Active Influx Of T Cells In B16f10 Melanoma Improves CD3 bsAb Therapy
We previously reported the therapeutic efficacy of a TRP1-targeting Fc-inert CD3 bsAb (CD3xTRP1) in a B16F10 melanoma model, which is described as an immunologically ‘cold’ tumor with endogenous expression of the TRP1 tumor surface antigen (7), using immunocompetent syngeneic mice (8). To investigate the dependency of this therapy on T-cell influx, we utilized C-X-C motif chemokine receptor 3 (CXCR3) knock-out (KO) mice, since CXCR3 is a key chemokine receptor for T-cell trafficking towards tumors (9). We inoculated wild-type (WT) or CXCR3 KO mice with B16F10 tumor cells and administered CD3xTRP1 at an early timepoint (day six and nine) (Fig. 1A). The anti-tumor activity of CD3xTRP1 was significantly impaired in CXCR3 KO mice compared to WT mice, indicating that the efficacy of the therapy depended on homing of T cells towards the tumor (Fig. 1B and S1A-B). Nevertheless, CD3xTRP1 treatment still delayed tumor outgrowth in CXCR3 KO mice, suggesting that resident T cells or other chemokine receptors were also involved (10). To understand the mechanism of action, we studied the TME four days after CD3xTRP1 treatment using 54-parameter spectral flow cytometry and OMIQ cluster analysis (fig. S1C-E). Following CD3xTRP1 therapy, we observed significantly reduced infiltration of CD8 T cells into tumors of CXCR3 KO mice compared to WT mice (Fig. 1C). Importantly, the infiltrated CD8 T cells expressed similar amounts of granzyme-B (GzmB) in both WT and CXCR3 KO mice, suggesting that they were equally cytotoxic (fig. S1F). Furthermore, a minor reduction in CD4 T-cell and NK-cell infiltrate, but no differences in the frequency and phenotype of myeloid cell were detected (fig. S1F-G). Together, these results imply that reduced T-cell infiltration into the tumors was responsible for the inefficient therapeutic effect of CD3xTRP1 in CXCR3 KO mice, suggesting that homing of T cells towards the tumor is necessary for efficacious CD3 bsAb therapy in this ‘cold’ tumor model.
We hypothesized that therapeutically increasing the frequency of intratumoral T cells could improve the treatment outcome of CD3xTRP1. As CD3 bsAbs can activate T cells independent of their TCR specificity, we chose a universal approach by using tumor-nonspecific T cells that do not recognize B16F10 melanoma. Using this strategy, the effect of increasing intratumoral T-cell numbers to amplify CD3 bsAb therapy was examined. Naïve tumor-nonspecific OT-1 T cells were transferred to B16F10-bearing mice and subjected to a prime-boost vaccination regimen with a synthetic long peptide (SLP) containing the chicken ovalbumin (OVA) CD8 T-cell epitope and toll-like receptor (TLR)7/8 agonist imiquimod and IL-2 as adjuvant, followed by administration of CD3xTRP1 (Fig. 1D). A late time point (day twelve and fifteen) was chosen for the CD3xTRP1 administration, which we hypothesized to better reflect the clinical situation in ‘cold’ tumors, where CD3 bsAb monotherapy has limited efficacy. Due to the late administration of CD3xTRP1, no effect of CD3 bsAb monotherapy was observed (Fig. 1E and S2A). However, combination of CD3xTRP1 with transfer of tumor-nonspecific OT-1 T cells and OVA SLP vaccination significantly delayed tumor outgrowth, and improved the survival rate and percentage of long-term survivors. TME analysis by flow cytometry two days after CD3 bsAb administration revealed that OVA peptide vaccination induced a strong increase in intratumoral OT-1 T cells despite the absence of their cognate antigen in the tumor (Fig. 1F and S2B-C). Of note, even though imiquimod was able to induce CXCR3 expression on T cells in the blood, we observed no effect on T-cell infiltration when combining the adjuvant (imiquimod and IL-2) with CD3xTRP1 (fig. S3A-B). These results show that vaccine-induced T-cell infiltration in the tumor, in combination with CD3 bsAb, augments survival irrespective of T-cell specificity.
