Abstract
A plethora of work implicates important effects of the vitamin A derivative retinoic acid (RA) in myeloid differentiation, whereas fewer studies explore the role of RA in lymphoid cells. Most work on lymphoid cells has focused on the influence of RA on CD4 T cells. Little information about the role of RA in CD8 T cell differentiation is available, and even less on cell-intrinsic effects in the CD8 T cell. This study explores the role of RA in effector and memory differentiation in a cell-intrinsic manner in the context of vaccinia virus infection. We observed the loss of the short-lived effector cell phenotype (reduced KLRG1+, T-bethi, granzyme Bhi), accompanied by an enhanced memory precursor phenotype at the effector (increased CD127hi, IL-2+) and contraction phases (increased CD127hi, IL-2+, eomesoderminhi) of the CD8 response in the absence of RA signaling. The lack of RA also increased the proportion of central memory CD8s. Collectively, these results introduce a new role for RA in CD8 T cell activation and differentiation. This new role may have significant implications for optimal vaccine design in which vitamin A supplementation is used to augment effector responses, but it may be to the detriment of the long-term central memory response.
Introduction
The morphogenic role of all-trans retinoic acid (RA), a vitamin A derivative, in development and differentiation was confirmed by White et al. (1) in 2007, using a zebra fish model to confirm RA patterning in the hindbrain. However, immunologists have studied it in various contexts of immune cell differentiation as early as the 1980s. Among myeloid cells, RA has been shown to allow for differentiation into mature macrophages or APCs. (2). This RA-mediated differentiation of dendritic cells (DCs) has been shown to skew them toward IL-12–producing DCs (3). RA also regulates isotype switching and plasma cell formation by B cells (4–6).
In the adaptive immune compartment, RA has been shown to promote regulatory CD4 T and CD8 T cell differentiation and stabilization (7–9). Further, RA has been shown to enhance inflammatory effector responses by CD4 helper T cells (10, 11). In CD8 T cells, an early study showed that increased expression of RA receptor γ increased the number of CD8 T cells (12). To our knowledge, no previous studies have looked at the cell-intrinsic role of RA signaling in CD8 T cell effector and memory formation in the context of virus infection.
Paramount to eliciting optimal protective immunity to infections is the generation of high-quality memory cells. Superior memory generation is a key component of vaccine design, as these cells can elicit optimal protection. In response to an acute viral insult, CD8 T cell responses go through three phases: the primary acute expansion phase to resolve the infection; the contraction phase to eliminate potentially harmful cytotoxic effectors; and a memory phase, in which self-renewing Ag-specific cells are maintained at low frequencies for extended periods (13). Upon activation in mice and humans post infection, CD8 T cells form highly differentiated short-lived effector cell (SLEC) and memory-precursor effector cell (MPEC) populations (14–16). The SLEC population is driven by inflammatory cytokines like IL-12 or type I IFNs and characterized by high T-bet expression, compared with MPECs, which have high eomesodermin expression, recently shown to be driven by Forkhead Box Protein 01 (FOX01) expression (17–23). SLECs are identified by surface expression of high killer cell lectin-like receptor subfamily G member 1 (KLRG1) and low IL-7Rα (CD127), whereas MPECs are identified by the expression of low KLRG1 and high IL-7Rα (17). The terminally differentiated SLEC population, with its high cytotoxic potential, is the desired population to resolve a viral infection, whereas the MPEC population is thought to differentiate into the long-lived memory population (24). Among the memory population, central memory cells (Tcm) are the most long lived and are characterized by robust recall potential, capacity for homeostatic proliferation, and homing to lymphoid organs. Effector memory cells are characterized by homing to peripheral sites and lower homeostatic turnover, while being the first to respond after re-exposure to infection (16, 25–27).
Acknowledging the role of RA in differentiation, as seen by studies in development and in other immune cell types, we hypothesized that RA would promote the differentiation of CD8 T cells to their terminally differentiated phenotype, SLECs. To test this hypothesis, we used a mouse model expressing a dominant negative RA receptor α (RARαDN) in the T cell compartment, and mixed bone marrow (BM) chimeric mice to measure CD8 T cell–intrinsic effects. To determine the effect of the absence of RA signaling in CD8 T cell differentiation, mice were infected with vaccinia virus, which induces a strong memory CD8 T cell response (28).
