Transnuclear TRP1-Specific CD8 T Cells with High or Low Affinity TCRs Show Equivalent Antitumor Activity

Infiltration of CD8 T cells correlates positively with patient outcomes, as shown for colorectal and other cancers (1, 2). Even in the absence of a successful endogenous antitumor immune response, immunotherapy for cancer has begun to deliver on its promise (3, 4). The success of checkpoint blockade therapies such as anti-CTLA-4 illustrates the importance of the CD8 T cell response in tumor regression. At baseline, many human patients with cancer have antitumor CD8 T cells that are anergic or kept in check by CD4 regulatory T cells (Treg). CTLA-4 blockade can activate antitumor T cells and elicit clinical responses in 25% of patients. Anti-CTLA-4 works synergistically with chemotherapy, radiotherapy, or other therapies that generate tumor lysis and uptake of tumor antigens by dendritic cells (5). Blocking antibodies to PD-1 and PDL1 (6, 7) produce similar outcomes. The endogenous T cell response to tumors correlates with overall survival in patients with cancer (1), and therapies aimed at boosting the T cell response are now in widespread use (3, 8).

Many tumor antigens are mutated self-antigens or antigens that are overexpressed. T cell receptors (TCR) on T cells that recognize such antigens are generally of low affinity (9), as high-affinity self-reactive cells are deleted in the thymus (10, 11) and suppressed in the tumor microenvironment (12). The type of CD8 T cell response necessary for optimal tumor rejection remains poorly characterized because of a paucity of animal models. Most of the available animal models make use of TCR transgenic animals, the construction of which is based on long-term T cell cultures from which the corresponding rearranged TCR segments are cloned and expressed as transgenes. Alternatively, tumor cell lines are engineered to express antigens for which TCR transgenics happen to be available. By their very nature, populations of T cells harvested from tumor-bearing animals or tumor-infiltrating lymphocytes are necessarily heterogeneous, polyclonal, and variable from one experiment to the next, exactly the reason why TCR transgenics are so widely used. Models of antitumor immunity typically rely on overexpression of ovalbumin (OVA) in combination with OT-I transgenic T cells, known to be of very high affinity. Pmel TCR transgenic mice contain CD8 T cells of more moderate affinity for the endogenous melanoma antigen tyrosinase-related protein 1 (TRP1), but results from these mice are difficult to compare with OVA/OT-I systems due to differences in the antigen.

Somatic cell nuclear transfer may be used to clone mice from the nuclei of antigen-specific lymphocytes (13–16). These transnuclear mice are derived from primary CD8 T cells harvested in the course of a natural immune response, without any prolonged tissue culture to select for high-affinity T cell variants in vitro. Transnuclear mice contain no genetic alterations other than the rearranged TCR genes, expressed from their endogenous loci and under physiologic control. Finally, creation of transnuclear mice can be done rapidly, requiring only 6 weeks from isolation of a selected lymphocyte population, accomplished through isolation of class I MHC tetramer-positive (tetramer⁺) cells, to birth of chimeric mice. A key advantage of the transnuclear technology is the ability to clone multiple lines of mice with the same antigenic specificity but with different TCR sequences and therefore different affinities (15). These mice can then be used as a ready source of lymphocytes of defined specificity for in vitro culture or for adoptive transfer experiments.

Most tumors are weakly immunogenic at best. The well-studied melanoma cell line B16 downregulates class I MHC and is poorly infiltrated by immune cells (17–19). CD8 T cells can reject B16 in vivo, but only when mice are first exposed to some form of immune manipulation, such as vaccination with irradiated B16, engineered to secrete the cytokine granulocyte macrophage colony-stimulating factor (GM-CSF; refs. 18, 20, 21). Because B16 does not induce a detectable CD8 T cell response under normal conditions, we used a peptide-based vaccine strategy to elicit a robust CD8 T cell response to the melanoma-specific antigen TRP1. This vaccine consists of a palmitoylated version of the peptide to increase its circulatory half-life, anti-CD40 antibody, and the TLR3 ligand polyinosine-polycytidylic acid (poly-IC) and is referred to as TRIVAX. When administered to mice, it elicits a robust population of tetramer⁺ CD8⁺ T cells that can reject established B16 tumors (22, 23).

