Cooperative action of CD8 T lymphocytes and natural killer cells controls tumour growth under conditions of restricted T-cell receptor diversity

Abstract

In mice expressing a transgenic T-cell receptor (TCR; TCRP1A) of DBA/2 origin with reactivity towards a cancer-germline antigen P1A, the number of TCRP1A CD8⁺ T cells in lymphoid organs is lower in DBA/2 than in B10.D2 or B10.D2(× DBA/2)F₁ mice. This reduction results from haemopoietic cell autonomous differences in the differentiation of the major histocompatibility complex class I-restricted TCRP1A thymocytes controlled by DBA/2 versus B10.D2-encoded genes. We report here that the lower number of TCRP1A CD8⁺ T cells in DBA/2 mice correlated with their poor resistance to P1A-expressing mastocytoma solid tumours. Functional potency of CD8⁺ cytolytic T lymphocytes (CTL) from the above strains was not compromised, but their number after expansion appeared to be influenced by their genetic background. Intriguingly, non-transgenic DBA/2 mice resisted P1A⁺ tumours more efficiently despite poor representation of P1A-specific CTL. This was partly the result of their more heterogeneous TCR repertoire, including reactivity to non-P1A tumour antigens because mice that had rejected a P1A⁺ tumour became resistant to a P1A⁻ variant of the tumour. Such ‘cross-resistance’ did not develop in the TCRP1A transgenic mice. Nonetheless, reconstitution of RAGº^/º mice with TCRP1A CD8⁺ T cells, with or without CD4⁺ T cells, or exclusive representation of TCRP1A CD8⁺ T cells in RAGº^/º TCRP1A transgenic mice efficiently resisted the growth of P1A-expressing tumours. Natural killer cells present at a higher number in RAGº^/º mice also contributed to tumour resistance, in part through an NKG2D-dependent mechanism. Hence, in the absence of a polyclonal T-cell repertoire, precursor frequencies of natural killer cells and tumour-specific CTL affect tumour resistance.

Introduction

CD8⁺ T lymphocytes are the main cytolytic effectors against cellular targets expressing cognate tumour antigen. Characteristically, an antigen-specific cytotoxic T-lymphocyte (CTL) response consists of three phases: (i) a clonal proliferation of antigen-specific T cells, (ii) acquisition of effector functions by T cells resulting in elimination of target cells, and (iii) a contraction of the effector phase into the maintenance phase, in which the surviving antigen-specific T cells form the long-lived memory CD8⁺ T-cell population.^1–4 The circumstances leading to full expression of the CTL effector programme as opposed to the partial activation or induction of tolerance in situations of peripheral T-cell stimulation are still poorly understood. After initial antigenic exposure, the transition through these three stages of CD8⁺ T cells is directed by an instructional programme that commits CD8⁺ T cells to proceed with their differentiation.^5–8 However, the initial parameters of affinity and kinetics of T-cell receptor (TCR) engagement by cognate antigenic peptide–major histocompatibility complex (MHC) complex^9–11 and antigen persistence¹² appear to determine whether clonal expansion leads to a full¹¹ or abortive¹² activation programme.

T-lymphocyte reactivity toward autologous MHC-associated tumour antigens comprises a particular situation in that many of the identified tumour-associated antigens correspond to peptides derived from unmutated self-proteins. Two major types of such antigens have been identified that derive from (i) tissue-specific differentiation antigens or (ii) a class of ‘cancer-germline’ gene products, that, akin to the first identified MAGE gene, are encoded on the X chromosome and have restricted expression in gametogenic tissue and tumour cells (reviewed in refs ^13,14). In the mouse, the P1A antigen¹⁵ characteristic of such cancer-germline genes, has been shown to be the major rejection antigen of DBA/2-derived mastocytoma P815¹⁶ and is known to encode a nonapeptide presented by H-2L^d (L^d) to CD8⁺ CTL that can kill P815 cells.¹⁷ As a tool to trace CD8⁺ T-cell reactivity toward the model cancer-germline antigen P1A, we generated mice expressing as a transgene the TCR (TCRP1A) from P1A-specific CTL clone P1.5 of DBA/2 origin¹⁷ in DBA/2, B10.D2 and RAG-deficient B10.D2 backgrounds. These strains of TCRP1A transgenic (tg) mice presented a varying frequency of TCRP1A T cells in thymus and the peripheral lymphoid organs. This was the result of poor selection of TCRP1A CD8⁺ T cells in the DBA/2 background that was regulated by a haemopoietic cell-autonomous genetic control of the patterns of MHC class I-restricted TCRP1A thymocyte differentiation, independent of the thymic microenvironment.¹⁸

Using these TCRP1A tg mice, we investigated the requirements for immune resistance to the growth of mastocytoma solid tumours, including the impact of T-cell precursor frequency on immune resistance to tumour. We observed that the lower number of TCRP1A CD8⁺ T lymphocytes in DBA/2 as compared with (DBA/2 × B10.D2)F₁ tg mice correlated with their poorer resistance to the growth of P1A-expressing tumours. We investigated further whether this lower resistance to tumour growth was the result of poor representation of TCRP1A CTL, functional defects in the TCRP1A CTL, or non-CTL intrinsic factors in DBA/2 tg mice. Assessment of the reactivity of TCRP1A CTL from the above strains in vitro and in vivo revealed that the functional potency of DBA/2 TCRP1A CTL was not compromised, but the number of expanded TCRP1A CTL appeared influenced by their genetic background. Moreover, reconstitution of RAG-1º^/º mice with TCRP1A CD8⁺ T cells, with or without CD4⁺ T cells, or exclusive representation of TCRP1A CD8⁺ T cells in TCRP1A RAG-1º^/º B10.D2 mice provided efficient resistance to the growth of P1A-expressing tumours. This resistance to tumour in RAG-1º^/º mice was also partly mediated by natural killer (NK) cells through an NKG2D-mediated mechanism. This study illustrates how, under conditions of restricted TCR diversity, tumour resistance can be affected by the variation in precursor frequencies for tumour-specific CD8⁺ T-cell effectors and NK cells, and that NK cells can contribute to tumour resistance.

Materials and methods

Mice

Mice heterozygous for the H-2L^d/P1A_35–43-specific TCR transgene (TCRP1A) on the DBA/2, B10.D2 and RAG-1º^/ºB10.D2 backgrounds (TCRP1A DBA/2, B10.D2 and RAG-1º^/ºB10.D2, respectively), TCRP1A B10.D2(× DBA/2) F₁ (litter from mice after four backcross generations on TCRP1A B10.D2 crossed with DBA/2 mice, and hereafter referred to as TCRP1A F₁) have been previously described.¹⁸ These mice as well as the wild-type (wt) DBA/2 (Charles River Breeding Laboratories, Les Oncins, France) and B10.D2 mice (Harlan France SARL, Gannat, France) were bred in the animal facility of the Centre d’Immunologie de Marseille-Luminy (Marseille, France) and used when between 5 and 8 weeks old. All animal experiments were in accordance with protocols approved by the French and European directives.

