|Home | About | Journals | Submit | Contact Us | Français|
Previously we showed that CD11c defines a novel subset of CD8+ T cells whose in vivo activity is therapeutic for arthritis; however, the mechanisms directing their development, identity of their precursors, and basis of their effector function remain unknown. Here we show that the novel subset develops from CD11csurface-CD8+ T cells and undergoes robust expansion in an antigen- and 4-1BB (CD137)-dependent manner. CD11c+CD8+ T cells exist in naive mice (<3%) with limited suppressive activity. Once activated, they suppress CD4+ T cells in vivo and in vitro. Suppression of CD4+ by CD11c+CD8+ T cells is related to an increase in IDO activity induced in competent cells via the GCN2 pathway. CD11c+CD8+ T cells are refractory to the effect of IDO but constrict in a novel 1-MT-dependent mechanism resulting in reversal of their suppressive effects. Thus, our data uncover, for the first time, the origin, development, and basis of the suppressive function of this novel CD11c+CD8+ T cell subpopulation that has many signature features of Tregs.
Regulatory T cells (Tregs) exert a protective role in ameliorating autoimmune diseases, allergic disorders, and transplant rejection by targeting the activity of effector T cells, thereby maintaining immune homeostasis. The best-studied Tregs include the CD4+CD25+ T cells but other cell types with regulatory functions have also been reported . Although CD8+ T cells were the first cells found to have immunosuppressive ability , comprehensive understanding of their cellular and molecular mechanisms has been impeded by a lack of defining markers. Recently, however, great advances have been made in establishing the regulatory function of CD8+ T cell subsets [3-5].
Recently, anti-4-1BB immunotherapy has attracted much attention due to its seemingly diverse roles . Whereas in vitro treatment of T cells by anti-4-1BB preferentially supports CD8+ over CD4+ T cells , in vivo anti-4-1BB administration suppresses many cell types including CD4+ T and B cells [8-12]. Precisely how in vivo actions of anti-4-1BB treatment target a particular cell type is not fully resolved. Based on models tested, increased IFN-γ [9,11,13], TNF-α , TGF-β [14,15], and IDO [8,10] have been identified as candidate molecules involved in the anti-4-1BB-mediated regulatory pathway. The search for a common immunosuppressive pathway or a candidate (if any), however, remains unfulfilled.
Previously we have identified a novel inducible form of a CD8+ T cell subset expressing the CD11c molecule associated with regulatory functions aided by IDO and IFN-γ as central to anti-4-1BB-mediated therapeutic effects . Recently further evidence for this concept has been reported [8,16]. However, in depth understanding of CD11c+CD8+ T cell-mediated immunosuppressive activity remain elusive, as are the origins, developmental/activation requirements, and other immunological characteristics of these cells.
We report here our findings on the development of CD11c+CD8+ T cells, factors guiding their differentiation into full-fledged immune regulators, and the molecular basis of their suppressive function. Our results suggest that these cells evolve from CD11csurface-CD8+ T cells and acquire significant immunosuppressive capacity when co-stimulated by 4-1BB in an IDO-and GCN2-dependent manner. These data expand our understanding of CD11c+CD8+ T cell development, differentiation, and homing events, and provide insight into the mechanisms by which these Tregs exert their effects.
We first determined how CD11c+CD8+ T cells develop, in that our previous studies [8,10,16] did not reveal much information concerning their precursors or the conditions favoring their expansion. We found that a small percentage (<3%) of freshly isolated CD8+ T lymphocytes (CD8β+) expressed surface CD11c in naïve mice (Fig. 1A). Absence of staining with 33D1 or DC-SIGN (Fig. 1A) confirmed that the observed dual staining was not the result of CD8+ T-DC conjugation. The authenticity of CD11c staining in this experiment was confirmed by separately incubating cells with an unlabeled anti-CD11c (clone N418) mAb for 60 min on ice before staining with the same clone of phycoerythrin-labeled anti-CD11c (Data not shown). Calculations revealed about 1.22 × 103 ± 0.09 CD11c+CD8+ T cells per young adult spleen; the highest numbers were seen in bone marrow (Fig. 1B). These cells displayed a naive phenotype as judged by CD25, CD44, and CD62L expression patterns (data not shown) and showed cell division (less than CD11c-CD8+ T cells) when stimulated in vitro with a combination of CD3 and 4-1BB Abs (Fig. 1C). Phenotypic analysis of gated CD11c+CD8+ and CD11c-CD8+ T cells from naïve mouse spleens showed increased expression of intergrins such as CD11b, CD18, enhanced CD28, CD36, and comparable basal CD11a, CD25, CD103, CD122, and Foxp3 (Fig. 1D).