We have used such vaccine-induced CD8 T cells as a source of donor nuclei to generate two independent lines of TRP1-specific transnuclear mice. In doing so, we have generated, for the first time, a pair of transnuclear mice with TCRs specific for the identical peptide–MHC combination. These TCRs are expressed under the control of their endogenous promoters and represent cells directly harvested from an immunized mouse upon staining with the appropriate class I MHC tetramers, without preselection in tissue culture, and differ in their affinity for TRP1 by approximately 100-fold. The T cells from these mice recognize an endogenous tumor antigen and allow a direct assessment of the role of TCR affinity in antitumor CD8 T cell responses. The generation of antitumor transnuclear mice provides a unique set of tools for the fields of tumor immunology and CD8 T cell development.

All animals were housed at the Whitehead Institute for Biomedical Research (Cambridge, MA) and were maintained according to protocols approved by the Massachusetts Institute of Technology (MIT) Committee on Animal Care. C57Bl/6, CD45.1 congenics, and OT-I transgenic mice were purchased from The Jackson Laboratory. RAG2^−/− (RAG2TN12) mice were purchased from Taconic.

Transnuclear mice were generated as previously described (14,15, 24). Briefly, CD8⁺ TRP1 tetramer⁺ cells were sorted by fluorescence-activated cell sorting (FACS) and used as a source of donor nuclei for somatic cell nuclear transfer (SCNT). The mitotic spindle was removed from mouse oocytes and replaced with donor nuclei. The nucleus-transplanted oocytes were then activated in medium containing strontium and trichostatin A (TSA), and allowed to develop in culture to the blastocyst stage. Because the live birth rate of SCNT blastocysts is close to zero, SCNT blastocysts were instead used to derive embryonic stem cell lines. These embryonic stem cell lines were then injected into wild-type B6xDBA F1 blastocysts and implanted into pseudopregnant females. The resulting chimeric pups were mated to C57Bl/6 females to establish the 6.15 (TRP1^low) and 6.17 (TRP1^high) lines. All animals used were backcrossed four to seven generations onto the C57BL/6 background.

TRP1^high and TRP1^low CD8 T cells were purified by positive selection using CD8 magnetic beads (Miltenyi Biotec) and used as a source of RNA. 5′-RACE was conducted according to the manufacturer's protocol (GeneRacer, #L1502-01; Invitrogen).

Recombinant protein expression, refolding of the H-2D^b complex with the SV9-P7* conditional ligand, and their subsequent tetramerization were accomplished by following established protocols (25–27). The peptide exchange reaction was initiated by UV irradiation (360 nm), and the resulting MHC tetramers were used directly to stain freshly prepared splenocytes as described previously. TRP1-altered peptide ligands were produced by Fmoc-based solid-phase peptide synthesis by the MIT Center for Cancer Research (Cambridge, MA) biopolymers facility. All peptides were dissolved in dimethyl sulfoxide (10 mg/mL) and stored at −20°C until further use.

Cells were harvested from spleen, peripheral blood, or thymus. Cell preparations were subjected to hypotonic lysis to remove erythrocytes, stained, and analyzed using a FACSCalibur (BD). Tetramers were prepared fresh from photocleavable stocks (25–27) or directly refolded with TRP1 heteroclytic peptide (TAPDNLGYM). All antibodies were from BD Pharmingen.

