Despite apparent immunocompetence in the host, lymphomas can arise nonetheless, evading the body's attempts at tumor immunosurveillance. Following transformation, the host's innate immune response is essential in controlling the dissemination and growth of metastatic disease. Immunosurveillance against spontaneously occurring lymphomas is not well understood. Understanding the molecular mechanisms that regulate divergent arms of the immune response is a key to develop effective immunotherapies for many diseases, including cancer.1 Recent studies have suggested an immunosurveillance function for a unique subpopulation of T cells that play an important role in innate immunity called natural killer T (NKT) cells.2 NKT cells are defined as T cells expressing markers on the surface shared with NK cells, and that rapidly secrete both Th1 and Th2 cytokines upon T-cell receptor (TCR) ligation.3, 4 Rather than recognizing peptides presented by MHC class I or class II molecules, NKT cells recognize lipid antigens presented by the MHC class I-like CD1d molecule.5 NKT cells are key players in the regulation of antitumor immunity, particularly in experimental models of tumor immunotherapy, including IL-12 or α-galactosylceramide (α-GalCer) administration.6 In a mouse model in which tumors spontaneously regress after initial growth and then recur, the negative regulation of tumor immunosurveillance by IL-13, possibly produced by CD4+ NKT cells, has been reported.7, 8 In a more recent study, the same group has shown that a reduction in pulmonary metastases in the CT26 nonregressor colon carcinoma model occurs following the elimination of CD4+ NKT cells, further illustrating an inhibitory role for NKT cells in antitumor immunity.9 To date, the role of invariant (Vα14Jα18+) NKT (iNKT) cells in the evasion of hematopoietic tumors from the host's innate antitumor response in vivo has not been investigated. This type of study is vital in increasing our understanding of possible mechanisms by which blood tumors can evade innate antitumor immunity.
The majority of NKT cells express an invariant (or “canonical”) TCRα chain rearrangement (Vα14Jα18 in mice and Vα24Jα18 in humans), the so-called “iNKT cells,” whereas other TCRα chain rearrangements are expressed in noncanonical NKT cells.3, 4, 10 As CD1d molecules are vital for NKT cell development,5 mice lacking the CD1d1 gene (CD1KO mice) are deficient in all CD1d-restricted NKT cells.11, 12 Mice lacking the Jα18 gene (Jα18KO mice) are deficient in iNKT cells, but the TCR repertoire of these mice includes CD1d-restricted NKT cells that use diverse αβ TCR and γδ TCR genes [variant NKT (vNKT)] cells.13 The vast majority of NKT cells (>90%) use Vα14Jα18 TCRα chain rearrangement in mice14 and Vα24Jα18 in humans,15 are present in a wide variety of mouse strains, absent in mice deficient in β2-microblobulin16 and are present in mice deficient in the MHC class I peptide transporter, TAP1.17 NKT cells secrete both proinflammatory (e.g., IFN-γ, TNF-α) and antiinflammatory (e.g., IL-4, IL-13) cytokines, depending on the disease or state of host, apart from possessing NK-like killing activity. Hence, they are considered to be immunoregulatory cells and, in that context, have been shown to be involved in the host's innate antitumor immune response. NKT cells have become a major focus for those who study the innate immune response to tumors and infectious disease.2 Several malignant human hematopoietic cell types express CD1d on their surface,18, 19 but the overall role of CD1d in antitumor immunity is not well understood. We have previously demonstrated that hematopoietic tumors like the murine T-cell lymphoma (L5178Y-R) shed glycolipids that mask CD1d-mediated antigen presentation to NKT cells.20 In the current report, we have studied the role of CD1d-specific iNKT cells in the control of CD1d+ hematopoietic tumors in vivo.
Male and female C57BL/6 wildtype (WT) mice were obtained from The Jackson Laboratory (Bar Harbor, ME). CD1d1-deficient (CD1KO) mice11 and Jα18KO mice13 were backcrossed 10 generations onto a C57BL/6 background, with the latter mice kindly provided by Dr. M. Taniguchi (Chiba University, Japan). The mice were bred in specific pathogen-free facilities at the Indiana University School of Medicine. All mice were age- and sex-matched and used between 6 and 12 weeks of age. All animal procedures were approved by the Indiana University School of Medicine Institutional Animal Care and Use Committee.