To determine the origin of CD11c+CD8+ T cells — whether they were differentiated from existing precursor CD11c+CD8+ T cells (Fig. 1A; left panel) or induced from conventional CD8+ T cells (CD11csurface-) to acquire surface CD11c — we conducted precursor analysis. CD8+ T cells from OT-I mice without further purification (Fig. 2A; upper panels) or after removal of CD11c+ cells (Fig. 2A; lower panels) were either cultured in vitro (Fig. 2B) or adoptively transferred into syngeneic 4-1BB-/- mice (Fig. 2C). Stimulation was achieved by adding SIINFEKL-pulsed APCs and anti-4-1BB (Fig. 2B) or administration of SIINFEKL/anti-4-1BB (Fig. 2C). After 6 days in culture (Fig. 2B) or 6 days after adoptive transfer (Fig. 2C), both total CD8+ T cells as well as CD11csurface-CD8+ T cells (Figs. 2B, C) produced CD11c+CD8+ T cells at similar levels. The marginally increased CD11c+CD8+ proportion in cultures containing total CD8+ (49.4%; Fig. 2C; upper far right panel), compared to levels achieved by CD11c-CD8+ cultures (38.3%; Fig. 2C; lower far right panel), appeared to be due to an extra 2.51% CD11c+CD8+ T cells in the total CD8+ cell fraction at initiation of culture/adoptive transfer. The 4-1BB-/- recipients were used in the experiment shown in Fig. 2C to ensure that administered anti-4-1BB bound only to adoptively transferred 4-1BB+ OT-I CD8+ T cell subsets.
To determine if anti-4-1BB signaling is a limiting factor in generating CD11c+CD8+ from CD11c-CD8+ T cells and whether other signals influence such expansion, various agents such as IL-2, IL-7, IL-15, PMA, ionomycin, and anti-CD28 either alone or in combination (over a wide range of dilutions) were added to CD11csurface-CD8+ T cell cultures with plate-bound anti-CD3 as signal 1. Although gain in surface CD11c expression by CD8+ T cells was noted in all cultures (albeit generally low), anti-4-1BB costimulated cultures showed highly increased CD11c surface expression (Figs. 2D, E), suggesting that 4-1BB signals play a critical role in the differentiation of CD11c+CD8+ from CD11c-CD8+ T cells.
To resolve whether CD4+ T cell loss in anti-4-1BB injected mice [8-11,14,15] is due to downstream receptor effects on CD4+ or occurs in a bystander mode mediated by CD11c+CD8+, we used a T-cell Ag-specific mouse model described in Fig. 2C. We found that markedly increased CD11c+CD8+ T cell proportions in mice given SIINFEKL and anti-4-1BB correlated with severely reduced CD4+ T cell numbers (Fig. 2F; lower left and right panels; Fig. 2G). That we transferred only 4-1BB+/+CD8+ (OT-I) cells and that reduced CD4+ T cells belong to recipient 4-1BB-/- mice (which are refractory to anti-4-1BB receptor mediated effects) suggest that these events occur in a bystander mode mediated by CD11c+CD8+ T cells. To understand if anti-4-1BB activated CD11c+CD8+ T cells share the phenotype of certain established regulatory CD8+ T cell subsets, we performed additional analyses. Purified CD11c+CD8+ T cells from spleens of SIINFEKL/anti-4-1BB group showed increased CD103, CD122, Fas, perforin, and granzymeB expression over control CD11c-CD8+ T cells purified from the same animal (Fig. 2H).
