Detection, prevention, and reversal of acquired resistance to immune checkpoint blockade therapy

Abstract

Strategies for blocking anti-apoptotic proteins or activating proapoptotic proteins in combination with immune checkpoint blockade therapy to prevent or reverse acquired resistance are described, providing methods to detect, prevent, and / or reverse acquired resistance to ICB therapy. Anti-melanoma therapy can be enhanced by administering an effective amount of an activator of a pro-apoptotic protein, such as a BAX activator, and / or a downregulator of an anti-apoptotic protein, such as a BH3 mimetic. Also described is a method of detecting acquired resistance to anti-melanoma therapy by assaying a sample of tumor DNA for deletion and / or amplification of genes indicative of acquired resistance.

Description

DETECTION, PREVENTION, AND REVERSAL OF ACQUIRED RESISTANCE TO IMMUNE CHECKPOINT BLOCKADE THERAPY

[0001] This application claims benefit of United States provisional patent application number 63 / 610,823, filed December 15, 2023, the entire contents of which are incorporated by reference into this application.REFERENCE TO A SEQUENCE LISTING

[0002] The content of the XML file of the sequence listing named “UCLA294_Seq”, which is 5 kb in size, created on December 12, 2024, and electronically submitted herewith the application, is incorporated herein by reference in its entirety.BACKGROUND

[0003] Immune checkpoint blockade (ICB) provides the foundation for curative therapy of a large number of malignancies (1-5), but resistance poses a major challenge to eliciting further survival benefits. In addition to primary or innate resistance, initial ICB responses, which can last for years, may ultimately result in disease progression. Although melanoma displays an initial response rate of -40-60%, the rates of acquired resistance have been estimated at -20-60%. To date, the number of clinical studies addressing the mechanisms of acquired resistance has not kept pace with the growing number of patients treated with ICB.

[0004] There remains a need for improved methods of detecting, preventing, and reversing acquired resistance to ICB therapy.SUMMARY

[0005] The methods described herein provide strategies for blocking anti-apoptotic proteins or activating proapoptotic proteins in combination with ICB therapy to prevent or reverse acquired resistance. Also described are methods to detect, prevent, and / or reverse acquired resistance to ICB therapy by blocking anti-apoptotic proteins or activating proapoptotic proteins.

[0006] In some embodiments, described herein is a method of enhancing anti-melanoma therapy in a subject in need thereof, the method comprising administering to the subject an effective amount of an activator of a pro-apoptotic protein and / or a downregulator of an anti- apoptotic protein. In some embodiments, the activator of a pro-apoptotic protein is a BAX activator. In some embodiments, the BAX activator is BTSA1.2. In some embodiments, the downregulator of an anti-apoptotic protein is a BH3 mimetic. BH3 mimetics are smallmolecule antagonists of the anti-apoptotic BCL-2 members that function as competitive inhibitors by binding to the hydrophobic cleft. In some embodiments, the BH3 mimetic is Venetoclax, Navitoclax, S63845, and / or AMG176. Myeloid leukemia 1 (MCL-1) is an antiapoptotic protein of the BCL-2 family that prevents apoptosis by binding to the pro- apoptotic BCL-2 proteins such as BAK and NOXA. Inhibitors of MCL-1 include, for example, ANG176, AMG397, ABBV-467, and PRT1419. In some embodiments, the administering comprises administering both a BAX activator and a BH3 mimetic.

[0007] In some embodiments, the subject is treated with immune checkpoint blockade (ICB) or immune checkpoint therapy (ICT) as anti-melanoma therapy. Immune checkpoint blockade (ICB) for the treatment of cancer typically comprises an anti-PD-1 or anti-PD-L1 agent. Optionally, the ICB treatment further comprises additional agents, such as anti- CTLA4 and / or anti-LAG3. Representative ICT agents include, but are not limited to, pembrolizumab, nivolumab, and ipilimumab. In some embodiments, the activator of a pro- apoptotic protein and / or a downregulator of an anti-apoptotic protein is administered concomitantly with, prior to, and / or subsequent to the administering of the one or more anti- PD-1 / L1 antibodies, and / or an anti-CTLA-4 agent.