Tumour transplantation and monitoring

Mastocytoma P815 (H-2^d, DBA/2 origin) sublines, namely, P511 expressing P1A (P1A⁺ P511) and P1.204 deficient in P1A (P1A⁻ P1.204) but similar for other costimulatory and MHC expression characteristics¹⁹ were transplanted (10⁶ cells subcutaneously) in TCRP1A DBA/2, TCRP1A F₁, TCRP1A RAG-1º^/ºB10.D2 or non-tg (wt) DBA/2 or RAG-1º^/ºB10.D2 mice. All these recipient mice (except for the immunodeficient RAG-1º^/ºB10.D2 mice) carried at least one copy of the DBA/2 alleles and did not, therefore, react to minor histocompatibility antigens on the P815 mastocytoma. The tumour growth was monitored every fourth day. The product of millimetre measurements of the two perpendicular diameters, measured with Vernier callipers, was used to estimate tumour size and the survival of mice was scored as days from transplantation until death or a tumour size of 400 mm² (at which time they were killed) within a period of at least 90 days. Both the tumours showed a well-localized and regularly shaped solid form, which facilitated accurate measurement. In some experiments, wt and TCRP1A tumour survivor mice were reinjected subcutaneously with equal numbers (10⁶) of P1A⁺ P511 on the right flank and P1A⁻ P1.204 on the left flank and the tumour growth was monitored. Survival curves were plotted using graphpad prism (GraphPad Software, San Diego, CA).

Cell preparation and T-cell stimulation in vitro

Cells were prepared from lymph nodes (LN) and spleens according to standard procedures. Irradiated T-depleted splenocytes were used as antigen-presenting cells (APCs). The T cells were purified from LN of TCRP1A DBA/2, F₁ and B10.D2 mice or wt mice by negative selection using rat anti-CD4 monoclonal antibody (mAb; H129.19.6) or anti-CD8 mAb (53-6-72) and anti-B220 mAb (RA3.6B2) supernatants and anti-rat immunoglobulin G (IgG) Dynabeads (Dynal, Oslo, Norway). Purified CD8⁺ or CD4⁺ T cells represented 90–98% of the enriched population. Naïve TCRP1A CD8⁺ T cells (0·5 × 10⁶) were stimulated in vitro for 3 days in 24-well plates (Costar, Becton Dickinson, Le Pont De Claix, France) with syngeneic wt T-depleted APCs (10⁶) loaded with different doses of relevant P1A_35–43 peptide (LPYLGWLVF)¹⁷ or an irrelevant lymphocytic choriomeningitis virus nucleoprotein NP_118–126 peptide. In some T-cell stimulation experiments, bone marrow-derived dendritic cells (DCs) were used as APCs. DCs were generated following the published procedure.²⁰ Stated briefly, the bone marrow cells of DBA/2 mice were cultured for 3 days in Dulbecco’s modified Eagle’s minimum essential medium supplemented with 10% fetal calf serum, antibiotics, 2 mm glutamine, 50 μm 2-mercaptoethanol and 30% conditioned medium from NIH3T3 cells containing granulocyte–macrophage colony-stimulating factor. They were then diluted 1 : 1 in the same medium and after an additional period of 10 days of culture, plastic non-adherent cells were washed, resuspended in supplemented RPMI-1640, and used as APCs after loading with P1A or irrelevant peptide in a 1 : 1 ratio with T cells.

Antibodies, surface and intracellular immunofluorescence staining

Antibodies used for surface immunofluorescence staining were: fluorescein isothiocyanate-labelled anti-CD69, anti-CD25, anti-CD44 and anti-CD62L; peridinin chlorophyll protein-Cychrome 5.5-labelled anti-CD8 and anti-CD69; phycoerythrin-labelled anti-NK1.1; and allophycocyanin-labelled anti-CD3, anti-CD8 and anti-CD62L (BD PharMingen, San Diego, CA). After a pre-incubation of cells with anti-FcRII/III mAb supernatant (2.4.G2) for 15 min at 4°, 0·5 × 10⁶ to 1 × 10⁶ cells were incubated with a desired combination of the indicated surface immunofluorescence staining antibodies with or without phycoerythrin-labelled H-2L^d/P1A_35–43 tetrameric reagent (L^d/P1A-tet)¹⁸ for 30 min at 4°. After two washes, cells were fixed in 1% paraformaldehyde and run on a FACSCalibur cytofluorometer (BD Biosciences, Mountain View, CA). Data were analysed using cellquest (BD Biosciences) or flowjo (Treestar Inc., Sunnyvale, CA) softwares.

CFSE staining

The T-cell divisions were quantified by flow cytometry using the intracytoplasmic stable 5-,6-carboxyfluorescein diacetate succinimidyl ester (CFSE) dye that exhibits sequential halving of the fluorescence intensity at each division step.²¹ Purified TCRP1A CD8⁺ T cells were incubated for 10 min at 37° with 5 μm CFSE (Molecular Probes, Eugene, OR) and washed before in vitro culture or injection in mice.

Interferon-γ production

Interferon-γ (IFN-γ) in T-cell culture supernatants harvested following activation on day 3 was determined by enzyme-linked immunosorbent assay (ELISA) using mAb AN18 with biotin-conjugated R46A2 (PharMingen, San Diego, CA). The ELISA was revealed enzymatically with avidin-alkaline phosphatase and its substrate p-nitrophenyl di-sodium phosphate (Sigma Co., St Louis, MO). Results are expressed as IFN-γ U/ml in the supernatant with reference to a standard curve obtained with recombinant IFN-γ (Genzyme Corp., Cambridge, MA).

Cytotoxicity assays

Cytolytic activity of T lymphocytes was assayed by incubating purified stimulated or unstimulated TCRP1A CD8⁺ T cells with P1A⁺ P511 or P1A⁻ P1.204 tumour target cells labelled with ⁵¹Cr (New England Nuclear, Boston, MA). Targets (10⁴) were incubated with effector cells at 37° and the ⁵¹Cr release was assayed after 5·5 hr.

