In vitro cultures with anti-CD3/anti-CD28-coated beads in the presence of high dose IL-2 can efficiently expand CD4⁺CD25⁺ Treg cells that suppress the proliferation of effector T cells. However, the small number of natural CD4⁺ Treg cells available for expansion limits the use of this approach. Therefore, we attempted to obtain Treg cells from the more abundant total CD4⁺ and CD8⁺ T cell populations from the mouse spleen. Freshly purified splenic CD8⁺ or Mo-Flow sorted CD4⁺ T cells from 7-8-week old mice were cultured with an expansion protocol consisting of anti-CD3/anti-CD28-coated beads and high dose IL-2. By the end of the 10-13 days of expansion, the number of cells had generally increased by over 1,000-fold. The cultured cells were phenotyped for a number of surface markers (Figure 1). The vast majority of the cultured cells from CD8⁺ T cells were positive for CD8 (>95%) and the activation marker CD25 (98%) at the end of the culture. Consistent with the activation of these cells, the percentages of CD62L⁺ cells gradually decreased and became very low near the end of the culture (around 10%). Similarly, the culture conditions can efficiently expand CD4⁺ T cells. At the end of the culture, the cultured cells remained CD4⁺ (97%) and became positive for the activation marker CD25 (99%).

The cultured CD8⁺ and CD4⁺ T cells were tested for their ability to inhibit the proliferation of CD4⁺CD25^- naïve T cells (Tn cells) using two different in vitro suppression assays. In the first assay, the naïve T cells were labeled with carboxyfluorescein succinimidyl ester (CFSE) and T cell proliferation was assessed by the dilution of CFSE signal using fluorescence-activated cell sorting (FACS) analysis. As shown in Figure 2a, the cultured CD8⁺ T cells efficiently suppressed proliferation of naïve CD4⁺CD25^- T cells. The suppressive activity of the cultured CD8⁺ T cells is dose-dependent and strong suppression can be seen at the 1:16 expanded CD8⁺ T to Tn cell ratio (Tr/Tn; Figure 2b). In the second suppression assay, T cell proliferation was measured by incorporation of [³H]thymidine. As shown in Figure 2c, the dose-dependent suppression activity of the CD8⁺ T cells was confirmed. Furthermore, the cultured CD8⁺ T cells did not proliferate in response to anti-CD3 and antigen presenting cell (APC) stimulation. This anergic phenotype is consistent with the observation on CD4⁺CD25⁺ Treg cells 1819. Finally, the cultured CD8⁺ cells appeared to suppress better than freshly isolated CD4⁺CD25⁺ Treg cells (Figure 2c; p < 10^-6). The cultured CD4⁺ T cells also had some suppressive function at the high Tr/Tn ratio of 1:1, while the suppressive activity for the cells gradually became undetectable, suggesting that the suppressive activity of the cultured CD8⁺ T cells was much higher than the CD4⁺ T cells cultured under the same conditions (Figure 1c). Therefore, most subsequent studies focused on the phenotype of the cultured CD8⁺ T cells.

To gain further insight into the phenotypes and functions of the cultured CD8⁺ and CD4⁺ T cells, we carried out microarray analyses using Affymetrix GeneChips that cover the whole mouse transcriptome (>45,000 transcripts). Five independent cultures of CD8⁺ T cells and three independent cultures of CD4⁺ T cells as well as two groups of control cells were included in the microarray analysis. The first group of control cells included two freshly isolated naïve CD8⁺ T cells and the second control group consisted of two CD8⁺ T cells activated by a low dose of soluble anti-CD3 and anti-CD28 (activation protocol). Naïve CD8⁺ T cells as well as activated CD8⁺ T cells do not possess suppression function. This data set was analyzed as described in Materials and methods and the results are summarized in Table 1. As expected, the expression of thousands of genes was changed by the expansion protocol and the activation protocol compared to naïve CD8⁺ T cells (Figure 3). Surprisingly, over 100 genes were changed by >10-fold and a few dozen genes were changed by 40- to 800-fold in the cultured CD8⁺ and CD4⁺ T cells compared to naïve CD8⁺ T cells.

