Summary Despite significant interest in therapeutic targeting of splicing, few chemical probes are available for the proteins involved in splicing. Here, we show that elaborated stereoisomeric acrylamide chemical probe EV96 and its analogues lead to a selective T cell state-dependent loss of interleukin 2-inducible T cell kinase (ITK) by targeting one of the core splicing factors SF3B1. Mechanistic investigations suggest that the state-dependency stems from a combination of differential protein turnover rates and availability of functional mRNA pools that can be depleted due to extensive alternative splicing. We further introduce a comprehensive list of proteins involved in splicing and leverage both cysteine- and protein-directed activity-based protein profiling (ABPP) data with electrophilic scout fragments to demonstrate covalent ligandability for many classes of splicing factors and splicing regulators in primary human T cells. Taken together, our findings show how chemical perturbation of splicing can lead to immune state-dependent changes in protein expression and provide evidence for the broad potential to target splicing factors with covalent chemistry.
Leveraging endogenous tumor-resident T-cells for immunotherapy using bispecific antibodies (BsAb) targeting CD20 and CD3 has emerged as a promising therapeutic strategy for patients with B-cell non-Hodgkin lymphomas. However, features associated with treatment response or resistance are unknown. To this end, we analyzed data from patients treated with epcoritamab-containing regimens in the EPCORE NHL-2 trial (NCT04663347). We observed downregulation of CD20 expression on B-cells following treatment initiation both in progressing patients and in patients achieving durable complete responses (CR), suggesting that CD20 downregulation does not universally predict resistance to BsAb-based therapy. Single-cell immune profiling of tumor biopsies obtained following one cycle of therapy revealed substantial clonal expansion of cytotoxic CD4+ and CD8+ T-cells in patients achieving CR, and an expansion of follicular helper and regulatory CD4+ T-cells in patients whose disease progressed. These results identify distinct tumor-resident T-cell profiles associated with response or resistance to BsAb therapy.
Additional file 1: Fig. S1. Analysis pipeline. (a) (yellow and red paths) The metagenomics reads post-quality filtering and removal of human reads were assembled into contig—metagenome assembled genomes (MAGs)—using metaSPAdes. Viral MAGs were identified using CheckV and VirSorter2 and taxonomic assignments were done using vConTACT2. Bacterial MAGs were binned with vamb and taxonomic assignment was done using GTDB-Tk. We then mapped the reads back to the taxonomically assigned viral or bacterial MAGs to generate the bacterial and viral profiles for downstream differential abundance analyses. (b) (brown paths) We identified and extracted the spacers from the metagenomic reads using Crass. Spacers with 90% sequence identity were clustered together. We then used these spacers to identify the shared spacers between individuals and within and across households. Reads with spacers shared between individuals were mapped to the bacterial MAGs that had taxonomic assignments. Bacterial species containing shared spacers were identified as being shared between individuals. Fig. S2. Top 30 most abundant bacterial taxa. Bacterial taxa were ranked by their mean relative abundance across samples. The top 30 bacterial taxa are shown in the boxplot with the x axis representing the relative abundance for each sample. Fig. S3. Viral MAGs identified. (a) Top viral MAG taxa identified as sorted by median relative abundance. The x-axis shows the relative abundance of the viral taxa of all samples while the y-axis indicates the viral taxa, listed as family/genus and phage species included within MAG clusters. (b) Viral MAGs differentially abundant between the high flu infection household vs control or low flu infection household vs control, identified with FDR cut-off as 0.05. The log2 fold changes are shown on the x axis with the blue/turquoise indicating flu infection household groups and gray the control household group. Fig. S4. Shared bacteria between flu infection households. The bar plots show the bacteria shared between individuals and how many pairs of individuals from the high flu infection, low flu infection and no flu infection households shared bacteria. Fig. S5. Mapping of the spacers to the bacteria shared between individuals for each sample. For all panels, the x axis represents the spacers mapped to the specific bacteria as indicated by the plot titles while the y axis represents the subject ID and timepoint of the sample. The colored dots mark households, as per Fig. 3. The shaded boxes indicate which family members had a direct connection based on the sharing of bacteria. Fig. S6. Sharing of Bacteria and flu infection. (a) Density and boxplot plot for percent of spacers shared at the individual level within and between households. The red line on the density plot indicates the cut-off where all the “between household” individual pairs were removed. (b) The connection network was generated based on the percent of shared spacers between individuals for the data above the cut-off in (a). The nodes represent individuals and the edges represent percent of shared spacers. Same color nodes represent individuals from the same household. (c) Dotplot for proportion of individuals in each household that were connected. Number of individuals in (b) in each household were divided by the total number of individuals in the households and compared across flu infection groups. The x axis indicates household code and the panels show the household from high, low, or no flu infection groups. Fig. S7. Proportion of shared spacers between individuals with different flu infection status. The box plot shows the proportion of shared spacers between any two individuals that were: (1) both positive for flu, (2) one positive for flu, one negative for flu or (3) both negative for flu. We compared the proportion of shared spacers between the three groups and Kruskal-Wallis test p values are shown between any two groups. * indicates p values <2.22e-16.