Adoptive T-cell transfer, NK cell depletion and NKG2D blocking in vivo

A total of 0·5 × 10⁶ to 5 × 10⁶ purified CD4⁺ T cells and/or TCRP1A CD8⁺ T cells were resuspended in 0·1 ml of phosphate-buffered saline and injected intravenously in RAG-1º^/º.B10.D2 recipients 7 days before P1A⁺ P511 or P1A⁻ P1.204 tumour cell injection. TCRP1A or T-cell-transferred RAG-1º^/º mice were injected intraperitoneally with 200 μg mAb per injection of anti-NK1.1 (PK136)²², an equivalent amount of isotype-matched mouse IgG2a/κ (Ti98 anti-TCRBM3.3 or Désiré-1 anti-KB5.C20TCR; courtesy of M.B.), or nothing on days −1 and +1 of tumour injection. In some experiments, TCRP1A or T-cell-transferred RAG-1º^/º mice were injected intraperitoneally with 200 μg mAb per injection of anti-NKG2D (MI-6;²³ courtesy of David Raulet, Berkeley, CA), an equivalent amount of isotype-matched rat IgG2a/λ (anti-B20.2; courtesy of Claude Grégoire, CIML, Marseille, France), or nothing on days −1, +1 and +8 of tumour injection.

Statistical analysis

Statistical significance of the differences between the test groups was analysed by one-way analysis of variance using instat software (GraphPad software). Significance between the survival curves was analysed by Log-rank test using graphpad prism (GraphPad Software). Values of P<0·05 were considered statistically significant.

Results

Frequency of TCRP1A CD8⁺ T cells in transgenic mice impacts P815 tumour resistance

Mice expressing TCRP1A transgene in DBA/2 and B10.D2 backgrounds differed in the frequency of TCRP1A CTL in the peripheral lymphoid organs (Fig. 1a). This was the result of poor selection of TCRP1A on CD8⁺ T cells in the DBA/2 background as we reported previously.¹⁸ To assess if the poor selection of TCRP1A CTL in DBA/2 mice had any effect on growth of P1A antigen-expressing tumour cells, we injected mastocytoma P815 sublines, namely, P1A⁺ P511 and control P1A⁻ P1.204 subcutaneously (10⁶ each) in TCRP1A DBA/2, TCRP1A B10.D2(× DBA/2)F₁ (to avoid any reactivity to minor histocompatibility antigens on P815 cells) and non-tg wt DBA/2 mice. The onset of P511 tumour outgrowth was noticed on the 4th day after tumour transplantation in both wt and TCRP1A DBA/2 mice. By day 8, in the wt mice the tumour grew to 50 mm² whereas in TCRP1A DBA/2 mice the tumour size was less than 25 mm² (Fig 1b). In TCRP1A F₁ mice, where representation of TCRP1A CD8⁺ T cells is higher compared with TCRP1A DBA/2¹⁸ (Fig 1a), the tumour onset was delayed until day 11 and the tumour reached a size of 50 mm² around day 15. There was no significant difference in the pattern of the growth of control tumour P1.204 in the TCRP1A groups but there was a higher incidence of tumour rejection in wt DBA/2 mice. The survival statistics of these mice injected with tumour cells (Fig. 1c) showed that the median survival time of wt DBA/2 mice injected with P511 was 60 days (50% survival) in contrast to 35 days (5% survival) for TCRP1A DBA/2 and 48 days (23% survival) for TCRP1A F₁ mice. The median survival time of mice injected with control tumour P1.204 was 24 days in TCRP1A DBA/2 mice (0% survival), 30 days in TCRP1A F₁ mice (8% survival), and 38 days in wt DBA/2 mice (30% survival). Hence, the enriched distribution of TCRP1A CD8⁺ T cells in the F₁ compared with DBA/2 TCRP1A tg mice correlated with a delay in the onset of growth of the P1A⁺ tumour and in an increased survival time. When compared with the wt DBA/2 mice, however, there was only a marginal delay in tumour growth in the TCRP1A F₁ mice, but no advantage in terms of survival.

Figure 1.

Comparison of tumour growth in DBA/2 wild-type (wt) mice and in TCRP1A transgenic mice on different genetic backgrounds. (a) Representation of CD8 versus CD4 T cells (left panel) and of L^d/P1A-tet⁺⁺ CD8 T cells (right panel) in a pool of all lymph nodes (LN) from wt DBA/2, TCRP1A DBA/2 and TCRP1A F₁ mice. Calculated values (means ± SD of three individual mice) for total L^d/P1A-tet⁺ CD8 T cells within the LN of these mouse lines are indicated in bold. Values for TCRP1A DBA/2 versus TCRP1A F₁ are significantly different (P<0·001). (b–d) Subcutaneous growth of 10⁶ P1A⁺ P511 or P1A⁻ P1.204 tumour cells in wt DBA/2, TCRP1A DBA/2 and TCRP1A F₁ mice was monitored. Kinetics of tumour growth (b) involved six or seven mice per strain and Kaplan–Meier survival curves (c) involved four experiments including that shown in (b). Growth curves are representative of four independent experiments. Survival was scored within a period of at least 90 days, mice being killed when tumour size reached 400 mm². Numbers in parenthesis depicts median survival in days and the number of surviving mice out of total mice analysed. In P1A⁺ P511 groups, survival in wt DBA/2 (P<0·0001) and TCRP1A F₁ (P=0·02) was significantly different compared with TCRP1A DBA/2. In P1A⁻ P1.204 groups, survival in wt DBA/2 was significantly different from TCRP1A DBA/2 (P<0·0001) and TCRP1A F₁ (P<0·004). (d) Tumour growth curves of P1A⁺ P511 and P1A⁻ P1.204 tumour cells (10⁶ each) on the right and the left flanks respectively in four surviving DBA/2 wt or one TCRP1A DBA/2 plus three TCRP1A F₁ mice that rejected the first injection of P511 are shown.

Next, wt and TCRP1A tg mice having rejected a first inoculation of P1A⁺ P511 tumour were reinjected with equal numbers of P1A⁺ P511 tumour cells on the right flank and P1A⁻ P1.204 cells on the left flank. This showed that wt mice efficiently rejected both the P1A⁺ and P1A⁻ tumours on the two flanks, but TCRP1A tg mice rejected only the P1A⁺ tumour with no restriction of P1A⁻ tumour growth (Fig 1d). This suggested that in the wt mice rejection of the P1A⁺ tumour was supported by the development of effector cells directed at the P1A antigen as well as at non-P1A antigen. In contrast, the TCR repertoire in TCRP1A tg mice being very restricted, with 50% (DBA/2 mice) to 70% (F₁ mice) of the CD8⁺ T cells expressing almost exclusively TCRP1A, no sensitization to non-P1A tumour antigen appeared to occur.

