Adoptive T cell therapy (ACT) with antitumor CTL is a promising and tailored treatment against cancer. We investigated the role played by the affinity and avidity of the interaction between the tumor and the CTL on the outcome of ACT against a mouse non-Hodgkin B cell lymphoma that expresses OVA as a model neoantigen. ACT was assessed under conditions where antitumor CTL expressed TCR of varying affinity for OVA. We also assessed conditions where the avidity of Ag recognition varied because the lymphoma cells expressed high or low levels of OVA. Efficient eradication of small tumor burdens was achieved by high- or low-affinity CTL. Tumors expressing low levels of OVA could also be eliminated. However, ACT against large tumor burdens was unsuccessful, accompanied by CTL deletion and functional impairment. This negative outcome was not prevented by lowering the affinity of the CTL or the expression of OVA in the lymphoma. Thus, tumor burden, rather than CTL affinity or avidity, appears to be the main determinant of ACT outcomes in our lymphoma model. Insofar as our results can be extrapolated to the clinical setting, they imply that the range of CTL and tumor-associated Ag combinations that may be effectively harnessed in ACT against lymphoma may be wider than generally assumed. CTL expressing low-affinity TCR may be effective against lymphoma, and lowly expressed tumor-associated Ag should be considered as potential targets, but tumor reduction should always be implemented before infusion of the CTL.
To bypass the hurdle of a tolerized immune system in tumor bearing-hosts, the infusion of “ready-made” immunity has emerged as a promising treatment option. One of these strategies is adoptive T cell therapy (ACT), which consists of the transfer of tumor-associated Ag (TAA)-specific CTL (1, 2). The outcomes of ACT in some clinical trials have been very encouraging, but ACT has yet to live up to its full potential, and intense efforts are under way to improve its overall efficacy. Success of ACT may be compromised as the result of a combination of factors that ultimately impact the efficacy of CTL-mediated tumor eradication. Importantly, in many patients, ACT fails because of the impairment of long-term expansion, reduced survival, and/or functional inactivation of the transferred CTL. Defining the parameters that influence the success of ACT is important to ensure the effective use of this therapy in the clinic. Mouse models of ACT can provide insights into the influence of such parameters, owing to the availability of sophisticated reagents and experimental approaches that can be studied in these settings, although translation of the conclusions of mouse studies requires validation in the clinic.
An important parameter thought to impact ACT outcomes is the avidity of the interaction between the transferred CTL and the tumor cells against which they are targeted. The functional avidity of a T cell is defined as the strength of the interaction between the T cell and the cell presenting the Ag and is governed by several factors, including the number of TCR expressed by the CTL, the affinity of binding between the TCR and the MHC-peptide complex(es) it recognizes, the expression of adhesion and costimulatory molecules that stabilize the immunological synapse, and the abundance of the Ag presented as MHC-peptide complexes on the surface of the target cell. It is widely believed that, for ACT to be effective, high-avidity CTL are required. Higher-affinity CTL clones exhibit superior activity compared with their lower-affinity counterparts (3–6); consequently, strategies to engineer high-affinity CTL for ACT therapy are in practice (7). To further boost the avidity of the interaction, ACT is generally directed against TAA with high expression in malignant cells, to the detriment of many other potential TAA that may be expressed and presented at lower levels. However, recent developments show that using CTL engineered to recognize TAA with high avidity attracts considerable risks. The use of CTL with artificially enhanced TCR affinity was associated with dangerous toxicity (8) caused by CTL recognizing the TAA, or a closely related Ag, on normal tissues. An extreme example is the fatal encephalitis and cardiotoxicity elicited following transfer of CTL engineered to target MAGE-A3 with high affinity (9, 10). Consequently, the assumption that clinically effective ACT requires manufacture of high-avidity CTL warrants careful examination, at least in preclinical models.
