Tumors foster an immunosuppressive microenvironment to evade the antitumor immune response. However, the influence of intratumoral immunosuppressive steroids on tumor-infiltrating natural killer (NK) cells and their implications for effective immunotherapy has remained largely unexplored. Here, we report that the functional enrichment of glucocorticoid cortisol signaling in the lung tumor microenvironment (TME) impairs NK cell anti-tumor cytotoxicity and exacerbates hypoxic stress. Cancer-associated fibroblasts (CAFs) and macrophages convert inactive cortisone to active cortisol, while T cells, fibroblasts, myeloid cells, macrophages, and cancer cells contribute to de novo steroid biosynthesis, collectively establishing a steroid-rich niche. Pharmacological inhibition of the glucocorticoid receptor (GR) in vivo alleviates cortisol-mediated immune suppression, resulting in reduced tumor growth and enhanced cytotoxicity of tumor-infiltrating NK cells. To overcome the cortisol-induced dysfunction of solid tumor targeting immunotherapy, we engineered chimeric antigen receptor (CAR) -NK cells specific to the Carcinoembryonic antigen-related cell adhesion molecule 5 (CEACAM5) (highly expressed in lung tumors) and rendered them cortisol-resistant by genetic deletion of the cortisol receptor gene NR3C1. In cortisol-rich niches, cortisol-resistant CAR-NK cells sustained antitumor cytotoxicity. Mechanistically, NR3C1 deletion relieved cortisol-mediated suppression of PI3K-AKT-NF-κB signaling, restored anti-tumor activity, and markedly reduced hypoxic stress. In lung metastasis models, cortisol-resistant CAR-NK cells achieved superior tumor control and significantly reduced tumor burden compared with conventional CAR-NK cells. Together, these findings identify local cortisol signaling as a critical barrier to solid tumor immunotherapy and establish cortisol-resistant CAR-NK cells as a promising strategy for targeting steroidogenic solid tumors, which can be combined with therapeutic glucocorticoids.

Chimeric antigen receptor (CAR) T-cell therapy has shown unprecedented success in haematological cancers but faces challenges in solid tumours. Although carcinoembryonic antigen-related cell adhesion molecule 5 (CEACAM5) is differentially expressed in many solid tumours, CEACAM5 CAR T-cells are ineffective. Here, we have studied the interaction of CEACAM5 targeting CAR primary T-cells with colorectal cancer (CRC) cells using fluorescence microscopy. We find that CRC cells glycocalyx is much thicker than the CAR T-cell and likely contributes to immune-escape. Oscillating calcium flux, a signature of non-sustained triggering and decreased killing, was observed when CAR T-cells interacted with CRC, which increased with increasing cell-seeding time. This was because CEACAM5 became increasingly unavailable on the CRC cell monolayer, as revealed by fluorescence imaging. Local proteolytic treatment with trypsin to disrupt the CRC cell monolayer, using a micropipette, increased CEACAM5 availability, decreased glycocalyx thickness, and restored sustained CAR T-cell calcium fluxes, increasing the killing of CRC cells. Our results reveal why CAR T-cells targeting CEACAM5 are ineffective and suggest possible routes for improved therapy. ### Competing Interest Statement The authors have declared no competing interest.

Sensitive cell surface proteomics studies have shown that the number of completely tumour-specific targets for adoptive cellular immunotherapy is extremely low. Even approved CAR T-cell targets appear to have expression in the central nervous system, leading to long-term neurological complications. We propose that this toxicity could be significantly improved by adoption of NOT-gates, which have been shown to limit CAR T-cell activity against healthy tissue expressing a second target that is absent on the tumour. Furthermore, the approach could also target essential, but non-specific proteins on tumour cells. The use of a NOT gate confers the specificity, whilst targeting the essential protein limits antigen escape. Here we explore the feasibility of such an approach for CAR T-cell targeting of primary myeloma. We show that none of the 45 most essential proteins are unique to the myeloma cell. However, whilst widely expressed, one of the most important proteins for myeloma cell survival, the transferrin receptor, could safely be targeted by a NOT-gate approach. Exploring co-expression patterns demonstrate 26 proteins that are not expressed on myeloma cells, but which are coexpressed with the transferrin receptor in all healthy tissues. We also describe a web app, NOTATER, which can be used by scientists with no bioinformatic capabilities to explore potential NOT-gate combinations in myeloma.