CD3xTRP1 Induces Spatial Relocation Of Vaccine-induced Tumor-nonspecific T Cells And Expression Of Effector Molecules
Next, we investigated the kinetics of T-cell distribution and activation for the combination treatment in more detail. To visualize these dynamics over time, we used transgenic OT-1xTbiLuc mice, containing OT-1 T cells that express two different types of luciferase to detect the presence of OT-1 cells by converting D-Luciferin and the activation of OT-1 cells by converting CycLuc1 (11). In this way, T-cell distribution and T-cell activation can be monitored simultaneously over time using a bioluminescence in vivo imaging system (IVIS). Moreover, to discern systemic effects from CD3 bsAb-mediated local effects, we investigated these dynamics in mice inoculated with TRP1-positive and -negative KPC3 pancreatic tumor cells at distinct flanks (Fig. 2A). In line with B16F10, KPC3 tumors also represent a ‘cold’ TME (12), making it a suitable tumor model to study combinations of vaccination and CD3 bsAb. We observed that only few naïve OT-1 cells infiltrated the tumor after adoptive transfer, while most homed towards lymph organs (Fig. 2B and fig. S4A-B). However, activation of OT-1 cells by OVA SLP vaccination resulted in systemic infiltration into both KPC3 and KPC3-TRP1 tumors, as well as into the spleen. The combination of OT-1 cell transfer, OVA vaccination and administration of CD3xTRP1, induced only a modest increase in OT-1 numbers, but a strong increase in OT-1 activation, selectively in the TRP1-positive tumors (Fig. 2B-D). In contrast, no CD3 bsAb-mediated effects were observed in the TRP1-negative tumors. Analysis of end-stage tumors by flow cytometry confirmed the increased OT-1 infiltration and activation specifically in the TRP1-positive tumors after the triple combination therapy (fig. S4B). We concluded that vaccine-stimulated T cells infiltrate tumors irrespective of their TCR-specificity and that tumoral expression of the antigen targeted by the CD3 bsAb is required for a further local burst of T-cell activation.
Next, we studied the role of the individual components of the combination therapy on T-cell infiltration and activation in mice bearing KPC3-TRP1 tumors in more depth using flow cytometry, immunohistochemistry (IHC) and Nanostring transcriptomics (Fig. 3A). In line with the in vivo imaging results, OT-1 transfer with OVA vaccination strongly increased OT-1 T-cell infiltration (Fig. 3B). High-dimensional flow cytometry analysis comparing the triple treatment to only OT-1 transfer and OVA vaccination revealed the appearance of a population of highly activated OT-1 T cells after local crosslinking via CD3xTRP1, reflected by a significant upregulation of GzmB production, and high expression of activation markers, such as 4-1BB, CD27, NKG2A, PD-1 and Tim3 (Fig. 3B-C). Ki-67 expression in OT-1 T cells was similar in mice receiving OT-1 transfer and OVA vaccination independent of subsequent CD3xTRP1 administration, indicating that OT-1 proliferation was induced by the vaccination and not by the CD3 bsAb at this time point. We observed similar findings in B16F10 tumors (fig. S5A). No significant differences were found in the activation status of OT-1 cells in the spleen, corroborating the notion that the CD3xTRP1-mediated activation burst is localized to the TME (fig. S5B). IHC staining for CD8 confirmed the presence of T cells within the tumor after OT-1 transfer and OVA vaccination (Fig. 3D-E [KPC3-TRP1] and S5C-D [B16F10]). Interestingly, these CD8 T cells were found mainly at the rim of the tumor, whereas administration of CD3xTRP1 facilitated further influx of CD8 T cells into the tumor nests. Transcriptomic analysis revealed a major upregulation of pro-inflammatory genes, such as Cxcl9, Cxcl10, Gzmb, Stat1 and Tap1 after CD3xTRP1 administration compared to untreated tumors (Fig. 3F-G). Inflammatory genes were even further upregulated in tumors treated with the triple combination therapy, accompanied by a greater expression of, for example, Cd3d, Cd3g, Cd69, Cxcr3, Gzmb and Ifng (Fig. 3H-I). In line with the increased presence and activation status of T cells observed by flow cytometry, we observed augmented expression of inflammatory genes in mice that received the triple combination therapy compared to only OT-1 transfer and OVA vaccination (fig. S5E). When using the nSolver analysis software to stratify the differential gene expression data into cell type or pathway representations, we found an increased score for most immune cell types and related pathways when comparing the triple combination therapy to the CD3xTRP1 monotherapy, indicating a stronger and broader immune response (fig. S5F-G).