[0008] Also described is a method of preventing or inhibiting acquired resistance to antimelanoma therapy in a subject in need thereof. In some embodiments, described herein is a method of ameliorating acquired resistance to immune checkpoint blockade therapy in a subject in need thereof, the method comprising administering to the subject an effective amount of an activator of a pro-apoptotic protein and / or a downregulator of an anti-apoptotic protein. In some embodiments, the method comprises administering to the subject an effective amount of an activator of a pro-apoptotic protein and / or a downregulator of an anti- apoptotic protein.

[0009] Additionally described is a method of detecting acquired resistance to anti-melanoma therapy in a subject. In some embodiments, the method comprises: (a) obtaining a biological sample from the subject, wherein the sample comprises tumoral DNA; (b) assaying the sample for deletion or amplification of genes listed in Table S4 and / or Table S5 of U.S. patent application 63 / 610,823, filed December 15, 2023; and (c) detecting acquired resistance. The assaying of step (b) indicates deletion or amplification of one or more genes listed in Table S4 and or Table S5 of U.S. patent application 63 / 610,823, filed December 15, 2023. In some embodiments, the deletion and / or amplification is measured relative to a sample obtained from the subject prior to, or at an earlier time point in, ICB therapy. In some embodiments, the sample comprises a tumor biopsy or liquid biopsy. A liquid biopsy comprises, for example, cell free tumor DNA or circulating tumor cells.BRIEF DESCRIPTION OF THE DRAWINGS

[0010] FIGS. 1A-1M show integrative genomic analysis to functionalize relapse-specific alterations identifies CNVs of apoptosis genes. (1A) Attribution frequencies of ON signatures in patient-matched baseline and DP CNVs. (1 B and 1C) Significantly enriched Reactome gene signatures of recurrently (> 4 of 17 patients) deleted (1 B) or amplified (1C) genes specifically detected in DP but not patient-matched baseline melanomas. (1 D) Venn diagram showing the numbers (overlapping and non-overlapping) of genes recurrently (> 3 of 17 patients) and specifically deleted in DP melanomas but not in patient-matched baseline melanomas (left circle) and of resister genes from CRISPR-Cas9 functional screens (right circle). Selected recurrent genes are shown with the number of patients affected. (1E) As in Figure 1 D, except for DP-specific amplified genes and sensitizer genes. (1 F) Venn diagram showing the numbers (overlapping and non-overlapping) baseline- (left circle) and DP- related (right circle) provisional SMGs. Listed are provisional SMGs also identified as sensitizer genes among baseline-specific SMGs and as resister genes among DP-specific provisional SMGs. (1G) GN and non-synonymous mutational status of the DP-specific SMGs B2M, JAK2 and TCF23. (1H) STRING protein-protein interaction network for recurrent DP- specific deleted or amplified cell-death genes. Widths and shades of the edges are proportional to the confidence scores of the interactions. Solid lines, interactions within cluster; dashed lines, across clusters. (11 and 1J) (Top) Cocultures showing mCherry+human melanoma cells (A375 in I; M407 in J) ± BAX knockdown as shown by Western blots (TUBULIN, loading control) and their killing by NY-ESO-1-TCR-transduced primary T cells (E:T, 1:3 in K and 1:5 in L). (Bottom) Cultures of above melanoma cells without T cells (T- cell media alone) or cocultures with non-transduced T cells. (1K) As in Figure 11, except including mCherry+, shBAX-1 A375 cells overexpressing a shBAX-1 -resistant mutant BAX (BAX Mut). 1(L) As in Figure 11, except including mCherry+M257 cells ± FLAG-DFFA CR (dominant-negative) overexpression and 1:100 E:T. (1M) As in Figure 1 J, except ± FAS knockdown and 1 :5 E:T. Data in Figure 1I-1M are representative of three replicates with T cells from the same donor. All data are represented as mean ± SEM and analyzed by the two-way A NOVA test.