Vitamin A is used in conjunction with various vaccines (29, 30), so it is imperative to understand how its biologically active metabolite, retinoic acid, stimulates the T cell response. These studies will elucidate the effect of RA in shaping the effector and memory response, helping to tailor the outcome of vaccination, in addition to providing essential information regarding optimal CD8 T cell activation and differentiation.
Materials and Methods
Mice
C57BL/6 and congenic B6-Ly5.2-Cr mice (Ly5.1/CD45.1+) were purchased from the National Cancer Institute (Bethesda, MD). C57BL/6 R26RAR403 (from this point referred to as RARαDN mice) were used by permission from Dr. Shanthini Sockanathan (Department of Neuroscience, The Johns Hopkins University School of Medicine, Baltimore, MD) (28). Wild-type (WT) and RARαDN mice used in all experiments are of the C57BL/6 strain. Mouse experiments were performed under the guidelines and approval of Dartmouth College Institutional Animal Care and Use Committee.
ALDEFLUOR assay to measure RA expression
We measured the levels of an enzyme necessary for all-trans RA production–aldehyde dehydrogenase (ALDH). ALDH is a cytosolic isoenzyme that contributes to the oxidation of retinol to retinoic acid. The ALDEFLUOR assay measures the activity of ALDH, which is a surrogate measure for RA in the cytosol (STEMCELL Technologies, Vancouver, BC, Canada). Living cells through passive diffusion take up uncharged ALDH substrate bound to a green fluorophore. The substrate is converted by intracellular ALDH into a negatively charged reaction product, which is retained inside cells, causing the cells expressing ALDH to become brightly fluorescent. The brightly fluorescent ALDH-expressing cells are detected by flow cytometer in the green channel (520–540 nm). As a control, half of each sample has an inhibitor of ALDH added to determine the background.
Chemical inhibition of RA signaling
A pan–retinoic acid antagonist (NRX194310) was obtained from Randolph J. Noelle (Department of Immunology, Dartmouth Medical School, Hanover, NH) (11). Each mouse was injected i.p. with 30 μg every other day, starting at day 1, until euthanization at day 10 post infection.
Vaccinia and Listeria infection
Vaccinia virus, Western Reserve strain (VV-WR), was obtained from Dr. William Green (Dartmouth Medical School, Lebanon, NH). Primary infection (1000 PFU) was administered via the intranasal (i.n.) route. Listeria monocytogenes expressing the VV-WR CD8 dominant epitope, VV-WR epitope B8R20–27 (TSYKFESV)/Kb (hereafter referred to as B8R), was kindly provided by Dr. Ross Kedl (University of Colorado, Denver CO) and administered (2 × 106 PFU) via the i.v. route.
Generation of mixed BM chimeric mice
BM from donor B6-Ly5.2-Cr mice (1 × 106 cells) and C57BL/6 RARαDN mice (1 × 106
+:CD45.2− were used in the study.Tissue preparation
Blood was processed and the RBCs lysed. The spleen and lymph nodes were mashed, and the RBCs were lysed. The lungs were digested with collagenase (2.33 mg/ml; Sigma-Aldrich, St. Louis, MO) and DNAse (0.2 mg/ml; Roche Diagnostics, Indianapolis, IN) for 30 min; then they were mashed and the RBCs lysed.
Flow cytometry
Intracellular cytokine staining
BrdU incorporation and staining
Mice were provided with 0.8 mg/ml BrdU (Sigma-Aldrich) in the drinking water 10 d prior to staining for BrdU incorporation. Splenocytes were stained with MHC/peptide tetramer for B8R and surface stained for CD8 and CD44, as described above, followed by fixation and permeabilization for BrdU staining. The BrdU staining kit from BD Biosciences was used to stain with anti-BrdU Ab, and the results were measured by flow cytometry.
Secondary responses
At >60 d post primary infection, the chimeric mice were challenged with 1 × 106 CFU of replication-defective L. monocytogenes expressing the VV-WR CD8 dominant epitope—B8R—via the i.v. route. At 6 d post challenge, the splenocytes were analyzed to measure cell expansion. Ag-specific CD8 T cell proportions and numbers were determined along with their phenotype, using the surface and intracellular markers described previously.
Statistical analysis
The distribution was analyzed for normality, and then p values were determined using a Student t test or a Mann–Whitney U test. Paired t tests were used in the bone marrow chimeric model to account for the same internal environment of the WT and RARαDN CD8 T cell populations. A one-way ANOVA followed by a Dunn posttest was carried out when more than two variables were compared. A p value <0.05 was considered significant.