To elucidate the molecular basis of the function of the cultured CD8⁺ T cells, we functionally annotated the 314 genes with >5-fold differences (including 113 genes with >10-fold differences) between the cultured and naïve CD8⁺ T cells (Table 2). The largest group of differentially expressed genes (17% for >5-fold difference and 31% for >10-fold difference) is, as expected, involved in immunity and defense. The genes with >10-fold differences are enriched by 6-fold compared to the frequency of this functional group in the genome (p = 7.7 × 10^-15). Other significantly enriched gene groups with considerable interest include those involved in apoptosis, cell cycle, cell proliferation and differentiation, and cell adhesion (Table 2). Twenty-three cell cycle genes were upregulated by >5-fold, including 11 genes that were upregulated by >10-fold in the cultured CD8⁺ T cells (Table 2). Twenty-one genes in the cell proliferation and differentiation category were upregulated and twenty-five upregulated genes belong to the apoptosis group. A number of these genes were selected for confirmation using a combination of real-time RT-PCR, flow cytometry and ELISA. All selected genes have been confirmed and will be discussed in more detail later.

The expression of a large number of transcription factors (TFs) was changed in the CD8⁺ and CD4⁺ T cells cultured using the expansion protocol (Table 3). Most of the differentially expressed TF genes were upregulated, while a small number were downregulated in the cultured cells. The expression patterns of the TF genes share some similarity but also have significant differences in the cultured CD8⁺, cultured CD4⁺, activated CD8⁺ T cells and naïve CD8⁺ T cells. Many of the TF genes still have unknown biological functions and their roles in T cells have not been investigated. However, several TF factors are known to be critical for the immune system and may play a role in gaining suppressive function for the cultured CD8⁺ T cells. The V-myc myelocytomatosis viral related oncogene, neuroblastoma derived (Mycn) is essential to cell proliferation and differentiation 20. This was the most upregulated TF gene (21-fold) in the cultured CD8⁺ T cells but not in cultured CD4⁺ (2-fold) or activated CD8⁺ (1-fold) T cells (Table 3). RT-PCR analyses confirmed the expression differences observed with the microarray analysis (Figure 4). This may be a key gene for the cultured CD8⁺ T cell phenotype. The Eomesodermin homolog (Eomes) is a T-box transcription factor that is highly homologous to T-bet. Eomes and T-bet may have cooperative or redundant functions in regulating the genes encoding IFN-γ and cytolitic molecules in CD8⁺ T cells 21, and determine the fate of effector and memory CD8⁺ T cells 22. Furthermore, they are responsible for inducing enhanced expression of Il2rb (CD122) 22, a marker for some CD8⁺ Treg cells 23. Eomes was upregulated four-fold in the cultured CD8⁺ T cells while it was downregulated five-fold in the cultured CD4⁺ T cells and was unchanged by our activation protocol (Table 3). The upregulation of Eomes may be responsible for the increased expression of IFN-γ, perforin, granzymes, CD122 and other genes in cultured CD8⁺ T cells. It could be a critical TF for the suppressive function of the cultured CD8⁺ T cells. Runt related transcription factor 2 (Runx2) may be another critical transcription factor. Runx2 was highly upregulated in the cultured CD8⁺ T cells (8.6-fold) and moderately upregulated in the cultured CD4⁺ (3.5-fold) and activated CD8⁺ (1.8-fold) T cells. Runx2 plays an important role in early T cell development 24. Over-expression of Runx2 increases the proportion of single positive CD8⁺ T cells 25. Other potentially important TFs include Litaf, Jun (AP1), Zbtb32 (Rog), Zfp608 and Rnf13, which had higher expression levels in the cultured CD8⁺ T cells than in the other three types of cells. The expression of Foxp3, which is an important TF for CD4⁺ Treg cells, was not detectable by RT-PCR (data not shown) in the CD8⁺ T cells cultured under this condition.

