This study shows that removal of a T cell subpopulation can evoke effective tumor immunity in otherwise nonresponding animals. Elimination of CD25-expressing T cells, which constitute 5–10% of peripheral CD4+ T cells in normal naive mice, elicited potent immune responses to syngeneic tumors in vivo and eradicated them. The responses were mediated by tumor-specific CD8+ CTLs and tumor-nonspecific CD4−8− cytotoxic cells akin to NK cells. Furthermore, in vitro culture of CD25+4+ T cell-depleted splenic cell suspensions prepared from tumor-unsensitized normal mice led to spontaneous generation of similar CD4−8− cytotoxic cells capable of killing a broad spectrum of tumors; reconstitution of CD25+4+ T cells inhibited the generation. In this culture, self-reactive CD25−4+ T cells responding to self peptides/class II MHC complexes on APCs spontaneously proliferated upon removal of CD25+4+ T cells, secreting large amounts of IL-2. The IL-2 thus produced appeared to be responsible for the generation of CD4−8− NK cells as lymphokine-activated killer cells, because direct addition of an equivalent amount of IL-2 to the culture of CD4−8− cells generated similar lymphokine-activated killer/NK cells, whereas coculture of normal CD4−8− cells with CD25−4+ T cells from IL-2-deficient mice did not. Thus, removal of immunoregulatory CD25+4+ T cells can abrogate immunological unresponsiveness to syngeneic tumors in vivo and in vitro, leading to spontaneous development of tumor-specific effector cells as well as tumor-nonspecific ones. This novel way of evoking tumor immunity would help to devise effective immunotherapy for cancer in humans.
There are accumulating demonstrations that many tumor Ags recognized by autologous CTLs are antigenically normal self-constituents (reviewed in ref. 1, 2). Furthermore, immunotherapy of cancer by vaccination with tumor Ags or transfusion of ex vivo propagated cytotoxic lymphocytes often leads to the appearance of autoimmunity because of antigenic cross-reactions between tumor Ags and normal tissue Ags (3, 4, 5). These findings indicate that a part of tumor immunity can be an autoimmunity, and suggest the possibility that the mechanisms maintaining immunologic tolerance to self-constituents may impede generation of effective immunity against autologous tumor cells. If it is the case, a particular way of breaking immunologic self-tolerance may evoke effective tumor immunity in otherwise nonresponding individuals.
Accumulating evidence indicates that not only clonal deletion or anergy, but also T cell-mediated control of self-reactive T cells contributes to the maintenance of immunologic self-tolerance (reviewed in Refs. 6, 7). Indeed, depletion of a CD4+ T cell subpopulation defined by an expression level of a particular cell surface molecule leads to spontaneous development of various autoimmune diseases in otherwise normal rodents. For example, removal of CD5high4+, CD45RB/Clow4+, or CD25+4+ T cells, without deliberate immunization with self Ags, produced in mice or rats autoimmune gastritis, thyroiditis, insulin-dependent diabetes, and other autoimmune diseases immunopathologically similar to human counterparts; reconstitution of the eliminated population prevented the autoimmunities (8, 9, 10, 11, 12, 13, 14, 15). This immunoregulatory activity of CD5high or CD45RB/Clow CD4+ T cells can be attributed to CD25+4+ T cells included in the CD5high or CD45RB/Clow 4+ T cell population (13, 16, 17). Recent studies have shown that the CD25+4+ population in normal naive mice is nonproliferative (anergic) to antigenic stimulation in vitro, and, upon stimulation, potently suppresses the activation/proliferation of other CD4+ or CD8+ T cells in an Ag-nonspecific manner through cell to cell interactions on APCs (16, 17). Furthermore, removal of CD25+4+ T cells not only elicits autoimmunity, but also enhances immune responses to non-self Ags such as allogeneic tissue grafts (13). These in vivo and in vitro findings make it likely that removal of the CD25+4+ immunoregulatory T cells may also evoke effective immune responses to autologous tumor cells in vivo and in vitro by activating effector cells of tumor immunity (18, 19).