Results
Measuring all-trans retinoic acid production following infection
We infected WT C57BL/6 (WT B6) mice i.n. with 1000 PFU of the VV-WR strain and measured the levels of an enzyme necessary for RA production. ALDH is a cytosolic isoenzyme that contributes to the oxidation of retinol to RA. The ALDEFLUOR assay measures the activity of ALDH, which is a surrogate measure for RA in the cytosol. We used this assay to measure the level of RA in the spleen, lung, and mediastinal (lung-draining) lymph nodes (mLNs), and in T cells, B cells, DCs, and macrophages. We detected increased RA expression upon infection with vaccinia virus at day 4 post infection in the lung and mLNs (Supplemental Fig. 1A). Increased expression of RA was also observed in B cells (CD19+) post infection in the mLNs and lungs (Supplemental Fig. 1B). CD11b+ cells increased RA production post infection in the spleen and lungs (Supplemental Fig. 1C). CD11c+ cells showed expression of RA, but only marginally increased production post infection (Supplemental Fig. 1D).
Reduced Ag-specific response in the absence of RA signaling in RARαDN transgenic mice
To determine the effect of RA signaling on the Ag-specific CD8 response, we infected WT B6 and RARαDN-expressing mice crossed with CD4cre mice, DNRARα*CD4cre (hereafter referred to as RARαDN), i.n. with 1000 PFU VV-WR, and measured the Ag-specific response at day 9 post infection, by staining with MHC/peptide tetramer folded with B8R. We saw a decrease in the B8R-specific CD8 populations both proportionally and by total cell number per spleen (Fig. 1A).
Significantly reduced Ag-specific CD8 response, which is skewed toward an MPEC phenotype, in the absence of RA signaling. At 9 d post i.n. VV-WR (1000 PFU) infection, splenocytes from WT and RARαDN mice were stained with MHC/peptide tetramer for the VV-WR B8R epitope and analyzed by flow cytometry. (A) Top panel, Representative plot showing B8R gate in the CD8 T cell population. Bottom panel, Percentages and cell count per spleen. B8R-tetramer+CD8+ splenocytes were surface stained for (B) CD127 and (C) IL-21R. (D) Spleen cells were incubated for 5 h with or without B8R peptide in the presence of brefeldin A. Intracellular staining was performed to determine IFN-γ and IL-2 production by the CD8 T cells. The percent of IL-2+ reported is gated on the IFN-γ+, and the mean fluorescence intensity (MFI) was calculated for IFN-γ+IL-2+ cells. Filled circles are WT and open circles are RARαDN. n = 6 per group tested. Experiment was repeated three times, showing similar data. For t tests performed, ***p < 0.0001.
To determine the effect of the reduced Ag-specific CD8 response on the viral titer, we measured the viral load in the lung 9 d post infection by plaque assay. No significant difference in viral load was observed in the lungs of WT B6 or RARαDN mice (Supplemental Fig. 2A). Because RARαDN was expressed in CD8 and CD4 T cells, we wanted to determine the effect of RARαDN expression on the CD4 response. Using an IFN-γ ELISPOT to measure responses to two CD4 epitopes (I1L and L4R) of VV-WR, we saw no significant difference in the CD4 response (Supplemental Fig. 2B). We also analyzed the Ab response and noted similar levels of neutralizing Ab (data not shown).
Increased MPEC skewing with selective defects in effector phenotype in RARαDN mice during acute infection
At the acute stage of the VV-WR infection, we determined the phenotype of the Ag-specific CD8 response by staining for surface and intracellular markers in Ag-specific CD8 T cells, identified by B8R tetramer or by IFN-γ production following cognate peptide stimulation. We saw a statistically significant increase in surface expression of the IL-7Rα (CD127) and IL-21R in the B8R-specific CD8 T cells in the RARαDN (Fig. 1B, 1C). This phenotype is indicative of an MPEC population that is high in CD127 and responsive to IL-21. To further confirm the MPEC phenotype, we measured IL-2 production in the IFN-γ–positive CD8 T cell population. We observed an increase in both the proportion of B8R-specific CD8 T cells producing IL-2 and the amount of IL-2 produced per cell in the RARαDN relative to the WT B6 group (Fig. 1D). We observed a population of cells that were IFN-γ−, which produced IL-2. We verified that this population was not an artifact, but statistical analysis of the WT and RARαDN samples confirmed that this population was the same in both tested groups. In addition, this population did not alter the observed higher proportion of IL-2+ cells in the B8R-specific or total CD8 population. When we examined molecules associated with effector activity, we saw significant decreases in KLRG1 and granzyme B levels (Fig. 2A, 2B). This finding was indicative of a reduced SLEC phenotype, but not all effector functions were impaired. IFN-γ production (Fig. 2C) was not impaired, and TNF-α production (Fig. 2D) was significantly increased in the RARαDN compared with the WT group.