In this study, we manipulated the affinity of the TCR and the expression level of their cognate TAA to assess the impact of varying the avidity of the CTL–tumor cell interaction on the efficacy of ACT in a preclinical setting. The tumor that we targeted in this study is the well-characterized non-Hodgkin mouse lymphoma model, Eμ-myc, engineered to express OVA as a model neo-TAA. Our previous analyses showed that this tumor is ignored by the host immune system (11); however, under conditions of ACT, it can be successfully eliminated from the host (12). Notably, in this model, ACT is effective when the tumor burden is small; however, once a specific tumor size is exceeded, ACT fails as a result of the functional impairment of the transferred CTL (12). Therefore, the Eμ-myc–OVA model enables parameters that dictate ACT outcomes following both successful and failed tumor eradication to be investigated in mice. Using CTL expressing TCR of high versus low affinity, in addition to examining ACT under conditions of high versus low TAA levels, we show that avidity is not a major determinant of ACT outcomes in this animal model.
Materials and Methods
Experimental mice were housed and bred under specific pathogen–free conditions at the Walter and Eliza Hall Institute (Parkville, VIC, Australia) and Bio21 Institute Animal Facilities. C57BL/6J (Ly5.2), C57BL/6-Pep3b (Ly5.1), Ly5.1 × OT-I (13), OT-3 (14), and gBT-1 (15) male mice aged 6–8 wk were used for all experiments. All experiments were conducted in accordance with guidelines provided by the National Health and Medical Research Council of Australia. Experimental procedures were approved by the Animal Ethics Committee, Melbourne Health Research Directorate, and The University of Melbourne.
Eμ-myc lymphomas expressing GFP and GFP-OVA were described previously (11). Tumors with different TAA expression levels were generated from the original Eμ-myc–GFP-OVA lymphoma by sorting tumor cells by flow cytometry based on their GFP expression into GFP high, intermediate, and low. Sorted cells were injected into naive recipient mice for further expansion. After 11 d tumor cells were extracted from lymph nodes, frozen and stored in liquid nitrogen until use. Tumors were reinfused into recipient mice by i.v. injection.
Generation of CTL
Lymphocytes were isolated from spleen and/or lymph node (LN) using mechanical disruption of organs through a 40-μm sieve (BD Biosciences). RBC were lysed using red cell removal buffer (0.168 M ammonium chloride). In some cases, cell suspensions were labeled with anti–MHC II (M5/114), anti-CD11b (M1/70), anti-F4/80 (F4/80), anti-RBC marker (Ter119), anti-Ly6C/G (RB6-8C5), and anti-CD4 (GK1.5) to enable negative depletion with anti-rat IgG-coupled magnetic beads (QIAGEN). The purity of OT-I and gBT-I cells was determined by staining with anti-CD8 (YTS169) and anti-TCR Vα2 (B20.1), whereas OT-3 cells were stained with anti-CD8 (YTS169) and anti-TCR Vβ5 (MR9-4), and analyzed by flow cytometry. Enriched transgenic T cells were routinely 85–98% pure. OT-I CTL (16) and OT-3 CTL were generated by culturing 10 × 106 OT-I splenocytes or OT-3 CD8+ enriched splenocytes with 20 × 106 SIINFEKL-pulsed splenocytes. gBT-I CTL were generated by culturing 20 × 106 gBT-I splenocytes with 20 × 106 irradiated (1500 cGy) SSIEFARL-pulsed splenocytes. After 4 d of culture, the resulting cells contained 80–90% CD8+ and TCR Vα2+ cells (OT-I), CD8+ and TCR Va2+ cells (gBT-I), and CD8+ and TCR Vβ5+ cells (OT-3).
CTL analysis following transfer into Eμ-myc lymphoma-bearing mice
CTL were adoptively transferred after i.v. inoculation of lymphoma cells, and the lung, spleen, or LN were harvested for analysis. For the lung, mice were perfused slowly with 10 ml PBS through the left ventricle of the heart. The lung tissue was cut into small pieces with scissors and digested in 1 ml medium containing 3.5 mg collagenase (Type III; Worthington Biochemicals) and 0.5 mg DNase (grade II; Boehringer-Mannheim) for 1 h at 37°C. LN, spleen, and digested lung were disrupted through a 70-μm cell strainer (BD Falcon).