<p>(A) CAR T cells infiltration and expression of LIGHT at the tumor site was quantified by flow cytometric analysis on day 14 post CAR T cell treatment. (B) (C) Quantification of CAR T cells per gram of AsPC1 tumor mass on day 14 post CAR T cell treatment. Plot is representative of 4 to 5 mice per treatment group. Immunohistochemistry staining of mesothelin-negative cancer cell line, Panc1. Negative control. (D) Immunohistochemistry staining of mesothelin-positive cancer cell line, MDA-MB-231. Positive control. (E) Immunohistochemistry staining of various PDX slides samples to validate their mesothelin expression. (F) Flow cytometric analysis of mesothelin expression in PDAC2, one of the PDX selected for in vivo model. (G) <i>In vitro</i> cytotoxicity assay of PDX PDAC2 cocultured with various CAR T cell constructs. Plot represents 2 independent experiments of separate donor T cells.</p>
Antibodies (Abs) agonizing the CD40 immune receptor hold promise for cancer treatment by activating anti-tumor immune responses.1–3 However, clinical implementation has been limited due to minimal on-target activity and significant toxicity.4–7 We previously investigated the bases if these issues and discovered that the interaction between the antibody (Ab) Fc domain and the inhibitory Fc-gamma receptor FcγRIIB is crucial for the in vivo activity of anti-CD40 antibodies, and that previous antibodies were not optimized for this interaction.8–10 To address this, we developed 2141-V11, a human anti-CD40 Ab engineered to enhance FcγRIIB binding. In several tumor models, 2141-V11 enhanced dendritic cell (DC) activation and antigen-specific T cell responses, showing superior antitumor activity compared to other clinical CD40 Abs. Additionally, intratumoral administration of 2141-V11 reduced systemic toxicity associated with CD40 agonism and mediated abscopal responses in non-injected tumors.10–12 Here, we present the results of a first-in-human, phase 1 study,13 investigating the safety and preliminary clinical activity of intratumoral administration of 2141-V11 (NCT04059588), coupled with in vivo mechanistic studies in a humanized mouse model for CD40 and FcγRs (hCD40/hFcγR mice).
Methods
Primary endpoints of the clinical trial included safety and maximum tolerated dose (MTD). Secondary objectives included preliminary clinical activity and correlative studies from biospecimens. Reverse translational studies were also performed in hCD40/hFcgR mice using a breast cancer model.
Results
A total of 12 patients with metastatic solid tumors and identifiable metastatic lesions in the skin, amenable to intratumoral injection, were enrolled in the study. 2141-V11 was well-tolerated without dose-limiting toxicities, all treatment related adverse events were mild (Grade 1–2). In ten evaluable patients with metastatic cancer, the overall response rate was 20%, with complete responses in two patients (melanoma and hormone positive breast carcinoma) and stable disease in six patients (figure 1A). 2141-V11 induced tumor regression in both injected and non-injected lesions (figure 1B-C and figure 2), with increased leukocyte infiltration and tertiary lymphoid structures (TLS) formation in post-treatment biopsies of complete responders (figure 3). In hCD40/hFcγR mice, 2141-V11 induced TLS formation in mice bearing orthotopic breast carcinoma, correlating with local and abscopal antitumor effects, systemic immune activation, and immune memory (figure 4).
Conclusions
Intratumoral administration of 2141-V11 is safe and effective, warranting phase 2 studies that are currently ongoing. Correlative and in vivo studies suggest TLS formation as a unique mechanism of action for this Fc-enhanced immunotherapy.
Acknowledgements
We acknowledge Jim Ackland for his consulting on regulatory affairs, and Dr. Sarah J. Schlesinger and Arlene Hurley for assisting with protocol updates and reports to the IRB. We thank the Pharmacy and Hospital staff, and of course the patients and families who contributed to this trial. We thank Carlo M. Sevilla and Alessandra E. Marino for their excellent technical assistance. We also thank all the members of the J.V.R. Laboratory of Molecular Genetics and Immunology for 804 helpful discussions and sharing experiment materials.
Trial Registration
NCT04059588.