Ag-specific CD8 T cells exhibited a reduced SLEC phenotype in the absence of RA signaling. WT and RARαDN mice infected by VV-WR (i.n. 1000 PFU) were analyzed 9 d post infection. (A) B8R-tetramer+CD8+ splenocytes were surface stained for KLRG1. (B–D) Spleen cells were incubated for 5 h with or without B8R peptide in the presence of brefeldin A. (B) Intracellular staining was performed to determine granzyme B (GrB) production in the IFN-γ+ CD8+ cells. The t tests were performed on values subtracted for background, using an isotype control for GrB (filled: WT, open bold: RARαDN, open thin: isotype). (C) Intracellular staining was performed to determine IFN-γ production in the CD8+ cells. Statistics were performed on values subtracted for background, using a no B8R peptide control IFN-γ. (D) Intracellular staining was performed to determine TNF-α production in the IFN-γ+ CD8+ cells. The percent of TNF-α+ reported is gated on the IFN-γ+, and the MFI was calculated for IFN-γ+TNF-α+ cells. B8R-tetramer+CD8+ splenocytes were stained for expression of (E) T-bet and (F) eomesodermin (filled: WT; open bold: RARαDN; open thin: isotype). Filled circles are WT and open circles are RARαDN. Experiment was repeated three times. n = 6 per group tested. For t tests performed, *p < 0.01, ***p < 0.0001.
We wanted to determine the effect of the absence of RA signaling on transcription factors associated with effector and memory CD8 T cell differentiation, so we stained for T-bet and eomesodermin, respectively. We observed a reduction in both T-bet and eomesodermin in the RARαDN mice (Fig. 2E, 2F).
Similar MPEC skewing observed after RA inhibitor treatment
The RARαDN mice expressed the RARαDN in all CD4 and CD8 cells, as T cells will express the cre recombinase when they express CD4, at the double-positive stage during thymic development. This finding raised the possibility that the MPEC phenotype was due to developmental defects during T cell development, rather than to the absence of RA signaling specifically during the priming stage of the CD8 response. To rule out the effect of the absence of RA during development, we used an antagonist of RARα, β, and γ (RA-I), which we administered before and during the priming stage (every other day from day −1 to day 9, post infection). We did not observe a significantly reduced B8R Ag-specific response in the RA-I–treated mice during the acute response (Fig. 3A), unlike that observed in the RARαDN mice (Fig. 1A).
Enhanced MPEC and reduced SLEC phenotype reproduced in mice treated with a chemical RA antagonist. WT and RA antagonist (RA-I)–treated mice were infected with 1000 PFU VV-WR i.n. At day 10 post infection, splenocytes were stained with MHC/peptide tetramer for (A) the VV-WR epitope B8R and CD8 and analyzed by flow cytometry for the presence of B8R-specific CD8 T cells. B8R-tetramer+CD8+ splenocytes were surface stained for (B) CD127 and (C) intracellular staining was done for B8R-specific IFN-γ and IL-2 production. The percent of IL-2+ reported is gated on the IFN-γ+ cells, and the MFI was calculated for IFN-γ+IL-2+ cells. (D) B8R-tetramer+CD8+ splenocytes were surface stained for KLRG1. (E) Spleen cells were incubated for 5 h with or without B8R peptide in the presence of brefeldin A. Intracellular staining was performed to determine GrB production in the IFN-γ+ CD8+ cells (filled: WT; open bold: RA-I; open thin: isotype). (A–C) n = 5 or 6 per group tested. Experiment was repeated three times, showing similar data. Combined data from two experiments with (D) n = 5 or 6, (E) n = 3 or 4 per group tested. Experiment was repeated three times, showing similar data. The t tests were performed on values subtracted for background, using an isotype control for GrB. Filled circles are WT and open circles are RA-I. For t tests performed, **p < 0.001, ***p < 0.0001.