Together, these results show that naïve OT-1 T cells, when primed by a vaccine, expand and home to all tumor sites. The addition of CD3xTRP1 leads to a strong local burst of tumoricidal effector functions and deeper infiltration selectively within tumors expressing the antigen targeted by the CD3 bsAb, leading to enhanced anti-tumor activity.
Although the binding affinity of the CD3 arm of CD3xTRP1 is lower than the TRP1-binding arm (13, 14), the CD3 bsAb could theoretically piggyback on T cells to reach tumor sites. The combination of T-cell transfer and vaccination with CD3xTRP1 might therefore affect the pharmacokinetics of CD3xTRP1. To evaluate the biodistribution in vivo, CD3xTRP1 was labelled with radioactive indium-111 and imaged using single-photon emission computerized tomography (SPECT) technology (fig. S6A). Analysis of the blood revealed that the majority of CD3xTRP1 was located in the cell-free serum fraction, whereas only a minor fraction was bound to cells (fig. S6B). Importantly, this ratio of distribution was comparable in all treatment groups. For the combinations with OVA vaccination, analysis of the tissues showed a tendency towards increased CD3xTRP1 distribution into CD3-rich organs, such as the spleen and lymph nodes (LNs) (fig. S6C-D). This was accompanied by a slightly increased accumulation of CD3xTRP1 in the KPC3 tumor, muscle and liver, which is most likely attributed to elevated amounts of T cells in these tissues as a result of the vaccination. However, we found no difference in CD3xTRP1 accumulation in the KPC3-TRP1 tumor as a result of vaccine-induced systemic T-cell distribution, suggesting that the minor differences in CD3 bsAb biodistribution in this combination therapy would not affect treatment outcomes.
Influx Of Endogenous CD8 T Cells Is Essential For Durable Treatment Responses
In the experiments described above, we utilized adoptively transferred OT-1 T cells to trace their homing and activation during the combination therapy. At the same time, OT-1 T cells are also effector cells in the treatment response. To remove the need for adoptive cell transfer and to determine the contribution of endogenous CD8 T cells to the anti-tumor response, we studied the influx of endogenous CD8 T cells (CD45.1−) in the triple combination therapy. OT-1 transfer, OVA vaccination and CD3xTRP1 treatment induced a strong increase in infiltration of not only OT-1 T cells, but also endogenous CD8 T cells in KPC-TRP1 tumors (Fig. 4A-B). One week after the last CD3xTRP1 administration in the triple combination therapy, approximately 35% of the total intratumoral CD8 T cells was derived from the endogenous CD8 T-cell pool in the KPC3-TRP1 tumor (Fig. 4B). Tetramer analysis in a separate TME study in B16F10 tumors showed that treatment with OVA vaccination and CD3xTRP1, in the absence of OT-1 transfer, resulted in a large proportion of infiltrating OVA-specific CD8 T cells from the endogenous repertoire, suggesting that the majority of the endogenous T cells for the triple combination are probably also OVA-specific (Fig. 4C). We wondered if infiltration of endogenous CD8 T cells was important for the treatment efficacy of the triple combination. To study this, we utilized B16F10-bearing WT and CXCR3 KO mice receiving the triple combination treatment protocol: adoptively transferred WT OT-1 cells, OVA vaccination and CD3xTRP1 (Fig. 4D). In contrast to endogenous T cells, the tumor infiltration of OT-1 T cells and other immune populations should not be affected in the CXCR3 KO mice. Indeed, similar frequencies of intratumoral OT-1 T cells were observed in the WT and CXCR3 KO mice, but significantly lower frequencies of endogenous CD8 T cells were detected in tumors of CXCR3 KO mice (fig. S7A-B). Importantly, despite normal influx of OT-1 T cells and other immune subsets, combination treatment in CXCR3 KO mice failed to induce durable responses, thereby implying that endogenous CD8 T-cell influx was essential for the anti-tumor response at later time points (Fig. 4E and S7C). Together, these results indicate that the influx of endogenous T cells in the tumor is crucial for durable treatment responses of the triple combination treatment.