[0011] FIGS. 2A-2G show that concurrence of functional relapse-specific alterations defines heterogeneity of AR mechanisms with convergent apoptotic de-sensitization. (2A) Correlation analysis of concurrent CNV-affected and non-synonymously mutated genes with odds ratios and p values by pair-wise Fisher’s exact test. (2B to 2D) Evolutionary trajectories of BRAF (2B), NRAS (2C), or NF1 (2D) mutated melanomas of indicated patients. Maximally parsimonious phylogeny based on somatic single-nucleotide variants (SNVs) and insertiondeletion (INDEL) mutations in patient-matched normal (NRM) plus baseline and DP tumortissues. Annotations of truncal driver genes and mutations as well as a DP-specific mutation in the driver gene, JAK1, are shown in text. DP tumors are annotated with resistance-driver genes and their CNV and somatic SNVs are listed. AMP, amplification; DEL, deletion. (2E to 2G) Co-immunofluorescence showing indicated resistance-driver protein levels within patient-matched baseline and DP tumors from two patients. Left, representative images; right, quantifications of the mean fluorescent intensities from eight fields. Ruler, 50 pm. Data are represented as mean ± SEM and analyzed by unpaired two-tailed Student’s t-test.

[0012] FIGS. 3A-3L show human and murine AR melanoma models recapitulate clinical relapse-specific CNVs that lower the apoptotic threshold. (3A) Schematic showing the timeline of deriving isogenic AR ICI sublines from the M486 P cell line via repeated selection by HLA- / antigen-specific T cells. (3B) Growth curves of individual YER P tumors treated with vehicle (Veh) or ICI (anti-PD-1 + anti-CTLA-4). Three regressing tumors were selected for analysis of acquired resistance (AR1-3) and development of an AR model (AR3). (3C to 3G) Cocultures showing the relative confluence of the M486 P cell line or indicated AR sublines exposed to HLA-A2.1- / NY-ESO-specific or control (non-transduced) T cells (E:T, 1:1). Data representative of three replicates with T cells from the same donor. (3H) Growth curves of YER P- or AR3 subline-derived tumors treated with Veh or anti-PD-1 + anti-CTLA-4 (n = 10 / group). Data representative of two replicates. (3I) Evolutionary trajectories of the M486 P cell line and its isogenic sublines, AR1-5. Maximally parsimonious phylogeny based on somatic SNVs and INDELs in patient- matched NRM PBMC, P and AR melanoma cells. Truncal driver genes, their CNVs and non-synonymous mutations are shown in black text. AR sublines are annotated with resistance-driver genes and their CNVs and / or non- synonymous mutations. AMP, amplification; DEL, deletion. (3J) Western blots showing the indicated proteins in the M486 P cell line and its isogenic AR sublines. TUBULIN, loading control. (3K) As in 3I, except for YER P and AR tumors. (3L) Western blots showing the indicated proteins in the YER P cell line and the isogenic AR3 tumor-derived cell line. Tubulin, loading control. Data in Figure 3C-3H are represented as mean ± SEM and analyzed by the two-way ANOVA test.

[0013] FIGS. 4A-4H show single-cell WGS uncovers preexistence and de novo subclonal resistance evolution. (4A and 4B) UMAP of single-cell CNV profiles, color-coded by subclones (4A) and by samples (4B) and derived from scWGS of the M486 P and isogenic sublines, AR3 and AR5. (4C) Evolutionary trajectory of subclonal consensus integer CNV profiles, showing root based on a diploid profile and relative branch lengths based on CNV events. Pie charts for each genetic subclone showing the proportions of M486 P, AR3, and AR5 samples / cells composing the indicated subclone. (4D) Heatmaps based on single-cell CNV profiles showing genetic subclones within M486 P cell line as well as AR3 and AR5sublines. Genes affected by clonal and subclonal CNVs, by subclonal CNVs and enriched in P, and by subclonal CNVs and enriched in AR, are all indicated with different shading intensities. (4E and 4F) As in Figure 4A and 4B, except derived from scWGS of YER vehicle- treated and ICI-treated AR3 tumors. (4G) As in Figure 4C, except for YER vehicle-treated and ICI-treated AR3 tumors. (4H) As in Figure 4D, except for YER vehicle-treated and ICI- treated AR3 tumors.