Ag-specific cells had a statistically significant increase in CD127 expression (Fig. 3B) and IL-2 production (Fig. 3C) in the RA-I group. We also observed a statistically significant increase in IL-21R expression on a per cell basis (data not shown). Further examination of these cells showed a significant reduction in KLRG1 expression (Fig. 3D), but an increase in granzyme B, IFN-γ, and TNF-α production (Fig. 3E, Supplemental Fig. 2C, 2D) in the RA-I–treated mice. These results indicate that the skewing toward an MPEC phenotype was not due to defects in CD8 T cell development caused by a lack of RA.
MPEC skewing and effector defects are CD8 T cell intrinsic
In the systems analyzed above, we compared WT B6 to either RARαDN or RA-I–treated mice. In neither case was the RA defect limited to the CD8 T cells. To determine the CD8 T cell–intrinsic effects of RA, we made mixed BM chimeric mice. We reconstituted the hematopoietic compartments of lethally irradiated B6 Ly5.2 hosts with a 1:1 mix of WT B6 and RARαDN BM. This experimental system resulted in the same mouse containing both WT and RARαDN CD8 T cells, identified by different congenic markers, so any differences observed could be considered cell intrinsic. Following a 50-d reconstitution period, we confirmed 1:1 reconstitution, and these mice were used in the proceeding experiments, referred to from this point as RARαDN chimeric mice.
We infected the RARαDN chimeric mice with 1000 PFU of VV-WR i.n. and determined the Ag-specific response at 10 d post infection. We did not see a significant difference in the total CD8 response (data not shown) or the B8R Ag-specific response (Fig. 4A). Analysis of surface markers by flow cytometry revealed an increase in CD127 and IL-21R (Fig. 4B, 4C). Thus, the surface MPEC phenotype seen in Figs. 1 and 3 in the previously used models was confirmed, showing this to be a cell-intrinsic phenotype. This phenotype was further corroborated by the increase in IL-2 production by B8R-specific IFN-γ–producing RARαDN CD8 T cells (Fig. 4D). No significant difference in the expression of eomesodermin was noted (Fig. 4E).
Cell-intrinsic skewing of CD8 T cells to MPEC in the absence of RA signaling. To test the cell-intrinsic requirements of RA in CD8 function, mixed BM chimeric mice (RARαDN) were infected with 1000 PFU VV-WR i.n. At day 10 post infection, splenocytes were stained with (A) MHC/peptide tetramer for the VV-WR epitope B8R and CD8 and analyzed by flow cytometry for B8R-specific CD8 T cells. B8R-tetramer+CD8+ splenocytes were surface stained for (B) CD127, (C) IL-21R (filled: WT, open bold: RARαDN). (D) Spleen cells were incubated for 5 h with or without B8R peptide in the presence of brefeldin A. Intracellular staining was performed for IFN-γ and IL-2. The percent of IL-2+ was calculated for the IFN-γ+ CD8 population. (E) B8R-tetramer+CD8+ splenocytes were stained for expression of eomesodermin (filled: WT; open bold: RARαDN; open thin: isotype). Filled circles are WT and open circles are RARαDN. n = 4 or 5 per group tested. Experiment was repeated two times, showing similar data. For t tests performed, ***p < 0.0001.
As seen in the RARαDN mice, the chimeric mice showed a significant reduction in SLEC phenotype by KLRG1, granzyme B, and T-bet expression in the RARαDN CD8 T cells (Fig. 5A–C). Unlike in the RARαDN mice, the chimeric mice showed an increase in IFN-γ (Fig. 5D), along with the previously observed increase in TNF-α in the RARαDN CD8 T cells (Fig. 5E). We further examined these cells for direct ex vivo granzyme B production and CD107 expression as an indicator of degranulation. We saw significant impairments in both (data not shown).
Selective reduction in SLEC phenotype in the absence of RA in chimeric mice. At day 10 post infection with VV-WR (i.n. 1000 PFU), splenocytes from RARαDN chimeric mice were analyzed. (A) B8R-tetramer+CD8+ splenocytes were surface stained for KLRG1. (B–E) Spleen cells were incubated for 5 h with or without B8R peptide in the presence of brefeldin A. (B) Intracellular staining was performed to measure GrB production in the IFN-γ+ CD8+ cells. The t tests were performed on values subtracted for background, using an isotype control for GrB. (C) B8R-tetramer+CD8+ splenocytes were stained for expression of T-bet. (Filled: WT; open bold: RARαDN) (D) Intracellular staining was performed to determine IFN-γ production in the CD8+ cells. Statistics were performed on values subtracted for background, using a no B8R peptide control. (E) Intracellular staining was performed to determine TNF-α production in the IFN-γ+ CD8+ cells. (A) Combined data from two experiments with n = 5 or 6; others are n = 3 or 4 per group tested. Experiment was repeated two times, showing similar data. **p < 0.001, ***p < 0.0001.
