Antagonistic effect of NK cells on alternatively activated monocytes: a contribution of NK cells to CTL generation

The crucial role of natural killer (NK) cells in bridging innate and adaptive immunity through modulation of the cytokine network is widely acknowledged. Indeed, following stimulation, through NKG2D receptor signaling1 or contact-dependent interactions with mature dendritic cells,2-5 NK cells are the major source of the type 1–polarizing cytokine interferon-γ (IFN-γ). Although the role of NK cells in T-cell activation can be dispensable in other models,6,7NK-derived IFN-γ has been described to maximize xenospecific cytotoxic T-lymphocyte (CTL) responses,8 to promote the sensitization of T cells to the effects of interleukin 2 (IL-2),9 to promote T helper 1 (Th1) cell development,10 and to activate macrophages.11,12 However, depending on the conditioning environment, NK cells can produce a variety of other type 1 (tumor necrosis factor [TNF-α/β], granulocyte-macrophage colony-stimulating factor [GM-CSF], macrophage inflammatory protein-α [MIP1α], IL-1, IL-8), type 2 (IL-3, IL-5, IL-10, IL-13), and type 3 (transforming growth factor-β [TGF-β]) cytokines,13,14 implicating a possible contribution to the development of the resident type 1/type 2 balance.

Besides an immunoregulatory role of NK cells through cytokine production, these cells might also perform accessory functions through cell-cell contact-dependent interaction with antigen-presenting cells (APCs). Both positive (activatory)15 and negative (lytic)16-24 regulatory interactions have been substantiated. In the latter condition, several APC surface molecules were found to be possibly involved in the lytic interaction, either activatory such as B7-1/B7-2,16,17,21,22CD40L,20 CD54,23 HLA,24 or inhibitory such as CD125 and major histocompatibility (MHC) class I molecules.26 As such, one might assume that the functional result of the interaction is determined by the sum of the opposing signals. Moreover, the outcome of NK/APC interaction seems to depend on the cytokine milieu.24,26 The implications of these NK/APC interactions for the adaptive immune response however remain speculative.

According to recent studies, different subsets of APCs can develop, dependent on the resident type 1/type 2 cytokine balance, namely classically activated APCs (DC1/M1 or type 1–associated) versus alternatively activated APCs (DC2/M2 or type 2–associated), and both are antagonistically regulated. DC1/M1 are the final targets and effectors of proinflammatory processes,27,28 whereas DC2/M2 appear to participate in anti-inflammatory processes, tolerance induction, and wound healing and express a distinct set of molecules and receptors used in innate immunity.29-31 Both subtypes of APC can, however, exert inducing and/or down-regulatory effects on adaptive immune responses.

For instance, during the process of malignant cancer, tumor immune escape can be mediated by the activity of tumor suppressor APCs that belong to either the DC1/M1 or DC2/M2 phenotype. Indeed, inflammatory macrophages or M1 can exert suppressive activities, for instance, by the secretion of nitric oxide (NO).32 However, many tumors secrete or induce IL-10 that steers immature dendritic cells toward the DC2 phenotype and are able to induce immune tolerance.33,34 Furthermore, immunosuppressive macrophages have been identified in the lymphoid organs of immunosuppressed, tumor-bearing mice, comprising a population of immature Gr1⁺/CD11b⁺ myeloid precursors.35-37 Exposure of these Gr1⁺/CD11b⁺ cells to IL-4 (type 2 cytokine) greatly increased the T-lymphocyte suppressive activity of this population, whereas type 1 cytokines induced their maturation into competent mature DC1.38,39 Considering the multiple possible effects of NK cells on DC/M either via soluble factors or by direct cellular interactions (see above) might envisage severe implications of NK-depletion or deficiencies on the cellular composition of the APCs that will concomitantly affect adaptive immune responses, including CTL activity.

In this study we have analyzed the possible involvement of NK cells (via either cytokine production and/or cytolytic activity) in determining the cellular composition of APCs and evaluated the resulting effects on antitumor CTL responses. Hereby, the BW-Sp3 lymphoma model was adopted because rejection of BW-Sp3 depends on eliciting strong CTL responses,40,41 and consequently this model allowed us to examine the involvement of NK cells in early CD8⁺-dependent immune reactions. To address the importance of NK cells for the early antitumor defense, α asialo GM 1 (α-ASGM-1)–treated mice were challenged with BW-Sp3 cells. Use of α-ASGM-1 to deplete NK cells was considered a well-founded option, considering our and other available information.42-44 Our results indicate that NK cells are a prerequisite for efficient CTL generation and that absence of NK cells favors the outgrowth of M2 cells. Subsequently, these M2 cells, associated with NK depletion, suppress the restimulation of memory CTLs. Finally, preferential engagement of M2 by activated NK (LAK) cells reveal that cytolytic interactions might be involved in the regulatory function of NK cells to drive M1-dependent CTL responses.