Cell populations and surface marker expression profiles were identified by flow cytometry with mAbs against CD8 (YTS 169.4), B220 (RA3-6B2), and TCR Vα2 (B20.1) (all generated at Walter and Eliza Hall Institute), Ly5.1 (A20.1), TCR Vβ5 (MR9-4), and PD1 (J43; all from eBioscience). Dead cells were excluded based on staining with propidium iodide (Sigma-Aldrich). Samples were analyzed using an LSR II, LSR Fortessa, or FACSCanto (BD Biosciences). Analysis was performed using FlowJo (Tree Star). Fluorescent CaliBRITE beads (BD Biosciences) were used to enable calculation of absolute cell numbers (17).
In vivo CTL killing assay
Spleen single-cell suspensions were pulsed with 30 ng/ml SIINFEKL (OT-I) or SSIEFARL (gBT-I) peptide (Auspep) and labeled with a high concentration (5 μM) of CSFE (Sigma-Aldrich) or not pulsed with peptide and labeled with a low concentration of CFSE (0.5 μM). In some experiments, CellTrace Violet (Invitrogen) was used instead of CFSE at the same concentrations. Recipient mice were injected i.v. with a combination of pulsed and unpulsed splenocytes in equal ratios (1 × 107 cells). Spleens were harvested 24 h later and analyzed by flow cytometry. The percentage of specific killing was calculated relative to animals that received only target cells.
In vitro CTL killing assay
CTL generated in vitro or CTL sorted to purity by flow cytometry were cultured with Eμ-myc–GFP (5 × 104) and Eμ-myc–GFP-OVA (5 × 104) target cells. Killing of Eμ-myc–GFP-OVA cells was analyzed by flow cytometry 24 h later.
Intracellular cytokine staining
Spleens were prepared as single-cell suspensions and incubated in the presence or absence of SIINFEKL peptide for 4 h at 37°C in the presence of brefeldin A (10 μg/ml) (GolgiPlug; BD Biosciences). After 4 h of incubation, the cells were stained with Ab for surface markers, treated with Fixation/Permeabilization buffer (BD Biosciences), and stained with anti–IFN-γ (XMG1; BioLegend) or anti–TNF-α (MP6-XT22; BioLegend) in Permeabilization buffer (BD Biosciences). Samples were analyzed by flow cytometry.
Ag presentation by Eμ-myc lymphoma
OT-I lymphocytes were enriched from OT-I–transgenic mice, labeled with CFSE, and cultured with titrated numbers of Eμ-myc lymphoma cells previously irradiated at 1500 rad. Proliferation of T cells was assessed based on CFSE dilution. CaliBRITE beads (BD Biosciences) were included to enable calculation of absolute cell numbers (17).
OT-3 CTL express TCR of lower affinity than OT-I CTL
OT-I TCR-transgenic T cells express a high-affinity TCR that recognizes the OVA-derived OVA257–264 peptide presented by H-2Kb (henceforth Kb-OVA). Similar to OT-I, OVA-specific CD8+ OT-3 T cells recognize Kb-OVA but with a 50-fold lower functional affinity (14). To compare the efficacy of CTL expressing TCR of differing affinity against the same Ag in a mouse model of ACT, we generated in vitro–cultured CTL from OT-3 or OT-I TCR-transgenic T cells. Analysis of anti–Kb-OVA tetramer binding as a measure of relative TCR affinity showed that OT-I CTL display significantly higher affinity for Kb-OVA than do OT-3 CTL. Indeed, tetramer binding in the latter was barely above that observed in CTL derived from intercrossed A9 and RIP-mOVA mice, in which negative selection leaves only low-affinity anti-OVA T cells in the periphery (18) (Fig. 1A). Evaluation of cytokine production by OT-I and OT-3 CTL following restimulation with 100 μg/ml SIINFEKL showed that OT-3 CTL display reduced IFN-γ and TNF-α secretion relative to OT-I CTL (Fig. 1B). Moreover, OT-3 CTL elicited significantly less killing of OVA-expressing tumor cells in vitro than did OT-I CTL (Fig. 1C). Together, these results confirm the lower TCR affinity of OT-3 CTL relative to OT-I CTL.