In this study, we demonstrate that elimination of CD25+ T cells, which constitute 5–10% of peripheral CD4+ T cells in normal mice and humans (13, 20, 21, 22, 23), indeed elicits potent tumor-specific immune responses to syngeneic tumors in vivo and eradicates them. Furthermore, in vitro culture of CD25+4+ T cell-depleted splenic cell suspensions prepared from tumor-unsensitized normal mice leads to spontaneous generation of active NK cells capable of killing a broad spectrum of tumors. Our results indicate that removal of immunoregulatory CD25+4+ T cells can break immunological unresponsiveness to autologous tumors in vivo and in vitro, leading to spontaneous development of tumor-specific effector cells as well as tumor-nonspecific ones in otherwise nonresponding individuals. This finding is instrumental in devising effective immunotherapy for cancer in humans.
Materials and Methods
BALB/c, BALB/c-nu/nu, C57BL/6 (B6), and C3H/He mice of 6 wk of age were purchased from Japan SLC (Shizuoka, Japan), maintained in our animal facility, and treated in accordance with the institutional guidelines for animal care. IL-2, IL-4, or IFN-γ gene knockout mice (IL-2−/−, IL-4−/−, and IFN-γ−/− mice, respectively (3)) were purchased from The Jackson Laboratory (Bar Harbor, ME) (24, 25, 26). B6-beige (bg/bg)3 mice were purchased from Clea Japan (Tokyo, Japan).
RL♂1 (BALB/c-derived radiation leukemia) (27, 28, 29, 30), B16 (B6-derived melanoma) (31), X5563 (C3H-derived plasmacytoma) (32), and P815 (DBA/2-derived mastocytoma) were gifts from Drs. E. Nakayama (Okayama University, Okayama, Japan), H. Fujiwara (Osaka University, Osaka, Japan), or T. Takahashi (Aichi Cancer Center, Nagoya, Japan).
FITC- or PE-labeled anti-CD25 (7D4) (33), anti-CD4 (GK1.5), anti-NK1.1 (2B4), anti-α/β-TCR (H57-597), anti-γ/δ-TCR (GL3), anti-CD24 (M1/69), or anti-CD16 Ab (2.4G2) was purchased from PharMingen (San Diego, CA). Hybridomas secreting anti-CD4 (RL172.4), anti-CD8 (3.155), or anti-CD25 (PC61) Ab were purchased from American Type Culture Collection (Manassas, VA) . For three-color staining, R-PE-Cy5-conjugated streptavidin (Dako/Japan, Kyoto, Japan) was used as the secondary reagent for biotinylated Abs. For in vivo injection of anti-CD25 mAb, PC61 (rat IgG1) (34) was made ascitic form in SCID mice and purified from the ascites by 40% ammonium sulfate precipitation twice. Purified rat IgG for control injections was purchased from Sigma (St. Louis, MO).
Preparation of T cell subpopulations
Spleen cells were depleted of CD4+, CD8+, or CD25+ cells, as previously described (13, 14). Briefly, spleen cells were incubated at 5 × 106/ml for 45 min at 37°C with the culture supernatant of the hybridoma cells secreting anti-CD4 (RL172.4), anti-CD8 (3.155), or anti-CD25 (7D4) Ab, or the mixture of the supernatants, and rabbit complement (C) 1/10 diluted at the final concentration (Cedarlane Laboratories, Ontario, Canada). The treatment was repeated twice. To purify CD4+ T cells, anti-CD8 plus C-treated cells were removed of B cells and adherent cells by panning on anti-mouse Ig-coated dishes, as previously described (13, 14). To purify CD25+CD4+ T cells, BALB/c spleen cells stained with PE-labeled anti-CD4 (H129.19) (PharMingen) and FITC-labeled anti-CD25 (7D4) were sorted by an EPICS ELITE cell sorter (Coulter Electronics, Miami, FL) with >90% purity, as previously described (16).