No difference in the antigen-specific activation and proliferation of CD8⁺ T cells from TCRP1A DBA/2 and TCRP1A B10.D2 mice

To test whether the genetic make-up of the TCRP1A tg mice impacted the functional abilities of TCRP1A CTL, we stimulated in vitro naïve LN CD8⁺ T cells from TCRP1A DBA/2 and TCRP1A B10.D2 mice with wt splenic APCs loaded exogenously with either P1A peptide or NP, a non-specific control peptide. TCRP1A CTL from TCRP1A tg mice exhibited a strong proliferative response to P1A as measured by CFSE divisions (Fig 2a). Although most of the cells divided at high concentrations (10⁻⁶ m) of P1A peptide by day 3 of stimulation, at a concentration of 10⁻⁸ m of the peptide a fraction of undivided cells was observed in L^d/P1A-tet⁺ T cells from TCRP1A DBA/2 and TCRP1A B10.D2 mice. No CFSE divisions were observed with a non-specific control peptide NP even at a concentration as high as 10⁻⁶ m (Fig. 2a) or in CTL from wt littermate controls as responders (data not shown). The L^d/P1A-tet⁺ T cells from both TCRP1A DBA/2 and B10.D2 backgrounds exhibited normal proliferative ability following exposure to P1A presented by splenic APCs. Following a 3-day stimulation with P1A peptide in vitro, TCRP1A CD8⁺ T cells also showed a dose-dependent up-regulated surface expression of CD44 for both TCRP1A DBA/2 and B10.D2 cells (Fig 2a).

Figure 2.

P1A peptide dose-dependent proliferation, activation and interferon-γ (IFN-γ) secretion by TCRP1A DBA/2 and B10.D2 CD8⁺ T cells. (A) CFSE-labelled naïve lymph node (LN) CD8 T cells from TCRP1A DBA/2 (left) and TCRP1A B10.D2 (right) mice were stimulated in vitro with wild-type (wt) splenic antigen-presenting cells loaded with the indicated doses of either P1A peptide or a control 10⁻⁶ m NP peptide and were analysed after 3 days by cytofluorimetry. Dot plots show, within L^d/P1A-tet⁺ CD8⁺ T cells, CFSE labelling versus CD44 staining. The numbers in the plots show % cells within the relevant quadrant. (b) IFN-γ detected in the supernatants of TCRP1A DBA/2 (empty bars) or B10.D2 (filled bars) CD8⁺ T cells cultured for 3 days as in (a). Values represent means ± SD of two independent experiments.

Production of IFN-γ was also induced specifically by the P1A peptide/APC in a dose-dependent fashion in both DBA/2 and B10.D2 TCRP1A CTL cultures (Fig. 2b). Hence, TCRP1A CD8⁺ T cells followed a T-cell activation and proliferation programme independent of the DBA/2 versus B10.D2 genetic background.

TCRP1A effectors develop in vitro at lower frequency in DBA/2 versus B10.D2 or F₁ backgrounds, but with equal cytolytic potential

To further test whether the genetic make-up of the TCRP1A tg mice affected the functional abilities of TCRP1A CTL, we stimulated in vitro the naïve LN CD8⁺ T cells from TCRP1A DBA/2, TCRP1A F₁, TCRP1A B10.D2 as well as TCRP1A RAG-1º^/º B10.D2 tg mice with bone marrow-derived DCs loaded with either P1A peptide or control NP peptide. The number of TCRP1A CD8⁺ T cells in the LN as detected by L^d/P1A-tetramer staining on day 0 were 1 × 10⁶ (53% of total CD8⁺ cells) in TCRP1A DBA/2, 3·2 × 10⁶ (67%) in TCRP1A F₁, 6 × 10⁶ (79%) in TCRP1A B10.D2 and 3·7 × 10⁶ (99%) in TCRP1A RAG-1º^/º B10.D2 mice (Fig. 3a). On stimulation with relevant P1A peptide for 40 hr, gated L^d/P1A-tet⁺ CD8⁺ T cells showed an up-regulated high-affinity interleukin-2 receptor α-chain CD25, an early T-cell activation marker (Fig. 3b). No significant difference in the mean fluorescence intensity of CD25 molecules on the activated TCRP1A T cells was noticed in TCRP1A DBA/2, F₁, B10.D2 and RAGº^/º B10.D2 mice. No CD25 activation (< 8%) was detected on stimulation with the irrelevant control NP peptide (Fig. 3b). Following specific stimulation with P1A, the CD8 molecule also showed an up-regulated profile in all groups (Fig. 3b).

Figure 3.

Comparative analysis of the activation in vitro of TCRP1A CD8⁺ T cells from different genetic origins. Naïve lymph node (LN) CD8⁺ T cells from TCRP1A DBA/2, F₁, B10.D2 and RAG-1º^/º B10.D2 transgenic (tg) mice were stimulated in vitro with bone marrow-derived dendritic cells loaded with 10⁻⁷ m P1A peptide or control NP peptide and analysed for expression of CD25 and for specific cytolytic activity. (a) Pre-culture (day 0) CD8 versus L^d/P1A-tet dot plots on gated CD8⁺ cells from the LN are shown. Numbers on left in the quadrants depict mean absolute number ± SD of L^d/P1A-tet⁺CD8⁺ T cells within LN/mouse from three independent experiments. Values for TCRP1A DBA/2 are significantly different from all the other groups (P<0·001). (b) CD25 histograms on gated L^d/P1A-tet⁺ CD8⁺ T cells in the dot plots on the left are shown following 40 hr of stimulation with control NP (left panel) and P1A (right panel) peptide. Numbers above the markers in the histograms show % positive cells along with the values for mean fluorescence intensities. One experiment representative of three performed independently with similar results is shown. (c) Per cent specific lysis of P1A⁺ P511 or P1A⁻ P1.204 tumour targets by CD8⁺ T cells from different mice is shown at different L^d/P1A-tet⁺ CD8⁺ effector to target ratios. One experiment representative of three performed independently with similar results is shown.

Moreover, after 3·5 days (84 hr) of P1A-specific stimulation, TCRP1A CTL from these different backgrounds became cytolytic effectors against P1A-expressing targets. TCRP1A T cells from tg DBA/2, B10.D2 and F₁ mice all showed 65–70% specific cytolytic activity at an effector to target ratio of 10, when calculated with respect to L^d/P1A-tet⁺ T cells (Fig. 3c). T cells from TCRP1A tg RAG-1º^/º B10.D2 mice at the same effector to target ratio showed 90% specific lysis against P1A⁺ P511 targets (Fig. 3C). No non-specific activation by NP was observed and non-specific lysis against P1A⁻ P1.204 targets was < 5%. The functional capability of the TCRP1A tg CTL was therefore not affected by the genetic background of their origin. Moreover, from a starting culture of 0·5 × 10⁶ purified CD8⁺ T cells stimulated in vitro with equal numbers of P1A peptide-loaded bone marrow-derived dendritic cells, a total of 0·6 × 10⁶ (DBA/2), 2·2 × 10⁶ (F₁), 3·8 × 10⁶ (B10.D2) and 1·4 × 10⁶ (RAG-1º^/º) CD8⁺ T cells were harvested following 84 hr of stimulation. Despite the lower T-cell yield in the cultures originating from TCRP1A DBA/2 mice, the CTL activity per L^d/P1A-tet⁺ T-cell unit in all mice was similar.