Specific pathogen-free female AKR (Thy1.1, H-2^k) mice were obtained from Harlan (Zeist, The Netherlands).

BW-Sp3 is derived from the spontaneous BW5147 T lymphoma (AKR origin, referred to as BW-O) by in vitro and in vivo passages, as described previously.45 The generation of BW-Sp3(B7-1) was described earlier.41 To steer macrophagelike RAW264.7 cells (TIB71; American Type Culture Collection [ATCC], Manassas, VA) to the M1 or M2 phenotype, they were cultured with 100 U/mL IFN-γ or 100 U/mL IL-4 and IL-10, respectively.

For the generation of CTLs, splenocytes from 2 mice were restimulated in 75-cm² flasks in the presence of 10⁷ irradiated (110 Gy) cancer cells (primary CTL cultures). Alternatively, AKR mice were immunized intraperitoneally 3 times with weekly intervals, with 2 × 10⁶ irradiated BW cells and restimulated on top of adherent cells isolated from primary CTL cultures (called secondary cultures). Five days later the cytolytic activity of the viable lymphocytes was tested in a classical¹¹¹In-release assay.46 

For NK depletion, mice were injected intravenously in the tail with 0.02 mL α-ASGM-1 antibodies (Wako Chemicals, Osaka, Japan), 24 hours prior to tumor inoculation, and repeated every 4 to 5 days. In NK depletion experiments, specific depletion was evidenced by flow-activated cell sorter (FACS) and spontaneous cytotoxic activity to YAC-1. Mice in groups of 6 were injected subcutaneously in the right flank with 2 × 10⁶ cancer cells in 0.2 mL phosphate-buffered saline (PBS). For every treatment, mortality of the hosts and tumor growth were followed up twice a week. To test spontaneous NK cell activity of tumor-inoculated hosts after tumor inoculation (3 or 6 weeks), viable splenocytes were isolated from mice with regressed versus progressing tumors and immediately, or on complement-mediated lysis of NK or CTL fractions, included in¹¹¹In-release cytotoxicity assays. To deplete NK or CD8⁺ cells, 10⁸ splenocytes were incubated with α-DX5 and α-ASGM-1 or α-CD8 (TIB211, ATCC), respectively, followed by complement-mediated lysis, using the supplier's protocol (Harlan Sera-lab, Loughborough, United Kingdom).

IL-2–activated LAK cells were prepared as described elsewhere.16 For the generation of IL-2/IL-12 LAK cells, recombinant mouse IL-12 (generously provided by Genetics Institute, Cambridge, MA) was added to the IL-2 LAK cultures at 100 U/mL, 72 hours after initiation of the cultures for another 48 hours. The adherent cell fraction was harvested with 0.01% EDTA (ethylenediaminetetraacetic acid) in PBS and was used as effector cell population in ¹¹¹In-release cytotoxicity assays.

The target cells were labeled with ¹¹¹In as described elsewhere.47 After extensive washing, 10⁴ target cells were incubated with 2-fold dilutions of effector cells in a total volume of 200 μL culture medium (CM) in 96-well round-bottom plates. All experiments were performed in triplicates and twice repeated. After 4 hours at 37°C, 100 μL supernatants was collected, and radiation was counted in a gamma-counter. The percentage of specific lysis was calculated as ([experimental release − spontaneous release]/[maximal release − spontaneous release]) × 100, where the spontaneous release was determined from labeled targets cells incubated without effector cells and maximal lysis from target cells in 2% sodium dodecyl sulfate (SDS).

Arginase activity was measured in cell lysates as previously described.48 One unit of enzyme activity is defined as the amount of enzyme that catalyzes the formation of 1 μmol of urea per minute.

NO was measured as nitrite by using the Griess reagent. Culture supernatant was mixed with 100 μL 1% sulfanilamide, 0.1% N-(1-naphthyl)ethylenediamine dihydrochloride, and 2.5% H₃PO₄. Absorbance was measured at 540 nm in a microplate reader.