Vaccination Just Prior To CD3xTRP1 Orchestrates A Broad Pro-inflammatory Innate And Adaptive Immune Response
Next, we examined the impact of the combination therapy of OVA vaccination and CD3xTRP1, in the absence of OT-1, on the local TME and mapped the immune landscape by flow cytometry, IHC and transcriptomics two days after administration of CD3xTRP1 (Fig. 5A). In contrast to CD3xTRP1 monotherapy, combination treatment led to increased tumor infiltration of CD8 T cells that were GzmB-positive (Fig. 5B and S8A-B). We compared the status of the intratumoral CD8 T cells after CD3xTRP1 monotherapy to those after combination treatment, examining differences between the natural T-cell pool and vaccine-recruited T cells. Unsupervised clustering of intratumoral CD8 T cells based on spectral flow cytometry showed eight clusters, varying from stem-cell like T cells (TCF1-positive in cluster 8) to differentiated effector cells (GzmB-positive in clusters 2 and 4) (Fig. 5C and S8C). CD3xTRP1 alone, but especially the combination therapy, increased the frequencies of differentiated effector cells in clusters 2 and 4 at the expense of TCF1-positive stem cell-like T cells in cluster 8 and the other GzmB-negative clusters 1 and 5. These data were further corroborated by the increased expression of activation markers 4-1BB, NKG2A, PD-1 and Tim-3 in the combination therapy group in two tumor models (fig. S8D for KPC3-TRP1 and S8E for B16F10). Furthermore, transcriptomic analysis revealed an upregulation of T-cell effector genes, such as Gzmk, Gzmb, Gzma, Ifng, Cxcr3 and Prf1 in the combination group compared to CD3xTRP1 monotherapy (Fig. 5D-E). This was accompanied by increased tumoral NK-cell and CD4 T-cell frequencies and elevated levels of GzmB-expressing NK cells (Fig. 5F). Significant alterations in the myeloid cell compartment within the tumor were also observed: pro-inflammatory M1-like macrophages dominated the TME of tumors treated with the combination treatment, whereas immunosuppressive M2-like macrophages were abundant in the TME of untreated and CD3xTRP1 monotherapy treated tumors (Fig. 5G for KPC3-TRP1 and S8F for B16F10). A similar increase in pro-inflammatory N1-like neutrophils at the expense of anti-inflammatory N2-like neutrophils was noticed in the combination treatment. We found these increases only in tumors that received both vaccination and CD3xTRP1 treatment, indicating that TLR stimulation by the adjuvant or local activation of intratumoral T cells by the CD3 bsAb alone, was not sufficient to induce such a broad pro-inflammatory remodeling of the TME (Fig. 5B-G and S8F). Finally, an increase in conventional dendritic cell type 1 (cDC1) was detected after combination treatment, which is associated with increased cross-presentation and a pro-inflammatory TME (15) (fig. S8G). This was supported by increased scores for tumor infiltration by multiple immune cells and increased gene expression of pathways involved in active immune responses (fig. S8H-I).
Altogether, these data demonstrate that combining tumor-nonspecific vaccination with CD3 bsAb induces profound infiltration of highly differentiated effector CD8 T cells and polarization towards a pro-inflammatory TME. This results in an effective coordinated immune response, encompassing both adaptive and innate immune cells, characteristics that are not observed in the CD3 bsAb monotherapy.
