfected OT-I CD45.1⁺ female mice 8–10 wk of age. The purity of CD8⁺ T cells was >90% as determined by flow cytometry. Isolated CD8⁺ T cells were activated for 48 h using solid-phase α-CD3 Ab (0.25 µg/ml, clone 17A2; eBioscience, San Diego, CA) and α-CD28 Ab (5 µg/ml, clone 37.51; eBioscience) in 10% RPMI medium containing 1% penicillin/streptomycin, 1% HEPES, 1% sodium pyruvate, 1% sodium bicarbonate, and β-mercaptoethanol (complete T cell media) with 20 U/ml recombinant human IL-2 (Roche, Basel, Switzerland) and 5 ng/ml recombinant murine IL-7 and 5 ng/ml recombinant murine IL-15 (both from PeproTech, Rocky Hill, NJ). Cells were collected and plated at a density of 3 × 10⁶ cells per 2 ml in poly-d-lysine plates (Thermo Fisher Scientific, Waltham, MA) coated with 20 µg/ml of RetroNectin (Takara, Shiga, Japan) and preloaded with retroviral supernatants. Cells were incubated for an additional 48 h. Transduction efficiency was determined by expression of Thy1.1 (CD90.1). Transduced cells were sorted with a FACSAria III sorter (BD Biosciences, San Jose, CA).

Splenic OT-I CD45.1⁺ and OT-I miR-139^−/− CD45.1⁺ CD8⁺ T cells were isolated by negative selection with magnetic beads (EasySep; STEMCELL Technologies) from uninfected female mice 8–10 wk of age. The purity of CD8⁺ T cells was >90% as determined by flow cytometry. Isolated CD8⁺ T cells were activated for 24 or 72 h using solid-phase α-CD3 Ab (0.25 µg/ml, clone 17A2; eBioscience) and α-CD28 Ab (5 µg/ml, clone 37.51; eBioscience) in 10% RPMI medium containing 1% penicillin/streptomycin, 1% HEPES, 1% sodium pyruvate, 1% sodium bicarbonate, and β-mercaptoethanol (complete T cell media) with 20 U/ml recombinant human IL-2 (Roche), 5 ng/ml recombinant murine IL-7 and 5 ng/ml recombinant murine IL-15 (both from PeproTech). For evaluation of miR-139 upregulation in response to TCR signal strength, splenic OT-I CD45.1⁺ CD8⁺ T cells were isolated as above and activated for 24 h in complete T cell media with 20 U/ml recombinant human IL-2 and 10 ng/ml of either OVA (SIINFEKL) peptide or the OVA peptide variants E1 (EIINFEKL) or R4 (SIIRFEKL) (29).

In vitro exhaustion was induced in CD8⁺ T cells as previously described (26). Briefly, splenic OT-I CD45.1⁺ and OT-I miR-139^−/− CD45.1⁺ CD8⁺ T cells were isolated by negative selection with magnetic beads (EasySep) from uninfected female mice 8–10 wk of age. Cells were cultured at 5 × 10⁵ CD8⁺ T cells either without cognate Ag stimulation, one-time stimulation with 10 ng/ml of OVA_(257-264) peptide for 48 h, or with daily peptide stimulation with 10 ng/ml of OVA_(257-264) peptide for 5 d. Cultures contained 5 ng/ml recombinant murine IL-7 and 5 ng/ml recombinant murine IL-15 throughout culture. For all conditions, after 48 h of culture, cells were washed twice (RPMI + 10% FBS medium containing 1% penicillin/streptomycin, 1% l-glutamine, 1% nonessential amino acids, 1% sodium pyruvate, and β-mercaptoethanol) and reseeded. On day 5, all conditions were harvested, and cells were counted. CD8⁺ T cells were stained as described below to measure inhibitory receptor and transcription factor expression, or they were stimulated with OVA_(257-264) peptide in the presence of GolgiPlug for 6 h followed by intracellular cytokine staining as described below to measure cytokine production.

For ex vivo and in vitro measurement of miR-139-3p and miR-139-5p expression, total RNA, including miRNA, was extracted using the miRNeasy Mini Kit (Qiagen, Germantown, MD) as per the manufacturer’s instructions. cDNA was synthesized from 100 ng of total RNA with the High Capacity cDNA Reverse Transcription Assay (Thermo Fisher Scientific). The expression of miR-139 was measured by quantitative real-time PCR (qRT-PCR) with the TaqMan mmu-miR-139-5p and mmu-miR-139-3p miRNA assays (Thermo Fisher Scientific). The expression of snoRNA-429 or RNU6-1 served as an endogenous control. All assays were run using a 7900 HT Real-Time PCR System. Expression was evaluated by the comparative cycling threshold method.