Immunotherapeutics have revolutionised the treatment of multiple myeloma (MM), but BCMA remains the only approved CAR T-cell target. Furthermore, despite impressive responses to these agents, relapses are still inevitable. There is thus an unmet need to expand our repertoire of targets. We have previously used proteomic approaches to demonstrate that classical CAR T-cell targets are inevitably expressed on healthy tissue, resulting in on-target/off-tumour toxicity. Further engineering is therefore required to enable better discrimination between healthy and malignant cells. We have previously identified SEMA4A as an attractive MM immunotherapeutic target, owing to its ubiquitous, obligate expression in primary MM. We also demonstrated that a SEMA4A-targeting CAR T-cell was highly effective in eliminating myeloma cells. However, SEMA4A has modest expression on monocytes, granulocytes, and CD34+ hematopoietic stem and progenitor cells (HSPCs), so on-target/off-tumour pancytopenia was a major risk of our original CAR T-cell. We therefore describe here our use of NOT-gating to enable much safer CAR T-cell targeting of SEMA4A in MM. To identify a suitable inhibitory CAR (iCAR) protein to pair with our primary SEMA4A CAR, we used a high-resolution mass-spectrometry-based proteomics dataset of primary human immune cells to identify cell-surface proteins that were absent on myeloma cells, but which were expressed at levels three-fold greater than SEMA4A, or more, on monocytes, granulocytes, and CD34+ cells. To prevent constitutive cis-inhibition of NOT-gated T-cells, we excluded all potential iCAR proteins also expressed on T-cells. We identified multiple potential iCAR partners, the most promising of which was the protein CLEC12A, based on its expression profile across the different lineages. We profiled multiple healthy and MM-associated bone marrow samples using flow cytometry to confirm the expression profile of this protein. As expected, we saw ubiquitous expression of CLEC12A on SEMA4A-positive monocytes, granulocytes and HSPCs, but a complete absence of expression on T-cells and on MM-associated plasma cells. Having identified CLEC12A as a high-quality iCAR protein partner, we proceeded to engineer the inhibitory machinery. We identified two single domain VHHs against CLEC12A. To confirm that they could bind the extracellular protein in its natural conformation and trigger downstream T-cell signalling, we overexpressed CLEC12A in K562 cells and co-cultured these with a second-generation CAR T-cell engineered with the VHHs to target CLEC12A. This led to robust T-cell activation, confirming the utility of the VHHs. We then cloned the VHHs upstream of the dual intracellular inhibitory regions of LIR-1 and PD-1 and co-expressed these iCARs in both Jurkat reporter cells and primary human T-cells expressing the SEMA4A targeting CAR. Parallel cell-surface expression of both constructs was validated through flow cytometry. Importantly, when co-cultured with CLEC12A-/SEMA4A+ target cells, there was robust T-cell activation and cell killing by the SEM4A-CAR/CLEC12A-iCAR demonstrating that the CLEC12A iCAR does not prevent cytotoxicity in the absence of its target. When co-cultured with CLEC12A+/SEMA4A+ target cells, T-cell activation was reduced, though not completely, indicating partial function of the CLEC12A iCAR. We reasoned that our failure to achieve complete inhibitory function was due to the relatively smaller size of the extracellular domain of CLEC12A relative to that of SEMA4A, leading to steric hindrance. We therefore experimented with different hinge lengths on the iCAR and targeting CAR to enhance performance. By increasing the iCAR/CAR hinge ratio, we were able to demonstrate significantly more inhibition of T-cell activation in the presence of CLEC12A on the target cell, whilst maintaining robust killing when CLEC12A was absent. Ongoing work is seeking to test our SEMA4A-CAR/CLEC12A-iCAR in vivo. In conclusion, we confirm that logic-gating can increase the targeting repertoire of CAR T-cells, and we demonstrate a NOT-gated CAR T-cell that can safely kill myeloma cells via their SEMA4A expression, while limiting on-target off-tumour toxicity of other immune cells. However, this study also illustrates that the increased targeting repertoire comes at the cost of increased engineering complexity.

