|Home | About | Journals | Submit | Contact Us | Français|
CD1d-restricted ‘NKT’ rapidly stimulate innate and adaptive immunity through production of Th1 and/or Th2 cytokines and induction of CD1d+ antigen-presenting cell (APC) maturation. However, therapeutic exploitation of NKT has been hampered by their paucity and defects in human disease. NKT:APC interactions can be modeled by direct stimulation of human APC through CD1d in vitro. We have now found that direct ligation with multiple CD1d mAbs also stimulated bioactive IL-12 release from CD1d+ but not CD1d KO murine splenocytes in vitro. Moreover, all CD1d mAbs tested also induced IL-12 as well as both IFN-γ and IFN-α in vivo from CD1d+ but not CD1d-deficient recipients. Unlike IFN-γ, CD1d-induced IFN-α was at least partially dependent on invariant NKT. Optimal resistance to infection with picornavirus encephalomyocarditis virus (EMCV) is known to require CD1d-dependent APC IL-12-induced IFN-γ as well as IFN-α. CD1d ligation in vivo enhanced systemic IL-12, IFN-γ, and IFN-α, and was protective against infection by EMCV, suggesting an alternative interpretation for previous results involving CD1d ‘blocking’ in other systems. Such protective responses, including elevations in Th1 cytokines, were also seen with CD1d FAb’2s in vivo, while an IgM mAb (with presumably minimal tissue penetration) was comparably effective at protection in vivo as well as cytokine induction both in vivo and in vitro. Although presumably acting immediately ‘downstream’, CD1d mAbs were protective later during infection than iNKT agonist α-galactosylceramide. These data indicate that NKT can be bypassed with CD1d-mediated induction of robust Th1 immunity, which may have therapeutic potential both directly and as adjuvant.
As well as adaptive immune responses mediated by classic antigen-specific T and B cells, the critical role of the innate immune system in protective responses against infectious challenge is now fully appreciated. NK cells, B-1 B cells, γδ T cells, and most recently a subset of “NKT” cells (T cells expressing of NK cell markers such as CD161) are now widely regarded as functioning as lymphoid components of the innate anti-pathogen immune system (1-5). The major CD1d-restricted subset of NKT cells recognizes glycolipids presented by non-polymorphic MHC class I-like CD1d (1-8). Primates have 5 CD1d molecules, whereas rodents have only 2 recently duplicated CD1d genes (6-8). CD1d is constitutively expressed on the surface of antigen presenting cells (APC) including B cells, thymocytes, and rodent (but not human) T cells and hepatocytes (6-10). Surface CD1d is inducible on these and certain other cell types, in some cases from intracellular stores (1;6-10). Some CD1d-restricted T cells utilize a unique invariant T cell antigen receptor (TCR) α chain rearrangement (murine Vα14Jα18) with restricted TCRβ chain repertoire (invariant NKT: iNKT) (1-8). These and other distinct CD1d-restricted T cell subsets can positively or negatively regulate ongoing as well as naïve adaptive immune responses through rapid production of large amounts of Th1- and / or Th2-type cytokines, including IFN-γ and IL-4, respectively (1-8;11-13). A specific iNKT glycolipid antigen, α-galactosylceramide (α-Galcer), was originally derived from marine sponge in an anti-cancer drug screen (1;5,6) and there are subsequently identified related bacterial analogues and other lipids, which include a first candidate endogenous ligand (14-20). Activation of iNKT cells by α-Galcer, widely used to exploit iNKT in vivo in rodents, induces a rapid mixed Th1 / Th2 systemic cytokine pattern and transient stimulation of both the innate and adaptive immune systems, including NK cells (1-8).
