Cutting Edge: Innate Memory CD8⁺ T Cells Are Distinct from Homeostatic Expanded CD8⁺ T Cells and Rapidly Respond to Primary Antigenic Stimuli

Conventional T cells arise as naive T cells through thymic selection and require further exposure to Ags to expand and acquire effector/memory function (1). Although conventional T cells are selected on thymic stroma by peptide–MHC complexes, nonconventional innate T cells have distinct selection pathways, spontaneously express markers typically found on memory T cells, and have the ability to respond rapidly upon stimulation (1–7). Nonconventional CD8⁺ T cells are distinguished by their memory-like attributes of the effector program, such as secretion of IFN-γ and expression of CD44 and CD122 (IL-2Rβ) (8, 9); thus, they are termed “innate memory” or “innate memory phenotype” (IMP) CD8⁺ T cells (1, 10–14). IMP CD8⁺ T cells may be selected independently of thymic MHC class I (MHCI)a, with a dominant role played by hematopoietic MHCI (11, 13). However, it was also suggested recently that innate memory-like CD8⁺ thymocytes are mainly selected by nonhematopoietic MHCIa (15), strongly alluding to alternative pathways for IMP CD8⁺ T cell development.

IMP CD8⁺ T cells exist in specific pathogen–free, as well as in germ-free, mice (16), independent of immunization and infection, and they are significantly increased in the absence of Tec kinase ITK and the transcriptional regulators CBP, KLF2, and Id3, as well as in mice carrying an ITK-binding mutant of Slp-76 (1, 10, 11, 13, 17–20). IL-4 was shown to be involved in the generation of these cells in a cell-extrinsic manner in the absence of ITK (12). Although it is known that IMP CD8⁺ T cells are distinct from Ag-induced memory CD8⁺ T cells, they share markers of T cells that have undergone homeostatic expansion (21); therefore, it is not clear whether IMP CD8⁺ T cells arise through homeostatic proliferation (HP). Functionally, IMP CD8⁺ T cells can facilitate the clearance of Listeria monocytogenes infection (10); however, it has been difficult to examine directly whether IMP CD8⁺ T cells can respond to Ag during an immune response because of the difficulty in discriminating IMP CD8⁺ T cells from conventional memory CD8⁺ T cells.

Wild-type (WT), MHCI^−/− (B2m^−/−: B6.129P2-B2m^tm1Unc/J), and Nude (B6.Cg-Foxn1^nu/J) mice were from the Jackson Laboratory (Bar Harbor, ME), on a C57BL/6 background. All experiments were approved by the Office of Research Protection’s Institutional Animal Care and Use Committee at Pennsylvania State and Cornell Universities.

Bone marrow from donor mice was injected i.v. into lethally irradiated recipients (6–9 wk old). T cell–depleted bone marrow was obtained by negative selection using biotinylated anti-CD3ε Ab (eBioscience, San Diego, CA) and magnetic column separation (Miltenyi Biotec, Cambridge, MA). Mice were kept in a specific pathogen–free environment and given acid water containing antibiotics (1 mg/ml Gentamicin Sulfate) for 8–9 wk post-bone marrow transplantation (BMT) prior to analysis.

Cell sorting was performed on a FACSAria Cell Sorter (BD Biosciences, San Jose, CA). IMP CD8⁺ T cells (TCRβ⁺CD8α⁺CD44^hi) were sorted from splenocytes of WT→MHCI^−/− chimeras 8 wk posttransplantation and 8-wk-old Itk^−/− mice. HP cells were generated by i.v. injection of naive CD8⁺ T cells (TCRβ⁺CD8α⁺CD44^lo) into Rag1^−/− recipients (0.5 × 10⁶/mouse), followed by sorting of TCRβ⁺CD8α⁺CD44^hi cells 8 wk posttransfer. mRNA from sorted cells was extracted, amplified, and used for microarray (GeneChip Mouse Genome 430 2.0 Array; Affymetrix, Santa Clara, CA) at the Cornell University Life Sciences Core Laboratories Center. Microarray data were analyzed using GeneSpring GX software (Agilent Technologies, Clara, CA). Robust Multi-array Average–normalized probe values were used to generate a correlation coefficient matrix and further converted to gene expression values with quantile normalization, followed by analysis of gene differential expression. Data have been deposited into the National Center for Biotechnology Information’s Gene Expression Omnibus repository (http://www.ncbi.nlm.nih.gov/gds) under accession number GSE41482. Quantitative real-time PCR was carried out using TaqMan probe sets (Applied Biosystems, Foster City, CA).