Flow cytometry staining was completed as previously described (5). In all stains, cells were pretreated with anti-CD16/32 (Fc Block, 2.4G2; BioLegend, San Diego, CA) for 15 min before continuing with surface staining. For surface stains, cells were stained for 20 min on ice. Cells were stained with the following fluorochrome-conjugated mAbs: anti-CD8a (clone 53-6.7), anti-CD45.1 (clone A20), anti-CD45.2 (clone 104), anti-Thy1.1 (clone HIS51) (all from eBioscience), anti-CD25 (clone PC61), anti-CD69 (clone H1.2F3), anti-CD44 (clone 1M7), anti-CD62L (clone MEL-14) (all from BD Biosciences), anti-KLRG1 (clone 2F1/KLRG1), anti–IL-7R/CD127 (clone A7R34), anti-PD-1 (clone 29F-IAI2) (all from BioLegend), anti-LAG3 (clone C9B7W), anti-CD160 (clone CNX46-3) (both from BD Biosciences), anti-TIGIT (clone GIGD7; eBioscience), and anti-TIM3 (clone RMT3-23; Invitrogen). Cells were also stained with cyanine 5.5–labeled annexin V (BD Biosciences) and, where indicated, allophycocyanin-labeled tetramers of H-2K^b MHC class I loaded with OVA_(257-264). After staining, cells were washed two times with HBSS containing 3% FBS and 0.02% sodium azide and fixed with 1% paraformaldehyde (PFA) solution. For annexin V staining, all buffers contained 2.5 mM CaCl₂. For staining of intracellular cytokines, cells were stimulated with the indicated peptides for 6 h at 37°C, 5% CO₂ in the presence of GolgiPlug (BD Biosciences), and mAb against CD107a (clone ID4B) or isotype control. Cells were surface stained as above, including fluorochrome-conjugated anti-CD107a Ab (clone ID4B) or the appropriate isotype control (both from BioLegend), then fixed overnight at 4°C with IC Fixation Buffer (eBioscience), washed using Perm/Wash buffer (eBioscience), and stained for intracellular cytokines for 45 min at 4°C. Fluorochrome-conjugated anti–IFN-γ Ab (clone XMG1.2), anti–TNF-α Ab (clone MP6-XT22), or the appropriate isotype controls (all from eBioscience) were used for intracellular stains. After staining, cells were washed twice with Perm/Wash buffer (eBioscience) and fixed with 1% PFA. For staining of transcription factors, cells were surface stained as above, then fixed for 1 h at 4°C with FoxP3 Fixation Buffer, washed using Perm/Wash buffer (eBioscience), and stained for transcription factors for 1 h at 4°C. The following Abs were used in combination with intracellular flow cytometry: anti–T-bet Ab (clone 4B10; BioLegend), anti-Eomes Ab (clone DAN11MAG; eBioscience), anti–T cell-specific factor-1 (TCF-1) Ab (clone C63D9; Cell Signaling Technology), anti-TOX Ab (clone TXRX10; eBioscience), anti-perforin Ab (clone S16009A; BioLegend), anti–granzyme B (clone QA16A02; BioLegend), or the appropriate isotype controls. After staining, cells were washed twice with Perm/Wash buffer (eBioscience) and fixed with 1% PFA. Anti–T-bet Ab staining was titrated using WT (T-bet⁺/⁺) and T-bet^+/− splenocytes to achieve a clear distinction between heterozygote and homozygote T-bet expression. All samples were collected with an LSRFortessa or LSR-X20 flow cytometer (BD Biosciences) and analyzed with FlowJo version 10 software (BD Biosciences, Ashland, OR).

For flow cytometry and qRT-PCR data analysis, the normality of the population distribution was assessed using the Shapiro–Wilk normality test in GraphPad Prism 8. Significant differences between normally distributed populations were assessed using a two-tailed, unpaired t test; significant differences between non-normally distributed populations were assessed using a two-tailed Mann–Whitney exact test. The tests performed are denoted in each figure legend, and subsequent p values are annotated in the associated figure.