The production of IL-2, GM-CSF, IL-4, IL-10, monocyte chemoattractant protein 1 (MCP-1), IFN-γ (BD Biosciences, San Diego, CA), and TNF-α (R&D Systems, Minneapolis, MN) was quantified by subjecting culture supernatants to commercially available sandwich enzyme-linked immunosorbent assay (ELISA) tests according to the manufacturers' protocols.

Cell samples comprising 10⁶ cells were incubated with the appropriate dilutions of antibodies, as suggested by the distributors, at 4°C for 30 minutes. Rat anti-CD11b–fluorescein isothiocyanate (FITC), anti-CD4–FITC, anti-CD8–FITC, anti-DX5–FITC, anti-CD80–FITC, anti-CD86–FITC, mouse anti-H-2D^d–FITC, anti-IE^d/IA^d–FITC, and hamster anti-CD69–FITC and anti–CTLA-4–phycoerythrin (PE) were purchased from BD Biosciences. Anti-NKG2–FITC was kindly provided by K. Van Beneden (UZ-Gent, Belgium). To prevent Fc receptor (FcR)–mediated binding of staining antibodies, cells were blocked with 2.4G2 (ATTC) derived F(ab)₂ antibodies prior to the addition of the indicated antibodies. For ASGM-1 staining, the IgG fraction was purified from antiserum (Wako Chemicals) and used in combination with FITC-conjugated mouse/rat/human-preadsorbed goat antirabbit IgG (BD Biosciences). As isotype controls for aspecific binding, control antibodies were purchased from the distributors and included during FACS staining. The stained cells were subjected to FACS analysis with the BDFACS Vantage. The results were analyzed with CellQuest software.

The statistical significance of the differences in subcutaneous tumor growth rate or percentage of specific release was assessed by a 2-sided analysis of variance (ANOVA) test, and significance of differences between experimental groups (presented as bars, mean ± SEM) was interpreted by an unpaired Student t test, applying GraphPad Prism for MAC software (GraphPad Software, San Diego, CA). Differences are considered as not significant forP ≥ .05, significant for P < .05, and very significant for P ≤ .01.

We used as a tumor model the immunogenic mouse T-cell lymphoma (BW-Sp3) that expresses high levels of MHC class I. Following subcutaneous inoculation, BW-Sp3 cells form primary tumors that will either progress or regress (Figure1).49 We have demonstrated previously that CD8⁺ T cells are the critical immune effectors involved in the control of BW-Sp3 tumorigenicity.40,41,49 To investigate the importance of NK cells in early local antitumor immune responses, AKR mice were treated with α-ASGM-1. In first instance, we verified the NK-specific effect of this treatment in the spleens of naive mice versus those of tumor-bearing mice 8 days after tumor injection, of which the latter are expected to contain activated tumor-specific CTL. Both the DX5⁺ and NKG2⁺ populations, which are almost exclusively ASGM-1⁺, are severely reduced by the α-ASGM-1 treatment (Table 1 and result not shown). The α-ASGM-1 depletion causes a slight reduction in the number of ASGM-1⁺/CD8⁺ T cells, but not CD4⁺T cells, in both the naive and the tumor-bearing condition, indicating that the affected T cells do not represent the activated pool (Table1). Indeed, IFN-γ–producing CD8⁺ T cells can only be detected in the ASGM-1⁻ population, dissociating the activated and ASGM-1⁺ phenotype (Figure2). This finding indicates that α-ASGM-1 depletion primordially affects NK cells but not activated T cells. In our tumor model this depletion increased tumor growth rate (P < .0001) and was sufficient to abrogate tumor regression in all recipients, resulting in early death of 50% of the hosts after 18 days (Figure 1). Thus, NK cells may be implicated in local antitumor responses against BW-Sp3.