Several genes encoding surface markers on NK cells were highly upregulated in the cultured CD8⁺ T cells (19-fold for CD244, 13-fold for Ly49e, 4.4-fold for NK1.1, 8.0-fold for NKG2A and 6-fold for NKG2D; Figure 5a) but not in the cultured CD4⁺ or activated CD8⁺ T cells. To confirm these findings, FACS analysis was carried out for a number of surface markers. As already mentioned, these cultured cells remained positive for CD8 (~99%) and negative for CD4 (Figure 1). They were activated T cells as indicated by the high expression levels of CD25 and CD69 as well as the low expression level of CD62L (Figure 1). Consistent with the low frequency of NKT cells among naïve CD8⁺ T cells, <1% of the CD8⁺ T cells were positive for these markers after three days of culture (Figure 5b), while the majority of the cells became positive for NK1.1 and CD244 after about 10 days of culture. The percentages of cells positive for the NK markers may vary from culture to culture. By day 10-13, 75-95% of the cells were normally positive for NK1.1 and CD244. NKG2A was upregulated by 8-fold in the cultured CD8⁺ T cells according to the microarray data (Figure 5a) and 25-30% of the cultured CD8⁺ T cells stained positive for NKG2A. Although CD94 and DX5 were not upregulated in the cultured CD8⁺ NKT-like cells according to the microarray data (Figure 5a), FACS analyses indicated that 15-30% of the cultured CD8⁺ T cells were positive for these NK markers. It is unclear if these discrepancies are due to an imperfect correlation between gene and protein expression. Since the vast majority of the cultured CD8⁺ T cells expressed NK markers, the cultured CD8⁺ T cells had similar phenotypes to NKT cells, which are defined as cells expressing both T cell and NK cell markers. Furthermore, these cells were negative for the α-galactosylceramide-loaded Cd1d tetramer (data not shown), suggesting that they were not Cd1d-restricted iNKT cells. It is unclear at this time what the source of these cultured CD8⁺ NKT-like cells was. As the CD8⁺ NKT-like cell precursors in the total CD8⁺ T cell pool were very rare, we believe that the cultured CD8⁺ NKT-like cells were probably expanded from the conventional CD8⁺ T cells, which acquired NK markers during the expansion. According to a recent classification of NKT cells 7, these cultured cells belong to the CD8⁺ NKT-like category of NKT cells.

Using the 318 genes that are upregulated by >5-fold in the cultured CD8⁺ NKT-like cells, we established molecular networks to understand the functional relationships of the genes upregulated in the CD8⁺ NKT-like cells. The largest network consists of genes involved in immunity and defense (Figure 6). This network highlights the importance of two central nodes: IL-10 and IFN-γ. Upon stimulation by anti-CD3/CD28 and IL-2, IL-10 and IFN-γ are highly upregulated (by 47- and 51-fold, respectively; Table 4). These proteins and pathways are known to influence the expression of many genes involved in immune responses, including those encoding the activation marker IL-2 receptor (Il2ra, or CD25), granzymes, the tumor necrosis factor (TNF) family genes, cytokines, chemokines and their receptors. Many of these genes are significantly upregulated in the CD8⁺ NKT-like cells (Tables 4 and 5). To confirm the microarray data, we used ELISA to measure the levels of secreted cytokines in the culture medium of CD8⁺ NKT-like cells and natural CD4⁺CD25⁺ Treg cells stimulated by anti-CD3 and APC (Figure 7a). Consistent with the microarray data, the cultured CD8⁺ NKT-like cells secreted more IL-10 and IFN-γ but a similar level of IL-4 when compared to fresh CD4⁺CD25⁺ Treg cells (Figure 7a). The secretion of IFN-γ was particularly high in the CD8⁺ NKT-like cells. The expression of IFN-γ and lack of expression of IL-4 are also consistent with the observation on other NKT-like cells 7. IL-10 and IFN-γ are immunosuppressive cytokines known to be involved in the suppressive function of CD4⁺ Treg cells and may contribute to the suppressive function of the expanded CD8⁺ NKT-like cells. Transforming growth factor (TGF)-β is another important immunosuppressive cytokine that might be important for the suppressive function of the CD8⁺ NKT-like cells. Our microarray and RT-PCR data (Figure 4) indicate that the TGF-β mRNA level was about two-fold higher in CD8⁺ NKT-like cells compared to naïve CD8⁺ cells. As TGF-β cannot be accurately measured from serum-containing culture medium, we performed blocking experiments using an anti-TGF-β antibody to assess the role of TGF-β. Our results (Figure S1 in Additional data file 1) indicate that TGF-β blockade cannot block the suppression function of the CD8⁺ NKT-like cells.