TCRP1A effectors develop also in tumour-draining LN with equal cytolytic potential, but at lower frequency in DBA/2 versus F₁ backgrounds

Strain differences in resistance to tumour growth could also be the result of limited migration of P1A-specific CD8⁺ T cells and poor activation and induction of CTL in vivo. We therefore isolated T cells from the tumour-draining LN from wt DBA/2, TCRP1A DBA/2, and TCRP1A F₁ mice on days 8, 16 and 24 of tumour growth. We analysed the total number of P1A-specific CD8⁺ T cells in the tumour-draining LN by L^d/P1A-tetramer staining as well as the activation status and direct CTL activity ex vivo. In the tumour-draining LN, a peak of L^d/P1A-tet⁺ CD8 T cells was observed in TCRP1A F₁ mice 7 days following tumour cell injection (3 × 10⁶), whereas in the TCRP1A DBA/2 mice the total number of L^d/P1A-tet⁺ CD8⁺ T cells (1 × 10⁶) was threefold less than the number found in F₁ with a peak around day 15 after tumour injection (Fig. 4a). L^d/P1A-tet⁺ cells were detected even in the tumour-draining LN of wt mice with a peak on day 7 (3 × 10⁴ cells), though the number was 100-fold less than in the tumour-draining LN of TCRP1A tg mice. The numbers of L^d/P1A-tet⁺ CD8⁺ T cells returned to the basal level by day 25. A significant increase in the proportion and the total number of L^d/P1A-reactive CD8⁺ T cells was observed in the tumour-draining LN of P1A⁺ tumour-bearing mice compared with the control LN draining the P1A⁻ tumours (Fig. 4a) or the contralateral control LN (data not shown).

Figure 4.

Analysis of Ld/P1A-specific CD8 T cells ex vivo from tumour-bearing mice. CD8⁺ T cells from the LN of P1A⁺ P511 or P1A⁻ P1.204 tumour-bearing wild-type (wt) DBA/2, TCRP1A DBA/2 and TCRP1A F₁ mice were analysed at the indicated time-points. (a) Kinetics of total L^d/P1A-tet⁺ T-cell yield in a pool of tumour-draining brachial and inguinal LN. Values are means ± SD of three independent experiments. The expression of activation markers CD44 and CD62 ligand is shown within the gated population of L^d/P1A-tet⁺ CD8⁺ cells (in the dot plots on left) in tumour-draining LN (TDLN) on day 16 in P1A⁺ P511-bearing (b) or control P1A⁻ P1.204-bearing (c) TCRP1A DBA/2 and TCRP1A F₁ mice. The numbers above the markers in the histograms depict % positive cells. (d) The specific cytolytic activity of CTL effectors ex vivo from tumour-draining brachial and inguinal LNs or contralateral LNs (CLN) of P1A⁺ P511 tumour-bearing wt DBA/2, TCRP1A DBA/2 and TCRP1A F₁ mice was tested against P1A⁺ P511 or P1A⁻ P1.204 tumour targets. Per cent specific lysis at different L^d/P1A-tet⁺ CD8⁺ effector to target ratios is shown. One experiment representative of three independent experiments is shown.

Accumulated L^d/P1A-reactive CD8⁺ T cells in the LN draining P1A⁺ tumour showed an antigen-experienced phenotype with a majority of cells being CD44^hi CD62L^lo in both DBA/2 or F₁ TCRP1A mice (Fig. 4b) compared with the CD44^lo CD62L^hi phenotype of the TCRP1A CD8⁺ T cells in the LN draining P1A⁻ tumour (Fig. 4c), or in the LN of mice that did not receive tumour cells (data not shown). Furthermore, TCRP1A CTL effectors from the P1A⁺ tumour-draining LN of TCRP1A DBA/2 or TCRP1A F₁ mice showed cytolytic activity ex vivo against P1A⁺ P511 targets (> 35% specific lysis at L^d/P1A-tet⁺ effector to target ratio = 10) on day 16 (Fig. 4d) or day 25 (data not shown).

The tumour antigen-reactive monoclonal CD8⁺ T lymphocytes from the tumour-draining LN of TCRP1A tg mice exhibited activated phenotype and cytolytic activity ex vivo. The relatively higher resistance of the TCRP1A F₁ as compared with the TCRP1A DBA/2 mice would therefore correlate with the higher frequency of L^d/P1A-tet⁺ CTL effectors in F₁ rather than with any differences in the specific killing activity of the effectors.

CD4 T-cell-independent restriction of tumour growth

To evaluate further whether the higher frequency of L^d/P1A-reactive CTL effectors have the potential to restrict the growth of tumour in the absence of any help from other T or B lymphocytes, we proceeded to monitor the growth of P1A⁺ P511 and P1A⁻ P1.204 tumour cells in TCRP1A RAG-1º^/º B10.D2 mice. In these mice, LN were composed of around 90% CD8⁺ T cells, all of which expressed the L^d/P1A-binding TCR.¹⁸ When 10⁶ P1A⁺ P511 tumour cells were inoculated subcutaneously in RAG-1º^/º mice, the tumour grew in all mice (Fig. 5). However, in RAG-1º^/º mice expressing the TCRP1A tg, the tumour failed to grow in the majority of animals (81%). The P1.204 P1A⁻ variant had similar growth patterns in TCRP1A tg or non-tg RAG-1º^/º B10.D2 mice and tumour grew in all mice (Fig. 5). The high expression of the TCRP1A CD8⁺ T cells in RAG-1º^/º B10.D2 mice therefore provided efficient resistance to the growth of P1A⁺ tumour without any requirement for help from other T or B lymphocytes.

Figure 5.

Restriction of tumour growth in TCRP1A RAG-1º^/º B10.D2 transgenic mice. Subcutaneous growth of 10⁶ P1A⁺ P511 or P1A⁻ P1.204 tumour cells in RAG-1º^/º and TCRP1A RAG-1º^/º B10.D2 mice was monitored. Kinetics of tumour growth is shown. Numbers on top depict the number of mice without any tumour growth out of total mice analysed. One experiment representative of seven independent experiments with at least five mice in each group with similar patterns of tumour growth is shown.

The possibility that some of the CD4⁺ T cells present in the RAG-1^+/+ mice may contribute to dampen the anti-tumour response by regulatory T-cell-like suppressor function was also evaluated by reconstitution experiments in RAG-1º^/º B10.D2 mice. The RAG-1º^/º mice were reconstituted with naïve TCRP1A CD8⁺ T cells (0·5 × 10⁶) alone or mixed with purified CD4⁺ T cells (5 × 10⁶) either from wt or from TCRP1A B10.D2 mice 7 days before subcutaneous P1A⁺ P511 tumour cell injection (1 × 10⁶). No tumours grew in mice reconstituted with CD8⁺ T cells or a mixture of CD8⁺ T cells and CD4⁺ T cells, whereas all mice without reconstitution showed vigorous tumour growth (Fig. 6). The CD4⁺ T cells apparently did not inhibit the tumour rejection capability of L^d/P1A-reactive CD8⁺ T cells from TCRP1A B10.D2 mice.