To further substantiate the possibility that NK cells are involved in the immune response to BW-Sp3, we monitored the general activation status of peripheral NK cells during subcutaneous tumor growth by incubating splenocytes, freshly isolated from tumor-challenged mice, with the NK-sensitive target YAC-1 and by evaluating lytic activity. We observed that in hosts in which the local tumor was regressing (Figure 3Ai,ii) or had regressed (Figure 3B), NK activity was higher than in naive mice. Furthermore, in spleens of mice with progressing tumors, 3 (Figure 3Aiii) or 6 weeks (Figure 3B) after tumor inoculation, NK activity was substantially depressed as compared with those of naive mice and regressors (Figure 3). The effect, observed with the total splenocyte fraction, was substantially reduced after removal of DX5⁺/ASGM-1⁺ cells from the effector population (Figure 3A). On the contrary, in vitro depletion of CD8⁺ splenocytes enriched for YAC-1–reactive effectors, corroborating that the lytic cells reside within the NK population. These results suggest that ex vivo NK activity correlates with BW-Sp3 tumor outcome.

Because BW-Sp3 tumor rejection relies on both CD8⁺ cytolytic T cells40,41,49 and NK cells (see earlier section), the role of NK cells in early antigen-specific CTL activity in tumor-bearing mice was investigated. Splenocytes of tumor-inoculated immunocompetent and NK-depleted mice were restimulated in vitro and subsequently tested for their CTL activity. Within 5 to 8 days after tumor inoculation, immunocompetent mice generated substantial levels of specific effector CTLs that progressively reached a maximal activity around day 12 (Figure4A). This time point coincides with the time when tumor regression is initiated (Figure 1). On the contrary, NK cell depletion completely prevents the generation of CTL effector function. Indeed, as shown in Figure 4A, the residual CTL activity of NK-depleted splenocytes does not exceed the CTL activity of restimulated naive spleens up to 12 days following tumor injection (P ≥ .5). The inferior CTL activity of in vitro–restimulated splenic cultures, derived from α-ASGM-1–treated mice, could be correlated with a lower percentage (P ≤ .01 for all time points except for day 5 (P = .06) (Figure 4B) and a lower level of activation of CD8⁺ T cells (Figure 4B). The latter was evidenced by a smaller fraction of CD8⁺CTLA4⁺ and CD8⁺CD69⁺ T lymphocytes in the α-ASGM-1–depleted conditions as compared with the immunocompetent conditions (21% versus 35% [day 8] and 18% versus 28% [day 12] of the CD8⁺ T cells are double-positive for CTLA4, whereas 12% versus 18% [day 8] and 17% versus 27% [day 12] are double-positive for CD69; Figure 4B). The absence of NK cells does not significantly influence the number of CD4⁺ splenocytes (P ≥ .06 for all time points) or their activation status (Figure 4C). Again, the relative contribution of ASGM-1⁺cells to the CD8⁺ T-cell pool is unchanged, regardless of the presence and expansion of antitumor CTLs (Table2), corroborating that BW-Sp3–activated CTLs are not typically characterized by ASGM-1 expression.

Because NK cells can be active producers of cytokines, it was appropriate to monitor the presence of type 1/type 2 cytokines in the different CTL cultures. In addition, because the chemokine MCP-1 was reported to fuel the development of type 2 immune responses, its production was also evaluated. As shown in Figure5A, on subcutaneous inoculation of BW-Sp3 cells, type 1 cytokines (ie, IFN-γ, TNF-α, IL-2, and GM-CSF) are progressively induced (.001 ≤ P < .03 in all cases) and reach a maximum within 8 days. In contrast in α-ASGM-1–treated tumor-bearing animals, the production of type 1 cytokines barely rises above background levels and is always significantly lower (.001 ≤P ≤ .04) than in the corresponding immunocompetent conditions (Figure 5A). However, the production of the type 2 cytokine IL-4 was significantly higher in cultures derived from NK-depleted mice, at least at 8 days (P = .03) to 12 days (P = .027) after tumor inoculation (Figure 5B). This finding was, however, not the case for IL-10, when we did not measure substantial differences in any situation (results not shown). Finally, as shown in Figure 5B, the constitutive production of the type 2–associated chemokine MCP-1 became progressively reduced in tumor-conditioned immune splenocytes from day 5 (P = .03) and day 8 (P = .005) through day 12 after receiving the tumor load (P < .001), and its expression remained high through day 12 in the absence of NK cells (P = .4) as compared with the naive control.