In vitro culture of lymphocytes and cytotoxicity assay
Spleen cells (5 × 106 cells in 2 ml) were cultured with or without 5 × 105 mitomycin C-treated tumor cells in DMEM containing 10% FCS in 24-well plates (Costar, Cambridge, MA). Five or seven days later, viable cells were harvested and used as effectors in the 51Cr release cytotoxicity assay, in which target cells (1 × 106) were labeled with 3.7 MBq of sodium chromate (DuPont/NEN, Wilmington, DE) for 60 min at 37°C, washed three times, and incubated at 1 × 104 cells/well with various numbers of effector cells in 96-well round-bottom plates (Costar) for 4 or 6 h at 37°C. The mean percentage specific lysis of triplicate culture was calculated from the radioactivity of the supernatants: percent specific lysis = 100 × [(cpm experimental release − cpm spontaneous release)/(cpm maximum release − cpm spontaneous release)]. Spontaneous release from the target cells incubated in medium alone was always <20% of the maximum release obtained by adding 1 N HCl to the labeled target cells.
To examine the activity of anti-CD4 (RL172.4), anti-CD8 (3.155), anti-α/β-TCR (H57-597), or anti-γ/δ-TCR Ab (GL3) to block the tumor killing mediated by cytotoxic cells, these mAbs as culture supernatants were added to the 4-h 51Cr release assay described above.
In vitro measurement of cell proliferation and IL-2 activity
Spleen cells (5 × 105/well in 96-well flat-bottom plates (Costar)) were cultured for 3 to 10 days, and pulsed with [3H]thymidine (37 kBq/well) (DuPont/NEN) for the last 6 h. IL-2 activity in the culture supernatants was assessed with the CTLL-2 cell line (35). Murine rIL-2 (3.89 × 106 U/mg) was a gift from Shionogi (Osaka, Japan).
To examine blocking of in vitro proliferation with anti-class II MHC Ab, graded concentrations of pan anti-class II MHC mAb CA4 (36) (a gift from Dr. O. Kanagawa, Washington University, St. Louis, MO) were added to the culture of CD25− spleen cells.
Histological and serological assessment of autoimmune disease
Tissues and organs (thyroid, lung, pancreas, stomach, adrenal gland, kidney, ovaries, or testes) were fixed in 10% Formalin and processed for hematoxylin and eosin staining. Gastritis was graded 0 to 2+, depending on macroscopic and histological severity: 0 = the gastric mucosa was histologically intact; 1+ = mild gastritis with histologically evident destruction of parietal cells and cellular infiltration of the gastric mucosa; 2+ = severe destruction of the gastric mucosa accompanying the formation of giant rugae due to compensatory hyperplasia of mucous-secreting cells (see Ref. 8 for the giant rugae) (13, 14). Thyroiditis was also graded 0 to 2+: 0 = the thyroid gland was histologically intact; 1+ = mild thyroiditis with histologically evident destruction of the thyroid follicles and interstitial infiltration of inflammatory cells; 2+ = severe destruction of the thyroid gland accompanying goiter formation (see Ref. 8 for the goiter) (13, 14). The ELISA (using alkaline phosphatase-conjugated secondary Ab and p-nitrophenyl disodium hexahydrate as the substrate) for detecting autoantibodies specific for the gastric parietal cell Ags or mouse thyroglobulins was previously described (13, 14).
In vivo induction of effective tumor immunity by removing CD25+4+ T cells
To determine whether elimination of CD25+4+ cells is able to elicit primary immune responses to syngeneic tumors, we transferred to BALB/c athymic nude mice BALB/c splenic cell suspensions depleted of CD25+ cells (hereafter designated to be CD25− cells (see the insets in Fig. 1,A for cytofluorometric profiles)), and then, on the same day, s.c. transplanted BALB/c-derived RL♂1 leukemia cells (Fig. 1,A). In the majority of mice, the tumors first grew and then regressed within 1 mo, allowing the hosts to survive a long-term (>80 days), whereas all the nude mice transferred with nondepleted spleen cells or the mixture of CD25− cells and CD4+ T cells (10% of which were CD25+ cells) in the ratio 3:1 died of tumor progression within 40 days (Fig. 1,B). Upon rechallenge with larger doses of RL♂1, the CD25− cell-transferred nude mice rejected the tumor inocula more rapidly and vigorously than the primary rejection, indicating that they had become immune to the tumor (Fig. 1 A, middle panel).