Cells were cultured in RPMI-1640 medium supplemented with 10% heat-inactivated FBS, 2 mmol/L l-glutamine, 100 U/mL penicillin G sodium, 100 μg/mL streptomycin sulfate, 1 mmol/L sodium pyruvate, 0.1 mmol/L nonessential amino acids, and 0.1 mmol/L 2-ME. Bone marrow–derived dendritic cells (BMDC) were obtained by culturing of bone marrow aspirates in RPMI media supplemented with recombinant murine GM-CSF (rmGM-CSF; Peprotech) and recombinant murine interleukin-4 (rmIL-4; Peprotech). Fresh media was added to the BMDC cultures every 2 days for 6 days. BMDCs were used at day 6 to 8 of differentiation, and were washed before use. CD8 T cells were isolated from pooled spleen and lymph nodes of TRP1 transnuclear or wild-type mice using positive selection on anti-CD8 magnetic beads (Dynabeads; Invitrogen). For cocultures, BMDCs were pulsed with the indicated peptides for 30 minutes before the addition of CD8 T cells at a ratio of 100,000 T cells to 50,000 BMDCs in round-bottomed 96-well culture plates. IL-2 and IFN-γ production were measured by ELISA of 24-, 48-, or 72-hour culture supernatants as indicated (BD Pharmingen). CCL5 (RANTES) and MIP-1β were measured by cytokine bead array (BD).

CD8 T cells from TRP1^high, TRP1^low, or OT-I transgenic mice were isolated by positive selection on anti-CD8 Dynabeads (Invitrogen; 114.62D) and cultured for 18 hours with BMDCs plus either TRP1 heteroclytic peptide or OVA peptide (SIINFEKL) at 1 μmol/L concentration. Activated CD8 cells were then counted mixed with B16 cells at 20:1 ratio and resuspended in three-dimensional (3D) culture matrix cocktail as follows: Dulbecco's PBS (D-PBS) containing 33% of Cultrex basement membrane extract (BME; Trevigen; 3445-005-01), 10% fetal calf serum (FCS), 1 mg/mL fibrinogen (Calbiochem; 341578), 1.4 U/mL thrombin (MP Biomedicals; 02194083), 1.25 × 10⁵ cells/mL of B16, and 2.5 × 10⁶ or 5.0 × 10⁶ cells/mL of CD8⁺ T cells. Fifty-five microliters of this cell+matrix cocktail was cast in flat-bottomed 96-well plates, incubated for 45 minutes to allow the matrix become solid and then overlaid with 100 μL of Dulbecco's modified Eagle medium (DMEM)+10% FCS medium on top of the solid gel matrix. Cultures were incubated for 4 days. Cells were reisolated in ice-cold PBS+5 mmol/L EDTA, washed, and resuspended in PBS+0.5% bovine serum albumin and stained using anti-CD8 and the viability dye 7-amino-actinomycin D (7-AAD) (BD Pharmingen). B16 cells were quantified as a ratio of 7-AAD⁺ to 7-AAD⁻ cells to determine the dead:live ratio.

B16 cells were a kind gift of Dr. Glenn Dranoff (Dana-Farber Cancer Institute, Boston, MA). B16 cells were cultured to 90% confluency, trypsinized, washed in Hank's balanced salt solution (HBSS), and resuspended in HBSS at 50,000 tumor cells per 250 μL volume. Cells were injected subcutaneously in the left flank of 6- to 12-week-old female C57BL/6 mice. In mice receiving no transferred T cells, palpable tumors arose 7 to 10 days postinoculation. Tumor diameter was measured in three dimensions, and mice were sacrificed when tumor volume reached 2 cm³. For adoptive transfers, unless otherwise stated, mice received 300,000 CD8 T cells in 150 μL PBS administered via tail vein injection within 24 hours of tumor inoculation.

Statistical analysis of survival curves was conducted using GraphPad Prism software. P values were determined using the Mantel–Cox log-rank test. Delays in tumor growth were obtained by comparison of median survival times.

Vaccination with palmitoylated TRP1 peptide, anti-CD40 antibody, and the TLR3 ligand poly-IC yields a robust population of anti-TRP1 CD8 T cells relatively quickly and provides a source of TRP1-specific lymphocytes for SCNT. SCNT is most successful using hybrid mouse strains as a source of donor nuclei. To this end, B6 × 129 F1 male agouti mice were injected intravenously with palmitoylated peptide TAPDNLGYM, anti-CD40, and poly-IC. The TRP1 peptide sequence used differed from native TRP1 in the ninth position (A to M substitution) to increase its binding to H-2D^b. This heteroclitic peptide was used to construct TRP1 tetramers, and robust TRP1 tetramer⁺ populations were observed in the spleens of immunized mice at 6 and 8 days postimmunization (Fig. 1A). These CD8⁺ TRP1 tetramer⁺ populations were sorted by FACS and used as a source of donor nuclei for 2 separate SCNT experiments. A total of 105 nuclear transfers were conducted, yielding five independent embryonic stem cell lines (Fig. 1B). Of these, two resulted in agouti chimeric mice containing TRP1 tetramer⁺ CD8⁺ T cells in peripheral blood (Fig. 1C).