Our earlier studies have indicated a physiologic interaction between NK cells and APCs.16 In the present study we also observed that the number of splenic adherent cells expressing CD11b was substantially higher in CTL cultures derived from NK-depleted animals (Figure 6A). Microscopic examination of the above mentioned CTL cultures, revealed a striking difference in the morphologic appearance of the adherent fractions from the CTL cultures derived from NK-depleted tumor-bearing mice as compared with those derived from immunocompetent tumor bearers (Figure 6B), suggesting a role of NK cells in the regulation of the phenotype and/or activation status of adherent splenic cells. This difference is also expressed by the fact that in our culture conditions, M2 have a substantial induction in CD11b expression level (Figure 6B). Moreover, ASGM-1 is uniformly expressed on both M1 and M2, excluding a preferential engagement of one population by α-ASGM-1 (Table 2). Hence, taking these observations into account together with our aforementioned results demonstrating a switch from a type 1– to a type 2–dominating cytokine response in the absence of NK cells, it was appropriate to test whether in the present experimental conditions different types (M1/M2) of macrophages are elicited. Thus, NO was measured in the supernatant from CTL cultures, and the adherent fractions were tested for arginase activity. The results compiled in Figure 7A show that the adherent fractions of CTL cultures, derived from BW-Sp3 tumor-bearing AKR mice, produce important quantities of NO, whereas, on in vivo depletion of NK cells, the NO production in the CTL cultures drops to background levels. Furthermore, in contrast to the up-regulation of NO production under immunocompetent conditions, the adherent fractions derived from NK-deficient cultures preferentially express the arginase pathway (Figure 7B). Indeed, the arginase activity is significantly higher in these cell populations as compared with populations derived from naive (P ≤ .02) or immune tumor-bearing (P ≤ .01) AKR. Hence, NK depletion of tumor-bearing mice favors the expansion of splenic M2 cells.

To determine whether the adherent cells, including M2 cells, identified in the NK-depleted cultures, can directly influence CTL activity, memory T cells from BW-Sp3–immunized mice were restimulated in the presence of the aforementioned adherent populations. Respecting physiologic T/monocyte proportions as they are present during primary CTL restimulation, the results depicted in Figure8A show that the adherent fraction of splenocytes from day 5 and day 8 tumor-bearing NK-deficient mice can completely block memory CTL restimulation (P < .0001). In contrast, although cultured adherent splenocytes from mice, injected with BW-Sp3 on day −5, suppress CTL activity with 71% ± 13% (P = .0007), this suppressive activity disappears in adherent splenocytes from mice injected with BW-Sp3 on day −8. Notably, cultured adherent cells from naive spleens can also block efficiently CTL effector function (P < .0001) (Figure8A). Furthermore, there was a perfect correlation between suppression of CTL activity and arginase activity in the adherent fractions of the secondary cultures (Figure 8B). To exclude that the suppressive activity solely depends on quantitative effects (the number of adherent cells is substantially higher in NK-depleted conditions), we repeated these experiments by restimulating BW-Sp3–specific memory T cells in the presence of equal numbers of adherent splenocytes derived from primary cultures. Again, the suppressive activity remains predominantly associated with the condition of NK depletion (Figure 8C) and correlated to arginase activity (Figure 8D). These results suggest that adherent splenic populations from both naive mice and NK-depleted tumor-bearing mice are activated during the secondary CTL cultures toward M2 cells that are highly suppressive on CTL activity. In contrast adherent splenic populations from tumor-bearing mice are less prone to develop into suppressive M2 cells.

As mentioned above, depletion of NK cells, besides favoring the development of M2 cells, leads to an increase in the number of adherent cells in the splenic cultures (Figure 6B). In earlier studies, we and others have demonstrated that autologous antigen-presenting cells can become targets for LAK-mediated cytolysis, provided the NK cells are properly activated.16,17 Thus, it was of interest to test whether M1 versus M2 cells differ in their susceptibility toward activated NK cells (Figure 9A). In the first instance, mouse macrophagelike RAW264.7 cells were steered toward a M1 (cultured in the presence of IFN-γ) or M2 (cultured in the presence of IL-4 and IL-10) phenotype, using the NO/arginase balance as a read-out system to verify the M1 versus M2 phenotype of the cytokine-treated RAW264.7 cells (Figure 9B). Subsequently, both RAW264.7 populations were tested as targets in LAK cell–mediated cytotoxicity experiments. The results depicted in Figure 9A indicate that steering of RAW264.7 to M1 strongly reduces the susceptibility of the targets to LAK lysis (23% specific lysis at an effector-target ratio of 40:1) as compared with Il-4/IL-10–treated targets (54% specific lysis at an effector-target ratio of 40:1;P < .0001). Because we described several antigens contributing to NK recognition in earlier studies,16 we evaluated the surface expression of MHC and B7 antigens on the differentially activated RAW264.7 cells by FACS. The expression of MHC class I and II is considerably higher on RAW264.7 (IFN-γ) as compared with RAW264.7 (IL-4/IL-10) (Table 3), supporting a role for MHC class I in conferring resistance to NK lysis. Moreover, the expression of NK (co)stimulatory B7-1 and B7-2 is twice as high on RAW264.7 (IL-4/IL-10) versus RAW 264.7 (IFN-γ) cells. Secondly, the adherent fractions of the different primary CTL cultures (see above) were also tested for their susceptibility to IL-12–activated LAK cells. As shown in Figure 9C, consistent results could be obtained: M2-like cells from NK-deficient conditions were significantly better lysed (P ≤ .0001) than M1-like cells from immunocompetent cultures. Hence, NK cells can be involved in preferential physical elimination of M2 in vivo.