A number of other secreted molecules were also highly upregulated in the CD8⁺ NKT-like cells based on the microarray data (Table 4). Many of these secreted molecules are known to have immunosuppressive function or potentially contribute to the suppression function. Perforin (Prf1) and granzymes are among the most noticeable. Perforin and granzyme expression is regulated by IFN-γ (Figure 6). Both natural and adaptive CD4⁺CD25⁺ Treg cells in human display perforin-dependent cytotoxicity against autologous target cells, suggesting that the perforin/granzyme pathway is one of the mechanisms that Treg cells can use to control immune responses 26. Prf1 was upregulated by 29-fold in the CD8⁺ NKT-like cells (with potent suppression activity; Table 4), but unchanged in cultured CD4⁺ cells (with weak suppression activity) and activated CD8⁺ T cells (without suppression activity). Several granzymes were highly upregulated (834-, 535-, 446-, 329-, 105-, 63-, 61- and 23-fold for granzymes D, E, C, G, B, F, K and A, respectively). These molecules were generally upregulated to a much lesser degree in the cultured CD4⁺ T cells and were unchanged in activated CD8⁺ T cells (except Gzmk). The large expression differences for Prf1 and selected granzymes were confirmed using RT-PCR (Figure 4).

Several secreted molecules can potentially be implicated in the immunosuppressive function. The most noticeable include Esm1, Spp1, Fgl2, Tnfrsf11b, Lgals3, Lgals1, and IL-24 (Table 4). Esm1 (endothelial cell-specific molecule 1) was upregulated in the CD8⁺ NKT-like cells by 75-fold and was only slightly increased in the cultured CD4⁺ T cells (2.8-fold) and was unchanged in the activated CD8⁺ T cells. The expression pattern was confirmed by RT-PCR (Figure 4). Esm1 is a proteoglycan mainly secreted by endothelial cells under the control of inflammatory cytokine. It binds to LFA-1 integrin on the surface of lymphocytes and monocytes 27 and therefore inhibits the binding of intercellular adhesion molecules (ICAMs) to LFA-1 and influences leukocyte adhesion and activation. Spp1 (secreted phosphoprotein 1) is better known as osteopondin. In addition to its well known function in bone formation, it functions as a cytokine and chemokine to regulate cell-cell and cell-tissue interaction. Much less well known is its function in suppressing T cells and activating B cells 28. Osteopondin is believed to be the most abundant protein secreted by activated T cells, which is consistent with our microarray data (7.4-fold higher expression in activated CD8⁺ T cells versus naïve T cells). Osteopondin was upregulated by 252-fold in the CD8⁺ NKT-like cells and 204-fold in the cultured CD4⁺ T cells based on the microarray data (Table 4). Based on the RT-PCR data, Osteopondin (Spp1) was greatly increased (by 25,000-fold) in the CD8⁺ NKT-like cells compared to the naïve CD8⁺ T cells (Figure 4). It is possible that Osteopondin contributed to the suppression activity of both the CD8⁺ and CD4⁺ T cells cultured using our protocol. Tnfrsf11b, also known as Osteoprotegerin (Opg), is a member of the TNF receptor superfamily. Opg is a decoy receptor of RANKL and inhibits the binding of RANKL (Receptor activator for nuclear factor κB ligand) to its receptor RANK. Opg is secreted as a disulfide-linked homodimer 29. Opg can inhibit the inflammatory effect of RANKL secreted by activated T cells 3031 and RANKL blockade can significantly prolong heart allograft survival 32. Opg was upregulated by 29-fold in the CD8⁺ NKT-like cells and unchanged in the cultured CD4⁺ and activated CD8⁺ T cells. The expression changes were also confirmed by RT-PCR (Figure 4). Fgl2 (Fibrinogen-like protein 2) is a member of the fibrinogen-related protein superfamily. In addition to its well established role in triggering thrombosis, it is known to be secreted by T cells under the control of IFN-γ 33. Fgl2 has been shown to exhibit immunomodulatory properties capable of inhibiting dendritic cells (DC) maturation and T cell proliferation stimulated by alloantigens or anti-CD3/anti-CD28 antibodies in a dose-dependent manner 34. Fgl2 was upregulated by 33-fold in the CD8⁺ NKT-like cells but was unchanged in the cultured CD4⁺ and activated CD8⁺ T cells. Thus, Fgl2 could be a critical factor for the suppression mechanism of the CD8⁺ NKT-like cells. Lgals3 and Lgals1, also known as Galectin (Gal)-3 and Gal1, are members of the beta-galactoside-binding gene family. They are multifunctional proteins implicated in a variety of biological functions, including tumor cell adhesion, proliferation, differentiation, angiogenesis, cancer progression and metastasis. It was recently shown that Gal3 secreted by tumor cells induces T cell apoptosis 35. The expression of Gal3 has been positively correlated with the level of apoptosis of tumor-associated lymphocytes 36. Treatment with the Gal3 gene is also beneficial against asthma in mice 37. Finally, IL-24 is a member of the IL-10 family of cytokines 38. Over-expression of IL-24 induces apoptosis in cancer cells 39. Therefore, IL-24 appears to be an immunosuppressive cytokine.