Effector CTL are phenotypically and functionally different from naive CD8⁺ T cells (30, 31). We and others have demonstrated that specific miRNAs are upregulated or downregulated upon activation of CD8⁺ T cells (3, 32, 33), such as the increased expression of miR-155 upon in vitro activation or in vivo in antiviral CTLs (3). We hypothesized that miRNA expression within the naive and effector CD8⁺ T cell compartment would reflect the phenotypic and functional differences of naive versus in vivo generated effector CD8⁺ T cells. To test this, we performed miRNA microarray expression profiling of miRNA in naive and effector OT-I CD8⁺ T cells [CD8⁺ T cells expressing a transgenic TCR recognizing the OVA_(257-264) epitope]. Our use of transgenic OT-I cells for these analyses ensured that any differences in miRNA expression were not a result of differences in TCR affinity or avidity. To generate effector CTLs in vivo, we performed adoptive transfers of FACS-sorted naive (CD44⁻CD62L⁺) CD45.1⁺ OT-I CD8⁺ T cells into CD45.2⁺ WT recipient mice, which were then infected with WSN-OVA influenza virus [an OVA_(257-264)-expressing strain of A/WSN/33 influenza virus]; a portion of sorted naive OT-I CD8⁺ T cells were kept for naive cell miRNA profiling analysis. On day 9 after infection, during the peak of the effector CD8⁺ T cell response to influenza virus infection, donor CD45.1⁺ OT-I CD8⁺ T cells were sorted from the lungs of CD45.2⁺ WT recipient mice. miRNA profiling of these naive and effector OT-I CD8⁺ T cells was performed.

miRNA expression analysis revealed that naive and effector CTLs had significantly different miRNA profiles (Supplemental Fig. 1A). We observed that 315 mature miRNAs were expressed above background level (miRNA with an average naive or effector sample detection p value ≤0.05) in naive CD8⁺ T cells, and 309 miRNA were expressed above background level in effector CTLs. Of these, 53 miRNAs were expressed only in naive and 47 miRNAs only in effectors (Supplemental Fig. 1B). We found that 51 miRNAs in effector CTLs demonstrated ≥2-fold upregulation compared with naive expression levels (FDR, 0.05; p ≤ 0.05), and 69 miRNAs demonstrated ≥2-fold downregulation compared with naive expression levels (FDR, 0.05; p ≤ 0.05) (Supplemental Fig. 2A, 2B). Of these, miR-126a demonstrated the highest upregulation (103-fold), followed by miR-182 (46-fold) and miR-183 (23-fold); miR-669c, miR-181a, miR-192, and miR-466 family members were the most downregulated, and all had >9-fold lower expression levels compared with naive cells (Supplemental Fig. 1C, 1D). Strikingly, the transition from a naive cell to an effector CTL was accompanied by a change of only 11 miRNAs exhibiting >10-fold changes (up or down) in expression levels. miR-155 expression levels in effector CTLs were 3.3-fold greater than in naive CD8⁺ T cells, confirming our previous findings by qRT-PCR (3).

miR-139 was one miRNA identified by microarray analysis to be expressed at higher levels in naive CD8⁺ T cells and downregulated in effector CTLs. Our findings were consistent with a previous study that identified the in vivo downregulation of miR-139 in KLRG1⁺CD127⁻ effector cells compared with KLRG1⁻CD127⁺ memory precursor cells from lymphocytic choriomeningitis virus (LCMV) Armstrong–infected mice 8 d after infection (23). Trifari et al. had also shown that overexpression of the full-length pri-miR-139 gene in P14 CD8⁺ T cells, transgenic CD8⁺ T cells that recognize the LCMV glycoprotein residues 33–41 (GP33), resulted in the decreased killing ability of activated CTLs in response to GP33-expressing LM infection. Taken together with our miRNA profiling data, we hypothesized that miR-139 expression was essential in regulating mRNAs that are critical in maintaining a naive or memory CD8⁺ T cell state. We therefore chose to assess the role of miR-139 in CTL differentiation in response to influenza virus infection.

miRNA profiling of influenza-specific effector CTLs identified that miR-139-5p was significantly downregulated (−6.89-fold; p = 0.001) (Supplemental Fig. 1D). To assess how quickly miR-139 was downregulated upon CD8⁺ T cell activation and to exclude that this downregulation was a result of the OT-I transgenic TCR or was cytokine driven, qRT-PCR analysis of both miR-139-3p (Fig. 1A) and miR-139-5p (Fig. 1B) was performed on splenic CD8⁺ T cells purified from C57BL/6 mice that were activated using solid-phase bound α-CD3/α-CD28 Abs (activated, gray bars) or cultured without stimulation (nonactivated, white bars) and cultured with or without IFN-γ, TNF-α, IFN-β1, or IL-1β for 24 or 72 h. For both miR-139-3p and -5p, only the addition of TNF-α was sufficient to drive the downregulation of miR-139 in nonactivated CD8⁺ T cells at the 24-h time point. Although TCR activation was sufficient to drive a high level of downregulation of miR-139-3p by 24 h, miR-139-5p was slower to decrease. Regardless of the addition of cytokines, however, both miR-139-3p and miR-139-5p expression levels were undetectable or barely detectable by qRT-PCR after 72 h of activation (Fig. 1A, 1B). Because both miR-139-3p and miR-139-5p were significantly downregulated upon TCR stimulation with α-CD3/α-CD28 Abs, we questioned whether the level of TCR stimulation impacted miR-139 expression as well. We therefore stimulated WT OT-I CD8⁺ T cells with the OVA peptide SIINFEKL or the altered peptide ligand (APL) variants SIIRFEKL (R4) or EIINFEKL (E1). E1 is a TCR antagonist or very weak APL agonist that weakly impacts TCR downregulation (∼15% of SIINFEKL), whereas R4 is a pure APL antagonist with nominal TCR downregulation capability (∼6% of SIINFEKL) (29). Within 24 h of stimulation, SIINFEKL stimulation resulted in downregulation of miR-139-3p to almost undetectable levels (Fig. 1C), whereas miR-139-5p was reduced by ∼50% (Fig. 1D) compared with unstimulated control cells. Although E1 stimulation resulted in a modest decrease in miR-139-3p expression (∼25% reduction; (Fig. 1C), it had no effect on miR-139-5p expression levels (Fig. 1D). R4 stimulation had no impact on miR-139-3p expression (Fig. 1C), but in fact it increased expression of miR-139-5p (Fig. 1D). Taken together, these data confirmed that miR-139 is downregulated upon TCR stimulation of CD8⁺ T cells.