Physiologically, CD1d-restricted T cells can augment or inhibit Th1 responses, including antitumor, autoimmune, and anti-pathogen responses, through a variety of mechanisms depending on context (1-8;21-28). The positive or negative contribution of CD1d-restricted T cells in Th1-like immune responses to pathogens depends upon the individual pathogen and resistance mechanisms involved. In particular, CD1d-restricted T cells appear to contribute to resistance against specific viral infections, but not others (22,23,25,26;28-40), and there is evidence for anti-viral roles of human iNKT (41,42). Optimal resistance to picornavirus diabetogenic encephalomyocarditis virus (EMCV-D) requires IL-12, IFN-γ, NK cells, and CD1d-restricted T cells (30,33,39). Similar results have been reported with herpes simplex viruses (HSV) (34,35), although this may be strain- or dose-specific (38). EMCV resistance involves the CD1d-dependent sequential induction of IL-12 and type 1 and 2 IFNs, leading to both innate and adaptive immune responses with NK and T cell activation (33,39). CD1d-restricted T cells also appear to stimulate CD8 T cell responses against respiratory syncytial virus (32), but the reverse has been found in the case of lymphocytic choriomeningitis virus (31) and immunity to certain viruses as well as other infections appears to be CD1d-independent (26,31,36-38,43-45). Also consistent with a critical role for NKT cells in resistance to specific viral and bacterial infections, multiple cases of MHC-like suppression of CD1d expression and antigen presentation to NKT cells by infections have been uncovered (46-53). In contrast, several unrelated infections including low dose HSV-1, coxsackie virus CVB3, HCV, and Listeria, can lead to up-regulation of local tissue CD1d (54-57), which could be reflective of immune-surveillance and/or alternative pathogen counter measures. Consistent with these activities, α-Galcer is transiently prophylactically protective against a wide variety of pathogens in rodent models (1-6;25,26,28,30,36;58,59), irrespective of physiological involvement of iNKT or other CD1d-restricted T cell populations in resistance.
Despite the potential for therapeutic exploitation of CD1d-restricted T cells, clinical progress to date has been hampered by the relative paucity of the iNKT subset in humans (6,8,22,23,41,42). Interestingly, CD1b, -c, or -d (but not CD1a) antibody cross-linking can activate CD1+ cells (60,61). We found that direct CD1d ligation can model human iNKT activation of APC leading to bioactive IL-12 production (62). We now show that ligation of murine CD1d with multiple mAb (IgM or IgG) is similarly active at inducing bioactive IL-12, IFN-γ, as well as IFN-α release, both in vitro and in vivo. Interestingly, of the three, IFN-α only, was partially dependent upon invariant NKT cells. Furthermore, we find that CD1d mAb treatment can alleviate acute EMCV-D disease in vivo. The protective treatment response was also found with CD1d mAb Fab’2 fragments and protection was associated with increased systemic IL-12, IFN-γ, and IFN-α. Invariant NKT agonist α-Galcer has been shown to be protective in EMCV-D and MCMV infections (30,36). However, α-Galcer exacerbated viral disease when given on the day after infection or later. In contrast, CD1d mAb administration was protective up to 2 days into infection.
Collaboratively, we have also found that CD1d mAb can induce anti-tumor responses in vivo, which are at least partially dependent upon IL-12 and IFN-γ (63,64). Together, therefore, direct CD1d ligation (either alone or with other stimuli) provides a potential Th1-type therapeutic option in infectious disease as well as cancer. Significantly, previous reports of CD1d antibody “blocking” CD1d-restricted T cell function in vivo (65-71) are re-interpreted based upon these data.
Rat anti-mouse CD1d IgG mAbs 1B1 (BD-Pharmingen), HB323 (ATCC), 19G11 (kindly provided by A. Bendelac), 3C11 (a protein-A as well as protein-G binding rat IgM (72), or control isotype mAbs (BD-Pharmingen) were bound to 96-well plates indirectly for correct orientation and to obscure Fc portions from inhibitory Fc receptors via protein-G, as previously described for human CD1d mAb (62). Control stimuli: LPS (1μg/ml) or α-Galcer (0.2μg/ml). 1×105 splenocytes / well of 96-well plates. p70 IL-12, IFN-γ, and IFN-α release determined in triplicate ELISA (Ab prs, Endogen; kits from R and D). Limit of detection was ~1 pg/ml. Results shown with standard deviations.