In the nude mice having rejected RL♂1, in vitro stimulation of their spleen cells with RL♂1 revealed two types of cytotoxic cells (type I and II in Fig. 2, A and B). The stimulated spleen cells from some mice specifically killed RL♂1, but not allogeneic tumor cells such as B6-derived B16 melanoma cells (type I), while similarly stimulated spleen cells from other mice killed not only RL♂1 but also B16 (type II). The former type required in vitro stimulation with RL♂1 to become detectable, whereas the latter did not. Furthermore, the former RL♂1-specific killing was mediated by CD8+ CTLs: elimination of CD8+ cells completely abolished the killing activity, but that of CD4+ cells did not (Fig. 2,C). On the other hand, the latter promiscuous killing activity appeared to be mainly mediated by CD4−8− cells because elimination of both CD4+ cells and CD8+ cells did not abrogate the activity (Fig. 2 C). In these nude mice having developed CD4−8− killer cells, the retention of the killer activity after eliminating CD8+ cells from the cultured spleen cells does not necessarily mean that they did not harbor CD8+ CTLs, because tumor-specific CD8+ CTL clones could be easily prepared in vitro from their spleens (29 , and our unpublished data).
These results taken together indicate that elimination of CD25+ T cells alone, without prior deliberate immunizations, can evoke potent immune responses to tumor cells in vivo by generating two types, at least, of effector cells, which are different in the cell surface phenotype, the specificity of tumor killing, and the requirement of Ag stimulation for their development.
In vivo induction of tumor immunity by administration of anti-CD25 mAb
To determine whether transient elimination of CD25+ cells from normal mice can also elicit immunity to syngeneic tumors, we administered anti-CD25 mAb (PC61) twice (on 4 and 2 days before tumor inoculation) to BALB/c or B6 mice, and subsequently inoculated RL♂1 or B16 tumor cells, respectively (Fig. 3). In the majority (>90%) of PC61-treated BALB/c mice, the s.c. inoculated RL♂1 first grew and then regressed within 1 mo, whereas all of the BALB/c mice treated with normal rat Ig as a control died of tumor progression within 1 mo. Likewise, administration of PC61 to B6 mice significantly suppressed the growth of B16 when compared with control B6 mice treated with normal rat IgG, allowing the former to survive a longer term (>60 days) compared with the latter (<40 days). This PC61 treatment reduced the number of peripheral CD25+4+ T cells to a quarter of control mice for nearly 1 mo (our manuscript in preparation). Furthermore, this in vitro anti-CD25 treatment evoked rejection of various other syngeneic tumors in various other strains (S. Onizuka, et al., manuscript in preparation).
In vitro induction of cytotoxic cells from normal spleen cells by removing CD25+4+ T cells
To examine whether similar cytotoxic lymphocytes as detected in vivo (Figs. 1 and 2) can also be generated in vitro by simply eliminating CD25+4+ T cells, CD25− splenic cell suspensions prepared from tumor-unsensitized normal BALB/c mice were cultured for several days with or without RL♂1 cells and assessed for killing activity (Fig. 4). Cells harvested on day 5 from the RL♂1-stimulated culture indeed exhibited significant killing activity against RL♂1 cells (Fig. 4,A). Interestingly, when CD25− spleen cells were cultured for 7 days, potent cytotoxic activity was detected even in the cultures not stimulated with RL♂1. By contrast, similarly cultured nondepleted spleen cells or the mixtures of CD25− cells and purified CD25+CD4+ T cells (in the ratio 25:1 (see the inset in Fig. 4,A)) showed no cytotoxicity. Furthermore, the cells harvested from the 7-day culture killed not only RL♂1, which is NK sensitive (27), but also NK-resistant tumor cells (such as P815) and allogeneic ones (such as X5563 and B16) (Fig. 4,B). This promiscuous killing activity could not be toward calf serum proteins adsorbed to the target cells during 51Cr labeling, because tumor cells passaged in vivo and 51Cr labeled in the medium containing normal mouse serum, instead of FCS, were lysed as well by CD25− cell-derived cytotoxic cells (data not shown). These findings were not confined to the BALB/c strain: CD25− cells from other strains (including B6 and C3H), when cultured for 1 wk without tumor cell stimulation, also gave rise to cytotoxic cells promiscuously killing a broad spectrum of syngeneic or allogeneic tumor cells (Fig. 4 B).