The Tap 2-deficient RMA/S T-cell lymphoma cell line21 was kindly provided by Drs. J. Yewdell and J. Bennink (National Institutes of Health, Bethesda, MD). These cells were transfected with the pcDNA3.1-neo vector alone (Invitrogen) (RMA/S-V) or the vector with a cd1d1 cDNA insert (RMA/S-CD1) by standard electroporation techniques. The transfectants were grown in RPMI 1640 medium supplemented with 10% FBS, 2 mM L-glutamine, 2-mercaptoethanol (5 × 10−5 M) and G418 (500 μg/ml) for 2–4 weeks before analysis and used as a bulk population. L-CD1 cells22 are CD1d1-transfected L cells, and were kindly provided by Dr. William Paul (National Institutes of Health, Bethesda, MD). This cell line was maintained in DMEM supplemented with 10% FBS, 2 mM L-glutamine and G418 (500 μg/ml). Control L cells (L-control, also provided by Dr. Paul) were cultured in the same medium as L-CD1 only without G418. Splenocytes harvested for cytokine assays were cultured in IMDM medium supplemented with 10% FBS and 2 mM L-glutamine in the presence of antibiotics. Splenocytes harvested for generating antitumor CTL were cultured in RPMI 1640 medium supplemented with 10% FBS, 2 mM L-glutamine, 2-mercaptoethanol (5 × 10−5 M), recombinant human IL-2 (100 U/ml) and antibiotics.
Antibodies and ELISA
The 1H6 (IgG2a) and 8F3 (IgG1) are mouse CD1d-specific mAb that were generated in our laboratory and have been described earlier.23, 24 PE-conjugated anti-CD4 (GK1.5), PE-conjugated anti-CD8 and FITC-conjugated anti-TCRβ were obtained from BD-PharMingen (San Diego, CA). PE-conjugated rabbit antimouse Ig antiserum was obtained from Dako (Carpinteria, CA). Recombinant mouse and human IL-2 were from PeproTech (Rocky Hill, NJ). Purified and biotinylated antibodies, specific for mouse IL-13 were obtained from R&D Systems (Minneapolis, MN), whereas those specific for mouse IFN-γ and GM-CSF were obtained from BD-PharMingen. The IL-13, IFN-γ and GM-CSF standards used in the ELISA were obtained from PeproTech.
C57BL/6 WT, CD1KO and Jα18KO mice were inoculated with 5 × 105 RMA/S-V or RMA/S-CD1 cells in 500 μl RPMI 1640 medium, intraperitonially. The mice were monitored for 10–60 days posttumor inoculation, depending on the experiment.
Cytokine production by splenocytes ex vivo
Single-cell suspensions of splenocytes from RMA/S-V- or RMA/S-CD1-inoculated C57BL/6 WT, CD1KO and Jα18KO mice were prepared from tumor-bearing mice on day 18 posttumor inoculation. The cells were cultured at a density of 3 × 106 cells/well in a 24-well plate along with irradiated (8,000 rads) RMA/S-V or RMA/S-CD1 cells (7.5 × 105 tumor cells/well). Two days after the coculture, supernatants were harvested and stored at −20°C, until used for cytokine measurement by ELISA as described earlier.23 All samples were analyzed in triplicate. Supernatants from RMA/S-V, RMA/S-CD1 and splenocytes alone were included in the ELISA.
Staining for flow cytometry was performed as described earlier.25 Briefly, a single cell suspension of splenocytes from C57BL/6 WT, CD1KO and Jα18KO mice inoculated 10 or 18 days previously with RMA/S-V or RMA/S-CD1 cells were treated for 30 min on ice with supernatant from the 2.4G2 hybridoma to block Fc receptors, followed by a FITC-conjugated antimouse TCRβ, or PE-conjugated antimouse CD4 or anti-CD8 mAb. After washing three times in HBSS containing 0.1% BSA and 0.02% azide (HBSS/BSA), the cells were fixed in 1% paraformaldehyde, washed three additional times, resuspended in HBSS/BSA and analyzed using a FACScan cytofluorograph (BD Biosciences, Mountain View, CA).