Mice were given 50μg i.p. each intact CD1d mAb 3C11, HB323, 19G11, 1B1, 30μg FAb’2 (prepared according to kit manufacturer’s directions; Pierce, Rockford, IL), isotype controls, LPS (100μg), or α-Galcer (2μg). Serum for measurement of cytokine in vivo was diluted 1:10 for assay and values corrected following ELISA as above. Data are means with SD or for individual animals, as shown. 5 - 7 week old male Th1-dominant relatively virus-resistant WT C57BL/6J or more sensitive N12 C57BL/6J CD1d KO mice deficient in both CD1d genes (33; to be available at Jackson Labs.; http://jaxmice.jax.org/query; Stock No. 008881) or lacking only iNKT cells (30; Jα18 / Jα281 KO mice, N10) or 10 week old more-sensitive Th2-biased male WT Balb/c mice were used. Mice were infected with 500 pfu of the diabetogenic strain of encephalomyocarditis virus (EMCV-D), essentially as previously described (30,33,39). Briefly, glucose tolerance tests were performed 5 - 7 days post-infection (depending on extent of paralysis) by injection of 2g/Kg glucose and blood was collected one hour later with glucosidase inhibitors for analysis by OneTouch basic glucometer (LifeScan Inc., Milpitas, CA). Encephalitis was assessed by semi-quantitative paralysis score (30,33): 1= no paralysis (to indicate number of animals / group), 2 = weakness in one limb, 3 = one completely paralyzed limb, 4 = weakness in two limbs, 5 = paralysis of two limbs, 6 = paralysis of ≥ three limbs.
We previously found that CD1d+ human monocytes and dendritic cell produce large amounts of IL-12 in response to CD1d mAb ligation (56). Therefore, splenocyte cultures were similarly tested for bioactive p70 IL-12 production in response to stimulation with plate-bound CD1d mAbs or controls. Splenocytes from C57BL/6 mice produced substantial IL-12 at 24 hours in response to 2 distinct CD1d IgG mAbs 1B1 and 19G11 when presented on protein-G coated plates and the levels further increased over 3 days (Fig. 1A, not shown). In contrast, isotype control mAbs did not produce IL-12 levels above background. Importantly, IL-12 levels induced by both CD1d mAbs were comparable to these induced by LPS (Fig. 1A). Similarly, IL-12 levels of WT splenocytes, but not CD1d KO cultures, were markedly increased by a third CD1d mAb, IgM 3C11, with a mean of ~400 pg/ml (Fig. 1B), comparable to the other CD1d mAbs (Fig. 1A,B), LPS (Fig. 1A,B), and to iNKT ligand α-Galcer (Fig. 1B). Consistent with activation of IL-12 production, substantial IFN–γ was also induced in CD1d mAb stimulated WT and not CD1d KO cultures (Fig. 1C). CD1d-stimulated IFN–γ production was comparable to that induced by α-Galcer, although less than that produced in response to LPS.
Therefore, these studies demonstrate efficient and specific induction of murine bioactive IL-12 and downstream activation of IFN–γ production in vitro by CD1d mAbs of IgG or IgM isotypes. Further studies with the 1B1 CD1d mAb versus isotype control confirmed these in vitro findings and, interestingly, extended the results of CD1d ligation of splenocytes to marked specific stimulation of potent anti-viral type 1 interferon (Fig. 1D-F).
To determine whether CD1d mAbs could similarly stimulate CD1d+ APC in vivo, mice were treated with CD1d mAbs and sera tested for cytokine production (Fig. 2A,B,C). Remarkably, intact CD1d IgM mAb 3C11 (Fig. 2A) as well as CD1d IgG mAb 1B1 (Fig. 2D) could specifically induce substantial systemic bioactive p70 IL-12 production. Notably, no such induction was noted from CD1d KO mice, thereby eliminating the possibility of selective LPS contamination of the CD1d mAb and further demonstrating specificity. However, mice lacking only the invariant NKT cell population, but retaining intact CD1d and CD1d-restricted “non-invariant” NKT (Jα18 KO mice), where able to mount strong IL-12 responses, as expected. Comparable IFN-γ responses were also found to be specifically induced by CD1d mAb in WT and Jα18 KO mice, although not CD1d KO animals (Fig. 2B). Interestingly, however, while WT mice produced a strong IFN-α response to CD1d mAb and CD1d KO mice (as expected) did not, Jα18 KO mice made only very modestly increased IFN-α relative to isotype controls (Fig. 2C). Jα18 KO mice also had a slightly higher baseline response (Fig. 2C). These results may indicate that unlike for IFN-γ, reciprocal iNKT responses (other than IFN-γ itself) are required for optimal IFN-α induction.