Figure 6.

Reconstitution of RAG-1º^/º mice with naïve TCRP1A CD8⁺ T cells affords resistance to P1A⁺ tumour growth even in the presence of CD4⁺ T cells. RAG-1º^/º B10.D2 mice were transferred with 0·5 × 10⁶ TCRP1A CD8⁺ T cells (purified CD4⁺ T-depleted and B-depleted cells from the lymph nodes of TCRP1A B10.D2) alone or mixed with 5 × 10⁶ CD4⁺ T cells 7 days before P1A⁺P511 tumour cell injection (1 × 10⁶, subcutaneous). CD4⁺ T cells were purified CD8⁺ T-depleted and B-depleted cells from the lymph nodes of either wild-type (wt) B10.D2 (experiment in second panel) or TCRP1A B10.D2 (experiment in third panel) mice. Kinetics of tumour growth is shown. Curves are representative of two independent experiments with six mice in each group with similar patterns of tumour growth.

Representation of NK cells in RAG-deficient and RAG-competent mice

To investigate the contribution of innate immune cells in tumour rejection, we evaluated the number of NK cells in RAGº^/º and RAG^+/+ mice. The representation of NK cells determined as NK1.1⁺ CD3^– cells is highly enriched in LN and spleen of RAG-1º^/º as compared with TCRP1A RAG-1º^/º and RAG-1^+/+ TCRP1A F₁ mice (Fig. 7). When total cell numbers are considered (Table 1), RAG-1º^/º and TCRP1A tg RAG-1º^/º LN have similar numbers of NK cells (0·5 × 10⁶ to 0·8 × 10⁶), whereas their number is lower in the TCRP1A tg F₁ RAG-1^+/+ mice. In the spleen, however, the total number of NK cells was in the same range for the three types of mice (5 × 10⁶ to 9 × 10⁶). In TCRP1A tg RAG-1º^/º mice, the total LN cell number was consistently higher in the P511-draining LN than in those draining the P1A-negative P1.204, leading to a threefold to fourfold local increase in the number of NK cells (Table 1).

Figure 7.

Representation of NK1.1⁺ cells in RAG-1^+/+ and RAG-1º^/º mice. Cells from the LN and spleen of F1, RAG-1º^/º B10.D2 and TCRP1A RAG-1º^/º B10.D2 mice were analysed by cytofluorimetry. CD3 versus NK11 dot plots on total cells are shown. The numbers in the dot plots depict % cells. One experiment representative of two experiments performed independently with similar results is shown.

Table 1.

Number of natural killer (NK) cells in lymph nodes and spleen of RAG-1 competent and deficient mice

Mice1	Non-draining contralateral LN (× 106)		Tumour-draining lymph node (× 106)		Spleen (× 106)
No tumour	Total	NK2	Total	NK	Total	NK
RAG-1+/+TCRP1A F1	20·5	0·07 ± 0·0173			81·5	9·37 ± 3·17
RAG-1º/º	0·71	0·49 ± 0·087*			11·5	6·58 ± 1·21
RAG-1º/ºTCRP1A	5·95	0·76 ± 0·206*			22·15	5·47 ± 1·69
P1A−P1.204
RAG-1º/º	0·45	0·29 ± 0·046	0·15	0·09 ± 0·044	10·0	8·23 ± 2·32
RAG-1º/º TCRP1A	3·60	0·47 ± 0·257	0·25	0·04 ± 0·011	20·0	5·46 ± 0·38
P1A+P511
RAG-1º/º	0·55	0·34 ± 0·043	0·125	0·09 ± 0·024	10·0	8·13 ± 0·26
RAG-1º/º TCRP1A	3·95	0·59 ± 0·136**	0·80	0·15 ± 0·026**	19·6	3·61 ± 0·62

¹Five- to eight-week-old age-matched RAG-1º^/º B10.D2, TCRP1A RAG-1º^/º B10.D2 or TCRP1A RAG-1^+/+ B10.D2(× DBA/2)F₁ mice were injected with saline or P1A⁻ P1.204 or P1A⁺ P511 tumour cells (1 × 10⁶, subcutaneously). On day 3 after tumour injection, cells from the lymph nodes and spleen of these mice were harvested and analysed for NK1.1 and CD3 distribution by cytofluorimetry.

²Total cell number multiplied by the fraction of NK1.1⁺ CD3⁻ cells.

³Values are mean ± SD of three independent experiments.

^*Significantly different from all the value for TCRP1A transgenic RAG-1^+/+ F₁, P<0·001.

^**Significantly different from the values for corresponding RAG-1º^/º mice or for TCRP1A transgenic RAG-1º^/º mice injected with P1A⁻ P1.204 tumour cells, P<0·001.

The rejection of P1A⁺ tumour in TCRP1A RAG-1º^/º tg mice may also be associated with an increased number of NK cells observed in their tumour-draining LN. The less efficient rejection afforded by the RAG-1^+/+ TCRP1A F₁ mice (Fig. 1) may be correlated with the lower number of NK cells in the LN of these mice (Table 1).

NK cells contribute to tumour resistance through an NKG2D-dependent mechanism

The P815 tumour lines P511 and P1.204 both express high levels of MHC class I, and fail to express some of the molecules, namely Rae-1 and H60, that are stress-inducible ligands for the NK cell activating receptor NKG2D. They express, however, the MULT-1 ligand for this receptor,²⁴ consistent with the binding of NKG2D-tetramers to both the P511 and P1.204 lines (results not shown).

Treatment of TCRP1A RAG-1º^/º mice or non-tg RAG-1º^/º mice reconstituted with TCRP1A CD8⁺ T cells with the depleting anti-NK1.1 mAb PK136 consistently abrogated the resistance of these mice to growth of the P511 tumour (Table 2). Treatment with the NKG2D-specific mAb MI-6, previously shown to block NKG2D–ligand interaction,²³ also led to P511 tumour outgrowth in all treated host mice, either TCRP1A RAG-1º^/º tg, or TCRP1A-reconstituted RAG-1º^/º mice (Table 2). In a fraction of the mice, injection of isotype control mAb with irrelevant specificity also led to reduced resistance of the mice to tumour growth (Table 2). It is not known at this point whether this effect is attributed to engagement of FcR that may transduce negative signals for interleukin-12 production by APCs²⁵ and has been shown to recruit the phosphatase SHIP-1 (an SH-2 domain-containing inositol-5-phosphate) to the membrane of NK cells, thereby inhibiting their cytolytic activity.²⁶ In spite of this ‘non-specific’ effect of mAb observed in a fraction of the mice (35% and 40% for mouse and rat control mAb, respectively), a clear additional effect was observed by treatment with the depleting anti-NK1.1 mAb or blocking anti-NKG2D mAb as 100% of the treated mice failed to reject the tumour (Table 2). Hence, the NK cells appear to contribute to tumour resistance through an NKG2D-dependent mechanism.