Characterization of cytotoxic cells generated in vitro upon removal of CD25+4+ T cells
As shown in Fig. 4, there was no significant difference in the killing activity or the killing spectrum of target tumor cells between CD25− cells in vitro stimulated with tumor cells for 7 days and those nonstimulated (see the inset in Fig. 4 A). We therefore analyzed in the following experiments the mechanism by which killer cells were generated from CD25− cells cultured for 1 wk without in vitro antigenic stimulation.
Fig. 5 A shows that depletion of either CD4+ cells or CD8+ cells, or both, from the cell suspensions harvested from 1-wk culture of CD25− BALB/c spleen cells did not abrogate the cytotoxic activity against RL♂1 or B16. The result indicates that the promiscuous killing activity was mainly in the CD4−8− population, although slight reduction of the killing activity by depleting CD8+ cells suggests that tumor-nonspecific CD8+ cytotoxic cells might also be generated in the culture.
We then examined whether the activity could be blocked by adding anti-CD4, CD8, α/β-TCR, or γ/δ-TCR Ab to the 51Cr release assay (Fig. 5 B). None of the Abs blocked the activity, in contrast to significant degrees of blocking exerted by anti-CD8 or anti-α/β-TCR Ab on allospecific CD8+ CTLs generated in vitro by stimulating normal B6 spleen cells with BALB/c-derived RL♂1.
The lymphocytes harvested from 1-wk culture of B cell-depleted CD25− spleen cells prepared from B6 mice were composed of CD4+ T cells (∼60%), CD8+ T cells (∼30%), and CD4−8− cells (∼10%) (Fig. 5,C). The CD4−8− population contained α/β-TCR+ cells (∼50%), γ/δ-TCR+ cells (∼20%), and NK1.1+ cells (∼30%) (Fig. 5 C). The population was mostly (>80%) B220+, and contained CD122 (IL-2R β-chain)+ cells (∼70%), CD16+ cells (∼5%), and CD24 (heat stable antigen (HSA))+ cells (∼5%) (data not shown).
Furthermore, culturing CD25− splenic cells prepared from B6-bg/bg mice, which are genetically defective in NK cell-killing capacity (37), failed to generate the promiscuous cytotoxic activity in the 1-wk culture, in contrast to intact generation of allo-specific CTL activity by B6-bg/bg mouse spleen cells stimulated with allogeneic RL♂1 cells (Fig. 6).
The mechanism of spontaneous in vitro generation of CD4−8− cytotoxic cells
In the course of this in vitro generation of cytotoxic cells, CD25− cells, especially CD25−4+ T cells, showed spontaneous and vigorous proliferation and secreted large amounts (∼500 U/ml) of IL-2, whereas noneliminated spleen cells did not (Fig. 7, A and B). This response was not against xenogeneic proteins in the culture medium, because the use of normal BALB/c serum, instead of FCS, did not alter the response (data not shown). On the other hand, the proliferative response was blocked by adding anti-class II MHC Ab to the culture, indicating that the proliferating CD4+ T cells were presumably responding to self peptides/class II MHC complexes on APCs (Fig. 7 C).