CD8 T cell–intrinsic increase in MPEC skewing with selective defects in effector phenotype is maintained during the contraction phase
When the VV-WR RARαDN chimeric mice were examined at 21 d post infection (contraction phase of the CD8 response) we saw a significant reduction in the B8R Ag-specific CD8 T cell response among the RARαDN CD8 T cells (Supplemental Fig. 3A). The B8R Ag-specific RARαDN CD8 T cells maintained the increase in CD127 and IL-21R (Supplemental Fig. 3B, 3C) seen at 10 d (Fig. 4) post infection. The increased production of IL-2 was also maintained during the contraction phase (Supplemental Fig. 3D). At the contraction phase, the expression of eomesodermin corroborated the MPEC phenotype seen with other markers by being significantly elevated in the absence of RA signaling (Supplemental Fig. 3E).
The SLEC phenotype remained diminished at the contraction phase, as seen by the reduction in KLRG1, granzyme B, and T-bet (Supplemental Fig. 3F–H), but the increase in IFN-γ (Supplemental Fig. 3I) was not maintained at the contraction phase, unlike the increased TNF-α expression (Supplemental Fig. 3J). This finding may indicate a differential requirement for RA signaling in IFN-γ production at the acute and contraction phases.
Enhanced central memory phenotype in the absence of RA signaling
We examined RARαDN chimeric mice at >60 d post infection (memory phase of the CD8 response) and saw a significant reduction in the proportion, but not total number, of the B8R-specific CD8 T cell response (Supplemental Fig. 4A). These memory cells were further examined and revealed increases in CD127, CD62L, CD27, and IL-2 production (Fig. 6A–D), which are all indicative of a Tcm phenotype. Analysis of transcription factors revealed a significant decrease in T-bet (Fig. 6E) along with an increase in eomesodermin expression (Fig. 6F).
Enhanced Tcm phenotype in the absence of RA signaling in chimeric mice. B8R-tetramer+CD8+ splenocytes from RARαDN chimeric mice at >60 d post infection with VV-WR (i.n. 1000 PFU) were surface stained for (A) CD127, (B) CD62L, and (C) CD27 (filled: WT, open bold: RARαDN). (D) Intracellular staining was performed to determine IL-2 production in the IFN-γ+ CD8+ cells. The percent of IL-2+ reported is gated on the IFN-γ+, and the MFI was calculated for IFN-γ+IL-2+ cells. B8R-tetramer+CD8+ splenocytes were stained for expression of (E) T-bet and (F) eomesodermin (filled: isotype [arrow]; open bold: RARαDN; open thin: WT). Filled circles are WT and open circles are RARαDN. n = 4 or 5 per group tested. Experiment was repeated four times, showing similar data. For t tests performed, **p < 0.001 and ***p < 0.0001.
Tcm preferentially home to lymphoid organs, so we determined the total B8R-specific CD8 T cell number in the spleen, lymph nodes, and lungs and did not see preferential homing to lymphoid organs, but impairment in homing to the lung (Fig. 7A). Further, Tcm cells exhibit increased homeostatic proliferation, so we examined BrdU incorporation after 10 d of BrdU administration in the drinking water, at the memory stage. We saw a significant increase in the proportion of BrdU+ cells in the absence of RA signaling (Fig. 7B).
Distribution, homeostatic turnover, and recall responses in memory CD8 T cells. At day >60 post 1000 PFU VV-WR i.n infection, RARαDN chimeric mice were analyzed. (A) Spleens, mLNs, and lungs were stained with MHC/peptide tetramer for the VV-WR epitope B8R and CD8 and analyzed by flow cytometry for B8R-specific CD8 T cells. (B) B8R-tetramer+CD8+ splenocytes were stained for BrdU incorporation after 10 d of BrdU uptake from the drinking water (filled: WT; open bold: RARαDN). (C) To test the recall potential of the B8R-tetramer+CD8+ T cells, at day >60 post infection, some mice remained unchallenged as the memory (MEM) mice and others were infected with 106 CFU of L. monocytogenes–B8R i.v. as the Recall mice. At 6 d post challenge, splenocytes were stained with B8R tetramer and anti-CD8 and analyzed by flow cytometry for B8R-specific CD8 T cells. n = 4 per group tested. Experiment was repeated (A) four times, (B) two times. (C) n = >5 per group tested. Experiment was repeated four times, showing similar data. **p < 0.001, ***p < 0.0001.