Table 2.

Natural killer (NK) cells contribute to tumour resistance by an NKG2D-dependent mechanism

Tumour growth	Treatment
		Mouse mAb1		Rat mAb2
Mice3	None	Anti-NK1.1	control	Anti-NKG2D	control
With tumour4/total	4/41*	22/22**	5/14	14/14***	4/10
	9·7%	100%	35%	100%	40%
Days5 ± SD	57 ± 12·4	14 ± 3·9	21 ± 3·8	21 ± 3·2	27 ± 3·5

¹PK-136 (anti-NK1.1) or its mouse isotype control immunoglobulin G2a (IgG2a) Ti98 or Désiré-1 was injected on day −1, +1 and +8 of tumour injection at 200 μg monoclonal antibody (mAb) per intraperitoneal injection.

²MI-6 (anti-NKG2D) or its rat isotype control IgG2a B20.2 was injected on day − 1, + 1 and + 10 of tumour injection at 200 μg mAb per intraperitoneal injection.

³Five- to eight–week-old age-matched TCRP1A RAG-1º^/º B10.D2 mice or RAG-1º^/º B10.D2 mice reconstituted with 5 × 10⁶ of purified TCRP1A CD8⁺ T cells 7 days before P1A⁺ P511 tumour injection.

⁴Number of mice with tumour burden reaching 100 mm² out of total.

⁵Day at which the tumour reached a size of 100 mm².

^*Significantly different from all other groups, P<0·001.

^**Significantly different from the groups of its isotype control or of anti-NKG2D mAb, P<0·001.

^***Significantly different from the group of its isotype control mAb, P<0·01.

Discussion

In this study, we were interested in dissecting the requirements for immune resistance to the growth of mastocytoma solid tumours. We examined the aspects of T-cell activation and the effect of the variation in the precursor frequency of CD8⁺ T cells and NK cells on tumour rejection in mice that expressed as a transgene the TCRP1A specific for the mouse tumour antigen P1A, shown to be the major rejection antigen of mastocytoma P815, or were reconstituted with TCRP1A CD8⁺ T cells. We were struck by the poor selection of TCRP1A CD8⁺ T cells in the DBA/2 background, which we determined to be controlled by DBA/2 thymocyte-intrinsic genetic factors.¹⁸ Because the lower numbers of TCRP1A CD8⁺ T cells as a result of this weak selection were associated with a poor resistance to the growth of P1A-expressing tumours in TCRP1A DBA/2 tg mice, we investigated the reactivity of CD8⁺ T lymphocytes from TCRP1A DBA/2 mice by comparing them with those from TCRP1A tg mice on other genetic backgrounds that showed better selection of TCRP1A CD8⁺ T cells, such as B10.D2 or B10.D2(× DBA/2)F₁. The results show that the functional potency of CD8⁺ T lymphocytes from TCRP1A DBA/2 mice was not compromised and was not significantly different from that of the TCRP1A CD8⁺ T cells on the other backgrounds. No differential influence of the DBA/2 versus B10.D2-encoded genes was observed on the functional potency of TCRP1A CD8⁺ T lymphocytes.

The antigen P1A is a self tumour-associated antigen. So, it was possible that endogenous expression of the P1A gene might have rendered T cells anergic. However, the present data rule out this possibility as purified TCRP1A CD8⁺ T cells showed strong proliferation, activation and induction to efficient CTL in a manner independent of their genetic background following exposure to P1A presented by splenic APCs or bone marrow-derived DCs (Figs 2 and 3). Our earlier study demonstrated that TCRP1A CD8⁺ T cells escape clonal deletion in the thymus and maintain a naïve phenotype as indicated by the absence of CD69, CD25 or CD44 expression or presence of CD62 ligand expression.¹⁸ With respect to the normal development of P1A-reactive tg T cells, our results are in agreement with a previous report on independently derived P1A-reactive TCR tg mice established in SW mice backcrossed to BALB/c.²⁷ Our data are also concurrent with the observation that the TCR tg mice are not more resistant than wt mice to a P1A⁺ plasmacytoma.²⁷

The failure of tumour rejection in most TCRP1A DBA/2 (95%) and in a majority of TCRP1A F₁ (77%) tg mice after an initial resistance during the establishment phase of tumour may be for the following reasons: (i) responder CD8⁺ T cells may become outnumbered by the rapidly growing tumour cells rather than presenting any functional defect, (ii) inhibition of CTL activity, (iii) lack of or insufficient CD4 T-cell help, (iv) presence of suppressive CD8⁺ or CD4⁺ T regulatory cells, (v) intra-clonal T-cell competition, and (vi) lack of or insufficient activation of an innate immune response.

Our observation that TCRP1A F₁ mice delayed tumour onset (Fig. 1) and TCRP1A RAGº^/º B10.D2 mice efficiently rejected tumour (Fig. 5) may suggest that the initial number of responder CD8⁺ T cells in the face of tumour–T-cell combat might be a crucial factor for tumour rejection. We did not see any inhibition of P1A-specific CTL activity ex vivo in the tumour-draining LN of TCRP1A tg mice (Fig. 4d). Moreover, in this model the tumour rejection was independent of CD4 T-cell help as RAGº^/º TCRP1A tg mice or RAGº^/º mice reconstituted with TCRP1A CD8⁺ T cells efficiently contained the growth of P1A⁺ tumours, selectively (Figs 5 and 6). The possibility that the anti-tumour immune response was dampened by CD4⁺ T cells or non-L^d/P1A-binding CD8⁺ T cells was not supported in reconstitution experiments involving mixtures of TCRP1A CD8⁺ T cells and CD4⁺ T cells (Fig. 6).