As we hypothesized that miR-139 expression was critical for maintaining the naive state of CD8⁺ T cells, we questioned if miR-139 played a functional role in the development of CD8⁺ T cells. Using flow cytometry, we assessed if miR-139 deficiency impacted T cell phenotype in the absence of infection. The spleens of 8-wk-old uninfected miR-139^+/+, miR-139^+/−, and miR-139^−/− female mice were evaluated to assess the frequency and phenotype of CD4⁺ and CD8⁺ T cells. Total cell counts from the spleens were comparable across all mouse strains. Flow cytometry identified that there were no substantial differences in the frequencies of CD4⁺ or CD8⁺ T cells in the periphery when miR-139 was absent (Fig. 1E). Furthermore, miR-139 deficiency did not affect the frequency of naïve CD4⁺ or CD8⁺ T cells (Fig. 1E). Thus, we conclude that miR-139 expression is not required for the normal development of CD4⁺ and CD8⁺ T cells, and its deficiency did not result in the skewing of the steady-state naive and memory phenotypes of CD4⁺ or CD8⁺ T cells.

To directly test if miR-139 impacts the generation of effector CTL responses to an active infection, we assessed CTL responses to acute influenza virus infection using two approaches. First, miR-139^−/− mice were infected with PR8 influenza virus. Next, to limit the effects of miR-139^−/− to a CD8⁺ T cell–intrinsic mechanism, miR-139^−/− OT-I CD8⁺ T cells were adoptively transferred into WT recipient host mice, followed by WSN-OVA influenza virus infection. Regardless of the approach, absence of miR-139 globally or exclusively in CD8⁺ T cells had no impact on the CTL response to influenza virus infection. When infected with PR8 influenza virus, miR139^−/− mice had the same average percentage body weight as WT control mice (Fig. 2A). At 10 d after infection with influenza virus, we used NP_(366-374) tetramer to identify the virus-specific CD8⁺ T cells within the lungs, spleens, and mediastinal lymph nodes (MLNs) of PR8-infected mice (Fig. 2B). The frequency of NP_(366-374)–specific CD8⁺ T cells in the lungs did not differ between WT and miR-139^−/− mice (Fig. 2C). In addition, no difference was found in either spleens or MLNs) as well (Fig. 2C). We also did not observe any differences in the frequency of proliferating virus-specific CD8⁺ T cells in the lungs or spleens as assessed by Ki-67 expression (Fig. 2D). The failure of miR-139 to affect the development of influenza-specific CTL responses was also confirmed in the adoptive transfer experiments, where donor OT-I cells and miR-139^−/− OT-I cells both generated similar CTL responses in terms of frequencies and numbers of cells in lungs, spleens, and MLNs (Fig. 2E–G). miR-139 deficiency did not affect CTL expansion, and there was also no evidence of impaired or enhanced trafficking, because the numbers of virus-specific CTLs were similar in the lungs, spleens, and MLNs of infected animals (Fig. 2G). Similar to PR8 infection, there were no differences in donor miR-139^−/− OT-I proliferation in either the lungs or spleens of WSN-OVA influenza virus–infected mice (Fig. 2H).