Low- and high-affinity CTL effectively eliminate small lymphomas
The capacity of OT-3 CTL to kill tumors in vivo was determined in a mouse model of ACT in which transferred CTL were specific for a TAA expressed by Eμ-myc lymphoma cells. In our previous studies, we demonstrated that adoptive transfer of high-affinity OT-I CTL effectively eliminates small lymphomas in an Ag-specific fashion (12). To compare the effectiveness of low-affinity CTL in this setting, mice were inoculated with Eμ-myc lymphomas expressing GFP alone (GFP tumor) or GFP together with OVA (GFP-OVA tumor) and 2 d later were transferred with OT-3 or OT-I CTL. Tumor burden was measured 2 d after CTL transfer (Fig. 2A). Lymphoma was clearly detectable in mice that did not receive CTL, as well as following transfer of CTL into GFP tumor-bearing hosts; however, a dramatic reduction in tumor burden was observed in mice receiving OT-I or OT-3 CTL (Fig. 2B, 2C). This result suggested that, provided the affinity of the TCR expressed by CTL is sufficiently high to enable target recognition, low-affinity CTL can be as effective an agent of anti-lymphoma ACT as high-affinity ones, at least in settings of low tumor burden.
Both high-affinity and low-affinity CTL are inactivated by large lymphomas
Next, we investigated ACT under conditions in which the tumor burden in mice receiving CTL is high. Our previous studies showed that a large proportion of high-affinity OT-I CTL is rapidly eliminated, with the surviving CTL becoming functionally impaired when used for ACT in mice bearing large Eμ-myc–GFP-OVA lymphomas. This occurs in the spleen (12) and LN (Supplemental Fig. 1). What we observed as CTL deletion cannot be explained by the accumulation of CTL in tissues outside the secondary lymphoid organs. We recovered equivalent numbers of transferred OT-I CTL from the lungs of Eμ-myc–GFP-OVA lymphoma-bearing mice relative to Eμ-myc–GFP lymphoma-bearing mice, with the OT-I CTL being functionally impaired in settings following the encounter of their cognate Ag (Supplemental Fig. 2). We examined the possibility that if CTL inactivation were a consequence of a high-affinity interaction with cognate Ag, it might be expected that low-affinity OT-3 CTL would remain active under these conditions. To test this hypothesis, we transferred OT-I or OT-3 CTL into mice that had received Eμ-myc lymphoma (GFP or GFP-OVA) 5 d previously and evaluated outcomes 2 d after CTL transfer (Fig. 3A). Neither OT-I CTL nor OT-3 CTL induced a significant reduction in GFP-OVA tumor burden relative to mice that received no CTL or that contained control GFP tumors (Fig. 3B, 3C). Next, the number and phenotype of the CTL recovered under these conditions were assessed. Consistent with our previous observations, there was a significant reduction in the number of OT-I CTL recovered from mice bearing OVA tumors relative to mice with control GFP tumors. In contrast, OT-3 CTL recovery was comparable in mice bearing GFP-OVA and GFP tumors (Fig. 3D). Phenotypically, both OT-I and OT-3 CTL displayed TCR Vβ5 downregulation and enhanced programmed cell death protein (PD)-1 expression in mice bearing GFP-OVA tumors, a phenotypic change that required Ag recognition because it was not observed in mice bearing GFP tumors (Fig. 3E) (12). Therefore, although recognition of Ag on mice bearing a large tumor burden did not affect the recovery of OT-3 CTL, functional impairment, based on reduced TCR and elevated PD-1 (19), still seemed to be occurring. To verify this, the killing activity of OT-I and OT-3 CTL was assessed by injection of SIINFEKL-pulsed target cells (normal splenocytes) 2 d after CTL transfer. All targets were killed within 24 h in the spleens of mice bearing GFP tumors, showing that OT-I and OT-3 retained a high and comparable capacity to lyse target cells (Fig. 3F). In contrast, for both OT-I and OT-3 CTL, target cell killing was severely impaired in mice bearing GFP-OVA lymphomas (Fig. 3F), despite the similar numbers of CTL present in GFP- and GFP-OVA–bearing mice (Fig. 3D). Therefore, both high-affinity OT-I and low-affinity OT-3 CTL are inactivated in vivo in an Ag-specific fashion following encounter of a high tumor burden.