The contribution of NK cells in the regulation of CTL activity has been documented in a number of experimental models, including xenospecific CTL generation,8,50 induction of influenza virus–specific CTLs,51 and priming of tumor-specific CTLs.52,53 Overall, in these studies the mechanisms underlying NK-mediated regulation of CTL activity were not exactly defined and remain controversial. Indeed, Smyth and Kelly8and Smyth et al50 demonstrated that, although CD4⁺ T cells are critical for a mouse antihuman xenospecific CD8⁺ CTL response, NK1.1⁺ cells play an IFN-γ–dependent accessory role in generating a maximal CTL response at the level of the lymph nodes. However, Kos and Engleman51 provided evidence that NK cells, but not NK-derived soluble factors, are strictly required for the generation of influenza virus–specific CTL following infection. Finally, in the case of CD4⁺–dependent priming of B16 tumor-specific CTLs, Terao et al53 found that depletion of NK cells before B16⁻ immunization abrogated tumor-specific CTL activity. Furthermore, they observed that NK cells had a promoting effect on priming of CD4⁺ cells but inhibited the priming of CD8⁺ T cells. Kurosawa et al52 refined the latter study and observed that part of the antitumor CTL activity in NK-depleted mice was restored by intraperitoneal injections of IL-2 and to a lower extent of IFN-γ.

Difficulties in analyzing the NK dependence of CTL activation, relies partially on the NK specificity of the antibodies that are widely applied to deplete NK cells in vivo. Indeed, NK/T cells express T-cell markers as well as NK cell markers, but several studies have in addition demonstrated that antigen-specific T cells can express any of the specific NK cell markers (DX5, NK1.1, ASGM-1, Ly49) at selective steps during CTL differentiation or activation.44,54-59Also ASGM-1, used in our study to specifically target NK cells for in vivo depletion, has had its controversies. Early studies demonstrated that cytotoxic treatment of splenocytes with anti-ASGM-1 specifically removed NK cells and was ineffective in removing alloreactive killer and helper T cells.43,58,60,61 Other studies on virus-specific CTLs have suggested an influence of α-ASGM-1 cytotoxic treatment on CTL activity, but these studies fail to address the direct effect on T-cell populations via FACS analysis and could be explained by the need for NK cells to accomplish full CTL activation.43 Finally, quite recently Kelly et al62 demonstrated in their tumor system (H-2^b) that α-ASGM-1 depletion has no direct effect on (EG7) tumor-specific CTL. Thus, considering that the effect of α-ASGM-1 treatment on CTL activity can rely on both the mouse strain and/or the antigen stimulus,61 we wanted to delineate the effect of α-ASGM-1 treatment in our tumor model. As such, we observed that up to 50% of AKR CD8⁺ T cells indeed expressed ASGM-1, regardless of their activation status (naive versus BW-Sp3 activated). But in vivo treatment of mice with α-ASGM-1 does not influence the total number of splenic CD8⁺ cells as was demonstrated before for allospecific CTLs. We detected a slight reduction in ASGM-1⁺CD8⁺ double-positive cells ex vivo, which, however, again was not correlated to immune activation and moreover disappeared after 5 days of antigen-specific restimulation in vitro. However, α-ASGM-1 treatment effectively removed the ASGM-1⁺DX5⁺ cells in the spleen, corroborating a dominant effect of α-ASGM-1–mediated in vivo depletion on the NK subset of the spleen.