The two-tailed Student t test was performed using GraphPad Prism v5.00 (GraphPad, San Diego, CA), with p < 0.05 considered statistically significant.

IMP CD8⁺ T cells were reported to be able to develop in the thymus independent of MHCI molecules on thymic stroma (11). We took advantage of this to generate mice carrying predominantly IMP or naive CD8⁺ T cells through BMTs using WT and MHCI^−/− (B2m^−/−) mice. The spleens of WT→MHCI^−/− mice had predominantly CD44^hiCD122⁺ IMP CD8⁺ T cells that rapidly produced IFN-γ upon PMA and ionomycin (P/I) stimulation, as reported previously (1, 10–14). In contrast, MHCI^−/−→WT mice had predominantly naive CD8⁺ T cells (Fig. 1A). As expected, hematopoietic MHCI expression is necessary and sufficient for development of IMP CD8⁺ T cells.

Further examination revealed that the thymus is dispensable for IMP CD8⁺ T cell development, because transplant of WT bone marrow into athymic Nude mice resulted in the development of cells similar to those seen in the WT→MHCI^−/− transplants (Fig. 1B, 1C). Of note, although endogenous CD8⁺ T cells in Nude mice are CD44^hi, they do not express CD122 and fail to produce IFN-γ rapidly in response to P/I (Fig. 1B).

When transferred into lymphopenic environments, naive CD8⁺ T cells undergo significant proliferation in an attempt to restore “normal” levels of T cells. Along with this proliferation, naive CD44^loCD8⁺ T cells acquire a phenotype resembling memory-like CD8⁺ T cells, a process termed “homeostatic expansion” (22, 23). Naive CD8⁺ T cells also proliferate and differentiate upon Ag recognition and activation, converting into long-lived Ag-specific memory CD8⁺ T cells that reside, in part, in bone marrow (24). Therefore, it is possible that the cells we observed are due to homeostatic expansion of a small number of naive T cells and/or proliferation of memory T cells in the transferred donor bone marrow. However, we found that depletion of T cells prior to BMT did not affect the development of IMP CD8⁺ T cells (Fig. 2). This suggests that IMP CD8⁺ T cells observed in the WT→MHCI^−/− chimeric model are the result of hematopoietic MHCI selection and development rather than homeostatic expansion of T cells from donor bone marrow.

To confirm that the IMP CD8⁺ T cells in the WT→MHCI^−/− chimeras are not the result of homeostatic expansion, we analyzed whole-genome gene-expression profiles of sorted IMP CD8⁺ T cells from WT→MHCI^−/− (WM) chimeras, Itk^−/− mice, and CD8⁺ T cells derived from homeostatic expansion of naive CD8⁺ T cells in Rag1^−/− recipients (Fig. 3A). Correlation coefficient matrix shows that gene expression between WM IMP CD8⁺ T cells and Itk^−/− IMP CD8⁺ T cells are highly correlated (correlation coefficient > 0.98), whereas both did not correlate as well with HP CD8⁺ T cells (correlation coefficient: ∼0.66–0.87) (Fig. 3B). Among 27,800 genes, compared with WM CD8⁺ T cells, ∼4,000 genes are upregulated and >4,000 genes are downregulated by >2-fold in HP CD8⁺ T cells (p < 0.05; p values were generated by asymptotic computation with Benjamini–Hochberg false discovery–rate correction), whereas only 21 genes of this type are found in Itk^−/− CD8⁺ T cells (Fig. 3C). We used quantitative real-time PCR to examine a few meaningful genes that differed between these cell types; consistent with the differential-expression profile identified by microarray (Supplemental Fig. 1A), we found upregulation of TNFR (Tnfarsf1a) and NFκB2 (Nfkb2) and downregulation of Eomesodermin (Eomes) and Bim (Bcl2l11) (Supplemental Fig. 1B) in HP CD8⁺ T cells, but not in Itk^−/− CD8⁺ T cells, compared with WM IMP CD8⁺ T cells. This whole-genome expression analysis strongly suggests that IMP CD8⁺ cells generated in WM chimeras highly resemble IMP CD8⁺ T cells in Itk^−/− mice and are distinct from those derived from homeostatic expansion.