Generation of B cell lymphoma with differential levels of TAA
We used a second approach to alter the avidity of the interaction between transferred CTL and tumor. In this case, we altered the expression level of OVA Ag presented per cell by the Eμ-myc lymphoma and assessed whether the lower presentation of TAA impacted ACT outcomes under conditions of small and large tumor burdens.
We purified Eμ-myc–GFP-OVA lymphoma cells with three levels of GFP expression and expanded them in recipient mice. When transferred to naive mice, all three tumors grew with similar kinetics and maintained stable levels of GFP expression (Fig. 4A). Expression of OVA in these cells correlates with that of GFP (11), so the three resulting tumors expressed low (OVAlow), intermediate (OVAint), or high (OVAhigh) levels of the model TAA. To verify this, we cultured naive OT-I CTL with titrated numbers of irradiated OVAhigh, OVAint, or OVAlow tumors and measured OT-I proliferation. Approximately 4- and 30-fold more OVAint and OVAlow tumor cells were required, respectively, to induce a level of OT-I proliferation similar to that elicited by OVAhigh cells (Fig. 4B). Likewise, the CTL/target cell ratio required to kill 50% of OVAlow cells in vitro during incubation of the tumor cells with OT-I CTL was 4-fold higher than the ratio required to kill an equivalent fraction of the OVAint or OVAhigh cells (Fig. 4C). Approximately 50% of OVAlow tumor cells were refractory to OT-I CTL killing in vitro, even at high CTL/target cell ratios, probably because their level of OVA presentation was just below the level of detection by OT-I CTL (Fig. 4C). Therefore, Eμ-myc–GFP-OVAhigh and OVAlow tumors provide model tumors of significantly different TAA levels, with the latter expressing the Ag at a level barely above the limit of detection by high-affinity CTL.
Tumors expressing low levels of TAA can be eliminated by ACT but elicit CTL inactivation under conditions of high lymphoma burden
First, we assessed how TAA expression impacted ACT under conditions of successful tumor eradication in mice bearing small OVAhigh or OVAlow lymphoma burden. Tumors were inoculated into recipient mice, and OT-I CTL were transferred i.v. 2 or 3 d later. ACT outcomes were evaluated 48 h later. Elimination of tumor cells expressing low levels of OVA was achieved in mice that received CTL when the tumor cells were inoculated 2 d previously (Fig. 5A, 5B). This suggested that the low expression of OVA did not impair these tumors from being eliminated by OT-I CTL, and only the cells that expressed the Ag below the level of detection survived, as they did in vitro (Fig. 4C). However, if the CTL were injected at day 3 following OVAlow tumor inoculation, we detected little elimination of OVAlow lymphomas by OT-I CTL (Fig. 5C). This was surprising and suggested that tumors with low TAA expression levels were also capable of rendering functional inactivation of the transferred CTL if the tumor burden reached a critical threshold.