In further experiments, CD1d mAb was compared to other control stimuli in vivo. As above, CD1d mAb, but not isotype control mAb, stimulated systemic IL-12, IFN-γ, and IFN-α production from WT mice. Control stimuli LPS and α–Galcer were also active in WT mice (Fig. 2D-F). However, there was no effect of CD1d mAb or α–Galcer in CD1d KO mice (Fig. 2D-F), demonstrating lack of LPS in the CD1d mAb preparations. Only LPS stimulated cytokines from CD1d KO mice, comparably to WT (Fig. 2D-F).
CD1d mAb could induce IL-12 from human APC in vitro (62) and murine IL-12 along with type 1 and type 2 IFNs both in vitro (Fig. 1) and systemically in vivo (Fig. 2). Therefore, mice challenged with EMCV were treated with CD1d mAb to determine whether such activation of APC could occur during infection in vivo and if such treatment could be protective, as previously shown for α-Galcer (30). Representative results from a series of EMCV-D infections are summarized in Figs. Figs.33 - -5.5. Incidence of abnormal glucose tolerance hyperglycemic response in relatively resistant ~ 7 week old (at day of infection) C57BL/6J wild type (WT) male mice was specifically eliminated by CD1d mAb treatment (Fig. 3A). As previously shown (30,33), CD1d KO mice had significantly higher incidence and severity of hyperglycemia (Fig. 3A, Fig. 4A). Similarly, CD1d mAb also specifically eliminated the relatively low incidence of mild paralysis in this series of infections (Fig. (Fig.3B,3B, ,4B4B).
Next, WT EMCV-D infected animals were treated with either IgM CD1d mAb 3C11 alone or with a second CD1d mAb, IgG2b 1B1, compared to appropriate isotype controls. Again, either or both CD1d mAb eliminated the relatively modest frequency of blood glucose increases observed (Fig. 3C). The severe paralytic disease in these younger more sensitive 5 week old WT mice was also partially but significantly ameliorated by CD1d mAb 3C11 in severity, with reduced incidence (Fig. 3D). Although combination of 2 CD1d mAb also eliminated modest level of hyperglycemia in this series of infections (Fig. 3C), they appeared to only marginally reduce paralysis severity and incidence was actually as high as isotype control-treated mice (Fig. 3D), which might represent a competition effect. Table 1 summarizes experiments assessing the protective effect of 2 CD1d mAb. As shown previously (30,33), CD1d KO mice were more susceptible to EMCV infection than WT mice in both diabetes and paralysis (Table 1). Notably, IgM CD1d mAb 3C11 could significantly and specifically reduce incidence of diabetes and paralysis (Table 1). IgG CD1d mAb 1B1 induced a similar level of paralysis protection as did 3C11, although was less active on glucose tolerance test (Table 1). Finally, when analyzed in aggregate with in built experiment-to-experiment variability, α-Galcer had only a modest effect in this series (Table 1), although statistically significant protection was seen in individual expt. (see below).
Resistance to EMCV depends on IL-12 induced IFN-γ, which can be produced by CD1d-restricted T cells (30,33). To determine if the protective effects of CD1d mAb treatment in vivo during EMCV-D infection reflected similar cytokine induction to that seen with uninfected animals in vivo (Fig. 2), further EMCV infections were performed and glucose responses, paralysis scores, and serum cytokines were determined (Fig. 4). CD1d mAb specifically reduced incidence of disease, as could α-Galcer (Fig. 4A,B). Fig. 4C-E show that both CD1d mAb and α-Galcer also stimulated substantial systemic production of bioactive IL-12, IFN-γ, and IFN-α from EMCV-infected WT mice. As previously reported, CD1d KO mice infected with EMCV were defective in cytokine responses (33,39).