We next assessed if lack of miR-139 altered the activation state of effector CD8⁺ T cells in the lungs following infection with either PR8 or WSN-OVA influenza virus. Both WT and miR-139–deficient CTLs expressed similar levels of CD25 and CD69, markers of T cell activation, ex vivo from the lungs (Fig. 3A, 3B). To assess the functional impact of miR-139 expression on CTL responses, we performed intracellular cytokine staining after peptide restimulation to determine IFN-γ and TNF-α production. Again, no differences were observed between WT and miR-139^−/− CTLs in terms of their ability to produce effector cytokines after the adoptive transfer of miR-139^−/− OT-I followed by WSN-OVA infection or in miR-139^−/− mice following PR8 infection (Fig. 3C–E). Because previous studies evaluating the role of miR-139 and T cells have reported that overexpression of miR-139 resulted in decreased perforin expression (23), we evaluated both granzyme B and perforin expression in OT-I CTLs ex vivo following WSN-OVA infection. When miR-139 expression was absent, we did not observe any changes in the frequency of granzyme B– or perforin-expressing CTLs (Fig. 3F, 3G), nor were there any significant differences in the relative abundance of these cytotoxic molecules within WT or miR-139^−/− OT-I CD8⁺ T cells (Fig. 3H, 3I). Therefore, in response to influenza virus infection, absence of miR-139 in CD8⁺ T cells had no impact on the generation of the CTL effector response or cytotoxicity.

Although miR-139 deficiency did not appear to limit the development of CD8⁺ T cells or the responsiveness of CD8⁺ T cells to influenza virus infection, we questioned whether there was a competitive advantage for cells to express miR-139 in the naive state. Naive WT (CD45.1⁺/CD90.1⁺) or miR-139^−/− (CD45.1⁺/CD90.2⁺) OT-I CD8⁺ T cells were enriched and mixed 1:1 (Fig. 3J, left) before being coinjected into WT C57BL/6 (CD45.2⁺/CD90.2⁺) host mice followed by WSN-OVA influenza virus infection. At day 10 after infection, evaluation of the donor OT-I population in the lungs of infected animals showed that WT OT-I cells significantly outcompeted miR-139^−/− OT-I CTL (Fig. 3J, right) with a >2-fold increase in relative frequency (Fig. 3K). These data indicate that although miR-139^−/− CD8⁺ T cells are capable of responding to influenza virus infection, the WT CD8⁺ T cell response has a competitive advantage when in the same host environment. Despite reduced expansion of these miR-139^−/− CD8⁺ T cells, they maintained equal levels of granzyme B and perforin expression compared with WT cells (Supplemental Fig. 4A).

We next examined if miR-139 deficiency impacts the development of memory CTLs in response to influenza virus infection. Similar to our previous findings in uninfected mice, we observed no differences in the memory CTL frequency or skewing of memory CD8⁺ T cells toward the central memory (CD44⁺CD62L⁺) or effector memory (CD44⁺CD62L⁻) phenotype as assessed by CD44 and CD62L expression (Supplemental Fig. 3A, 3D). Taken together, our data demonstrate that the expression of miR-139 is dispensable for the development of both effector and memory CD8⁺ T cell responses to acute infection with influenza virus.

The transcription factor Eomes is important to the development of memory CTLs (34) and T cell exhaustion (24). Previous studies on the role of miR-139 and T cells have reported that overexpression of miR-139 resulted in decreased Eomes expression (23). We therefore reasoned that the absence of miR-139 should promote Eomes expression. We evaluated the frequency and level (median fluorescence intensity [MFI]) of Eomes during the effector phase of the CTL response to influenza virus infection; we simultaneously assessed T-bet expression levels because the balance of expression between the two transcription factors is important in the development of CD8⁺ T cell responses to acute viral infection (35). In both the spleen and the lung, we did not observe any differences in the frequency or expression levels of Eomes and T-bet in influenza virus–specific CD8⁺ T cells upon either PR8 infection of miR-139^−/− mice or WSN-OVA infection following miR-139^−/− OT-I adoptive transfer (Fig. 4A–G). This may explain the lack of any impact on effector or memory CTL development as we originally hypothesized.

In previous studies assessing the role of miR-155 in the regulation of CD8⁺ T cells, we have observed that deletion versus overexpression of an miRNA can result in alteration of miRNA target selection (3, 5). To confirm if this was the case with miR-139–targeted regulation of CTLs, we used retroviral transduction to overexpress miR-139 in OT-I CD8⁺ T cells (miR-139 OE). Control transduced or miR-139 OE OT-I CD8⁺ T cells were FACS sorted based on the expression of the selectable marker CD90.1/Thy1.1, and 10⁴ OT-I cells were adoptively transferred into WT recipient mice (CD90.2/Thy1.2) followed by infection with WSN-OVA as in previous experiments. At day 9 after infection, we evaluated the donor OT-I cells in the lungs of influenza virus–infected mice. Similar to our results from miR-139^−/− OT-I donor transfer experiments, we did not observe any change in the frequency of donor OT-I cells between the control and miR-139 OE groups (Supplemental Fig. 4B), CD69 expression (Supplemental Fig. 4C), or cytokine production (Supplemental Fig. 4D). Unlike what was previously noted in T cell responses to LM (23), we observed no change in the expression levels of either Eomes or T-bet in miR-139^−/− OE CD8⁺ T cells responding to influenza virus infection (Fig. 4H, 4I).