To specifically address whether CTL against a low-avidity TAA were functionally inactivated by a large tumor burden, mice bearing OVAhigh, OVAlow, or control GFP tumors for 5 d were inoculated with OT-I CTL, and the tumor cells in the spleen were enumerated 2 d later (Fig. 6A). Neither OVAhigh or OVAlow tumor underwent a reduction in size following CTL transfer, indicating that ACT is unsuccessful under these conditions (Fig. 6B, 6C). The number, phenotype, and function of the OT-I CTL persisting under these conditions were evaluated. Mice bearing OVAhigh or OVAlow tumors contained significantly fewer OT-I CTL than those bearing the control GFP lymphoma (Fig. 6D). Therefore, the level of TAA presentation had little influence on the loss of CTL that ensues following encounter of a high tumor burden. The phenotype of the antitumor CTL remaining in the mice bearing OVAhigh lymphoma was TCR Vα2low PD-1high, whereas the CTL remaining in mice bearing OVAlow lymphomas retained almost normal TCR Vα2 levels but were PD-1high (Fig. 6E). Homogeneous PD-1 upregulation confirmed that the number of OVAlow tumor cells presenting Ag, as well as the amount of Ag expressed by those cells, exceeded the threshold required to elicit high PD-1 expression by most, if not all, of the transferred OT-I CTL. Cytotoxic activity of the CTL remaining in OVAhigh and OVAlow lymphoma-bearing mice was evaluated in vivo by injecting tumor-bearing hosts with SIINFEKL-pulsed target cells 24 h after ACT. Target cell killing was reduced in both OVAhigh and OVAlow lymphoma-bearing mice compared with active CTL killing in GFP-lymphoma bearing mice (Fig. 6F). This suggested that, in both settings, the CTL were functionally impaired; however, it remained a possibility that the reduced killing was due to the lower number of CTL present in the OVA-tumor bearing hosts and not to a CTL-intrinsic impairment. To test this, ex vivo effector CTL activity was examined by purifying CTL from tumor-bearing hosts and culturing them in vitro with SIINFEKL-pulsed target cells. OT-I CTL from OVAhigh and OVAlow tumor-bearing mice were functionally impaired, requiring a 4–8-fold higher target:T cell ratio than their counterparts purified from GFP tumor-bearing hosts to kill an equivalent number of cells (Fig. 6G). Furthermore, measurement of IFN-γ production by OT-I CTL isolated from OVAhigh or OVAlow tumor-bearing mice was reduced compared with OT-I CTL isolated from GFP-tumor bearing mice (Fig. 6H), again demonstrating intrinsic CTL impairment. In summary, large OVA-expressing tumors promote CTL deletion and impairment, irrespective of the level of TAA presentation.
The current consensus in the ACT field is that for this therapy to be effective, the infused CTL must be of high avidity. This limits the choice of TAA targeted by the CTL to those highly expressed by tumors. Efforts are also directed to produce CTL derived from rare, high-affinity intratumoral lymphocytes or to manufacture CTL engineered to express high-affinity TCR or chimeric AgRs (7). We assessed the impact of CTL avidity on ACT outcomes in a mouse model of non-Hodgkin lymphoma where OVA is expressed as a neoantigen. We show that avidity is not the major determinant that dictates the fate of transferred CTL or the efficacy of tumor eradication in this preclinical setting.
Low-affinity OT-3 CTL could eliminate small tumor burdens as efficiently as high-affinity OT-I CTL. This was surprising given the poor killing capacity of OT-3 CTL in vitro. In vivo conditions may elevate OT-3 CTL killing due to the cytokine milieu, the architecture of the lymphoid organs, and/or additional cell populations that promote their killing capacity. As such, in vitro assessment of CTL killing in some cases may be misleading and may not accurately predict in vivo outcomes. More importantly, our results indicate that the requirement to artificially engineer CTL TCR for higher-affinity recognition of MHC-peptide complexes may not be as critical as previously considered. Indeed, in some settings, increasing TCR affinity can impair, rather than enhance, CTL activity (20–22). Although it seems counterintuitive, this is likely due to the reduced ability of high-affinity CTL to undergo serial triggering, where one MHC-peptide complex sequentially engages several TCR to achieve a critical activation threshold (23–25). Indeed, our data support findings from both mouse (26) and human (27) antimelanoma CTL where CTL with higher-avidity TCR do not elicit more potent antitumor activity than those with lower-avidity TCR. An advantage of using CTL expressing low-affinity TCR is that the risk for toxic recognition of low-abundance TAA in normal tissues may be averted. This is a question that remains to be tested in experimental settings where CTL expressing high- versus low-affinity TCR can recognize tumor and normal cells.