Both lines of TRP1 transnuclear mice gave germline transmission, and the F1 progeny showed skewing of the CD4:CD8 ratio, with a large percentage of TRP1 tetramer⁺ cells at baseline (Fig. 1D). Tetramer⁺ CD8 T cells were sorted from each mouse line and subjected to 5′-RACE to sequence the TCR-α and -β genes (Supplementary Fig. S1). The 6.15 mice had consistently fewer TRP1 tetramer⁺ cells than their 6.17 counterparts, and the tetramer staining was dimmer, as judged by a reduction in mean fluorescent intensity. TCR, CD3, and CD8 surface expression were normal on 6.15 cells (data not shown), suggesting that the lower intensity tetramer staining could be due to weaker binding of tetramers to the TCR. To measure TCR avidity, splenocytes from 6.15 or 6.17 mice were incubated with TRP1 tetramer at saturating conditions, washed, and incubated for various periods of time to allow dissociation of the TCR–peptide–MHC complexes (Fig. 2A and B). The 6.15 cells rapidly lost tetramer staining, such that by 30 minutes all cells were tetramer-negative (tetramer⁻) and indistinguishable from background. The 6.17 cells lost fluorescence over time, but a significant fraction remained tetramer⁺ even after overnight incubation. Thus, CD8 T cells from 6.17 mice show slower tetramer dissociation than do 6.15 cells (t_1/2 = 41 vs. 4.9 minutes); we therefore reclassified these transnuclear mouse lines as TRP1^high and TRP1^low, respectively.

The degree to which cultured CD8 T cells can produce cytokines in response to properly presented serial dilutions of peptide correlates with antitumor function (28). We cultured isolated CD8 T cells from TRP1 transnuclear mice with BMDCs pulsed with different amounts of TRP1 peptide, either the heteroclitic variant used for tetramer staining or the native TRP1 sequence. TRP1^low T cells were unable to respond to the heteroclitic TRP1 peptide at concentrations less than 100 pg/mL, whereas TRP1^high T cells secreted IFN-γ in response to as little as 1 pg/mL peptide under the same conditions (Fig. 2C). Both TRP1^high and TRP1^low cells produced IFN-γ in response to native TRP1 peptide, but higher concentrations of peptide were required (Fig. 2D). In addition to IFN-γ, we also measured chemokine production. TRP1^low cells produced RANTES and MIP-1α even when exposed to concentrations of peptide (Fig. 2E) that yielded negligible IL-2 or IFN-γ production, confirming previous studies showing a hierarchy of effector functions based on TCR affinity (29).

A panel of TRP1-altered peptide variants was generated (Fig. 3A) that differed from the heteroclitic reference peptide by single amino acid substitutions. No substitutions were made in anchor residue positions. Sorted CD8 cells from TRP1^high and TRP1^low mice were incubated with BMDCs pulsed with the altered peptide ligands to assess IL-2 and IFN-γ production (Fig. 3B). TRP1^high T cells were more promiscuous than TRP1^low T cells. One peptide (A1) generated similar levels of both IL-2 and IFN-γ from both TRP1^high and TRP1^low CD8 cells, showing that the lower cytokine production typically observed from cells from TRP1^low mice is not an intrinsic limit on the ability of these cells to produce cytokine.

We compared the surface profile of naïve CD8 cells from age-matched TRP1 transnuclear mice and found that TRP1^low cells show higher expression of NK1.1 and KLRG1 than TRP1^high or wild-type control cells (Fig. 4A). No differences were seen in Tim3 or CD25 staining, suggesting that natural killer (NK) receptor upregulation is not a function of global activation state, but rather an intrinsic difference established by the clonotypic TCRs.