A large number of surface molecules were highly upregulated in the CD8⁺ NKT-like cells, while a few surface molecules were down regulated (Table 5). Many of the upregulated molecules have been implicated in immunosuppressive function. Most notably, many of the genes are related to IFN-γ and some belong to the TNF family receptors and ligands. The expression patterns for these genes are clearly different among the cells cultured under different conditions or different cell types cultured under the same condition. The overall pattern seems to correlate well with their cellular functions. The genes already implicated in suppressive function or having suppressive potential were highly upregulated in the CD8⁺ NKT-like cells, which have potent suppression activity, while these genes were only moderately upregulated or unchanged in the cultured CD4⁺ T cells and activated CD8⁺ T cells, which have only weak or no suppression activity.

Ifitm1 (Interferon induced transmembrane protein 1) is the most upregulated surface molecule in the CD8⁺ NKT-like cells (90-fold increase compared to naïve CD8⁺ T cells; Table 5). This gene is not upregulated by the conventional activation protocol and upregulated to a much lesser degree in the cultured CD4⁺ T cells. Ifitm1 has been shown to be a key molecule in the anti-proliferative function of IFN-γ 40. Two other interferon-induced transmembrane genes (Ifitm2 and Ifitm3) were also highly upregulated in the CD8⁺ NKT-like cells (45- and 24-fold, respectively). It is highly likely that these proteins are involved in the suppressive function of the CD8⁺ NKT-like cells.

Lilrb4 (Leukocyte immunoglobulin-like receptor, subfamily B, member 4) is a member of the leukocyte immunoglobulin-like receptor (LIR) family. The encoded protein belongs to the subfamily B class of LIR receptors with a transmembrane domain, extracellular immunoglobulin domains, and cytoplasmic immunoreceptor tyrosine-based inhibitory motifs. The receptor expressed on immune cells binds to MHC class I molecules on antigen-presenting cells and transduces a negative signal that inhibits stimulation of an immune response. The receptor can also function in antigen capture and presentation. It may be involved in controlling inflammatory responses and cytotoxicity to help focus the immune response and limit autoreactivity. This gene was highly upregulated in both the CD8⁺ NKT-like cells and cultured CD4⁺ T cells.

Havcr2 (Hepatitis A virus cellular receptor 2), more commonly known as Tim3, was upregulated by 36-fold in the CD8⁺ NKT-like cells and 5-fold in the cultured CD4⁺ T cells compared to naïve CD8⁺ T cells. Tim3^-/- mice have exacerbated diabetes due partly to a defect in CD4⁺CD25⁺ Treg cell function 41. Therefore, Tim3 may be important for CD8⁺ NKT-like cell suppression function. Tnfrsf9, also known as 4-1BB and CD137, was highly upregulated in the CD8⁺ NKT-like cells (30-fold) and only slightly upregulated in the cultured CD4⁺ T cells (5-fold). 4-1BB is a costimulatory molecule that may be very important for Treg cell function. 4-1BB-primed CD8⁺ T cells possess suppressive function 42 and an agonist monoclonal antibody specific for 4-1BB can mitigate autoimmunity 4344454647. 4-1BB^-/- mice exhibit enhanced T cell proliferation 48. GITR (Tnfrsf18) is an important molecule for CD4⁺ Treg cell function. GITR was upregulated 5-fold in the CD8⁺ NKT-like cells and 10-fold in the cultured CD4⁺ T cells. Clearly, GITR could contribute to the suppressive function of both CD4⁺ and CD8⁺ Treg cells. Other upregulated costimulatory molecules, such as Pdcd1 (5.6-fold), PDL2 (4.6-fold) and Icos (4.3-fold), Ctla4 (5-fold) and CD28 (3.9-fold), may be good candidate molecules involved in the suppressive function. CD24a is one of the few costimulatory molecules that was downregulated in the CD8⁺ NKT-like cells (0.03-fold) but unchanged in the cultured CD4⁺ T cells (1.2-fold).