The BMT approach suggests a method to generate Ag-naive, Ag-specific IMP CD8⁺ T cells. To study the primary response of Ag-specific IMP CD8⁺ T cells, we used OTI mice (expressing OVA-specific TCR on CD8⁺ T cells) as bone marrow donors to generate Ag-naive OVA-specific IMP CD8⁺ T cells in MHCI^−/− recipients. Reciprocally, OTI-transgenic mice that lack MHCI^−/− (MHCI^−/−OTI) were used as donors to generate OVA-specific naive CD8⁺ T cells in WT recipients, with OTI→WT chimeras as controls. Similar to the results with BMT in nontransgenic backgrounds, OTI→MHCI^−/− chimeras had predominantly TCR transgene–positive OVA-specific CD44^hiCD122⁺ IMP CD8⁺ T cells that rapidly produced IFN-γ. Analogously, MHCI^−/−OTI→WT chimeras had predominantly TCR transgene–positive OVA-specific CD44^loCD122⁻ naive CD8⁺ T cells (Fig. 4A, 4B). Using this model, we examined whether Ag-naive, Ag-specific IMP CD8⁺ T cells can respond rapidly to Ag in the absence of prior antigenic exposure. OTI-transgenic IMP and naive CD8⁺ T cells were stimulated in vitro with OVA or OVA_257–264 epitope of OVA recognized by the OTI TCR, without prior primary Ag exposure in vivo. We found that OVA/OVA_257–264-specific IFN-γ and TNF-α production by CD44^hi CD8⁺ T cells peaked on the third day poststimulation in OTI IMP CD8⁺ T cells (OTI→MHCI^−/−) (Fig. 4C), suggesting that these cells were highly functional IFN-γ/TNF-α double producers (Fig. 4C) and indicating that Ag-naive OVA-specific IMP CD8⁺ T cells can respond rapidly in a highly functional manner to specific Ag in the absence of primary antigenic exposure. In contrast, naive OTI cells (MHCI^−/−OTI→WT) did not show significant cytokine response until >6 d of stimulation, as would be expected in vivo.

The function of IMP CD8⁺ T cells, as well as the mechanism behind their development, has elicited significant interest. Recently, Rafei et al. (15) showed that Ag-naive OTI-transgenic CD44^hiCD8⁺ thymocytes, selected on thymic MHCIa and that have never left thymus, can respond promptly to OVA peptide with cytokine secretion. In addition, Haluszczak et al. (21) showed that a proportion of functional Ag-specific peripheral CD8⁺ T cells in germ-free mice have a memory-like phenotype, although one cannot rule out that these cells are Ag experienced. Using bone marrow chimeras and transgenic approaches, we show for the first time, to our knowledge, that we can develop significant numbers of Ag-naive, Ag-specific IMP CD8⁺ T cells that are similar to those that develop in the absence of ITK but are different from those that develop as the result of homeostatic expansion, although they have similar phenotypes. Our work provides a model and opens the door for more detailed analysis of the development, and, more importantly, function of such cells.

Our findings that endogenous CD8⁺ T cells in Nude mice have a partial innate memory phenotype suggests that perhaps other signals may be required for the complete development and/or maturation of these cells, which may be indirectly dependent on the thymic structure or the Foxn1 gene. This could involve a precursor that traffics through the thymus prior to our BMT. In the absence of ITK or the transcription factor KLF2, the increase in innate memory-like CD8⁺ T cells was suggested to be due to the influence of IL-4 produced by NKT-like cells in these mice (12). It remains to be determined what role IL-4 signaling plays in the development of these hematopoietic MHCI-dependent IMP CD8⁺ T cells. Our preliminary analysis indicates that IMP CD8⁺ T cells differentially express TNFR (lower than HP cells) and Bim and the transcription factor Eomesodermin (higher than HP cells); the latter was observed previously in IMP CD8⁺ T cells (25). Future experiments will distinguish the role of such factors in the development of these cells.

In the immune system, innate immune cells initiate a relatively Ag-nonspecific response to primary infection, allowing cells of the adaptive immune system to develop more specific and exquisite Ag responses, along with the generation of immune memory (26, 27). In particular, Ag-specific memory T cell responses are thought to be generated only after antigenic exposure (28). However, our work suggests that Ag-naive innate memory CD8⁺ T cells can generate potent memory-like Ag-specific responses. We suggest that such IMP CD8⁺ T cells evolved to respond rapidly early in infections in an Ag-specific fashion, assisting the innate immune response and allowing priming of the Ag-specific adaptive immune response. These cells may occupy an essential niche in the early immune response to primary infection.

We thank Meg Potter and Amie Wood for animal care, Shailaja Hegde and Gabriel Balmus for assistance with mouse irradiation, Nicole Bem and Rod Getchell for assistance with flow cytometry, Lavanya Sayam for assistance in cell sorting, and Jennifer D. Mosher for assistance with microarray.

The authors have no financial conflicts of interest.