A functional test of Tcm cells is their ability to recall in response to a secondary challenge. Thus, we i.v. administered 2 × 106 CFU of L. monocytogenes expressing the B8R epitope to a group of chimeric memory mice (labeled Recall) and left a group of mice unchallenged (labeled MEM). The Recall mice showed a robust expansion in the number of B8R Ag-specific CD8 cells, compared with the MEM mice, and both WT and RARαDN CD8 T cell compartments responded similarly (Fig. 7C). Next we examined the phenotype of the secondary effector cells. The RARαDN CD8 T cells had an increase in CD127 and IL-2 expression (Supplemental Fig. 4B, 4C) together with reduced KLRG1, granzyme B, and T-bet expression (Supplemental Fig. 4D–F), consistent with phenotypic differences observed prior to secondary challenge.
Discussion
We have shown that RA signaling is essential for optimal effector and effector memory differentiation by CD8 T cells. We see an increase in the MPEC phenotype at the expense of the SLEC phenotype in the primary effector and recall stages of the responses, as observed by increased CD127hi, eomesoderminhi, and IL-2+ CD8 T cells and reduced KLRG1+, T-bethi, and granzyme Bhi CD8 T cells. This suboptimal differentiation to effector T cells resulted in more MPECs and an enhanced central memory population. The central memory population was characterized by the surface expression of CD62L, CD27, and CD127; elevated IL-2 production; and increased homeostatic proliferation, measured by BrdU incorporation.
Previous work using conditional (VavCre) knockout of RARγ showed that the absence of RARγ signaling in the hematopoietic and endothelial compartments resulted in reduced Ag-specific CD8 T cells at the effector and memory time points, whereas the CD4 T cells were unaffected (31). A plethora of recent work allows for the appreciation of RA signaling in the functions of many other immune cells, which may influence the outcomes observed in CD8 T cells. Therefore, some of the data from this previous study may be a result of extrinsic effects of the absence of RA signaling on other immune cell types. In addition, this study focused on a hematopoietic and endothelial knockout of the RARγ receptor-mediated signaling, which may enhance signaling through other RARs via the increase in availability of retinoic X receptors, which are necessary for all RARs to bind the response elements on DNA (32). In our model, the mice express an RARαDN; thus, a repressor is bound constitutively to the RA response element and inhibits signaling via all the RARs. Thus, to our knowledge this is the first study to determine the cell-intrinsic effect of RA signaling on effector and memory CD8 T cell differentiation.
Our hypothesis that RA is essential for optimal effector differentiation is supported by our observation that RA is expressed in the lung, lymph nodes, and spleens and is increased after virus infection. This finding may indicate a requirement during the priming of the response. The expression was restricted to professional APCs, DCs (CD11c+), macrophages (CD11b+), and B cells (CD19+), indicating that it may augment the established three signals for T cell activation (TCR signal, costimulation, and inflammatory signals) (33–36). Previous studies have shown that the NF of activated T cells family of transcription factors and calcium mobilization are enhanced by vitamin A and RA signaling, respectively (10, 37, 38). The NF of activated T cells family of transcription factors and calcium mobilization are essential features of T cell activation, and their dependence on RA signaling supports the concept that RA is an important component of optimal T cell activation. Hall et al. (10) examined the role of RA in promoting the mTORC1 and AKT signaling pathways during T cell activation. They showed that the absence of RA diminished these pathways, cellular cytokine production, and T cell proliferation. Their data are consistent with our findings of decreased SLECs, which is the expected result of weaker TCR signaling in the absence of RA responsiveness. Taken together, local upregulation of RA by APCs during the initiation of an immune response, and promotion of effector CD8 T cell generation, act to enhance effector T cell generation, which is beneficial for the infected host.
Although we observed impairments in effector functions in the absence of RA signaling, they were limited to KLRG1 upregulation and granzyme B production. In contrast, TNF-α production was significantly increased in the acute and memory phases. This outcome may not be surprising, as RA has been previously shown to negatively regulate TNF-α mRNA stability (39). Thus, the robust increase in TNF-α production by the CD8 T cell, in our system, may result from the absence of posttranscriptional regulation of TNF-α by RA. IFN-γ was impaired in the absence of RA signaling during the acute phase of the response, which is supported by work in CD4 T cells showing impairment in IFN-γ production in effector cells (10, 11). This impairment in IFN-γ production was not carried forward into the contraction and memory phases, showing that RA significantly influenced IFN-γ production only in effector cells.