CD5 expression correlates with intensity of TCR signaling and is tightly regulated during T cell development (30). TRP1^low cells also expressed lower levels of CD5 in both the periphery and during thymic development, whereas TRP1^high cells showed a peak of CD5 expression in CD4⁺CD8⁺ thymocytes (Fig. 4B). TRP1^high cells show a decrease in CD4⁺CD8⁺ thymocytes and an increase in TCR β and CD3ϵ expression at the CD4⁺CD8⁺ stage, indicative of accelerated thymic development, as is typical of mice with prerearranged TCR genes. TRP1^low cells did not show accelerated thymic development. Overall, thymic development in both TRP1 transnuclear mouse strains is near normal, and the T cells produced have a naïve phenotype.

To determine whether TRP1 cells could be cytotoxic, we activated TRP1^high and TRP1^low CD8 T cells with anti-CD3/CD28–coated beads for 7 days and cultured with TRP1 A1 peptide-pulsed GFP⁺ B cells mixed with unlabeled B cells at a 1:1 ratio. Both high and low affinity TRP1 cells displayed specific killing in a dose-dependent fashion, with TRP1^low cells slightly outperforming TRP1^high cells (Fig. 5A). To determine whether TRP1 transnuclear cells could exert cytotoxic function against tumor cells, we established 3D cocultures using B16 cells and CD8 T cells seeded together in Matrigel. B16, whether grown in culture or explanted from a mouse, expresses TRP1 protein as evidenced by immunoblotting (Fig. 5B). Coculture of B16 and TRP1 transnuclear cells showed an increase in the ratio of dead/live tumor cells in groups that received either TRP1^low or TRP1^high CD8 T cells. These results show that both lines of TRP1 transnuclear cells can be cytotoxic under 3D culture conditions (Fig. 5C). Again, TRP1^low cells tended to show greater cytotoxicity than TRP1^high cells although these differences did not reach statistical significance.

To assess their tumor-protective function in vivo, naïve CD8 T cells isolated from TRP1^high or TRP1^low mice were transferred into wild-type recipients simultaneously with subcutaneous inoculation with B16 melanoma. The B16 dose of 50,000 cells was empirically determined such that tumors would first become palpable at day 7 to 10 post-inoculation and 50% of untreated mice would require euthanasia by 21 days post-inoculation. Under these conditions, administration of naïve TRP1^high cells showed a modest 2-day difference in median survival, whereas TRP1^low cells had no discernible effect on tumor growth (Fig. 6A). Administration of multiple doses of naïve CD8 T cells (at days 0, 3, and 5 post-tumor inoculation) gave similar results to single dose trials, with TRP1^high cells conferring a 2-day survival benefit and TRP1^low cells having no discernible effect (data not shown).

We next analyzed the effect of ex vivo activation of the TRP1 cells before adoptive transfer. CD8 T cells from TRP1^high or TRP1^low mice were cultured with BMDCs and TRP1 peptide for 18 hours and transferred into wild-type hosts at day 1 postinoculation with B16. Under these conditions, both TRP1^high and TRP1^low cells conferred a 4-day increase in mean survival (Fig. 6B). Thus TRP1^low cells acquire antitumor properties equal to their high-affinity counterparts upon ex vivo activation. As further confirmation, CD8 T cells from the same TRP1^low donor mouse were adoptively transferred either naïve, or after culture with peptide-pulsed BMDCs (Fig. 6C). Although activation of polyclonal CD8 T cells has a modest effect, possibly due to the selective expansion of TRP1-specific cells from the polyclonal pool, activated TRP1^low cells conferred an additional 5-day increase in median survival when compared with polyclonal activated CD8 T cells (Fig. 6C).