The in vitro generation of CD4−8− cytotoxic cells appeared to require the presence of these self-reactive CD25−4+ T cells, because depletion of CD4+ T cells from the CD25− population before culture abrogated the generation (Fig. 7,D, treatment d), whereas removal of CD8+ T cells did not (Fig. 7,D, treatment c). On the other hand, addition of IL-2 (1000 U/ml) to the culture of such CD4+ cell-depleted CD25− cells (including CD25−8+ cells and CD4−8− cells) or CD4−8− cells (Fig. 7,D, treatment f and g, respectively) generated marked cytotoxic activity, which was comparable with the activity generated by culturing uneliminated spleen cells with the same dose of IL-2 (Fig. 7 D, treatment h).
To further examine the key role of IL-2 secreted by CD4+ T cells in generating CD4−8− cytotoxic cells, CD25−4+ T cells prepared from IL-2−/−, IL-4−/−, or IFN-γ−/− mice on BALB/c background were mixed with an equal number of CD4−8− cells from normal BALB/c mice and then assessed for cytotoxic activity after 1-wk culture (Fig. 8). CD25−4+ T cells from IL-2−/− mice failed to generate the activity, whereas those from IL-4−/− or IFN-γ−/− mice generated nearly equivalent levels of killer activity as with CD25−4+ T cells from normal BALB/c mice.
Thus, elimination of CD25+ cells leads to spontaneous activation/proliferation of CD25−4+ self-reactive T cells; and large amounts of IL-2 produced by them may be responsible for the generation of CD4−8− cytotoxic cells, which are similar to LAK cells (or IL-2-activated NK cells (Fig. 7 D, treatment h)) in their promiscuous cytotoxic activity and the critical dependency of their generation upon IL-2 (38, 39, 40, 41, 42).
T cell subpopulations required for autoimmunity
The CD25− cell-transferred nude mice that had rejected tumor transplants (as shown in Fig. 1,A) developed histologically evident autoimmune diseases, including gastritis (100% incidence, n = 14), thyroiditis (42.9%), and oophoritis (64.3%) accompanying respective autoantibodies (such as those specific for gastric parietal cells, thyroglobulins, or oocytes (8, 13, 14)) (Fig. 9). Transfer of in vitro cultured CD25− cells also produced similar autoimmune diseases at similar incidences, whereas transfer of CD4−8+ or CD4−8− cells, or both, prepared by eliminating CD4+ cells from the cultured CD25− cells, did not. Thus, CD4+ T cells were indispensable for producing these organ-specific autoimmune diseases; and CD8+ or CD4−8− cells alone were unable to cause the autoimmunity.
In contrast to these CD25− cell-inoculated BALB/c nude mice, normal BALB/c or B6 mice treated with PC61 (Fig. 3) did not develop histologically or serologically evident autoimmunity (S.Y., J.S., S.S., unpublished data).
The main conclusion in this study is that removal of immunoregulatory CD25+4+ T cells can break immunological unresponsiveness to syngeneic tumor cells in vivo and in vitro. The result indicates that tumor immunity and autoimmunity share a common regulatory basis, because removal of the population also elicits autoimmunity in vivo (Fig. 9) and in vitro (Fig. 7 A).