Chimeric antigen receptor (CAR) T-cell therapy is a highly effective novel treatment in haematological malignancies that has shown promise as a therapeutic option in multiple myeloma. However, widespread adoption of CAR T-cell therapy in myeloma has been hindered by the challenge of unbiased target antigen identification and selection. As activation of CAR T-cells requires minimal antigen on the cell surface, a major risk of toxicity is destruction of healthy tissue expressing the target protein, i.e. on-target, off-tumour toxicity. Indeed, examination of the myeloma surface proteome demonstrated that there was no single target that was completely unique to myeloma cells. One approach to achieve target specificity is to require simultaneous expression of two proteins on the target cells, so-called AND-gate targeting. To identify potential AND-gate combinations for myeloma, we devised an algorithm to prioritise pairings that exhibited pan-myeloma expression and no overlapping expression in vital healthy tissue, as predicted by proteomics. Through this approach, we identified over 600 combinations. To minimise the risk of exhaustion or priming by CAR T-cells, any combination whereby one of the two antigens was expressed in T-cells was also excluded, leading to the prioritisation of 144 candidate pairings. This demonstrates the potential for AND-gating to expand the repertoire of CAR T-cell targets for myeloma. We evaluated one of these candidate pairings, TMPRSS11E and TNFRSF8, in vitro . Activation in the Jurkat cell line co-expressing a suboptimal CAR against TNFRSF8 and a chimeric costimulatory receptor (CCR) against TMPRSS11E was markedly enhanced following co-culture with a dual-target positive myeloma cell line compared with single-target positive K562, demonstrating improved discrimination between tumour and non-tumour cells.

Whilst chimeric antigen receptor (CAR) T cells have shown remarkable success, further novel targets are required. However, our previous cell surface proteomics work in multiple myeloma (MM) showed that target choice is highly limited by an absence of truly tumour-specific antigens, and often very low on-tumour expression of target proteins in some primary samples. One approach to improve tumour specificity is to use combinatorial antigen recognition, or “AND-gates”, whereby a CAR T cell must recognise two different antigens for full activation and target cell elimination. It has also been reported that this approach might overcome low target expression. We therefore decided to explore the potential for AND-gate targeting in MM and identify a suitable target pairing. To determine a comprehensive list of viable AND-gate combinations for MM, we first integrated our MM proteomics data with a normal tissue proteomics dataset. From 777 extracellular proteins expressed in primary MM, we identified a total of 287,890 possible pairwise combinations. We next excluded any pairs with overlapping expression in vital, healthy tissues, reducing potential combinations to 664. Additional filtering steps to remove difficult-to-target proteins and combinations where either target was expressed on T cells - thus avoiding T cell fratricide - resulted in a final list of 92 AND-gate targets. From our prioritised candidates, we selected TNFRSF8 and TMPRSS11E as an exemplar pairing. By mass spectrometry, both TNFRSF8 and TMPRSS11E were expressed across all 8 profiled primary MM samples at comparable levels to existing CAR T cell targets, such as GPRC5D, and had no predicted overlapping expression in normal tissue. To test our AND-gate combination, we designed a ‘split-CAR’ system in which TNFRSF8 was targeted by a 1 st generation CAR with an intracellular signalling domain, but without a co-stimulatory receptor (CAR-only construct), and TMPRSS11E was targeted by a construct with a chimeric co-stimulatory receptor (CCR), but lacking an intracellular signalling domain (CCR-only construct). Constructs were initially validated in a Jurkat cell line model of T cell activation. Jurkats were transformed with CCR-only, CAR-only, or both constructs, as well as with full 2 nd generation CARs targeting TNFRSF8 or TMPRSS11E. These cells were co-cultured with the TNFRSF8+TMPRSS11E+ human MM cell line XG1 or the TNFRSF8+TMPRSS11E- erythroleukemic cell line K562. As anticipated, neither the CCR-only nor the CAR-only Jurkats were activated, with no production of IFNγ or IL2 ( Fig. 1A). Whilst Jurkats expressing the full 2 nd generation TMPRSS11E CAR were activated in co-culture with XG1, Jurkats expressing the TNFRSF8 2 nd generation CAR were not, despite expression of TNFRSF8 on XG1. Jurkats co-expressing both components of the split-CAR were markedly activated when co-cultured with TNFRSF8+TMPRSS11E+ XG1 cells (but not with TNFRSF8+TMPRSS11E- K562 cells), as demonstrated by a large increase in expression of CD69, IFNγ, and IL2 ( Fig. 1A). Thus, the split-CAR system showed a very high degree of specificity, requiring both antigens to be co-expressed on the target cell. Moreover, it was activated even when full 2 nd generation CARs against TNFRSF8 were not. Next, we sought to investigate the cytolytic activity of our split-CAR constructs in donor T cells. In a cytotoxicity assay, T cells expressing the CAR-only or CCR-only construct did not exhibit significant killing of XG1 cells ( Fig. 1B). In parallel with the Jurkat reporter assays, only 2 nd generation TMPRSS11E CAR T cells, and not 2 nd generation TNFRSF8 CAR T cells, were cytotoxic to XG1. Finally, donor T cells expressing both components of the split-CAR demonstrated potent killing of dual-positive XG1 cells ( Fig. 1B), but not single-positive K562 cells. We have thus developed an analytical pipeline for the discovery of AND-gate CAR T cell targets for MM, also applicable to other tumours. This approach led to the development of an extremely potent, but highly selective, split-CAR targeting TNFRSF8 and TMPRSS11E. Importantly, our split-CAR targeted very low levels of TNFRSF8 antigen that were insufficient to trigger activation of a 2 nd generation TNFRSF8 CAR-T. In summary, our work suggests that AND-gate targeting can increase not just specificity, but potency, even against very low expression tumour targets in MM. Link to preprint https://doi.org/10.1101/2023.04.04.535580