Genes for several TNF family proteins, such as Fas-L, RANKL, TRAIL and OX40, were all upregulated in the CD8⁺ NKT-like cells and the cultured CD4⁺ T cells. All these molecules could be implicated in the suppression function. Fas-L expression was 25-fold higher in the CD8⁺ NKT-like cells compared to naïve CD8⁺ T cells while CD4⁺ T cells cultured in the same condition did not upregulate Fas-L. Fas/Fas-L is one of the two pathways of lymphocyte-mediated cell killing 49.

Several lymphocyte receptors were upregulated - Gpr105 (65-fold), Ptger3 (13.1-fold), Ptger2 (4.3-fold), Fcer1g (13.7-fold) - and a few other receptors were downregulated by the expansion protocol (Table 4). Prostaglandin E receptor 3 (subtype EP3; Ptger3) may have pro-inflammatory or anti-inflammatory properties depending on the physiological condition 50. EP2, EP3 and EP4 receptors have been shown to be important for the immunosuppressive function of PGE2 51. Cytokine and chemokine receptors were up- or downregulated by the expansion protocol (Table 5). Notably, Il2ra (CD25) was highly upregulated in the CD8⁺ NKT-like and cultured CD4⁺ T cells as well as activated CD8⁺ T cells. These expected results were confirmed by FACS analysis (Figure 1). Ccr5 and Ccr2 were highly upregulated in the CD8⁺ NKT-like cells but unchanged or slightly upregulated in the cultured CD4⁺ T cells and activated CD8⁺ T cells. Ccr5 has been shown to be important for the suppression function of CD4⁺CD25⁺ T cells 52. CCR5 and several other CCR members (CCR2) were highly upregulated in the CD8⁺ NKT-like cells (Table 3) and they should be excellent candidate molecules for further studies.

A large number of adhesion molecules were highly upregulated in the CD8⁺ NKT-like cells and only slightly upregulated/unchanged or even decreased in the cultured CD4⁺ T cells (Table 5). Several of these genes (Tjp1, Emilin2, Nov) were confirmed by RT-PCR (Figure 4) and Sell (CD62L) was confirmed by FACS analysis (Figure 1).

Immunosuppressive cytokines can enhance the suppressive activity of Treg cells. For example, naturally occurring CD4⁺CD25⁺ Treg cells use a combination of IL-10 and TGF-β to suppress immune responses 5354555657. As IL-10 and IFN-γ are two key nodes of the molecular network modified in the CD8⁺ NKT-like cells (Figure 6), we further tested their role in the function and generation of these cells using two different approaches. First, an anti-IL-10 antibody was used in the suppression assay to determine whether IL-10 neutralization could block or reduce the suppression function of the CD8⁺ NKT-like cells. Blockade using anti-IL-10 could slightly reduce the suppression but could not completely block the suppressive function of the CD8⁺ NKT-like cells (data not shown). Since the levels of IL-10 and IFN-γ secreted by the CD8⁺ NKT-like cells was very high, antibody blocking may not have been efficient. Furthermore, the antibody blockade experiment evaluated only the role of these cytokines in the suppression function but did not allow us to assess the role of these molecules in the generation of the CD8⁺ NKT-like cells. Therefore, we tested the potential roles of IL-10 and IFN-γ in CD8⁺ NKT-like function/generation using IL-10^-/- and IFN-γ^-/- mice. CD8⁺ NKT-like cells can be cultured using naïve CD8⁺ T cells from both knockout mice; however, at the later stages of culture, the viability of cells from IFN-γ^-/- mice is not as good as those from wild-type mice and IL-10^-/- mice. Addition of IFN-γ in the culture medium can improve the viability of the cultured cells (Figure S2 in Additional data file 1). Although the CD8⁺ NKT-like cells cultured from both knockout mice had good suppressive activity at a high ratio (Tr/Tn = 1:1) of suppressor (CD8⁺ NKT-like) to responder (naïve CD4⁺ T cells), the reduction in suppressive activity for the CD8⁺ NKT-like cells cultured with naïve CD8⁺ T cells from both knockout mice became very clear at lower suppressor to responder T cell ratios (Figure 7b), which provides a more accurate estimate of the suppressive activity. At the 1:16 Tr/Tn ratio, the suppressive activity of CD8⁺ NKT-like cells cultured from IL-10^-/- mice was reduced to 30-35% of that for CD8⁺ NKT-like cells cultured from wild-type B6 mice (Figure 7b), suggesting that the vast majority of the suppressive activity of the CD8⁺ NKT-like cells can be attributed to the IL-10 pathway. Similarly, almost 50% of the CD8⁺ NKT-like suppression could be attributed to the IFN-γ pathway (Figure 7b). These results together suggest that IL-10 and IFN-γ play an important role in the in vitro generation and function of CD8⁺ NKT-like cells.