It has been postulated that one of the elements impeding effective tumor immunity in tumor-bearing hosts may be concomitant development of a T cell population suppressing the generation or the action of tumor-killing effector cells (43, 44, 45). Although some of such suppressor T cells were shown to be CD4+, they have been elusive from further characterization because of the lack of reliable markers specific for them (46). Our results indicate that these suppressive T cells, at least in part, can be CD25+4+ T cells. The CD25+4+ immunoregulatory T cells, however, bear several characteristics distinct from the suppressor T cells to date reported in tumor immunity (43, 44). First, removal of CD25+4+ T cells before tumor implantation was effective in evoking specific tumor immunity in vivo (Figs. 1 and 3) and in generating cytotoxic cells in vitro (Fig. 4). Second, their removal elicited not only tumor immunity, but also various autoimmunities in vivo (Fig. 9) and self-reactivity in vitro (Fig. 7). The removal also enhanced immune responses to allografts or xenogeneic proteins (13). Third, they are continuously produced by the normal thymus and already functional in the thymus, as previously shown (14, 16, 17); and their depletion from thymocyte suspensions indeed generated CD4−8− NK-like killer cells in vitro (our unpublished data) and autoimmunity in vivo (17). Furthermore, these CD25+4+ T cells/thymocytes are naturally anergic (nonproliferative) to antigenic stimulation in vitro, and, upon stimulation, potently suppress the activation/proliferation of other CD4+ or CD8+ T cells in an Ag-nonspecific manner through cell to cell interactions on APCs (16, 17). These findings when taken together indicate that the CD25+4+ immunoregulatory T cells present in normal naive mice may be engaged in continuously up-regulating the activation thresholds of other T cells, thereby impeding effective generation of tumor immunity while inhibiting autoimmunity (16, 17, 18, 19). To further elucidate how they suppress tumor immunity (and autoimmunity), the natural ligands physiologically stimulating them must be characterized. Judging from their activated phenotype (e.g., being CD25+, CD45RBlow, CD44high, and CD5high (8, 9, 11, 13, 16, 17)) in normal naive mice, the CD25+4+ immunoregulatory T cells might be inherently reactive with, and continuously stimulated by self Ags in the normal internal environment, although they themselves may be harmless because of their anergic property (16, 17).
Removal of CD25+4+ immunoregulatory cells led to the activation of not only tumor-specific CD8+ CTLs (and presumably tumor-specific CD4+ Th cells), but also tumor-nonspecific CD4−8− effector cells in vivo, and predominantly the latter in vitro. It also led to in vitro vigorous proliferation of CD4+ self-reactive T cells presumably responding to self peptides/class II MHC expressed on APCs, as in the autologous or syngeneic mixed lymphocyte reaction (47, 48). For the following reasons, it is the large amounts of IL-2 formed by such CD4+ self-reactive T cells that may be responsible for the generation of the CD4−8− effector cells, at least in vitro. First, similar cytotoxic cells could be generated by adding an equivalent amount of IL-2 to the culture of normal spleen cells or CD4−8− cells. Second, CD25−4+ T cells from IL-2-deficient mice failed to generate the cytotoxic cells from the CD4−8− population in normal mice. Third, the TCR− CD4−8− cytotoxic cells generated by culturing CD25− cells killed not only NK-sensitive target cells (such as RL♂1 (27)), but also NK-resistant ones (such as P815), as IL-2-induced LAK cells can kill a broader spectrum of tumors than NK cells (49). The majority of such CD25− cell-derived killer cells indeed appeared to be NK cells in terms of their cell surface phenotype (e.g., NK1.1+, B220+), marked IL-2 dependency in their generation, and the lack of tumor-killing activity in beige mice (38, 39, 40, 41, 42). Taken together, these results indicate that active NK cells as LAK cells can be easily generated in vitro by simply eliminating a T cell subpopulation, without exogenous IL-2.
It remains to be determined, however, how in vivo removal of CD25+4+ T cells leads to the generation of two distinct types of effector cells (tumor-specific CD8+ CTLs and tumor-nonspecific CD4−8− NK-like cells (Fig. 2)), or whether CD4+ self-reactive T cells play key roles in the in vivo generation of tumor immunity as well by secreting large amounts of IL-2. By in vivo administration of anti-CD25 Ab (Fig. 3), CD4−8− LAK/NK-like cells indeed appeared within 1 wk, followed by dominant development of tumor-specific CD8+ CTLs, which seemed to play main roles in ultimate tumor rejection (our manuscript in preparation). The result suggests that in vivo removal of CD25+4+ cells may first activate LAK/NK cells by activating CD4+ self-reactive T cells and instigating their IL-2 secretion; and tumor-nonspecific killing by such LAK/NK cells and consequent release of tumor Ags, together with the IL-2 secreted by CD4+ self-reactive T cells, may contribute to the subsequent development of tumor-specific CD8+ CTLs. Alternatively, although mutually not exclusive, the removal of CD25+4+ cells may enhance the development of both LAK/NK cells (through activating CD4+ self-reactive T cells) and CD8+ CTLs (and tumor-specific CD4+ Th cells) at the same time, because the immunoregulatory CD25+4+ T cells can directly suppress the activation/proliferation of both CD4+ T cells and CD8+ T cells, at least in vitro (16, 17). Further analysis of the in vivo interplay of the two types of effector cells in tumor immunity may contribute to our devising effective ways to generate them in vitro, especially tumor-specific CD8+ CTLs.