To assess whether circulating CD1d-specific antibodies could be detected, serum samples were collected from C57BL/6 WT, CD1KO and Jα18KO mice inoculated 18 or 60 days previously with RMA/S-V or RMA/S-CD1 cells. Murine L control and L-CD1 cells were treated with various dilutions of these serum samples, or normal mouse serum as a negative control (Sigma-Aldrich, St. Louis, MO). The antimouse CD1d mAb 1H623 served as the positive control. After three washes, the cells were incubated with a PE-conjugated rabbit antimouse Ig F(ab)′2 obtained from Dako (Carpinteria, CA). The cells were washed, fixed and analyzed by flow cytometry as mentioned earlier.
The cytotoxic activity of splenocytes from tumor-bearing mice against RMA/S-V or RMA/S-CD1 cells was measured in a standard 5 hr 51Cr release assay.25 Tumor-specific CTL were generated from splenocytes harvested on day 18 from RMA/S-V- or RMA/S-CD1-inoculated C57BL/6 WT and CD1KO mice, cultured in the presence of irradiated tumor cells and recombinant human IL-2 (100 U/ml) for 5 days. 51Cr-labeled RMA/S-V or RMA/S-CD1 cells were used as targets. To determine tumor-specific CTL activity, the nonspecific killing of cultured splenocytes (CTL) from control (i.e., nontumor-bearing) mice was subtracted from that derived from tumor-bearing mice. The percentage of specific CTL activity was calculated as [(experimental release − spontaneous release)/(maximum release − spontaneous release)] × 100.
ALAK killing assay
Adherent lymphokine activated killer (ALAK) cells from C57BL/6 WT and CD1KO mice were generated as described earlier.26 The cells were used as effectors against 51Cr-labeled RMA/S-V or RMA/S-CD1 cell targets in a standard 5 hr 51Cr release assay. The percentage of ALAK-specific killing activity was calculated as for CTL, described earlier.
CD1d-Ig dimer staining of iNKT and other lymphocytes
Recombinant soluble mouse dimeric CD1d-Ig (Pharmingen) was loaded with 40 M excess of purified α-GalCer, according to the manufacturer's instructions. α-GalCer was synthesized as described.27 In brief, lipids resuspended in PBS (pH 7.4) were mixed with the dimers and incubated at 37°C overnight. Liver mononuclear cells (LMNC) from C57BL/6 WT mice inoculated with or without RMA/S-V or RMA/S-CD1 cells on day 10 and 18 posttumor inoculation were treated with 2.4G2 hybridoma supernatant before staining for cytofluorography with empty or lipid-loaded CD1d1 dimers, APC-conjugated antimouse-TCRβ (BioLegend, San Diego, CA), PE-conjugated antimouse CD4, CD8 or NK1.1 mAb (Pharmingen). The dimers were directly labeled with phycoerythrin, using a mouse IgG1-specific Zenon labeling kit, following the instructions of the manufacturer (Molecular Probes, Eugene, OR).
The Log-rank and Student's t tests were performed using GraphPad PRISM software (version 3.00 for Windows; GraphPad, San Diego, CA). Probability value less than 0.05 was considered significant. The error bars in the bar graphs show the standard deviation from the mean.