Studies using transgenic CD8 T cells from Pmel mice have shown that decreasing the number of transferred CD8 T cells leads to increased proliferation rates and increased antitumor effects of the transferred cells (31). In Fig. 6A–C, we administered a constant dose of 3 million TRP1 cells per mouse. Peptide-activated TRP1^high cells retained their antitumor activity at lower doses, and showed optimal antitumor effect at a dose of 300,000 to 500,000 cells (Fig. 6D); the 300,000 T cell dose was used for all subsequent experiments. To assess possible synergy between TRP1^high and TRP1^low cells, purified CD8 T cells from each transnuclear mouse line were peptide-activated ex vivo and transferred into wild-type hosts, either alone or mixed at a 1:1 ratio. The total number of transferred T cells remained constant. The combination of TRP1^high plus TRP1^low cells did not confer additional survival benefit beyond what could be seen with either population alone (Fig. 6E).

The cumulative effect of activated TRP1 transnuclear cells against subcutaneous B16 caused a delay, but did not prevent tumor growth. This eventual failure to control the tumor could be due to several factors: (i) incomplete activation of the TRP1 transnuclear cells ex vivo; (ii) failure to traffic to the tumor; (iii) immunosuppression in the tumor microenvironment; or (iv) loss of the TRP1 epitope. To address some of these issues, we first cultured TRP1^high cells with anti-CD3/CD28–coated beads and found that 40% of mice survived (Fig. 6F). We next transferred naïve TRP1^high or TRP1^low cells into RAG-deficient mice to induce homeostatic proliferation of the transferred T cells and to observe TRP1 transnuclear CD8 T cell function in the absence of CD4⁺ Tregs. When challenged with B16, the RAG-deficient mice reconstituted with either TRP1^high or TRP1^low cells showed a 7-day delay in tumor growth (Fig. 7A). The tumors harvested from RAG-deficient mice in Fig. 7 were melanotic, showing that melanin production was intact. Immunoblotting for TRP1 protein showed no decrease in overall TRP1 levels (data not shown). A more likely explanation for the eventual failure of CD8 T cells to control tumor growth is exhaustion. In RAG-deficient hosts, which lack endogenous T cells, the tumor-infiltrating CD8 T cells are derived entirely from the transferred TRP1 transnuclear population. Both TRP1^high and TRP1^low cells can migrate to the tumor, yet even when they accumulate to the extent seen in RAG-deficient hosts, they eventually fail to control tumor outgrowth. Surface expression of PD-1 and LAG3 was increased on TRP1 transnuclear cells harvested from end-stage tumors (Fig. 7B), consistent with exhaustion (32, 33).

The role of TCR affinity in the antitumor T cell response remains a topic of debate. On the one hand, the endogenous pool of antitumor T cells is generally of low affinity, presumably because T cells bearing high-affinity TCRs that might recognize antigens on tumors are deleted in the course of thymic development, as most such antigens represent self-antigens expressed at supranormal levels. Tumor-specific CD8 T cells express TCRs of much lower affinity than typical antiviral T cells, as was recently quantified for 24 different human TCRs (9). The repertoire of antitumor TCRs is limited not only by thymic deletion of high-affinity self-reactive clones but also by negative regulatory mechanisms in the tumor environment. However, high-affinity antitumor T cells can be generated in experimental mouse models or can be expanded by ex vivo culturing of human samples. The increased use of chimeric antigen receptors (34, 35), usually selected to display higher affinities for their cognate antigens than those of typical endogenous TCRs, demands a better understanding of these supraphysiologic binding affinities: how does increased TCR affinity affect functional outcomes such as cytokine production, cytotoxicity, and memory formation? The role of TCR affinity with respect to tumor destruction and long-term antitumor memory is therefore an important question and at a minimum would require the comparison of T cells that recognize the same antigen with different affinities.