In the context of LM bacterial infection, miR-139–overexpressing CD8⁺ T cells have been shown to have enhanced CTL responses and downregulated Eomes expression (23). Therefore, we hypothesized that miR-139 deficiency may reduce the ability of CTLs to respond to LM infection compared with influenza virus infection. WT or miR-139^−/− OT-I CD8⁺ T cells were adoptively transferred into C57BL/6 host mice, followed by infection with LM-OVA. At day 10 after infection, CTL responses were evaluated in the spleens of infected animals. We observed that when miR-139 was deficient, donor CTL responses were significantly reduced (9.48% versus 3.8% of CD8⁺ T cells; p = 0.0078) (Fig. 5A, 5B), despite increased expression of the proliferation marker Ki-67 by miR-139^−/− CTL (54% versus 33%; p = 0.0032). When evaluating the differentiation of donor CTLs into short-lived effector cells in response to LM-OVA, we observed no differences in the frequency of cells that had upregulated KLRG-1 (Fig. 5D–F). However, the function of these cells was significantly reduced upon peptide stimulation (Fig. 5G), with reduced polyclonal functionality by miR-139^−/− CTLs (47.7% versus 38%; p = 0.0235) and overall reduced IFN-γ–producing cells (81% versus 70.8%; p ≤ 0.0001). Because miR-139 overexpression was reported to reduce Eomes expression levels in response to LM (23), we compared T-bet and Eomes expression in WT versus miR-139–deficient CTLs. Similar to responses during influenza virus infection, we did not observe any changes in the frequency of T-bet– or Eomes-expressing cells (Fig. 5J) or in the expression levels of these transcription factors (Fig. 5K, 5L). Taken together, these data indicate that miR-139 deficiency results in reduced expansion of LM-OVA–specific CTLs as well as reduced functionality of these cells but does not appear to alter their differentiation.

During acute viral infection, CD8⁺ T cells clear virally infected cells and can protect the host from reinfection through the generation of functional memory CD8⁺ T cells. However, in cases of chronic infection such as murine infection with LCMV clone 13 and HIV infection in humans, CD8⁺ T cells enter a different differentiation pathway and become exhausted (36). T cell exhaustion is characterized by the increased coexpression of inhibitor receptors (including PD-1, LAG3, TIM-3, CD160, and TIGIT), decreased proliferative capacity, and loss of cytokine production (37). T cell exhaustion is also accompanied by significant changes in the transcriptome, including changes in the expression levels of T-bet and Eomes (38). High levels of Eomes expression have previously been implicated in the development or promotion of T cell exhaustion in response to chronic viral infection and cancer (24, 25). Despite the lack of miR-139–mediated regulation of Eomes in influenza virus infection, we reasoned that lack of miR-139 may enhance Eomes expression and therefore promote T cell exhaustion under conditions of repeated TCR stimulation. Using a novel method to develop T cell exhaustion in vitro (26), we assessed the development of T cell exhaustion in miR-139^−/− CD8⁺ T cells. As expected, repeated stimulation of WT CD8⁺ T cells with peptide resulted in the loss of IL-2 and TNF-α production (Fig. 6A); in agreement with our in vivo studies, there were no differences in cytokine production by “effector” miR-139^−/− CD8⁺ T cells compared with WT cells after a single round of peptide stimulation (Fig. 6A). Importantly, we observed a similar loss of polyfunctionality by miR-139^−/− CD8⁺ T cells compared with WT cells upon repeated peptide stimulation (Fig. 6A, 6B). Repeated stimulation of both WT and miR-139^−/− CD8⁺ T cells resulted in equivalently high frequencies and expression levels of PD-1, LAG3, TIM3, CD160, and TIGIT (Fig. 6C).

To assess if miR-139 deficiency resulted in altered Eomes expression under exhaustive conditions, we evaluated both Eomes and T-bet expression in nonstimulated, single-stimulated effectors, and repeat stimulation exhausted CTLs. In all conditions, we did not observe any changes in the frequency or quantity of Eomes or T-bet expression (Fig. 7A, 7B). Recent studies have highlighted roles for the transcription factors TCF-1 and TOX in the development of stemlike progenitor CTLs or terminally exhausted CTLs, respectively (25, 40). To evaluate if the absence of miR-139 increased or decreased the prevalence of TOX⁺TCF-1⁻ terminally exhausted and TOX⁻TCF1⁺ progenitor exhausted CTLs, we evaluated the frequency of these after repeat stimulation. Repeated stimulation of WT and miR-139^−/− CD8⁺ T cells resulted in an equivalent downregulation of TCF-1 expression (Fig. 7C, 7D) and upregulation of TOX (Fig. 7C, 7E), whereas single-peptide stimulation failed to promote TOX expression in either WT or miR-139^−/− CD8⁺ T cells, in agreement with their characterization as effector cells. Taken together, these data in combination with the in vivo influenza virus studies demonstrate that miR-139 expression is dispensable for the generation of effector CD8⁺ T cell responses.