The observations (i) that wt DBA/2 mice showed a rejection response in 50% of mice against P1A-expressing tumour and in 30% of mice against P1A-deficient tumour, and (ii) that wt mice that had rejected the P1A⁺ P511 tumour had the capacity to reject a rechallenge of P1A⁺ tumour on one flank and P1A⁻ tumour on the opposite flank, whereas TCRP1A tg mice rejected only the rechallenge of the P1A⁺ tumour (Fig. 1) may be argued in two ways. There may be another major rejection antigen in addition to P1A shared by P511 P1A⁺ and P1A⁻ variants. Indeed, a second major tumour-specific antigen recognized by CTL has been reported on P815.²⁸ In this context, the availability of the polyclonal T-cell repertoire and the multiple antigen determinants may have helped the efficient rejection of P1A⁺ and P1A⁻ tumours in the wt mice. Other studies also suggest a benefit for multiple antigen determinants, even in the face of CD8⁺ T-cell immunodominance.^29,30 Alternatively, it may be argued that in the polyclonal T-cell repertoire of the wt mice, immunization against one epitope may provide subsequent cross-protection against other epitopes, whereas in the tg mice enriched with monoclonal TCRP1A CTL, the response against a single epitope may fail to provide cross-protection against other epitopes. This may also explain why in the event of generation of antigen-loss tumour escape variants in the face of antigen-specific immunoselection/editing, a monoclonal environment fails in some tumour models. Notably, although monoclonality is not absolute in RAG^+/+ TCRP1A tg mice (Fig. 1a), the homogeneous expression of the tg TCRP1A β-chain may also contribute to restrict the CD4 and CD8 T-cell repertoires in the TCRP1A DBA/2 and F₁ mice.

The influence of clonal precursor frequency on the development and survival of effector and memory T cells is complex.^31,32 In the present study, the threefold difference in the initial number of TCRP1A CD8⁺ T cells between DBA/2 and F₁ mice was reflected in the higher yield of these cells in CTL culture. This factor was also maintained at the peak of the expansion phase (Fig. 4– day 16), as well as after the contraction phase (Fig. 4– day 24) of the in vivo response to the P1A⁺ P511 tumour. Therefore, both the initial delay in the onset of tumour growth and the increased frequency of F₁ mice rejecting the tumour appear to correlate with the frequency of precursor and effector TCRP1A CD8 T cells in the F₁ as compared with the DBA/2 TCRP1A mice (Fig. 1). This result is consistent with others implicating the number of tumour-reactive CD8⁺ T-cell precursors as a limiting factor in tumour resistance.^33,34

In the RAG-1º^/º TCRP1A tg mice, however, this correlation does not seem to hold true because the number of TCRP1A CD8⁺ T cells is similar to that found in the TCRP1A F₁ tg mice, yet only mice on the RAG-1º^/º background fully resist P511 tumour growth (Fig. 5). Tumour growth is so efficiently contained in these mice that none is detected in the large majority (81%) of mice. The RAG-1º^/º mice reconstituted with TCRP1A CD8⁺ T cells also showed efficient (100%) tumour rejection (Fig. 6). This may in part be the result of efficient homeostatic proliferation of T cells in RAGº^/º empty hosts.

In wt mice, comparison between the frequency of L^d/P1A binding CD8⁺ T cells and tumour containment could not be simply addressed because antigens other than P1A appear to elicit a response as indicated by the secondary rejection of the P1A⁻ P1.204 tumour in wt mice that rejected the P1A⁺ P511 tumour (Fig. 1). Furthermore, the precursor frequency may be inversely correlated with the extent of clonal expansion and may control the commitment to the CD8⁺ T-cell memory lineage.³¹ Intriguingly, comparison of expansion and survival of monoclonal versus polyclonal CD4⁺ T-cell precursors in response to antigen further suggests the existence of clone-specific survival signals as well as intra-clonal competition that would limit clonal size and favour heterogeneity over monoclonality.³² The existence of such control mechanisms for CD8⁺ T cells may also contribute to the higher resistance of wt as compared with TCRP1A tg mice to tumour growth. It does not, however, account for the exquisite tumour resistance of the monoclonal RAG-1º^/º TCRP1A tg mice. Furthermore, in other tumour models mice tg for a tumour antigen-reactive TCR were selectively resistant to the relevant tumour, whereas wt mice succumbed to tumour challenge.³⁵

The basis for the paradoxical situation in RAG-1º^/º mice in which CD8⁺ T-cell monoclonality and absence of CD4 T-cell help or regulation afford full tumour resistance was traced to the presence of increased number of NK cells in the tumour-draining LN of these mice (Fig. 7 and Table 1). NK cells contributed to tumour resistance and an NKG2D-dependent mechanism appeared to be involved (Table 2). A complementary response from the innate components of the immune system was necessary for an effective anti-tumour antigen-specific T-cell response. However, NK cells alone in RAG-1º^/º mice that contain a similar number of NK cells to TCRP1A RAG-1º^/º tg mice failed to reject P1A⁺ tumour. Tumour resistance by NK cells appears to require an activation/recruitment signal by the activated tumour antigen-specific CD8⁺ T cells.³⁶ The exact mechanism of this interaction between the effector T cells and NK cells still remains unclear.

In summary, the differential tumour rejection can be ascribed to the following differences among the various strains of mice investigated in this study. The TCRP1A tg B10.D2(× DBA/2) F₁ mice resist tumour growth better than TCRP1A tg DBA/2 mice because of the higher frequency of TCRP1A CD8⁺ T cells in F₁ mice being controlled genetically.¹⁸ The wt DBA/2 mice resist tumour growth better than TCRP1A tg DBA/2 or F₁ mice because of the more diverse TCR repertoire of the wt mice, as evident from their resistance to P1A⁻ tumour, and also most likely because of the absence of intra-clonal competition in these mice. In the TCRP1A tg mice, as a result of clonal competition between TCRP1A CD8⁺ T cells and endogenously rearranged TCR-expressing T cells, there is probably a preferential expansion of TCRP1A tg T cells and, therefore, no activation of the other T cells capable of reacting to the P1A⁻ tumour variants. The TCRP1A tg RAGº^/º mice resist tumour growth better than TCRP1A tg RAG^+/+ mice because of a combination of higher frequency of TCRP1A CD8⁺ T cells and the anti-tumour contribution of NK cells. We have not investigated in this study the role of tumour microenvironmental factors such as immunosuppressive cytokines, or other non-T-cell intrinsic factors that may affect tumour rejection.

Altogether these results suggest that the frequency of tumour antigen-specific T-cell precursors and the number of antigen determinants contribute to the efficiency of adaptive T-cell responses to tumours, and that the efficiency of tumour elimination can depend on some privileged interaction between the effector T cells and innate components of the immune system.

Glossary

Abbreviations:

APCs

antigen-presenting cells

CTL

CD8⁺ cytolytic T lymphocyte

DCs

dendritic cells

ELISA

enzyme-linked immunosorbent assay

IFN-γ

interferon-γ

LN

lymph node

mAb

monoclonal antibody

MFI

mean fluorescence intensity

MHC

major histocompatibility complex

NK

natural killer

NKG2D

natural-killer group 2 member D

TCR

T-cell receptor

TCRP1A F₁

TCRP1A B10.D2(× DBA/2)F₁

TCRP1A

L^d/P1A-specific TCR transgene

tg

transgenic

wt

wild-type

Disclosures

The authors have no potential conflicts of interest.