To verify if the minor CD11c+CD8+ T cells found in naive mice (Fig. 1A; left panel) also possess regulatory activity that we observed with in vivo generated counterparts (Fig. 2F), we purified CD11c+CD8+ and CD11c-CD8+ T cell subsets from naive mice (Fig. 3A) and performed co-culture assay using syngeneic CD4+CD25- T cells as responders. As shown in Fig. 3B, CD11c+CD8+ T cells inhibited proliferation of CD4+CD25- T cells on stimulation with anti-CD3 and anti-4-1BB Abs but only at higher effector : responder (1:1) ratios. In contrast, control CD11c-CD8+ T cells did not exhibit a suppressive effect at any of the effector : responder ratios tested (Fig. 3B). To address whether prior activation increases their suppressive activity, we purified CD11c+CD8+ and CD11c-CD8+ T cells from spleens of mice treated 7 days previously with anti-CD3/anti-4-1BB and added them to in vitro cultures containing freshly isolated naive syngeneic CD4+CD25- T cells (Fig. 3C). The dose of in vivo anti-CD3 used in this experiment was pretitrated to obtain optimal T cell activation without cell depletion. Flow cytometric analysis of 3-day cultures showed dose-dependent and significantly reduced CD4+ T cell proportions (3.83-, 2.64-, and 1.5- fold decrease at 1 : 1, 1 : 0.5, and 1 : 0.25 effector : responder ratios respectively; Fig. 3D). In contrast, addition of CD11c-CD8+ to CD4+ T cell cultures did not exhibit any significant suppressive activity except for a 2.13-fold decrease at 1 : 1 effector : responder ratios (Fig. 3E). Collectively, these data suggest that freshly isolated CD11c+CD8+ T cells, although they have a mild regulatory phenotype, are enhanced several-fold on activation in vivo by Ag and anti-4-1BB.
In the experiment described in Figs. 3D and E, we used naïve CD4+ T cells as targets in coculture assays driven by anti-CD3-mediated polyclonal activation. This raises the question that suppression of naïve CD4+CD25- by CD11c+CD8+ T cells is biased as only one of the interacting partners is activated while the other is not. To address the issue if CD11c+CD8+ can also suppress activated CD4+ T cells, we developed a unique in vivo assay system where both CD4+ and CD8+ T cells receive equal in vivo Ag-specific signals. Purified CD4+CD25- T cells from OT-II and purified CD8+ T cells from OT-I mice when adoptively transferred together at equal numbers into syngeneic mice 4-1BB-/- recipients, only mice receiving SIINFEKL and anti-4-1BB but not ISQAVHAAHAEINEAGR and anti-4-1BB showed suppression of both donor-derived (Vα2+) as well as host derived (Vα2-) CD4+ T cells (Fig. 4; compare upper and middle panels) despite presence of OT-I specific CD8+ T cells in all cases. When SIINFEKL and ISQAVHAAHAEINEAGR peptides were given together to induce activation signals in both OT-I CD8+ and OT-II CD4+ T cell populations, treatment with anti-4-1BB mAb again caused suppression of CD4+ T cells (Fig. 4; lower panels). The latter experiment strongly suggested that anti-4-1BB-mediated CD11c+CD8+ T cell directed suppression of CD4+ T cells occurs irrespective of the latter's state of activation.
To explore further the regulatory activity of CD11c+CD8+ T cells in a clinical setting, we relied on recently reported findings that treatment with anti-4-1BB attenuates disease symptoms associated with hapten-induced colitis  and that adoptive transfer of CD4+ T cells from 4-1BB-/- mice into SCID mice aggravated symptoms and increased mortality . We sorted CD11c-CD8+ and CD11c+CD8+ T cell subsets (from mice previously treated with TNBS/anti-4-1BB) and adoptively transferred them into colitis-prone 4-1BB-/- mice to compare their regulatory functions (Fig. 5A). To increase the efficacy of transferred CD11c+CD8+ T cells, the mice were treated with anti-4-1BB. Reversal of mortality due to colitis in 4-1BB-/- mice (Fig. 5B) confirmed the regulatory activity of transferred CD11c+CD8+ T cells. Coinciding with reduced mortality, a 10-fold loss of CD4+ T cells in mice receiving adoptively transferred CD11c+CD8+ T cells was seen (Fig. 5C; compare panels second from left with far right), reinforcing the data described in Figs. 2F and 3D,E showing that these cells suppress CD4+ T cells. Analysis of mice receiving CD11c-CD8+ T cells, although they showed a decrease in CD4+ T cell proportions (Fig. 5C; compare panels left with third from left), showed a reversal in mortality that was not as striking as that seen in mice receiving CD11c+CD8+ T cells (Fig. 5B). The 2-fold loss of CD4+ T cells in mice receiving CD11c-CD8+ T cells in this experiment appeared to be due to expansion of CD11c+CD8+ T cells from CD11c-CD8+ T cells as was shown to occur in Figs. 1A-C (lower panels). Given the importance of CD4+ T cells in TNBS-induced colitis [19-21] and since CD11c+CD8+ therapy increased survival of 4-1BB-/- mice coinciding with a massive 10-fold loss of CD4+ T cells (Fig. 5C; compare panels second from left with far right), we looked for increased CD4+ T cells in these mice. Analysis of TNBS-treated 4-1BB-/- mice showed about 3-fold higher proportions of CD4+ T cells when compared to ethanol-treated controls (Fig. 5C).