Previously, microarray profiling of in vitro activated CD8⁺ T cells was performed by Wu et al. (41). However, because Ag presentation and the local inflammatory environment during an active infection can have a major impact on CTL differentiation (42), we directly assessed the miRNA expression in in vivo generated CTLs. We analyzed the miRNA expression profiles of naive CD8⁺ T cells and compared them with in vivo generated effector antiviral CTLs taken directly from the site of infection, the lungs, to establish the relative dynamics of the CTL miRNome. Of the 362 mature miRNAs detected by microarray analysis, large-scale changes in effector CTLs are rather modest, with only 11 miRNAs of the miRNome changing >10-fold, whereas 120 miRNAs change by >2-fold. This suggests that the overall miRNome is relatively stable in CD8⁺ T cells, with large-scale changes potentially associated with cellular identity and lineage. The relatively modest changes in the majority of miRNAs meanwhile appear sufficient to impart effector CTL differentiation, function, and expansion.

We found that the most upregulated miRNAs in effector CTLs were miR-126a, miR-182, miR-183, miR-185, and miR-31, whereas miRNA miR-669c, miR-181, miR-378b, and miR-192 were the most downregulated. In vitro activation of mouse CD4⁺ T cells showed the upregulation of miR-155 concurrent with the downregulation of miR-150, miR-146, miR-142, and miR-16 (32), expression patterns that are distinct from our influenza virus–specific CD8⁺ effector cells. Previous studies have reported that ex vivo subsets of human CD8⁺ T cells demonstrated unique expression patterns of miRNAs, noting in particular the upregulation of miR-21, miR-155, and miR-146a and the downregulation of miR-19b, miR-20a, miR-92, and miR-26a in differentiated effector cells (43). Our profiling of mouse antiviral effector CTLs shared the upregulation of miR-21, miR-155, and miR-146a and downregulation of miR-92 and miR-26a. Surprisingly, we found that the second most downregulated miRNA in effector CTLs was miR-181a (down by >10-fold), which is known to regulate TCR signaling by targeting phosphatases (44). At first glance, this seems counterintuitive because decreased miR-181a would result in reduced TCR activation, which would likely impact CTL killing ability. However, miR-181a is downregulated after TCR signaling in thymocytes (44), and its downregulation in in vivo effector CTLs could result from TCR engagement. In effector CTLs, this downregulation may signify an increasing threshold of TCR activation at the peak of the CTL response and may relate to the forthcoming CTL contraction phase. Kinetic studies of miR-181a expression at early stages of effector CTL generation could address this and reveal if this downregulation in CTLs occurs at the later stage of the response or whether miR-181a may have some unanticipated targets and effects in the CTL response. Interestingly, miR-150, which has been shown to be required for optimal effector CTL responses to LM (45), was downregulated in effector CTLs in our profiling, which has also been reported with in vitro activation of CTLs (23). These studies highlight the contextual expression of miRNA; understanding the dynamic nature of the miRNA expression landscape during CTL differentiation and immune responses in acute or chronic infections and cancer and identifying the unique miRNAs associated with the different phases of CTL immunity may prove important for the enhancement of cellular immunotherapy approaches such as chimeric Ag receptor T cells or tumor-infiltrating lymphocyte cancer therapies.

Our study calls into question the functional contribution of individual miRNAs over the course of the CTL response to infection despite evidence of the modulation of a particular miRNA’s expression. Microarray analysis identified miR-139 as one miRNA highly downregulated in ex vivo effector antiviral CTLs, downregulation that was confirmed by qRT-PCR of in vitro activated or cytokine-treated CD8⁺ T cells. In our study, we have directly tested the functional contribution of miR-139 expression in the development of T cells and maintenance of their naive state in the periphery, as well as in the development of effector CTLs and memory CD8⁺ T cells following influenza virus infection. In two model systems (global deficiency and CD8⁺ T cell intrinsic deficiency), the absence of miR-139 did not affect the frequency, numbers, or phenotype of virus-specific CTLs at either acute primary infection or a memory time point of 60 d after infection. We also did not observe any differences in weight loss in these animals, indicating that miR-139 does not significantly contribute to influenza-induced morbidity. However, in a head-to-head competition setting, deficiency of miR-139 led to reduced expansion compared with WT CD8⁺ T cells.