Although intuitively the highest affinity T cells would seem to be the most efficacious at tumor regression, the situation is likely more complicated. In human cytomegalovirus (CMV) infection, a population of T cells with such low affinity that they fail to bind MHC tetramers, still produces more IFN-γ and is more cytotoxic than higher affinity T cells from that same individual (36). Several lines of evidence suggest that a CD8 T cell's fate is determined, or programmed, by its initial encounter with antigen (37–39). Duration and strength of antigenic stimulation are inversely correlated with CD8 memory cell differentiation during acute viral infection (38, 39). Low-affinity T cells can increase their functional avidity by upregulating expression of CD8. Conversely, high-affinity T cells downregulate CD8 expression to avoid antigen-induced cell death (39, 40). Upon encountering abundant antigen, high-avidity CD8 T cells cease proliferation and initiate apoptosis, although they can still exert cytotoxic function before death (18).

Adoptive cell transfer therapy, as pioneered by Rosenberg and colleagues, selects for the highest affinity CD8 T cells and aims for maximal activation in vitro before infusion into the patient. This strategy has been remarkably successful in a limited clinical population (41–43), and suggests that high-affinity, maximally activated CD8 T cells confer the maximal antitumor response. Likewise, comparison of two different T cell lines specific for the same model antigen overexpressed in the pancreas showed that the higher affinity clone 4 cells were better able to expand in the pancreatic environment and cause tumor destruction, as measured by increased serum glucose (44) although low-affinity clone 1 cells were able to mediate tumor destruction when paired with antigen-specific CD4 cells (45). However, when the same clone 4 cells were injected into mice with hemagglutinin (HA)-expressing renal cell carcinomas, the antitumor effects of high-affinity clone 4 cells were curtailed by suppressive mechanisms in the tumor microenvironment (46). Sequence analysis of TCRs from T cells induced by effective versus ineffective vaccine strategies shows that higher affinity T cells were induced by the effective vaccines, although the K_D values measured for each group differed by less than 2-fold (47).

On the other hand, several studies have suggested that an optimal CD8 T cell response uses low- to moderate-affinity TCRs. Retroviral transduction of T cells with a panel of TCRs of known gradations in affinity showed that the highest affinity TCRs were absent from tumor-infiltrating CD8 T cells, although these TCRs could be found on tumor-infiltrating CD4 cells (48). A panel of NY-ESO1–specific T cell lines engineered to express TCRs with affinities ranging from low to supraphysiologic showed that optimal cytotoxicity and calcium flux occurred using TCRs with intermediate K_D values of 1 to 5 μmol/L with a decline in functionality for higher affinity TCRs (49, 50). A similar ceiling was observed using vaccination with peptide mimetics with a range of affinities for a given T cell clone (51). A recent study evaluated a large panel of human anti-melanoma cell lines for peptide sensitivity, production of various cytokines, and TCR affinity as defined by the kinetics of multimer binding. Although peptide sensitivity and production of IL-2, TNF-α, and IFN-γ correlated with tumor cell killing, multimer binding failed to correlate with any of the functional parameters measured (28), suggesting that TCR affinity alone could not predict antitumor responses. This raises the interesting possibility that attributes other than affinity, yet associated with a particular TCR clonotype, may determine the functional property of a T cell. Using tumors transduced with altered peptide ligands for OT-I CD8 T cell recognition, low-affinity TCR interactions generated enhanced cytokine production during both primary and secondary responses, but the lower affinity T cells were more susceptible to TGF-β–mediated suppression (52). An interesting study compared the TCR affinity of the endogenous CD8 response to OVA-expressing glioma as a function of the vaccine injection site (53). Mice vaccinated at sites proximal to the tumor generated lower affinity CD8 T cells than mice vaccinated at more distant sites, which could implicate suppressive mechanisms that specifically limit high-affinity CD8 priming in the tumor-draining lymph nodes (53).