The above results showed that intact CD1d mAb could reduce disease in mice challenged with EMCV. It was possible that distinct CD1d mAb recognizing separate epitopes could have different efficacy and that the presence of Fc portions, which could bind to inhibitory Fc receptors in vivo, reduced efficacy of the treatment. Therefore, several CD1d mAbs were compared and divalent mAb FAb’2 were prepared and used compared to the corresponding intact mAb during EMCV infection in vivo (Fig. 5). Fig. 5A shows that a CD1d mAb FAb’2 as well as all three intact CD1d mAb tested suppressed EMCV induced paralysis when treated on the day of infection. Glucose tolerance testing revealed that CD1d mAb FAb’2 and the intact CD1d mAb IgG could also modestly reduce hyperglycemia incidence in this expt., although to a lesser extent (Fig. 5B) than paralysis (Fig. 5A).
In a further experiment (Fig. 5C), 2 different CD1d mAb FAb’2 were directly compared to isotype and isotype FAb’2, this time treating on Day 1 post-infection. The results confirmed and extended the previous data and showed that 2 independent CD1d mAb could protect against EMCV infection. Again, there was little difference in efficacy between intact mAb and FAb’2 (Fig. 5C).
The systemic cytokine responses of mice were also monitored. Levels of IL-12 and type 1 and type 2 IFNs were measured (Fig. 5D-F). Infection of mice with EMCV-D and treatment with CD1d mAb FAb’2 as well as intact mAbs in vivo also resulted in markedly elevated levels of IL-12, IFN-γ, and acute anti-viral cytokine IFN-α as measured on Day 3 post-infection (Day 2 after CD1d mAb treatment) (Fig. 5D-F) and later time points (not shown). Initial systemic cytokine levels were most increased by the more effective CD1d mAb 1B1 and corresponding FAb’2 (Fig. 5D-F). Subsequently, by Day 7 post-infection, CD1d mAb as well as FAb’2 specifically maintained enhanced systemic IFNγ, IFNα, and IL-12 levels, although these were all reduced relative to the earlier time point and control infected mice cytokines had least cytokine levels by Day 7 post-infection (not shown).
Given the therapeutic potential of intact IgM as well as IgG CD1d mAb and FAb’2 fragments, we finally asked whether mice could be effectively treated with CD1d mAb later than Day 1 post-infection. For comparison, mice were administered invariant NKT agonist α-Galcer, known to be protective up to the time of EMCV infection (30). Surprisingly, even 1 day post-infection α-Galcer was no longer protective but actually exacerbated viral disease (Fig. 6A) without enhancing protective IL-12 relative to no virus or to virus + isotype control mAb (Fig. 6B). A more severe increase in disease was found with treatment on subsequent days (Fig. 6C,D).
Apparently, systemic invariant NKT activation could be deleterious in ongoing viral infection. However, when CD1d mAb were administered up to 2 days post infection they were protective (Fig. 7A). Mice with CD1d mAb on Day 3 post-infection were apparently more sick than virus + isotype mAb controls (Fig. 7A). Therefore, bypassing NKT in this acute viral infection appears to permit later intervention than α-Galcer. To determine how CD1d mAb protected post-infection, systemic cytokines were measured. As shown in Figs. 2A-F and 7B-D, systemic IFNγ, IFNα, and IL-12 were all induced to high levels by IgG 1B1 as well as IgM CD1d mAb 3C11 in the absence of viral infection. Similar or higher levels of these cytokines were found when CD1d mAb treatment was on the day following infection with EMCV-D, declining with administration time into infection, although closer to the time of assay (all measured on Day 5 post-infection; Fig. 7B-D), as well as with time post treatment (not shown). Intermediate level cytokines were found following day 2 treatment, which was still effective at reducing disease (Fig. 7A). Finally, CD1d mAb on day 3 post-infection induced minimal if any cytokines relative to controls (Fig. 7B-D) and had no protective effect (Fig. 7A), perhaps reflecting damage to or focus of the immune system.