Unlike in response to influenza virus infection, single transfers of miR-139^−/− OT-I CD8⁺ T cells revealed that deficiency of miR-139 resulted in a reduced frequency and number of OT-I CD8⁺ T cells. Furthermore, the functional ability of these cells was reduced. Interestingly, we did observe increased frequency of Ki-67⁺ cells in the miR-139^−/− CD8⁺ T cells despite an overall reduced OT-I response to LM infection at day 10 (Fig. 5) and at an earlier time point of day 6 (data not shown). This could suggest that miR-139 deficiency promotes increased sensitivity to activation-induced cell death in response to Listeria infection. It also could be reflective of prolonged bacterial burden in mice that received miR-139^−/− OT-I CD8⁺ T cells as opposed to WT OT-I CD8⁺ T cells promoting cell cycling, but because bacterial burden could not be successfully assessed, additional studies would be required to determine if one of these two hypotheses was responsible for the increased frequency of Ki-67⁺ cells. In a previous study by Trifari et al., overexpression of miR-139 in CD8⁺ T cells (mediated by retroviral transduction) resulted in decreased expression of both Eomes and perforin in CD8⁺ T cells responding to Listeria infection (23). In response to influenza virus infection, we did not observe any changes in either Eomes or perforin expression when miR-139 was absent or overexpressed. We also did not observe any changes in Eomes expression when miR-139–deficient CTLs were responding to Listeria infection, despite observed reduced T cell expansion and functionality. This discrepancy in potential miRNA targets reaffirms that miRNA-mediated regulation of mRNA in T cells is context dependent (9) and is in line with our previous studies indicating changes in target selection in systems of miRNA overexpression versus deficiency (3, 5).

Our previous findings have shown that knocking out a single miRNA that is highly upregulated upon activation of CD8⁺ T cells, miR-155, has profound effects on the ability of CD8⁺ T cells to effectively control influenza virus and LM infection (3). Others have also demonstrated the effect that miR-155 deficiency has in response to LM infections (46), LCMV (46, 47), MHV-68 (48), and melanoma tumors (47), showing the high impact the deletion of a single miRNA can have on CD8⁺ T cell responses. For this study, we used the readily available technique of microarray profiling to evaluate changes in miRNA expression between the naive and influenza virus–specific CD8⁺ T cells. A major limitation of this method, however, is that we can only directly compare expression levels of the same miRNA between two conditions, and we are unable to draw any conclusions based on the relative abundance of different miRNAs. In future studies, the use of miRNA-deficient mice and RNA-sequencing quantification will provide key insight into whether a threshold of miRNA expression exists for which expression levels below this threshold hold negligible biological consequences for miRNA deficiency. Whether universally applicable thresholds can be determined or whether target abundance dictates a wide range of such thresholds remains to be seen. There is also the possibility of miRNA redundancy wherein other miRNAs may take over the regulation of miR-139 targets upon miR-139 deletion. TargetScan (version 7.2, March 2018) predicts 2104 potential targets of miR-139-3p and 312 potential targets of miR-139-5p (all with a cumulative weighted context++ score [49] <0); yet, only 45 predicted targets have a score <0.5. TargetScan analysis of the top-ranked predicted target of miR-139-3p, CDKN1A, predicts that this gene also contains conserved sites for miR-208-3p, miR-499-5p, miR-291-3p, miR-294-3p, and several others. Future studies could evaluate if other miRNAs are upregulated in response to miR-139 deficiency, and Argonaute-based cross-linked immunoprecipitation followed by next-generation sequencing (8) could be used to confirm miRNA–target interaction.

Taken together, our study provides greater insight into the dynamics of miRNA expression in CD8⁺ T cells by providing a snapshot of the expression profiles of naive and in vivo generated influenza virus–specific CD8⁺ T cells at the peak of the adaptive immune response to infection. Importantly, our findings identify an miRNA, miR-139, that, although significantly downregulated in both in vivo and in vitro activated CTLs, is dispensable for the development of functional CTL responses in response to acute influenza virus infection. We have also demonstrated through the use of an in vitro exhaustion model system that eliminating miR-139 neither enhances nor limits the development of T cell exhaustion. Our study therefore highlights the discrepancies that can be observed in miRNA targets between overexpression and knockout studies, and it reinforces the importance of considering the impact of contextual miRNA-mediated regulation.

We acknowledge the help and support of the Drexel University College of Medicine Animal Facility, the Sanford Burnham Prebys Animal Facility, the Sanford Burnham Prebys Flow Cytometry Core, the Thomas Jefferson University Cancer Genomics Facility, the Erasmus University Medical Center EDC staff, and the Erasmus University Medical Center Department of Immunology Flow Cytometry core in the completion of this work.

The authors have no financial conflicts of interest.