References
Alexandroff AB, Jackson AM, Paterson T, Haley JL, Ross JA, Longo DL, Murphy WJ, James K, Taub DD. Role for CD40-CD40 ligand interactions in the immune response to solid tumours. Mol Immunol 2000;37:515–26. doi: 10.1016/s0161-5890(00)00079-1. Beatty GL, Chiorean EG, Fishman MP, et al. CD40 agonists alter tumor stroma and show efficacy against pancreatic carcinoma in mice and humans. Science 2011;331:1612–6. doi: 10.1126/science.1198443. Beatty GL, Li Y, Long KB. Cancer immunotherapy: activating innate and adaptive immunity through CD40 agonists. Expert Rev Anticancer Ther 2017;17:175–86. doi: 10.1080/14737140.2017.1270208. Chowdhury F, Johnson PW, Glennie MJ, Williams AP. Ex vivo assays of dendritic cell activation and cytokine profiles as predictors of in vivo effects in an anti-human CD40 monoclonal antibody ChiLob 7/4 phase I trial. Cancer Immunol Res 2014;2:229–40. doi: 10.1158/2326-6066.CIR-13-0070. Furman RR, Forero-Torres A, Shustov A, Drachman JG. A phase I study of dacetuzumab (SGN-40, a humanized anti-CD40 monoclonal antibody) in patients with chronic lymphocytic leukemia. Leuk Lymphoma 2010;51:228–35. doi: 10.3109/10428190903440946. Irenaeus SMM, Nielsen D, Ellmark P, et al. First-in-human study with intratumoral administration of a CD40 agonistic antibody, ADC-1013, in advanced solid malignancies. Int J Cancer 2019;145:1189–99. doi: 10.1002/ijc.32141. Bajor DL, Mick R, Riese MJ, et al. Long-term outcomes of a phase I study of agonist CD40 antibody and CTLA-4 blockade in patients with metastatic melanoma. Oncoimmunology 2018;7:e1468956. doi: 10.1080/2162402X.2018.1468956. Li F, Ravetch JV. Inhibitory Fcgamma receptor engagement drives adjuvant and anti-tumor activities of agonistic CD40 antibodies. Science 2011;333:1030–4. doi: 10.1126/science.1206954. Li F, Ravetch JV. Antitumor activities of agonistic anti-TNFR antibodies require differential FcgammaRIIB coengagement in vivo. Proc Natl Acad Sci U S A 2013;110:19501–6. doi: 10.1073/pnas.1319502110. Knorr DA, Dahan R, Ravetch JV. Toxicity of an Fc-engineered anti-CD40 antibody is abrogated by intratumoral injection and results in durable antitumor immunity. Proc Natl Acad Sci U S A 2018;115:11048–53. doi: 10.1073/pnas.1810566115. Dahan R, Barnhart BC, Li F, Yamniuk AP, Korman AJ, Ravetch JV. Therapeutic activity of agonistic, human anti-CD40 monoclonal antibodies requires selective FcgammaR engagement. Cancer Cell 2016;29:820–31. doi: 10.1016/j.ccell.2016.05.001. Garris CS, Wong JL, Ravetch JV, Knorr DA. Dendritic cell targeting with Fc-enhanced CD40 antibody agonists induces durable antitumor immunity in humanized mouse models of bladder cancer. Sci Transl Med 2021;13. doi: 10.1126/scitranslmed.abd1346. Osorio J, Knorr D, Weitzenfeld P, et al. Intratumoral Fc-optimized agonistic CD40 antibody induces tumor rejection and systemic antitumor immunity in patients with metastatic cancer. Res Sq 2024. doi: 10.21203/rs.3.rs-4244833/v1.
Ethics Approval
Studies were approved by the institutional review board (IRB) of the Rockefeller University (DKN-0993).
<p>(A) Violin plots showing quality metrics of single cells included in downstream analysis. nCount RNA: number of RNA unique molecular identifiers (UMI); nFeature RNA: number of detected genes; nCount_ADT: number of antibody UMI; nFeature_ADT: number of detected antibodies; percent.mt: percentage of mitochondrial gene expression; HTO_margin: difference between signals for the hashtag with the highest signal and the hashtag with the second highest signal. (B) (C) Weighted-nearest neighbor (WNN) UMAP of single cells colored by condition (CAR T cell construct and timepoint) (top) and CD4+ vs CD8+ cell type (bottom). Expression of T cell markers (CD3, CD4, and CD8) at the RNA (top) and antibody-derived tag (ADT) level projected on WNN UMAPs. (D) Expression of the RNA (top) and ADT (bottom) markers of T cell type, proliferation, activation, cytotoxicity, and cytokines support CITE-seq cluster annotation. Blue is control CAR T cell and Red is LIGHT-CAR T cell.</p>
<p>(A) Transgene expression of the CAR constructs after retroviral transduction of human T cells by flow cytometry. CD19 CAR T cells were used as irrelevant CAR T cell control. (B-D) Healthy human donor-derived mesothelin-targeted CAR T cells were cocultured with tumor cells expressing GFP and firefly luciferase at different effector to tumor ratios. Bioluminescence was measured 72 hours later and plotted as a percentage of the signal detected with tumor cells alone (max bioluminescence signal). Associated expression of mesothelin with the corresponding mesothelin-positive cell lines are shown. JMN (left), SW620 (middle), and MDA-MB-231 (right). Plots represent 3 independent experiments. Data errors were analyzed with mean ± SEM. (E) HVEM expression in PDAC cell lines (CAPAN2, AsPC1, BxPC3, Panc1) (F) Flow cytometric analysis of overexpression constructs derived from Panc1 wildtype. Panc1 Mesothelin overexpression, Panc1 LTβR overexpression, and Panc1 Mesothelin and LTβR Overexpression</p>