The results presented herein unequivocally link NK activity and CTL-mediated antitumor responses and furthermore provide evidence for novel regulatory mechanisms possibly implicated in the NK/CTL interaction. Indeed our analysis of the involvement of NK cells in the BW-Sp3 tumor model revealed the following main findings: (1) Progression of BW-Sp3 tumors leads to both impaired anticancer cell CTL and NK cell activities, and depletion of NK cells during the early phase of BW-Sp3 tumor development impairs the development of CTL responses and increases tumor progression; (2) NK cell depletion during early BW-Sp3 tumor formation alters the tumor-elicited splenic cytokine production in favor of a type 2 cytokine microenvironment and promotes the development of alternatively activated splenic macrophages (M2 cells); (3) M2-enriched splenic adherent cells suppress potently memory CTL activity; (4) M2 cells are more susceptible to NK-mediated lysis as compared with classically activated macrophages (M1 cells).

The induction of type 1 cytokines by BW-Sp3 cancer cells, as documented in this and previous reports,41 may reflect the inherent capacity of these malignant cells to activate the innate immune system. Indeed BW-Sp3 cells express a ligand (ie, Rae-1; J.A.V.G., unpublished observation, 2001) for the recently identified stimulatory lectinlike NKG2D receptor that is expressed by NK cells and activated macrophages in mice.1,63,64 Interaction between NKG2D and ligands (Rae-1, H60) expressing tumor cells triggers NK cytotoxicity and IFN-γ secretion in vitro1,63 and induces potent priming of cytotoxic T cells and sensitivation of NK cells in vivo.64,65 Accordingly a direct interaction between BW-Sp3 cancer cells and innate cells (NK cells, activated macrophages) via Rae-1/NKG2D might be a primary stimulus that induces the cascade of cellular immune responses leading to type 1–associated adaptive tumor immunity (including generation of CTLs). Consequently, experimental NK cell depletion or NK cell inactivation during tumor progression may, at least in our tumor model, influence profoundly efficient generation of CTLs and tumor immunity. In this scenario the link between CTL and NK activity would rely primarily on the production of NK-derived type 1 cytokines as was suggested in other studies.8,52 However, the evidence that was so far provided for a direct role of NK-derived IFN-γ and/or other cytokines on CTL generation is debatable (see above). In view of our observations that NK depletion triggers the development of splenic M2 populations that exert suppressive activities on CTL activity, it is conceivable that alternative or parallel pathways involving such cells may also contribute to NK-dependent CTL responses.

The differential generation of M1 versus M2 cells depends on the resident type 1/type 2 cytokine balance that, in turn, can be determined by the genetic background of the host. Indeed in mice certain strains are more prone to develop type 1 or type 2 cytokine responses and, respectively, M1 or M2 cells.66 In this context it is appropriate to mention that the mouse AKR strain used in this study is strongly type 2 oriented (J.A.V.G., unpublished results, 2001) and, hence, biased toward the development of M2 cells. According to our results, provision of strong type 1 cytokine triggering signals, such as NKG2D ligand–expressing BW-Sp3 cells, may overcome a constitutive type 2–oriented immune system, resulting in the local production of type 1 cytokines (Figure 5) and hereby associated NO-producing M1 cells (Figure 7). In the absence of NK cells this increased production of type 1 cytokines drops to background levels, and concomitantly arginase-producing M2 cells dominate over NO-producing M1 cells. This shift in cell phenotype in the absence of NK cells is further supported by a striking difference in adherent cell morphotype and by a marked increase in expression level and number of CD11b⁺ cells (Figure 6). Hence, in our experimental tumor model NK cells play a crucial role in determining whether an ongoing antitumor response will support alternative or classical activation of macrophages. In agreement with other reports, the occurrence of M2 cells correlated with a limited production of type 1 cytokines and a sustained (IL-10) or increased (IL-4) production of type 2 cytokines.27,66-68 Our cytokine pattern analysis further revealed an intriguing role for MCP-1, a C-C(β) chemokine that was shown to be implicated in the recruitment of tumor-associated macrophages in several experimental setups.69 Indeed, in primary CTL cultures the production of MCP-1 in contrast to type 1 cytokines decreased drastically in function of time after tumor development. In the absence of NK cells, MCP-1 levels remained high throughout the tumor-bearing state and reached similar levels as naive AKR splenocytes. There are substantial data supporting a role for MCP-1 in polarization of type 1/type 2 responses. For instance, MCP-1 was found to specifically enhance the developing immune responses toward a Th2-type reaction by increasing IL-4 production.70 