The accessibility of cell surface proteins makes them tractable for targeting by cancer immunotherapy, but identifying suitable targets remains challenging. Here we describe plasma membrane profiling of primary human myeloma cells to identify an unprecedented number of cell surface proteins of a primary cancer. We used a novel approach to prioritize immunotherapy targets and identified a cell surface protein not previously implicated in myeloma, semaphorin-4A (SEMA4A). Using knock-down by short-hairpin RNA and CRISPR/nuclease-dead Cas9 (dCas9), we show that expression of SEMA4A is essential for normal myeloma cell growth in vitro, indicating that myeloma cells cannot downregulate the protein to avoid detection. We further show that SEMA4A would not be identified as a myeloma therapeutic target by standard CRISPR/Cas9 knockout screens because of exon skipping. Finally, we potently and selectively targeted SEMA4A with a novel antibody-drug conjugate in vitro and in vivo.

Introduction: Multiple myeloma remains an incurable disease with a significant variation in therapeutic response. Choice of treatment is largely determined by prior lines of therapy and the patient's fitness. However, myeloma is a highly heterogeneous disease and this approach risks failing to deliver the right drug at the right time to an individual patient. To begin to address this, we have adopted machine learning approaches to predict individual responses to specific anti-myeloma therapies. We have previously shown that it is possible to select rationally between bortezomib- and lenalidomide-based therapy using a gene expression signature. Gene expression models are difficult to implement clinically, so more recently we presented a five gene mutational signature to identify carfilzomib-specific responses. Although easier to translate, such molecular signatures can still be hindered clinically by slow processing and technical failures. Because flow cytometry is well established in the diagnostic laboratory and is sensitive enough to detect minimal residual disease, we set out to develop an immunophenotypic signature of carfilzomib-specific responsiveness in primary myeloma. Methods We originally developed our mutational signature on whole exome sequencing of diagnostic samples from the CARDAMON clinical trial. For the current study, we performed RNA sequencing on 80 of these. Patients were divided into carfilzomib-responsive and non-responsive groups according to their mutational signature. Genes were filtered for those encoding cell surface proteins. A second round of filtering selected only those genes whose RNA expression correlated well with protein expression as determined by plasma membrane fractionation and mass spectrometry. Genes differentially expressed between the two groups were then detected using DeSEQ2. Logistic regression on these candidate genes was performed to assess whether they could be used to correctly classify patients for carfilzomib responsiveness and hence form a predictive flow cytometric panel. Results 76 patients from the CARDAMON trial had RNAseq that passed quality control. 128 genes encoded cell surface proteins, were reliable predictors of cell surface protein expression (R2 > 0.8), and were differentially expressed between carfilzomib-responsive and non-responsive patients. No single cell surface protein alone could correctly classify patients. However, using multiple logistic regression, six genes whose expression could be detected by flow cytometry (CLEC7A, STRA6, TMPRSS11E, CD27, GPR176, TK1, LPXN) together predicted carfilzomib responsiveness with an AUC of 0.77. Independent validation of these genes by RNA-Seq in the CoMMpass dataset showed that they could predict carfilzomib responsiveness with an AUC of 0.803 (Figure 1A). Indeed, survival for those patients who were predicted to be carfilzomib-responsive by our potential flow panel was significantly higher than those predicted to be resistant (median 34 months vs 15months ; p= 0.0032), figure 1B. Conclusions By combining transcriptomic and proteomic approaches, we have generated a panel of six cell surface proteins that are amenable to flow cytometry and can predict carfilzomib-responsiveness. Prospective validation of these markers is underway. Figure 1View largeDownload PPTFigure 1View largeDownload PPT Close modal