C57BL/6 (B6), B6.IL-10^-/- and B6.IFN-γ^-/- mice were purchased from Jackson Lab (Bar Harbor, ME, USA) and housed/bred under specific pathogen-free conditions at the Medical College of Georgia Animal Barrier Facility. This study was approved by the Medical College of Georgia IACUC committee.

CD8⁺ T cells were enriched from spleens by negative selection using an AutoMACS from Miltenyi Biotec (Auburn, CA, USA). The resulting CD8⁺ cell purity was around 90-95%. High purity CD4⁺ T cells were obtained by sorting splenic T cells using a Mo-Flow cytometer (Dako, Carpinteria, CA, USA). Flow cytometric analyses were performed on a FACScalibur™ flow cytometer with CELLQuest™ software (Becton Dickinson, Franklin Lakes, NJ, USA).

AutoMACS-purified naïve CD8⁺ or Mo-Flow sorted CD4⁺ T cells at 10⁶ cells/ml were cultured with anti-CD3 and anti-CD28 coupled to 4.5 mm paramagnetic Dynal beads (Invitrogen, Carlsbad, CA, USA) supplemented with 2,000 IU/ml rIL-2 (PeproTech, Rocky Hill, NJ, USA) in complete medium, which consisted of 10% heat-inactivated fetal bovine serum (Sigma-Aldrich, St. Louis, MO, USA), nonessential amino acids, 0.5 mM sodium pyruvate, 5 mM Hepes, 1 mM glutaMax I (all from Invitrogen, Carlsbad, CA, USA), and 55 μM mercaptoethanol (Sigma-Aldrich) in DMEM base. The cultures were monitored daily and maintained at 1-1.5 × 10⁶ cells/ml by diluting with IL-2-supplemented complete medium for 8-13 days. At the end of the culture, the anti-CD3 and anti-CD28 beads were removed using a Dynal MPC-L magnet (Dynal Biotech), and the cells were routinely assayed for CD8, CD4, CD62L, and CD25 expression by flow cytometry and for in vitro suppression assays.

Two different types of suppression assays were performed in this study. Most of the studies were performed using the CFSE system. Briefly, naïve CD4⁺CD25^- responder T cells were labeled with 2.5 μM CFSE. These labeled responder cells (1 × 10⁵) were cocultured at 37°C with different numbers of suppressor cells (CD8⁺ NKT-like cells, cultured CD4⁺ T cells or CD4⁺CD25⁺ Treg cells) in the presence of 1.5 × 10⁵ irradiated (2,000 rads) splenic APC (T cell depleted spleen cells) and 1.5 μg/ml anti-CD3 in a U-bottomed 96-well plate. After 72 h of culture, responder T cell proliferation was assessed by determining the dilution of CFSE using flow cytometry. The second type of suppression assay has identical culture conditions but the responder T cells were not labeled. The cultures were pulsed with 1 μCi/well [³H]thymidine for the last 16 h of 72 h culture and the level of proliferation was assessed by [³H]thymidine incorporation using scintillation counting after cell harvesting.

Cells were cultured in 96-well plates with 1 × 10⁵ CD4⁺CD25^- T cells/well, 1.5 × 10⁵/well irradiated splenic APC and 1.5 μg/ml anti-CD3 in the presence or absence of 1 × 10⁵/well CD8⁺ NKT-like cells or fresh CD4⁺CD25⁺ Treg cells. The level of IL-10, IFN-γ, IL-4 and IL-2 in the culture supernatant was determined by ELISA kits purchased from R&D Systems (Minneapolis, MN, USA).