In the current study, data from the chimeric model show that the influence of RA signaling on eomesodermin begins at the contraction phase and is carried into the memory phase. Work from the Reiner group (40) has shown that CD8 T cells use T-bet and eomesodermin redundantly for effector functions, but eomesodermin is essential to promote memory survival and the central memory phenotype. Therefore, our observed increase in eomesodermin at the contraction phase and the memory phase may explain the enhanced MPEC and central memory phenotypes, respectively. As the absence of RA signaling induces this increase in eomesodermin, RA may have a direct or an indirect role in eomesodermin expression.
We observed an increase in IL-2 production by CD8 T cells from the acute to the memory phase in the absence of RA signaling. This finding is likely a direct effect of the absence of RA signaling, as there is an RA response element in the promoter region of IL-2, and studies have shown that RA suppresses IL-2 expression (41, 42). This increase in IL-2 production combined with the increased eomesodermin further confirms the MPEC and central memory phenotype.
Although Tcm are known to home preferentially to lymphoid organs, we do not observe a significant accumulation of CD8 T cells in the absence of RA signaling in the lymph nodes (25). CD62L expression by WT CD8 T cells may be sufficient for effective lymphoid homing, and further expression by the RARαDNs may not serve for better homing. Despite similarities in lymphoid homing, we see defective lung homing, as evidenced by the reduced Ag-specific CD8 T cell numbers in the lung. Previous studies have shown that cells primed to a respiratory infection become imprinted to home to the lung (43). In addition, a recent study showed that CD8 T cells primed with RA supplementation resulted in increased memory cells in mucosal sites, which supports our findings of defective homing to the lung in the absence of RA signaling (44).
Despite the significantly enhanced central memory phenotype of RARαDN CD8 T cells, WT CD8 T cells expanded as robustly as the RARαDN during a recall response. During the primary response, we observed a reduced number of effector cells, which implies that optimal RA signaling may be essential for robust effector differentiation, and this may also be true during secondary responses. Owing to constitutive expression of the RARαDN construct in CD8 T cells, RA signaling was absent during the recall response, which may prevent the central memory skewed RARαDN memory cells from expanding to their full potential.
Collectively, our data show that RA signaling is important during the priming of the response to promote effector differentiation. Deficiency in RA signaling may result in defaulting of the effector population to the MPEC phenotype, leading to an increase in the central memory population, which also requires RA to differentiate to secondary effectors when rechallenged.
To our knowledge, this is the first time that the effect of RA signaling has been isolated to the CD8 T cell compartment during a viral infection. These findings implicate RA in optimal effector function, and these studies are consistent with findings from the Greenberg group (44), showing the adjuvant effects of RA during vaccination against viral infections, which exhibit enhanced effector responses. Therefore, our study has an important bearing on vaccine trials involving supplementation of vitamin A or retinoic acid during priming, implying this may favor effector CD8 T cell differentiation but may not result in an optimal central memory response.
Note added in proof. Another paper reporting similar findings was published while this paper was under review (45).
Disclosures
The authors have no financial conflicts of interest.
Acknowledgments
We thank Dr. Patricia Ernst (Department of Genetics at the Geisel School of Medicine at Dartmouth, Lebanon, NH), who provided advice on improvements to the bone marrow chimera protocol; and Jie Deng and Dr. James D. Gorham (Department of Microbiology and Immunology at the Geisel School of Medicine at Dartmouth) for use of the ALDEFLUOR Assay Kit.
Footnotes
This work was supported in part by National Institutes of Health Grants AI069943 and CA103642.
The online version of this article contains supplemental material.
Abbreviations used in this article:
- ALDH
- aldehyde dehydrogenase
- BM
- bone marrow
- B8R
- VV-WR epitope B8R20–27 (TSYKFESV)/Kb
- DC
- dendritic cell
- i.n.
- intranasal(ly)
- KLRG1
- killer cell lectin-like receptor subfamily G member 1
- mLN
- mediastinal (lung-draining) lymph node
- MPEC
- memory-precursor effector cell
- RA
- retinoic acid
- RARαDN
- dominant negative RA receptor α
- SLEC
- short-lived effector cell
- Tcm
- central memory cell
- VV-WR
- vaccinia virus, Western Reserve strain
- WT
- wild-type.
- Received July 13, 2012.
- Accepted December 19, 2012.
- Copyright © 2013 by The American Association of Immunologists, Inc.