The increased susceptibility of CD1d KO mice to EMCV-D and other infections (22,23,25,26,30,33) as well as against tumors (1,2,5,21,23,24,27), has identified roles for CD1d-restricted T cells in protective Th1 immune responses in vivo. Resistance to EMCV-D, as with multiple other viruses, is dependent upon early IFN-γ production and IL-12 induction leading to NK cell activation (22,23,25,26,28,33,36). Consistent with these observations, α-Galcer is at least prophylactically protective against EMCV-D and many other infections and in vitro studies with α-Galcer have shown that IL-12 production by dendritic cells is stimulated by invariant NKT IFN-γ production and NKT:APC CD40 : CD154 interactions (1-6;73,74). Taken together, these observations suggest that the rapid physiological activation of dendritic cells to produce IL-12 in response to specific acute infections may be particularly dependent on CD1d-restricted T cells.
We recently found that direct CD1d stimulation can induce activation and maturation of human DC and monocytes in vitro (62). We have now found that murine CD1d ligation can also induce bioactive IL-12 both in vitro and in vivo as well as apparently mimic physiological IL-12 dependent responses in vivo. As expected, CD1d KO mice did not respond to CD1d mAb and iNKT-deficient mice responded with normal levels of both IL-12 and IFN-γ production. Interestingly, however, Jα18 KO mice IFN-α levels were intermediate, suggesting that downstream of IL-12 there is indirect amplification of IFN-α responses by iNKT, possibly through pDC which in rodents are CD1d+ (6), and which may sensitize by up-regulating CD1d in Listeria infection in response to IFN-β (57), or directly from iNKT, as previously shown for IFN-β from CD4+ iNKT (75).
Single IgG or IgM mCD1d mAb were comparably effective, apparently IgM mAb were able to activate APC systemically to induce IL-12 and provide protection. Multiple CD1d mAb to different epitopes had similar effects on IL-12 induction in vitro. However, combining 2 mAb in vitro actually reduced this protective effect. This could reflect the presence of inhibitory Fc receptors on monocytic cells, which were minimized through binding CD1d mAb via protein-G in vitro, or antagonistic effects of multiple mAb on CD1d signaling. Similarly, 2 mAb were actually somewhat less effective than 1, in vivo, where Fc mediated clearance may have been more important. Preparation of therapeutic quantities of IgG Fab’2 presumably resulted in circumventing FcR-dependent issues, but did not result in a significant increase in protective effect in vivo. Instead, the IgM 3C11 was as potent or better than the IgG mAbs, intact or as FAb’2. This may reflect the potency of multivalent IgM ligation resulting in as strong effects systemically or IgG dilution into tissue.
Previous studies have shown that systemic injection of CD1d mAb can alter CD1d-restricted T cell-dependent immunity (65-71). In some cases, no doubt, blocking of such responses may be the primary explanation, as also contributing where type 2 / ‘non-invariant’ NKT can suppress anti-tumor responses (63,64). However, the ability of CD1d mAb to induce potent systemic Th1 responses clearly has the potential to directly influence such responses independently of CD1d-restricted T cells.
A physiological role for CD1d-restricted T cells appears be to rapidly integrate signals from CD1d, cytokines, and the innate immune system and to influence both the innate response and the decision for Th1 or Th2-type adaptive responses based upon the nature of the antigen challenge. There are relatively few iNKT in humans, and these are further reduced and defective in a wide range of chronic diseases (6,8,22,23,41,42). Bypassing iNKT via direct therapeutic modulation of CD1d, feasibly in the context of other appropriate innate immune signals, may therefore be an approach to optimize natural and vaccine induced anti-pathogen and anti-tumor Th1-biased immune responses.
Drs. A. Bendelac and S. Porcelli kindly provided some of the CD1d mAb. We thank Dr. N. Bigley for generously providing her expertise and EMCV-D for us to set up the virus in vivo challenge system in our Lab., as well as our colleagues, especially Drs. M. Brenner, S. Porcelli, M. Smyth, J. Stein-Streilein, and S.B. Wilson for suggestions and advice, and Kirin Pharma, Ltd. for α-Galcer.
This study was supported by grants from NIH (R01 DK066917; M.E.), DF/HCC Prostate Cancer SPORE (P50 CA90381; S.B., M.E.), the Prostate Cancer Foundation (M.E.), and the Hershey Family Prostate Cancer Research Fund (S.B.).