Total RNA was extracted from cultured or fresh T cells using the Qiashredder column and RNeasy Mini kit (Qiagen Inc., Valencia, CA, USA). All RNA extracted was analyzed for quantity and quality using the Agilent 2100 Bioanalyzer system (Agilent Technologies, Palo Alto, CA, USA). Gene expression profiling was performed using the mouse genome 430 2.0 chips (GeneChip™, Affymetrix, Santa Clara, CA, USA). An aliquot of 1 μg of total RNA was converted into double-stranded cDNA (ds-cDNA) by using SuperScript Choice System (Gibco BRL Life Technologies, Carlsbad, CA, USA) with an oligo-dT primer containing a T7 RNA polymerase promoter (Genset, San Diego, CA, USA). After second-strand synthesis, the reaction mixture was extracted with phenol-chloroform-isoamyl alcohol, and ds-cDNA was recovered by ethanol precipitation. In vitro transcription was performed on the above ds-cDNA using the Enzo RNA transcript Labeling kit. Biotin-labeled cRNA was purified by using an RNeasy affinity column (Qiagen), and fragmented randomly to sizes ranging from 35-200 bases by incubating at 94°C for 35 minutes. The hybridization solutions contained 100 mM MES [2-(N-morpholino)ethanesulfonic acid], 1 M Na⁺, 20 mM EDTA, and 0.01% Tween 20. The final concentration of fragmented cRNA was 0.05 μg/μl in hybridization solution. Target for hybridization was prepared by combining 40 μl of fragmented transcript with sonicated herring sperm DNA (0.1 mg/ml), bovine serum albumin and 5 nM control oligonucleotide in a buffer containing 1.0 M NaCl, 10 mM Tris.HCl (pH7.6), and 0.005% Triton X-100. Target was hybridized for 16 h at 45°C to a set of oligonuceotide arrays (Affymetrix). Arrays were then washed at 50°C with stringent solution, then again at 30°C with non-stringent washes. Arrays were then stained with streptavidin-phycoerythrin (Invitrogen). DNA chips were read at a resolution of 3 μm with a Hewlett-Packard GeneArray Scanner and were analyzed with the GENECHIP software (Affymetrix GCOS 1.1). Both the CEL and DAT files for each hybridization have been uploaded to our server running GeneTraffic v3.2 (Iobion Informatics LLC, La Jolla, CA, USA).

Microarray data were first normalized using RMA 68 and normalized data were subsequently analyzed using the LIMMA 69 package in R. All groups were compared pairwise, and the resulting p-values were adjusted using the pFDR of Storey 70 and the qvalue package in R. We considered all probesets with a q-value ≤0.01 as being significant. The heatmap was constructed using the heatmap.2 function in R. Differentially expressed genes showing more than five-fold change between CD8⁺ NKT-like and naïve CD8⁺ T cells were annotated with respect to their involvement in biological processes and pathways using the PANTHER (Protein ANalysis THrough Evolutionary Relationships) classification system 71. Of the 314 genes, biological processes for 257 were found by PANTHER. One-tailed p-values for enrichment of particular biological processes were obtained using the standard Fisher's exact test, to determine if the observed number of counts exceeded the expected counts. Gene lists from each significantly enriched (p < 0.05) biological process were further analyzed by pathway analysis. A network depicting the molecular interactions between the given set of genes was constructed using Pathway Studio Version 5.0 software and ResNet mammalian database (Ariadne Genomics, Rockville, MD, USA). ResNet is a very large database of molecular interactions and biological relationships extracted from the biomedical literature. The found interactions between the given set of genes were manually curated by reading sentences from which the relationship or interaction was derived. Only manually curated interactions were used for the final pathway visualization.

An aliquot of total RNA (2 μg per sample) were arrayed in 96-well plates and then converted to cDNA using a High Capacity cDNA Reverse Transcription Kit (Applied Biosystems, Foster City, CA, USA) and a programmable thermal cycler (Applied Biosystems). The cDNA products were diluted and an aliquot of cDNA equivalent to 10 ng total RNA was used for quantitative RT-PCR performed using ready-to-use TaqMan gene expression assays from Applied Biosystems. 18srRNA and GAPDH were used as endogenous controls for normalizing RNA concentration. RT-PCR was performed in 384-well plates with the ABI 7900HT Fast Real-Time PCR System (Applied Biosystems). Standard thermal cycling conditions (10 minutes at 95°C, 40 cycles for 15 s at 95°C, 1 minute at 60°C) was used for all genes. All samples were analyzed on the same plate and each sample was analyzed in duplicate. Cycle threshold (C_T) values for each test gene and 18SrRNA and GAPDH were obtained for each sample using the SDS2.3 and analyzed with RQ Manager 1.2 software (Applied Biosystems). Differences in C_T values between a test gene and 18srRNA (ΔC_T) for each sample were calculated and used for statistical analyses.