A Wide Variety Of Vaccines Can Enhance The Anti-tumor Activity CD3 bsAb Therapy
Finally, we investigated the broad applicability of vaccines with CD3 bsAb administration. Three different antigens were selected as SLP vaccine to test in the B16F10 model: the tumor-nonspecific OVA and the tumor-specific altered-self Gp100 peptide (14), which were both supplemented with imiquimod and IL-2 as adjuvant, and the lymphocytic choriomeningitis virus (LCMV)-derived tumor-nonspecific Gp34 peptide (18), which was supplemented with TLR9 agonist CpG as adjuvant (Fig. 6). Treatment with these three vaccines prior to CD3xTRP1 administration resulted in vaccine-specific CD8 T-cell responses, delayed tumor outgrowth and increased survival proportions, irrespective if these vaccines target a tumor-specific or -nonspecific antigen (Fig. 6A-C and S9A-B for OVA/Gp100, Fig. 6D-E and S10A-B for Gp34). These data suggest that any T-cell priming vaccine could be used to enhance the therapeutic efficacy of CD3 bsAbs.
We therefore hypothesized that not only peptide vaccines, but also viral infections could improve CD3 bsAb therapy, due to the induction of strong T-cell responses to viral antigens. T-cell responses against acutely infecting viruses or attenuated viruses, such as those in vaccines, are generally not affected by functional exhaustion and would thus be ideal candidates for combination with CD3 bsAb treatment (19). To test this idea, we selected the LCMV-Armstrong and HKx31 influenza strains. We provided a low dose LCMV-Armstrong infection, which is known to be swiftly cleared and for its capacity to induce long-lasting effector T cells, 30 days prior to tumor inoculation (Fig. 6D) (19, 20). Indeed, LCMV Gp33-specific CD8 T cells with an effector memory phenotype and expressing CXCR3 were observed up to 63 days after LCMV infection (fig. S10C-D). Importantly, combination of low dose LCMV infection with CD3xTRP1 significantly enhanced long term survival of mice (Fig. 6E), similar to SLP vaccinations. Then, we combined CD3xTRP1 therapy with a HKx31 influenza viral infection (Fig. 6D) (21). Analysis of T cells in the blood revealed an induction of T-cell responses after HKx31 infection, which were boosted after administration of inactivated HKx31 virus, mimicking a vaccination approach (fig. S11A-B). In line with previous results obtained with LCMV, mice that received the HKx31 viral infection and inactivated HKx31 boost in combination with CD3xTRP1 showed increased survival when compared to the other treatment groups (Fig. 6F and S11C).
Finally, we used a COVID-19 mRNA vaccine to augment CD3 bsAb therapy. During the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) pandemic, these vaccines have been FDA-approved and widely administered, making them an ideal candidate to translate this concept to a clinical setting. It was recently reported that, in addition to neutralizing antibodies, COVID-19 mRNA also raises T-cell responses (22). COVID-19 mRNA vaccines were first used in the virus prime-boost scheme of Fig. 6D, but this did not lead to improved survival (fig. S12A-B). Given the design of the vaccine to primarily raise antibody responses and the relatively weak adjuvant effect of these capsulated mRNA (23), we hypothesized that a fast prime-boost schedule, as applied for the peptide vaccines (Fig. 6A), and addition of the TLR9 ligand CpG would result in enhanced treatment responses. Indeed, this set-up induced a potent SARS-CoV-2 specific T-cell response with an effector phenotype, measured five days after the booster vaccination in blood (fig. S12C-D). More importantly, combination of the COVID-19 mRNA vaccine with CD3xTRP1 resulted in enhanced survival, thereby providing proof-of-concept for a clinically relevant treatment combination (Fig. 6G and S12E).
Together, these results show that CD3 bsAb therapy for solid cancers can be transformed into an effective treatment by recruiting any source of activated CD8 T cells via either tumor-(non)specific SLP, COVID-19 mRNA or viral vaccine modalities.