Besides playing an indirect role in directing type 1–oriented responses and consequently altering the macrophage activation status toward the M1 phenotype, NK cells may also directly control the M1/M2 balance through their selective cytolytic activity. Indeed, our results indicate that M2 cells are better targets for activated NK (LAK) cells than M1 cells. In earlier studies, we and others have shown that NK cells can successfully engage autologous antigen-presenting cells, provided that the appropriate activation signals are present.16-18,71 This process can involve both soluble factors such as IL-1216 and/or optimal ratios of NK-associated activatory and inhibitory signals.16,17,20-26,71,72 Accordingly, we observed in the present study that as compared with M1, M2 cells express lower levels of inhibitory MHC class I, likely involved in conferring resistance to NK lysis26,73; higher expression patterns of B7-costimulatory molecules, implicated in reverting MHC class I–mediated inhibition of NK lytic activity16,17; and higher levels of activating CD11b (CD18) adhesion molecules, formerly described to confer NK sensitivity.71 Thus, M2 cells express a pattern of surface molecules allowing successful engagement by NK cells. As such, our results confer a positive effect of NK-mediated elimination of M2 cells to CTL generation. This could be expected in view of the capacity of M2 cells to counteract proinflammatory signaling, which is crucial during the early phase of CTL responses.

The M2-enriched splenic adherent populations that develop in our experimental conditions suppressed potently the in vitro generation of tumor-specific CTLs. Although M1 cells were amply documented to suppress antitumor responses,74-76 more recent investigations reported on an inhibitory population of adherent CD11b/Gr-1 double-positive cells that cause reversible impairment of tumor-specific CD8⁺ T-cell responses on direct cell-to-cell contact. Immune suppression depended entirely on chronic GM-CSF production that recruited these immature myeloid progenitors to the spleen. Exposure to IL-4 alone increased inhibitory activity, whereas Th1 cytokines sustained classic activation to APC.37-39,77Besides those functional similarities, the myeloid suppressor cells show some important differences from M2 described by others and us.39 The researchers acknowledge the dissimilarities between their Gr-1⁺/CD11b⁺–adherent population and M2, stressing the immature phenotype of their suppressor myeloid cells. Moreover, we can add some additional deviating features: First, our activatory, but not our suppressor, macrophages support the production of GM-CSF in the CTL cultures. This finding coincides with a study that demonstrates that the presence of GM-CSF before vaccination can induce a Th2 response, whereas administration of GM-CSF after vaccination will predominantly enhance Th1 immunity.78Secondly, none of the studies report the production of arginase in their suppressor cells. Finally, the M2 cells in our model do not express F4/80 (A.B.G., unpublished result, 2001) but do express Fas-L, FcRγ (A.B.G., unpublished result, 2001), and B7 in contrast to the myeloid precursor.39 Further characterization of the phenotype of the suppressor cells and mechanism of suppression is currently under investigation.

Collectively, our results corroborate the crucial role of NK cells in the development and maintenance of antitumor CTLs. Our study further points to a new regulatory mechanism possibly implicated in the NK/CTL interaction, namely NK-mediated control of suppressive M2-like populations. This control function may rely on both NK-mediated regulatory (ie, initiation of type 1–oriented signaling) as well as NK-mediated cytolytic (ie, elimination of M2 cells) activities. Hereby, it should be emphasized again that our used strain is strongly type 2 oriented and as such more suitable to uncover NK cell/M2/CTL interactions. Indeed, other mouse strains such as C57bl/6 tend to support type 1 cytokine responses that do not favor M2 cell development.66 Incidentally, numerous studies on NK cell–mediated regulation were performed in the C57bl/6 (B6) background in which anti-NK1.1 antibodies are applicable for NK cell depletion. However, the human immune system tends to be type 2 oriented,79-83 and, furthermore, poor NK cell function is frequently observed in patients with advanced tumors.84,85Hence, a regulatory pathway linking NK cells, M2 cells, and CTLs may be as well operative in human cancer.

We thank L. Brijs, M. Gobert, E. Vercauteren, and E. Omasta for excellent technical assistance; K. Van Beneden for providing anti-NKG2 antibodies; and both Hoffman-LaRoche, and Genetics Institute, for supplying the recombinant IL-12 used for this work.