Besides TGF-β , increased IFN-γ [8,9,11] is the hallmark of in vivo anti-4-1BB therapy. Microarray analysis of CD11c+CD8+ T cells, obtained from model described in Figure 2C, also revealed increased expression of IFN-γ, among others, in CD11c+CD8+ T cells (Supplemental Figs. S1A-C). We previously suggested that anti-4-1BB-mediated IFN-γ induces IDO whose downstream effects suppress CD4+ T cell number and function .
Since loss of host-derived CD4+ T cells was observed in 4-1BB-/- recipients receiving wild-type CD8+ T cells and anti-4-1BB treatment (Figs. 2F,G), we evaluated if IDO pathway is involved in this passive model. CD11c+ DCs purified from the 4-1BB-/- mice treated with OT-I CD8+ T cells/SIINFEKL/anti-4-1BB showed increased IDO expression (Fig. 6A) which may have resulted in a bystander manner. To understand whether suppressive effects mediated by CD11c+CD8+ T cells are IDO dependent, we looked at CHOP (GADD 153) expression. Levels of CHOP rise due to nutritional deficiency, as might occur when tryptophan levels are affected by increased IDO . Since proportions of CD4+ T cells were highly reduced in anti-4-1BB treated mice; we studied CHOP expression in the remainder of these cells. As shown in Fig. 6B, CHOP expression in the placebo group was increased over the control and 1-MT-treated groups. Coinciding with reduced CHOP expression in the 1-MT group (Figs. 6B), significant restoration of CD4+ T cell proportions (8.3-fold recovery; Fig. 6C; upper and lower panels) was noted. Interestingly, reversal of CD4+ T cell suppression in these mice, due to 1-MT, correlated with significantly decreased proportions of CD11c+CD8+ T cells (Compare 17.5% in placebo vs. 6.28% in 1-MT group; Fig. 6C). This effect of 1-MT was only noted in mice receiving both SIINFEKL and anti-4-1BB but not in those receiving SIINFEKL alone (Fig. 6C), suggesting that 1-MT treatment affects CD11c+CD8+ expansion in an Ag-dependent manner. To further understand the role of IDO in anti-4-1BB/CD11c+CD8+-mediated suppression of CD4+ T cells, we evaluated IDO-/- and GCN2-/- mice. As observed in experiments involving IDO blockade by 1-MT, mice lacking endogenous IDO also significantly protected CD4+ proportions from the effects of CD11c+CD8+ cells (Fig. 6D). The transcription factor GCN2 is located downstream of IDO and mice lacking endogenous GCN2 are impervious to IDO-mediated effects [22,23]. When GCN2-/- mice were adoptively transferred with OT-I CD8+ T cells and additionally treated with SIINFEKL and anti-4-1BB, also did not show suppression of CD4+ T cell proportions (Fig. 6D). Taken together these results suggest that anti-4-1BB-mediated suppression of CD4+ T cells is critically dependent on IDO and GCN2 pathways and is restored significantly when CD11c+CD8+ cells constrict (in this case by 1-MT) or when the expression of above molecules is lacking.
To further understand the reasons why treatment with 1-MT constricted CD11c+CD8+ T cells (Fig. 6C; compare panels fourth from left and fifth from left), we looked into the possibility that DC function is affected by such treatment. To precisely study the effect of 1-MT on DCs, we slightly modified the protocol developed by Aguague et al . In our model (Fig. 7A), purified splenic DCs were first exposed to 1-MT or anti-4-1BB, pulsed with SIINFEKL, mixed with freshly isolated OT-I CD8+ T cells, and adoptively transferred into fresh syngeneic 4-1BB-/- mice. In vitro pulsing of DCs with SINNFEKL rather than injection was used to avoid diffusion of injected SIINFEKL among APCs in vivo and obscure our aim, i.e., to study the effects of 1-MT mainly on Ag-bearing DCs and to recognize the consequences on adoptively transferred and partnering Ag-specific CD8+ T cells. We found that DCs pre-exposed to 1-MT (but not unmanipulated or anti-4-1BB) when adoptively transferred, greatly reduced differentiation/expansion of CD11c+CD8+ T cells (compare 5.65% in 1-MT group with 39.6% in control group: Fig. 7B, upper panels). Correlating with decreased expansion of CD11c+CD8+ T cells, proportions of CD4+ T cells in these mice recovered significantly (compare 10.4% in 1-MT group with 0.94% in control group; Fig. 7B, lower panels). These results show that 1-MT alters DC function which presumably might have affected their capacity to optimally prime CD8+ T cells such that the expansion and suppressive function of the CD8+ T cells are impeded.