Enhanced survival of NKT cell-deficient mice bearing a CD1d+ T-cell lymphoma
The role of iNKT cells in the control of CD1d+ hematopoietic tumors in vivo is unknown. To address this question, CD1d was overexpressed in the murine T-cell lymphoma RMA/S.21 As measured by staining with the 8F3 mouse CD1d-specific mAb,24 there was approximately 62% more CD1d on the surface of RMA/S cells upon transfection with the cd1d1 cDNA (Fig. 1). C57BL/6 WT, CD1d1-(CD1KO) and Jα18-deficient (Jα18KO) mice were inoculated with 5 × 105 vector- or CD1d1-transfected cells (RMA/S-V and RMA/S-CD1, respectively), and monitored for tumor growth and survival for 60 days. Ascites tumors began to develop as early as day 22 in WT mice. Within 5–8 days beyond this time, the mice exhibited severe morbidity and rapid mortality regardless of the tumor. The majority of WT mice inoculated with either tumor died within 40 days, with no difference in the mean survival time (MST) or overall survival rate (Fig. 2a; Table I). In contrast, CD1KO and Jα18KO mice inoculated with RMA/S-CD1 were substantially delayed in ascites development and survived longer than those inoculated with the control tumor (Figs. 2b and 2c; p<0.0001). Interestingly, approximately 25% of the mice exhibited fibrotic tumor development involving the abdominal muscles (data not shown). This type of tumor was predominantly found in WT rather than CD1KO or Jα18KO mice, and is consistent with a high level of IL-13 produced by NKT cells.7 More than 90% of the WT mice inoculated with RMA/S-V or RMAS-CD1 died with a MST of 31 days and 35 days (p = 0.342), respectively (Fig. 2a). Ninety percentage of the RMA/S-V-bearing CD1KO mice died, whereas only 65% of the RMA/S-CD1-inoculated CD1KO mice died by day 60 posttumor inoculation (MST of 32 and 47 days, respectively) (Fig. 2b; p<0.0001). A comparable difference in survival was also observed in Jα18KO mice (Fig. 2c). These differences in MST between WT and mutant mice were significant with RMA/S-CD1 (but not RMA/S-V) cells (Fig. 2; Table I). The percent survival of RMA/S-V cells bearing WT, CD1KO and Jα18KO mice was not significantly different (Fig. 2d, Table II), but the percent survival of WT mice inoculated with RMA/S-CD1 cells was significantly lower than CD1KO (p<0.0001) or Jα18KO mice (p<0.0001), and the MST of the mutant mice was 12–14 days longer than WT mice (Fig. 2e, Table I and II). There was no significant difference in survival between RMA/S-CD1-bearing CD1KO and Jα18KO mice (p = 0.221; Fig. 2e, Table I and II). These results suggest that CD1d+ T-cell lymphomas can evade the host's innate antitumor immune response by inducing an immunosuppressive response that is NKT cell-dependent.
Table I. Summary of Survival Data of C57BL/6 WT, CD1KO and Jα18KO Mice Inoculated With RMA/S-V or RMA/S-CD1 Cells (n = 20 Mice/Group)1
Enhanced survival of CD1d+ lymphoma-bearing NKT cell-deficient mice correlates with increased proinflammatory cytokine production
In a nonhematopoietic tissue-derived tumor model system, it was reported that NKT cells exerted an inhibitory effect on antitumor immunosurveillance through the actions of antiinflammatory cytokines.7 Thus, the status of pro and antiinflammatory cytokine production in WT and mutant mice inoculated with RMA/S-V and RMA/S-CD1 cells was analyzed. As our in vivo experiments indicated that ascites tumor development was detectable by day 22 to 26 postinoculation, RMA/S-V- and RMA/S-CD1-bearing WT, CD1KO and Jα18KO mice (3 mice/group) were sacrificed on day 18. Splenocytes harvested from these mice were cocultured with the respective irradiated tumors, and production of IFN-γ, GM-CSF and IL-13 into the supernatant was measured (Fig. 3). Splenocytes from tumor-bearing mice cultured in the absence of irradiated tumor cells did not secrete any cytokines; hence, the cytokines secreted by those from control mice in the presence of irradiated tumor cells was considered as background (top bar in each pair) (Fig. 3). Secretion of the proinflammatory cytokine IFN-γ was found to be higher in RMAS-CD1-bearing CD1KO (p<0.05) and Jα18KO (p<0.05), as compared to WT mice (Fig. 3). Interestingly, even between CD1KO and Jα18KO mice, there was a significant difference in IFN-γ production (p<0.05), although this was not reflected in a biological difference (i.e., MST). RMA/S-V-inoculated mice produced much lower levels of IFN-γ, but differences amongst the groups were comparable to those animals receiving RMA/S-CD1 (Fig. 3). Secretion of GM-CSF was comparable to IFN-γ production and differences could be observed in all groups, but these did not ultimately translate into contrasting survival rates in CD1KO or Jα18KO mice. Splenocytes from WT mice that received RMA/S-CD1 cells secreted 400-fold higher levels of the antiinflammatory cytokine IL-13 than the unstimulated control, whereas splenocytes from RMA/S-V-inoculated animals produced 150-fold more IL-13 at a level of 150-fold above baseline (Fig. 3). In contrast, IL-13 production by splenocytes from RMA/S-V-inoculated CD1KO and Jα18KO mice was approximately 350-fold greater than control, substantially more than WT mice. The difference in the secretion of IL-13 in splenocytes from WT mice inoculated with RMA/S-CD1 was much higher than that observed in either CD1KO or Jα18KO mice (p<0.05), and was not apparent in mice bearing RMA/S-V tumors (p > 0.2; Fig. 3). TGFβ secretion was only detectable on day 10, and WT mice inoculated with RMA/S-V secreted slightly more than the other groups (data not shown).