In the TRP1 transnuclear system, we show that an approximately 10-fold difference in tetramer dissociation rates and a 100-fold difference in peptide responsiveness have almost no effect on the overall antitumor function of the transferred CD8 T cells. How do the low-affinity T cells manage to perform as well as their high-affinity counterparts? Although tetramer dissociation and IFN-γ production showed clear distinctions between the two TRP1-specific transnuclear lines, we observed much less of a difference when comparing cytotoxic function in a 3D culture model. Likewise, both TRP1^high and TRP1^low cells produced the chemokines RANTES and MIP-1α when stimulated with picomolar concentrations of native TRP1 peptide, conditions under which no IL-2 or IFN-γ could be detected in supernatants from the TRP1^low cultures. Because the concentration of antigen presented in a tumor-bearing mouse is likely to be small, the fact that TRP1^low cells produce monocyte-attracting chemokines in response to low doses of peptide could be relevant for how TRP1^low CD8 T cells influence the tumor microenvironment. Likewise, the increased IL-2 production from TRP1^high cells could enhance Treg recruitment or proliferation in the tumor microenvironment, although such an effect would be eliminated in RAG-deficient hosts, which lack Tregs. Another likely possibility is increased activation-induced cell death of TRP1^high cells in the tumor environment, which could limit the efficacy of high-affinity CD8 T cells (54, 55). Indeed, TRP1^high tumor-infiltrating cells show reduced CD8 expression, indicative of a more activated state.

Here, we have generated a pair of mice with T cells specific for the same MHC–peptide combination but differing in affinity by one to two orders of magnitude. The TCR α and TCR β chains are under the control of their endogenous promoters, and expressed at physiologic levels. Both TCRs were selected by directly harvesting tetramer⁺ CD8 cells from a vaccinated mouse without any selection in tissue culture, and thus are representative of the normal pool of activated CD8 T cells that arises during an immune response. As such, we anticipate that the TRP1 transnuclear mice reported here may be useful for the studies of T cell development and T cell activation.

Although TRP1 is a self-antigen expressed in healthy melanocytes, we did not observe vitiligo in TRP1^high or TRP1^low mice on a RAG-proficient background, suggesting a possible role for CD4⁺ Tregs. Tregs are absent in TRP1^highRAG2^−/− mice, and these mice develop a mild spontaneous vitiligo around 3 months of age, affecting primarily the periorbital region. The Tregs that arise in TRP1 transnuclear mice are constrained to use a single TCR Vβ8.2 or TCR Vβ5, respectively, thus limiting the diversity of potential Tregs specificities in these mice. Nevertheless, the Tregs that develop are apparently adequate to prevent autoimmunity even in the presence of a high fraction of autoreactive CD8 T cells.

TRP1 can serve as a tumor rejection antigen, and we show a modest antitumor effect of anti-TRP1 CD8 T cells as a monotherapy. Although the effects are small, the delay in tumor growth and slight increase in overall survival parallels the human situation, where monotherapies such as anti-CTLA-4 also show modest clinical benefit (56). A future challenge for tumor immunology will be to determine which therapies synergize to achieve maximum clinical effect. CD8 T cell exhaustion is commonly observed in human patients, and must be overcome for immunotherapy to succeed. The TRP1 mice offer an excellent preclinical platform for testing the immunomodulatory effects of potential new drugs or drug combinations.

No potential conflicts of interest were disclosed.

Conception and design: S.K. Dougan, M. Dougan, J. Kim, J.A. Turner, R. Jaenisch, E. Celis, H.L. Ploegh

Development of methodology: S.K. Dougan, S. Ogata, H.-I. Cho, E. Celis

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): S.K. Dougan, M. Dougan, S. Ogata

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): S.K. Dougan, M. Dougan, J. Kim, J.A. Turner, H.L. Ploegh

Writing, review, and/or revision of the manuscript: S.K. Dougan, M. Dougan, J. Kim, J.A. Turner, E. Celis, H.L. Ploegh

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): J. Kim, J.A. Turner, H.-I. Cho

Study supervision: R. Jaenisch

The authors thank Patti Wisniewski and Chad Araneo for cell sorting, and John Jackson for mouse husbandry.

S.K. Dougan was funded by the Cancer Research Institute and by the Bushrod H. Campbell and Adah F. Hall Charity Fund and Harold Whitworth Pierce Charitable Trust. S. Ogata was funded by Janssen Pharmaceutica NV. E. Celis, H.L. Ploegh, and R. Jaenisch are funded by grants from the NIH. S.K. Dougan and H.L. Ploegh are funded by Janssen Pharmaceuticals, Inc. and by the AACR-Pancreatic Cancer Action Network.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.