Our previous data showed that anti-4-1BB stimulation is key for CD11c+CD8+ T cell induction [8,10] but it remained unclear at what stage and in which CD8+ subpopulation anti-4-1BB triggers development toward the novel CD11c+CD8+ T cell phenotype. We found that naturally occurring CD11c+CD8+ T cells (<3%) in naive mice are unlikely to be precursors of CD11c+CD8+ T cells. These naïve CD11c+CD8+ T cells expressed differential levels of various integrins and other molecules including a few that are shared by certain established classes of regulatory CD8+ T cells . Supporting the above idea is our observation that CD11csurface-CD8+ T cells successfully differentiated into CD11c+CD8+ T cells and that the naturally existing CD11c+CD8+ T cells (which do not expand further despite inflammation) might have been generated de novo. The parallel development of CD11c-CD8+ and CD11c+CD8+ T cells after anti-4-1BB stimulation signifies an important role for these subsets in immune responses because CD11c+CD8+ T cells but not CD11c-CD8+ T cells restricted CD4+ T cell expansion and reversed colitis-associated mortality. It is unclear why there are functional differences between these two CD8+ subpopulations, considering that one of them is the precursor of the other. Microarray analysis failed to identify an exclusive marker to distinguish CD11c+CD8+ from CD11c-CD8+ T cells although 320 genes underwent significant change in CD11c+CD8+ over CD11c-CD8+ T cells (Supplemental Figs. S1A-C). Phenotypic analysis of purified CD11c+CD8+ although co-expressed molecules previously been shown to present on certain regulatory CD8+ T cell subsets such as CD103, CD122, Foxp3 etc , we strongly believe that these cells are distinct because of their dependence on 4-1BB signal to expand and acquire regulatory properties. Interestingly, although the anti-4-1BB signal expands CD11c+CD8+ T cells, endogenous 4-1BB does not appear to be important for their development, in that 4-1BB-/- mice showed baseline levels of these cells which expanded further (only moderately) during inflammation (SIINFEKL alone or TNBS alone groups) but without associated regulatory properties. Thus, expansion of CD11c+CD8+ T cells and their acquisition of regulatory activity is a tightly governed process and unless activated by Ag and anti-4-1BB (but not by cytokines or anti-CD28), these cells do not exert a suppressive function.
Our findings and conclusions about the critical suppressive function of CD11c+CD8+ T cells do not detract from our previous finding [8,10] that IDO is critical for this activity as shown by 1-MT treatment. IDO is an enzyme catalyzing the initial and rate-limiting step in the catabolism of tryptophan along the kynurenine pathway . Increased IDO expression in APCs correlates with minimal T cell proliferation, enhanced apoptosis, and weak responses in vivo . With respect to our previous suggestion that IDO-mediated CD4+ T cell suppression occurs when they collaborate with Ag-bearing IDO+ cells (10), our current results are somewhat different in that lost CD4+ T cells belonged to recipient 4-1BB-/- mice (as we transferred only 4-1BB+/+ OT-I CD8+ T cells). In this scenario, it is unlikely that 4-1BB-/-IDO+-derived APCs present SIINFEKL peptide to incompatible CD4+ T cells to cause their deletion. Thus, the observed loss of CD4+ T cells caused by increased IDO in mice treated with SIINFEKL/anti-4-1BB is likely due to depleted tryptophan levels in that IDO blockade resulted in significant recovery of these cells. Also, the nutrient depletion indicator CHOP (also an indicator of GCN2 activity; 22, 23), which is up regulated when levels of IDO rise [22, 23], was found to be elevated in placebo-treated, CD4+ T cell fraction, suggesting that they are poised for death apparently due to the tryptophan deficiency induced by increased IDO. Experiments involving IDO-/- and GCN2-/- mice, where CD4+ T cell proportions were protected, further confirm the critical roles these molecules play in anti-4-1BB/CD11c+CD8+ T cell-mediated regulatory effects.