Rapid reduction of T cells in tumor-bearing WT(but not NKT cell-deficient) mice
To understand whether there were changes in T-cell subpopulations in RMA/S-V- or RMA/S-CD1-bearing mice that could account for the differences in survival between WT and mutant mice, splenocytes were harvested on days 10 or 18 posttumor inoculation and stained for TCRβ, CD4 and CD8, and analyzed by cytofluorography. In WT mice bearing either RMA/S-V or RMA/S-CD1 tumor cells, an elevation in CD4+ T-cell numbers over control mice on day 10 (up to 30%) was observed, with a slight increase in CD8+ T cells (Fig. 4). By day 18, CD4+ T-cell numbers decreased to approximately 70%, and CD8+ T cells were reduced to about 80% of normal. In contrast, in CD1KO and Jα18KO mice, only minor changes in these T-cell subpopulations occurred. These results suggest that the initial elevation (and subsequent decline) in CD4+ T-cell numbers in WT mice contributed to their enhanced susceptibility to tumor-induced death as compared to CD1KO and Jα18KO mice. It is important to note that the CD4+/CD8+ T-cell ratios were reduced from almost 2:1 in control animals to 1:1 in all tumor-bearing mice (Table III).
Table III. Summary of Splenic CD4/CD8 Ratios in C57BL/6 WT, CD1KO and Jα18KO Mice Inoculated With (Or Without: Control) RMA/S-V or RMA/S-CD1 Cells (n = 6 Mice/Group)
10 and 18 are days posttumor inoculation. The experiments were terminated on day 10 or 18 posttumor inoculation.
C57BL/6 WT, CD1KO and Jα18KO mice were inoculated with 5 × 105 RMA/S-V or RMA/S-CD1 tumor cells, intraperitoneally.
Increase in invariant NKT (iNKT) cell numbers in livers of tumor bearing WT mice
As we did not observe differences in the survival rate between RMA/S-V- and RMA/S-CD1 tumor-bearing WT mice, iNKT cell numbers were assessed in LMNC from tumor-bearing mice on days 10 and 18 postinoculation. LMNC were stained for α-GalCer/CD1d-dimer-positive cells (iNKT cells), NK cells, CD4 and CD8 T cells, and analyzed by cytofluorography. In mice bearing either RMA/S-V or RMA/S-CD1 tumor cells, an elevation in iNKT cell numbers by 3-fold was observed (Table IV), but no changes in iNKT cell numbers were observed on day 10 (data not shown). Additionally, there was a concomitant reduction in CD4+ and CD8+ T cell as well as NK cell numbers over control mice on day 18 (Table IV). The overall number of LMNC in each tumor-bearing mouse was 3–4 fold more than in control mice (Table IV). Therefore, these results further suggest that iNKT cells may be playing a direct role in suppressing the host's immune response to a T-cell lymphoma.
Table IV. Summary of Lymphocyte Profiles in Liver Mononuclear Cells (LMNC)From C57BL/6 WT Mice Inoculated With (Or Without: Control) RMA/S-V or RMA/S-CD1 Cells (n = 4 Mice/Group) on Day 18 Posttumor Inoculation
Anti-CD1d-specific cytotoxic T cells and antibodies do notplay a role in the enhanced survival of RMA/S-CD1-bearing CD1KO mice
A simple explanation as to why CD1KO mice inoculated with CD1d+ tumors lived substantially longer than WT mice is that the former mice generated CD1d-restricted CTL and/or CD1d-specific antibodies. Splenocytes from RMA/S-V- or RMA/S-CD1-bearing WT, CD1KO and Jα18KO mice (3 mice/group) were harvested on day 18 and cultured with irradiated tumor cells that were identical to the ones that they had been primed with under standard mixed lymphocyte reaction (MLR) conditions.28 On day 5 of culture, these stimulated splenocytes were used as effectors against RMA/S-V or RMA/S-CD1 target cells in a 5 hr 51Cr release assay.25 Interestingly, splenocytes from either RMA/S-V- and RMA/S-CD1-bearing WT or CD1KO mice killed the targets at a similar level with approximately 20% lysis at an E:T ratio of 50:1 (Figs. 5a and 5b). Therefore, there was no difference in the antitumor effector cells generated in either WT or CD1KO mice, and CD1KO mice did not develop CD1d-specific CTL to any significant degree.