Identification of TAA expressed at sufficient levels to enable high-avidity recognition of tumor cells is a challenge because such Ags are likely immunoedited during tumor growth (28). However, recent technological advances have dramatically increased the ability to identify numerous TAA expressed at relatively low levels in tumors isolated from individual patients (29). Given the consensus that such TAA do not represent suitable targets for effective ACT, many may be overlooked as candidates for ACT targeting without direct evaluation. To investigate this in a preclinical setting, we established two distinct mouse lymphoma models varying solely with regard to their levels of expression of the model TAA OVA. Eμ-myc–GFP-OVAhigh was similar to our previously described Eμ-myc–GFP-OVA tumor (11, 12), whereas Eμ-myc–GFP-OVAlow expressed significantly lower levels of TAA such that it elicited poor proliferation of naive OT-I T cells in vitro. Indeed, half of these tumor cells escaped killing by high-affinity OT-I CTL, presumably because their level of OVA expression was below the threshold of recognition. Yet, ACT achieved a significant elimination of Eμ-myc–GFP-OVAlow cells in conditions of small tumor burden. Based on our results in these mouse models of ACT, we propose that a combination of two or more CTL against a corresponding number of low-abundance TAA may be as efficient as using a single CTL against a high-abundance TAA. Therefore, tumors with residual TAA levels might be effectively targeted by ACT, a conclusion that would expand the number of TAA that could potentially be targeted by manufactured CTL. Naturally, such conclusions await validation in clinical settings before they can be translated into therapies against human cancer.
Neither affinity nor avidity was a major determinant of successful ACT outcomes against small lymphomas in our experimental system. We also assessed whether the avidity of the tumor–CTL interaction was a critical determinant contributing to failed ACT outcomes. We reported previously that, when ACT is attempted for the eradication of a high Eμ-myc–OVA lymphoma burden, the therapy fails because a large number of injected CTL undergo rapid death; although the surviving CTL expand, they acquire a dysfunctional phenotype (12). In this model, the inactivation of the CTL is strictly dependent on recognition of Ag presented by the Eμ-myc tumor cells themselves and not by tumor-associated dendritic cells, myeloid-derived suppressor cells, or other APC of the host (12). We investigated whether CTL impairment stems from a high-avidity interaction between the CTL and tumor. We hypothesized that this may be the case because high-avidity interactions have the potential to evoke T cell apoptosis (6, 30, 31); however, our analysis highlights that avidity is not the major determinant of CTL impairment in our model. Both low- and high-TCR affinity CTL were rendered functionally impaired when encountering a high tumor burden. Notably, low-affinity CTL did not undergo the significant loss in cell number that was experienced by their high-affinity TCR counterparts; nevertheless, they were rendered incapable of eliciting robust CTL activity in animals bearing large tumor burdens. Likewise, lowering the avidity of the CTL–tumor cell interaction by reducing the expression of the targeted TAA (in OVAlow cells) did not prevent functional inactivation of CTL encountering large tumors. The results of these experiments concur with our previous findings (12) showing that failed ACT occurs due to high-frequency or simultaneous interactions between the transferred CTL and the targeted tumor cells in conditions of high tumor density, irrespective of the strength of the interaction. An important role for target cell density in the maintenance of CTL activity is not without precedence in the literature. Recently, similar observations were described in settings of hepatic immunity, where the encounter of a high number of Ag-expressing liver cells elicited CTL impairment (32–34).
In summary, given the reduced risk for toxicity associated with the use of lower-avidity CTL, our results prompt investigation of their application in at least some clinical settings, for instance when the risk for recognition of healthy tissue is high. However, an important conclusion of our previous and present results is that any ACT protocol might benefit from reduction of tumor bulk prior to infusion of the CTL. We demonstrated a major improvement in ACT outcomes targeting large Eμ-myc–OVA lymphomas when using this combined therapy (12). If the conclusions of our analysis of ACT based on the mouse Eμ-myc–OVA lymphoma model holds true in humans, combining ACT with strategies to ablate tumor size prior to CTL injection, rather than manipulating parameters of tumor and CTL avidity, may have the highest likelihood of ensuring robust tumor eradication with reduced risk for toxicity. Validation of these predictions in the human system awaits testing in clinical settings.
This work was supported by a Leukemia and Lymphoma Society Scholar Award (to J.A.V.) and by research fellowships and research grants from the National Health and Medical Research Council of Australia (to J.A.V.). G.S. was funded by a scholarship from National Commission for Scientific and Technological Research of Chile-Becas, Chile.
The online version of this article contains supplemental material.
The authors have no financial conflicts of interest.