For reasons that are not entirely clear, the increased IDO expression did not affect the proportions of CD11c+CD8+ T cells, signifying their resistance to IDO-mediated effects. Equally perplexing was the finding that suppression of IDO activity by 1-MT reduced CD11c+CD8+ T cells. This seemingly dual nature of 1-MT in the suppression of both function and proportions of CD11c+CD8+ T cells and protection of CD4+ T cells is not currently clear. Taken together, it is compelling to assume that CD11c+CD8+ T cells survive better under conditions of increased IDO but succumb when its activity is inhibited. However, we reject this notion and consider that an as yet unidentified signal is responsible for the loss of CD11c+CD8+ T cells in the 1-MT treated group. That IDO-/- or GCN2-/- mice showed normal expansion of CD11c+CD8+ T cells (data not shown) confirms that endogenous IDO or GCN2 are not directly responsible for their loss, and substantiates our above belief. The possibility that 1-MT treatment “altered” APC function independent of IDO  affecting CD11c+CD8+ T cell expansion and function cannot be ruled out. Therefore “altered DCs”, due to 1-MT treatment, on SIINFEKL uptake and subsequent interaction with Ag-specific CD8+ T cells (OT-I here), might have led to CD11c+CD8+ T cell tapering or non-development. We found this expectation to be true as production of CD11c+CD8+ T cells in mice treated with 1-MT primed-SIINFEKL loaded DCs was significantly reduced, whereas CD4+ numbers were largely recovered. How exactly 1-MT affects DC function at the molecular level, whether such treatment transforms these cells into less potent entities, and the circumstances leading to ablation of CD11c+CD8+ T cell numbers/function are under investigation in our laboratories. Taken together, we propose that CD11c+CD8+ T cells support production of IDO such that in an ensuing nutritionally (tryptophan) deficient milieu, the CD4+ T cells constrict in a bystander mode rather than Ag bearing-IDO+ APCs collaborating with CD4+ T cells to cause their deletion. We further postulate that CD11c+CD8+ T cells are impervious to IDO activity but themselves are sensitive to 1-MT effects mediated by Ag bearing “altered DCs.”
In summary, these results provide the first molecular description of a precursor population that differentiates into CD11c+CD8+ T cells, conditions favoring development of a suppressive function, and the basis of their immunoregulatory effects. There are, however, still many questions concerning the molecular function of CD11c+CD8+ T cells. We do not know the signaling events linking the downstream 4-1BB receptor actions and the exclusive nature of these cells and to what extent the CD11c+CD8+ T cells demonstrate plasticity in their regulatory profiles when conditioned in different stimulatory environments. Advances in understanding of how CD11c+CD8+ T cells regulate suppressive functions may help develop novel therapeutic avenues to target many intractable inflammatory and malignant diseases.
All animal experiments have been approved by the Louisiana State University Health Sciences Center Animal Facility and the Immunomodulation Research Center Institutional Animal Care and Use Committees.
BALB/c and C57BL/6 (B6) mice were purchased from the National Cancer Institute (Frederick, MD). The 4-1BB-/- mice have been described . The OT-I (H-2kb-restricted) TCR transgenic mice were a gift from Dr. Michael Croft (La Jolla Institute for Allergy and Immunology, San Diego, CA). The IDO-/- and GCN2-/- mice were gifts of Dr. Andrew L. Mellor (Medical College of Georgia, Augusta, GA) and Dr. David Ron (The Skirball Institute New York School of Medicine, New York, NY).
Unless otherwise stated, all antibodies were purchased from eBioscience (San Diego, CA). FcR block (clone 2.4G2; produced in house) was added to all surface staining mixtures. Flow cytometry was performed using a FACS Calibur (BD Biosciences, San Diego, CA).