To analyze whether CD1d-specific antibodies were generated in tumor-bearing mice, serum samples were collected from WT, CD1KO and Jα18KO mice (3 mice/group), which received either RMA/S-V or RMA/S-CD1 on days 18 and 60 posttumor inoculation. Note that because almost all of the RMA/S-V-bearing mice had died by day 60, we were only able to analyze the serum from RMA/S-CD1-bearing mice. The serum samples were analyzed by flow cytometry, using control and mouse CD1d1-transfected L cells. As a positive control, the antimouse CD1d mAb 1H623 was used. Normal mouse serum served as the negative control. Under no conditions were we able to detect any circulating CD1d-specific antibody activity, whereas 1H6 stained the CD1d1-transfected (but not vector control) cells at a high level (data not shown). Therefore, taken together, these results suggest that neither CD1d-specific CTL nor anti-CD1d antibodies were generated in tumor-bearing WT or mutant mice.
CD1d-transfected RMA/S cells are more resistant to ALAKactivity from WT (but not CD1KO) mice
It has been reported earlier that CD1d1 expression on RMA/S cells results in their enhanced resistance to lysis by ALAK activity as compared to the vector control.29 To determine whether the inhibition of killing activity of NK cells by CD1d-expressing cells played a role in the enhanced susceptibility of WT mice to tumor-induced death, ALAK cells were generated from splenocytes harvested from C57BL/6 WT or CD1KO mice and were used as effectors with 51Cr-labeled RMA/S-V or RMA/S-CD1 target cells in a standard killing assay. RMA/S-V cells were killed at a very high level by ALAK from WT mice. In agreement with the observations of Kane et al.,29 RMA/S-CD1 cells were more resistant to ALAK-mediated killing (Fig. 6a). Interestingly, such resistance was not detected when ALAK from CD1KO mice were used (Fig. 6b), suggesting that NK cells can be “calibrated” against CD1d.
NKT cells have the ability to very rapidly secrete large amounts of Th1 and Th2 cytokines upon activation.30, 31, 32 A number of studies have analyzed the role of this T-cell subpopulation in antitumor immunity, infectious diseases and autoimmunity. These immunoregulatory T cells that bridge the innate and acquired immune responses recognize CD1d-bound glycolipid ligands, such as α-Galcer,33 glycosyl ceramides from the cell wall of Sphingomonas34, 35, 36, 37 and the endogenous lysosomal glycosphingolipid, isoglobotrihexosylceramide (iGb3).38 In the context of antitumor immunity, beneficial effects of invariant NKT (iNKT) cells against methylcholanthrene-induced fibrosarcomas following the adoptive transfer of α-GalCer-reactive NKT cells39 and by α-GalCer alone in a B16F10 melanoma model,40 have been shown. Recently, and in contrast to those two model systems, it has been shown in a immunogenic murine regressor tumor model that CD4+ NKT cells play an immunosuppressive role.7, 8 The same group also exhibited that the number of lung metastases in a murine CT26 colon tumor model system is reduced in CD1KO mice.9 In our study reported here, it was found that NKT cell-deficient mice survived longer than their WT counterparts, indicating (like that observed in the latter solid tumor models mentioned earlier) that NKT cells can play an immunosuppressive role in mice bearing hematopoietic tumors. Interestingly, this was noticed only when the lymphoma cells overexpressed CD1d; hence, the quantitative downregulation of CD1d on the cell surface is potentially one of the ways by which CD1d+ lymphomas can evade the host's antitumor immunosurveillance mechanisms. This adds further complexity to the ability of other CD1d+ tumors to evade the host's innate antitumor immune response, as the latter case includes the shedding of inhibitory glycolipids that can bind to CD1d and impair the activation of effector functions of NKT cells.20 In the current study, although CD1KO and Jα18KO mice were similar in that they survived longer than WT mice bearing CD1d+ tumors, these two groups were still