A critical issue in our study is how tumor immunity can be evoked without autoimmunity when the CD25+4+ immunoregulatory T cells are manipulated. The present and previous studies by us and others have shown that, upon introduction of abnormal control of self-reactive T cells, the intensity and the range of the autoimmune responses (i.e., which self Ags are more prone to be aggressed, or which self-reactive clones are more prone to be activated) depend on the effector T cell subpopulations involved, the degree and duration of depleting CD25+4+ T cells, and, most importantly, the host genetic background, which includes MHC as well as non-MHC genes (7, 13, 18, 50, 51, 52, 53). For example, when CD4+ T cells are depleted from in vitro cultured CD25− cells and the remaining cells, including CD8+ or CD4−8− cytotoxic cells, are adoptively transferred, no histologically and serologically evident autoimmune disease developed in the recipients (Fig. 9), presumably because CD4+ effector T cells are required for the development of this organ-specific autoimmunity as in many other models (8, 9, 10, 11, 12, 13, 14, 50, 51, 52, 53, 54). Furthermore, in BALB/c mice, which are genetically prone to develop organ-specific autoimmune diseases (in particular autoimmune gastritis (13, 14, 18, 50, 51, 52, 53)), limiting the period or the degree of depleting CD25+ T cells, for example, by in vivo administration of anti-CD25 Ab for a short period, could evoke effective tumor immunity without autoimmunity (Fig. 3, and our manuscript in preparation). In genetically autoimmune-resistant B6 mice (18, 50, 51, 52, 53), on the other hand, complete depletion of CD25+4+ T cells from the cell inocula (as shown in Fig. 1 A) led to tumor rejection, but failed to cause any autoimmune disease in B6 nude mice (our unpublished result). To further differentiate tumor immunity from autoimmunity, it is required to elucidate the molecular basis of the CD25+ T cell-mediated immunoregulation and the host genes determining the susceptibility/resistance to autoimmunity.
Thus, with a caveat of possible autoimmunity in certain situations (see discussion above), the present findings in mice may help to devise a novel immunotherapy for cancer in humans or to make the current immunotherapies more effective. For example, administration of anti-CD25 Ab to cancer-bearing hosts for a limited period may evoke or enhance tumor immunity. Removal of CD25+4+ T cells from PBL or tumor-infiltrating T cells before their in vitro culture with IL-2 may lead to production of more potent or a larger number of cytotoxic cells, including CTLs and LAK/NK cells (40, 45, 55). Furthermore, such in vivo or in vitro elimination of the CD25+4+ immunoregulatory T cells may enhance tumor immunity when combined with the current attempts to augment immunogenicity of tumor cells, for example, by cytokine gene transduction in tumor cells, or vaccination with tumor Ags/peptides or Ag-pulsed dendritic cells (56, 57).
We thank Ms. E. Moriizumi for histology; Dr. N. Sorimachi for staining reagents and discussion; and Drs. T. Takahashi, E. Nakayama, and H. Fujiwara for gift of tumor cells. This work was reported at the annual meeting of Japanese Cancer Society on Oct. 1, 1998, and Keio University International Symposia for Life Sciences and Medicine on Jan. 21, 1999.
This work was supported by grants-in-aid from the Ministry of Education, Science, Sports, and Culture; the Ministry of Human Welfare (for Longevity Sciences 10-01); and the Organization for Pharmaceutical Safety and Research (OPSR) of Japan.
Abbreviations used in this paper: bg, beige; C, complement; LAK, lymphokine-activated killer.