Single cell suspensions of spleens and/or lymph nodes were prepared and passed through nylon-wool columns (Polysciences, Inc., Warrington, PA). The nylon-wool non-adherent cells were loaded on a density gradient (Optiprep, Oslo, Norway) to separate low-density from high-density cells. The high-density cell fraction (pellet) collected from the gradient was incubated with biotin-labeled Abs to TER-119, CD4, B220, 33D1, and DC-SIGN (eBioscience). After an additional step involving anti-biotin microbeads (Miltenyi Biotec, Auburn, CA), cells were passed through magnetic columns (Miltenyi Biotec). The unbound cell fraction was collected and further enriched using paramagnetic CD8 beads using the manufacturer's guidelines (Dynal Mouse CD8 kit, InVitrogen, Carlsbad, CA). The positively selected CD8+ T cells were separated from the magnetic beads by incubation in “bead release” buffer provided in the kit (InVitrogen). The bead-free CD8+ T cells were incubated with anti-CD11c microbeads (Miltenyi Biotec) followed by separation using magnetic columns (Miltenyi Biotec). The flow-through (CD11c-CD8+) and column-bound (CD11c+CD8+) fractions were collected and used for various down-stream applications. In some cases, partially purified total CD8+ T cells (InVitrogen) were labeled with fluorochromes and sorted (FACS Aria, BD Biosciences). For in vitro assays, purified CD8+ T cell subsets (1 to5 × 105/ml) were cultured in 48-well tissue culture plates (Costar, Cambridge, MA). Activation was started by adding the indicated agonists.
A total of 5 × 106 purified CD8+ T cells from OT-I mice were injected into the tail veins of wild-type or syngeneic 4-1BB-/- or IDO-/- or GCN2-/- mice along with 0.5 mg SIINFEKL (OVA257-264 peptide; AnaSpec, Inc., San Jose, CA; purity >90%). On the day of peptide administration and the day after that, mice were given i.v. injections of 100 μg of anti-4-1BB (rat IgG1, clone 3E1; a gift of Dr. Robert S. Mittler, Emory Vaccine Center, Atlanta, GA). Six to seven days after the initial treatment, the mice were killed and single cell suspensions of spleens were prepared and subjected to flow cytometry.
CD11csurface-CD8+ T cells were purified as described and seeded in 24-well tissue culture plates (Costar) coated with anti-CD3 (10 μg/ml; eBioscience). Where indicated, cultures were supplemented with graded concentrations of recombinant IL-2, IL-7, and IL-15 (all from Peprotech, Rocky Hill, NJ), PMA (50 ng/ml; Sigma-Aldrich, St. Louis, MO), ionomycin (500 ng/ml; Sigma-Aldrich), anti-CD28 (2 μg/ml; eBioscience), and anti-4-1BB (5 μg/ml). Seventy three hours later, the cells were washed and used.
In vivo CD11c+CD8+ T cells were generated by administering i.v. anti-4-1BB (100 μg/mouse/day × 2 days) either alone or with functional grade anti-CD3 (clone 145.2C11; 500 ng; eBioscience). On the day of the experiment, splenic CD8+ T cell subsets were purified and studied.
To induce colitis, 0.5 mg of hapten reagent TNBS (Sigma-Aldrich) in 100 μl 50% ethanol (to break the intestinal epithelial barrier) was slowly administered into the lumen of the colon via a catheter fitted onto a 1-ml syringe on days 1 and 7 . Tissues and cells were assessed three days later (day 10). Control mice received 50% ethanol (EtoH) alone as described above. Some mice were treated i.p. with cIg or anti-4-1BB on days 1 through 4 and once on day 7 (with respect to the TNBS enema) at a dose of 100 μg/mouse. Mortality due to colitis was monitored for 10 days. At the end of the experiment, tissues, cells, and blood samples were collected from treated mice for additional analyses.
Cells (1 × 106) were first surface stained with PE-labeled anti-mouse Vα2 followed by fixation and permeabilization for 30 min at 4°C. Cells were subsequently incubated with anti-CHOP (0.2 μg; Catalogue No. sc-7351; Santa Cruz Biotechnology, Santa Cruz, CA). Bound antibody was revealed by incubation with anti-mouse IgG1-FITC (BD Biosciences). To prevent anti-mouse IgG1-FITC binding to endogenous IgG1 that may have been produced due to in vivo treatment, the permeabilized cells were preincubated with purified anti-mouse IgG1 (50 μg; 60 min; Bethyl Laboratories, Inc., Montgomery, TX) before addition of anti-CHOP antibody.