different in the context of cytokine production, suggesting that other CD1d-restricted T cells (i.e., other than iNKT cells) participated in these responses. This parallels our observations in mice infected with LCMV41 with CD1KO mice secreting more cytokines than WT mice. It was expected that RMA/S-CD1 cells in WT mice should be specifically cleared by NKT cells leading to their longer survival, but mice bearing these tumors died at the same rate as those inoculated with RMA/S-V cells. Consistent with this, a 3-fold increase in iNKT cell numbers in both RMA/S-V and RMA/S-CD1 tumor-bearing WT mice was observed. As RMA/S cells killed the mice very rapidly, this suggested the direct involvement of iNKT-derived (or induced) negative factors playing a major role in survival. Based on the in vivo response in NKT cell-deficient mice, CD1d on RMA/S-CD1 (not RMA/S-V) cells could stimulate an antitumor immune response [even though the latter possesses endogenous CD1d (Fig. 1)]. This suggests qualitative defects in the RMA/S-derived CD1d. Hence, it is very possible that these tumors can evade the host immune response by altering the functional expression of CD1d via a mechanism different from the one we found for the L5178Y-R T-cell lymphoma.20
Survival of WT and CD1KO mice bearing RMA/S-V21 was the same, which might appear to be contrary to earlier reports with a growth-regressor-recurrence tumor (15-12RM) or CT26 models in BALB/c background mice.7, 9 However, there were differences in RMA/S cells that overexpressed CD1d1, similar to the observations made in 15-12RM and CT26 models, but these latter two tumors are apparently not CD1d+, and it is not clear if the CD1d1-restricted T cells in those models are iNKT and/or other T cells. Our work reported here clearly shows that one cannot generalize an antitumor role for NKT cells, and depending upon the tumor (CD1d+ or otherwise), suggests that IL-13 produced by cells other than NKT cells can inhibit the host's antitumor immune response. It is important here to reiterate that there are, in a general sense, two major subpopulations of CD1d-restricted T cells. The first, canonical NKT cells express an invariant TCRα chain (iNKT). The second subpopulation consists of all other CD1d-restricted T cells (variant or “vNKT”). It is important to note that CD1KO mice have neither subpopulation, whereas Jα18KO mice only lack iNKT cells. It will be important to determine which subpopulation(s) of NKT cells is (are) mediating the effects observed in the interesting reports, using the 15-12RM and CT26 tumors described by Berzofsky, Ostrand-Rosenberg et al.7, 8, 9, 42 The results of those studies can then be compared to our work with RMA/S cells in which roles for iNKT and/or vNKT cells could be discerned. The importance of this latter point was recently illustrated in an acute Trypanosoma cruzi infection, whereby the NKT cell subsets provided distinct functions.43 In that study, most of the Jα18KO mice died, displaying a severe pathology with prominent spleen, liver, and skeletal muscle inflammatory infiltrates composed of NK cells, dendritic cells, B cells and T cells that was in contrast to the mild disease in WT and CD1KO mice.43
NKT cells have been shown to secrete proinflammatory cytokines during infections; these mediators are able to stimulate components of the innate and adaptive responses that eliminate the pathogen.44, 45, 46, 47, 48, 49, 50 In contrast, during other infections, NKT cells have been shown to secrete antiinflammatory cytokines that limit the infection-induced pathology,51 and we have found elevations in both Th1 and Th2 cytokines in NKT cell-deficient mice following an LCMV infection, suggesting that there is a level of CD1d-dependent control over this response.41 It is still unknown how or why NKT cells play dual roles during certain infections, with an increase in proinflammatory cytokines controlling a pathogen, whereas in other infections, antiinflammatory cytokines prevent infection-induced tissue damage.