The CD4+ T cells (from naive B6 mice) were purified using T cell enrichment columns (R and D Systems, Minneapolis, MN). The negatively selected CD4+ T cells were processed further to remove CD25+ cells using immunomagnetic beads (Miltenyi Biotec). The CD11c+CD8+ and CD11c-CD8+ T cells were purified from naive B6 mice as described above. For co-culture assay, 5 × 103/well purified CFSE-labeled (10 μM; In Vitrogen) CD4+CD25- T cells and CD11c+CD8+ or CD11c+CD8+ T cells were used at different effector : responder ratios in 48-well tissue culture plates coated with anti-CD3 (10 μg/ml). Costimulation was started by adding soluble anti-4-1BB (5 μg/ml). On the indicated days, cells were washed from cultures and flow cytometry was performed.
CD4+CD25- and CD8+ T cells from OT-II and OT-I mice were purified by protocols described elsewhere in the section. Purified OT-I CD8+ and OT-II CD4+CD25- T cells (5 × 106 each cell type) were injected i.v. into sex- and age-matched syngeneic 4-1BB-/- recipients. The injected mice were divided into three groups. To group one was injected SIINFEKL (0.5 mg; i.v.), ISQAVHAAHAEINEAGR (0.5 mg; i.v.) to the second group, and a mixture of SIINFEKL and ISQAVHAAHAEINEAGR (0.5 mg each peptide; i.v.) to the third group. All groups received anti-4-1BB (100 μg/day × 2 days; i.v.). Seven days later, mice were sacrificed, spleens excised, single cell suspensions prepared, and flow cytometry performed.
Spleens were minced and digested with two 15-min incubations at 37°C with collagenase IV (Sigma-Aldrich), pelleted, resuspended in 35% BSA (Sigma-Aldrich), overlaid with 2 ml of RPMI-1640, and centrifuged for 20 min at 7,500 g. The low-density floating fraction was collected, washed, and incubated for 90 min in a petri dish. The non adherent cells were removed by pipetting. The monolayer was incubated for an additional hour to remove additional small round lymphocytes. After overnight culture, most DCs became non adherent and were harvested. For further purification and enrichment of DCs, cells were isolated using anti-CD11c microbeads (Miltenyi Biotec). Purified DCs were cultured for 24 hr in the presence of 1-MT (1 mM) or anti-4-1BB (5 μg/ml). DCs were then washed and loaded with SIINFEKL (0.5 mg; 6 hr/37°C/5% CO2). After washing to remove unbound SIINFEKL, the cells were mixed with equal numbers of freshly isolated OT-I CD8+ T cells (5 × 106), and injected into the tail veins of syngeneic 4-1BB-/- mice. Control mice received unmanipulated DCs loaded with SIINFEKL plus OT-I CD8+ T cells in similar fashion. The recipient mice were treated additionally with anti-4-1BB (100 μg/day × 2 days) and sacrificed 7 days later for analysis.
For in vivo inhibition of IDO, slow releasing 1-MT pellets (20 mg/day; Innovative Research of America, Sarasota, FL) were placed under the skin (back of the neck) of mice. Control mice received placebo pellets (Innovative Research of America) in a similar fashion. Seven days later, spleen cells were collected and subjected to flow cytometry.
Total RNA was isolated from spleens using TRIzol (Invitrogen) as described by the manufacturer. IDO mRNA was determined by RT-PCR. In brief, 2 μg of RNA was reverse-transcribed using random hexamer primers specific for mouse IDO (forward, 5′-CACTGTATCCAGTGCAGTAG-3′; reverse 5′-ACCATTCACACACTCGTTAT-3′) and GAPDH (forward, 5′- GAACGGGAAGCTTGTCATCAA-3′; reverse, 5′-CTAAGCAGTTGGTGGTGCAG-3′). The PCR products were stained with ethidium bromide after electrophoresis on 2% agarose gel.
This study was supported by grants from NIH RO1-EY013325; the Arthritis Foundation (Innovative Research Award); the National Cancer Center, Korea (NCC-0890830-2 and NCC-0810720-2); Korean Science and Engineering Foundatioin (Stem Cell-M10641000040 and Discovery of Global New Drug-M10870060009); Korean Research Foundation (KRF-2005-084-E00001); and Korea Health 21 R&D (A050260). We thank Beatriz Finkel-Jimenez for cell sorting and Dr. Augusto C. Ochoa for critical advice. We also thank Dr. David Ron (The Skirball Institute New York School of Medicine, New York, NY) for a generous gift of GCN2-/- breeders and Dr. David H. Munn (Medical College of Georgia, Augusta, GA) for helpful discussions.
Conflict of financial or commercial interest: None