In the results described here, pro and antiinflammatory cytokines appear to play an important role in the early (or delayed) death in tumor-bearing mice (Fig 7). CD1d-expressing RMA/S cells induced the production of very high levels of IL-13 in WT mice compared to either CD1KO and Jα18KO mice, supporting a direct role for iNKT cells in the production of IL-13 as suggested in an earlier study.7 In addition, by day 18 posttumor inoculation, IL-13 production by Jα18KO mice was approximately 2-fold less than observed in WT mice, suggesting that iNKT cells are producing (and/or inducing) a major portion of IL-13. Earlier reports have suggested that in a murine regressor tumor model, tumor immunosurveillance was suppressed by IL-13 secreted by CD4+ NKT cells,7 and in an in vivo murine colon carcinoma model, the administration of an IL-13 inhibitor to WT mice reduced the number of lung metastases to the level observed in CD1KO mice.9 Proinflammatory cytokines, such as IFN-γ and GM-CSF, correlated with the longer survival observed in tumor-bearing NKT cell-deficient mice in the current study, with tumor-bearing NKT cell-deficient mice secreting higher levels of GM-CSF than WT mice. GM-CSF has been suggested to play a major role in increasing the immunogenicity of tumors and overcoming the suppressive mechanisms of peripheral tolerance. This was evident in models where tumors overexpressed GM-CSF.52, 53, 54 The levels of IFN-γ also appear to be crucial in determining whether there is early death or longer survival in tumor-bearing mice. By day 18 posttumor inoculation, although significant levels of IFN-γ could be observed in all three groups of mice bearing RMA/S-CD1 tumors, the high level of IL-13 production in WT mice very likely contributed to the more rapid death of these mice (as compared to the NKT cell-deficient mice that lived substantially longer). Consistent with our results, increased resistance to tumor recurrence was detected in CD1KO mice with increased IFN-γ and reduced IL-13 as compared to WT tumor-bearing mice.7 Balance and counter-balance between Th1 (IFNγ) and Th2 (IL-13) cytokines play a major role in the antitumor immune response. We have observed fibrotic tumors in approximately 25% of the mice inoculated with RMA/S-V cells (data not shown), and this is likely related to the substantial levels of IL-13 produced by these mice. It is well known that IL-13 is a potent mediator of tissue fibrosis in both schistosomiasis and asthma, suggesting that this cytokine is a key regulator of the extracellular matrix (reviewed in55).
Although the cytolytic activity of NK cells is controlled by the MHC class I status of the targets, one of the observations made by Kärre et al.56 suggested that there is calibration of NK cell killing activity, illustrated by the observation that NK cells from β2mKO mice were tolerant to β2mKO-derived (but not MHC class I+) targets.56 Likewise, in the current study, we have found that (unlike those from WT mice), the cytotoxicity of CD1KO ALAK is not impaired by CD1d. This suggests that ALAK from CD1KO mice are different than those from WT mice, although in vivo-activated NK cell generation induced by a virus infection are comparable.57
We were able to detect significant differences in vivo regarding tumor growth and mouse survival (Fig. 2), although the overall level of CD1d on the cell surface of RMA/S-CD1 cells was about 62% greater (as measured by the 8F3 antibody24) than the RMA/S-V line (Fig. 1). Thus, how does this difference manifest itself in the observations reported here? When we stained these two lines with a panel of 10 mouse CD1d-specific mAb (which have distinct epitope specificities for the murine CD1d1 molecule,24 and data not shown), it was apparent that the staining patterns were different when compared to that observed with CD1d-transfected murine L cells (data not shown). This suggested that there were both quantitative and especially “qualitative” differences between the two tumor-cell lines; namely, the RMA/S lipids that are presented by CD1d to NKT cells may be inhibitory in nature. With the lack of iNKT cells in RMA/S-CD1 tumor-bearing CD1d- and Jα18-deficient mice, this resulted in longer survival of the latter two types of mice as compared to WT (i.e., NKT cell-sufficient) mice. Thus, this suggests that NKT cells respond as negative immunoregulatory cells under these conditions, and is consistent with the in vivo tumor-recurrence model reported by Berzofsky et al.,7, 8 as well as our report concerning the inability of the murine T-cell lymphoma L5178Y-R to be recognized by NKT cells in vitro.20
In summary, we have shown that elevated functional levels of CD1d on a T-cell lymphoma (RMA/S) can promote antitumor immunity in NKT cell-deficient mice, as compared to those tumors with less. However, in WT (i.e., NKT cell-sufficient) mice, a protective role for CD1d was not evident. Thus, targeting the NKT cell-dependent polarization of the host's antitumor immune response could be a rational approach and novel treatment in patients bearing these types of tumors.
The authors thank J. Yewdell, J. Bennink, W. Paul and M. Taniguchi for reagents and mice. D. Jay, C. Willard and K. Gillett provided expert technical assistance.