[0014] FIGS. 5A-5G show that restoring apoptotic sensitivity re-sensitizes AR melanoma to cytotoxic T cells and ICI therapy. (5A to 5C) Western blots of M486 P and AR sublines with single (A), double (B) or triple overexpression or OE (C) of P53, BAD and TNFR1. (5D) Cocultures of M486 P and AR sublines with indicated gene OE with HLA- / antigen-specific (top) and control (bottom) T cells at 2:1 E:T. Data representative of three replicates with T cells from the same healthy donor. (5E and 5F) Western blots (5E) and growth curves (5F) of the YER P cell line or AR3cl and tumors, respectively, ± Bad and Fas OE (vs. empty vector). Tumor-bearing mice were treated with vehicle (Veh) or ICI (anti-PD-1 + anti-CTLA- 4). (5G) Growth curves of the YER P or AR3cl tumors treated with Veh, ICI (anti-PD-1 + anti- CTLA-4), venetoclax (Vene), or Vene + ICI. All data are represented as mean ± SEM and analyzed by the two-way ANOVA test.

[0015] FIG. 6 presents a set of radiographic images showing in situ and de novo acquired resistance to combination ICI. (Top) PET-CT images at the indicated timepoints for patient #11 (50-year-old female) who presented with cutaneous melanoma metastatic to the right mediastinum / lung and liver and underwent therapy with ipilimumab plus nivolumab. Treatment resulted in a complete metabolic response (center PET-CT at ~9 months into treatment). At ~2.5 years after treatment, a mediastinal mass emerged during active treatment. (Middle) CT images at the indicated timepoints for patient #15 (73-year-old female ) who presented with cutaneous melanoma metastatic to the splenic soft tissue bed. She started pembrolizumab treatment with initial clinical response (center CT at 6 months into treatment). At ~2 years into treatment, a mass at the splenic soft tissue bed emerged during active treatment. (Bottom) PET-CT images for patient #8 (60-year-old male) who presented with large acral primary melanoma and popliteal lymph node metastasis and underwent therapy with ipilimumab plus nivolumab, which resulted in a near complete response. At ~11 months into treatment, he developed a metastatic acral melanoma to the subcutaneous skin.

[0016] FIGS. 7A-7E illustrate CRISPR-Cas9 screen-derived cancer genes that resist and sensitize to T-cell cytotoxicity and clinical melanoma-derived DP-specific CNVs that converge on apoptosis-regulatory genes. (7A) Distribution of the numbers of resister genes and the weights of supportive evidence (shown as the numbers of independent CRISPR-Cas9 screen measurements from 23 publications). (7B) Significantly enriched Reactome gene signatures of 519 resister gene hits in Figure 7 A. (7C) As in Figure 7A, except for sensitizer genes. (7D) As in Figure 7B, except for 877 sensitizer genes. (7E) Circos plots showing DP-specific deletions and amplifications overlapping and converging on apoptosis- regulatory genes (locations indicated by arrows). Outermost layer, chromosome regions. Each inner layer represents a distinct DP tumor.

[0017] FIGS. 8A-8E show that down-regulating BAX, DFFB or FAS resists melanoma killing by CD8 T cells. (8A and 8B) As in Figures 11 and 1 J, except data showing cocultures results with T cells from two additional and distinct donors. (8C) mCherry+ A375 cells ± BAX overexpression, as shown by Western blots, and their killing by NY-ESO-1-TCR-transduced primary T cells (E:T, 1 :3). (8D) As in Figure 1 L, except data showing cocultures results with T cells from two additional and distinct donors. (8E) As in Figure 1M, except data showing cocultures results with T cells from two additional and distinct donors.

[0018] Data in Figure 8 represent 2-3 replicates with different T cell donors in each replicate. Data are represented as mean ± SEM and analyzed by the two-way ANOVA test.

[0019] FIGS. 9A-9B show frequencies of putative functional gene deletions or co-mutations in the acquired ICI-resistant melanoma cohort versus therapy-naive melanoma cohorts. (9A) Frequencies of deletions in pro-apoptotic genes and amplifications in anti-apoptotic genes in distinct melanoma cohorts: (i) TCGA-SKCM (n = 367 melanomas from 367 patients), (ii) pre- ICI melanomas (n = 56 tumors from 55 patients), and (iii) post-ICI melanomas from the current cohort (n = 20 tumors from 17 patients). (9B) As in 9A, except for concurrence frequencies of significant co-mutated gene pairs shown in Figure 2A.

[0020] FIGS. 10A-10H show AR Human and murine melanoma models. (10A) Crystal violet staining (left) and IncuCyte quantification of bright-field cell confluence (right) of the M486 P cell line and AR sublines, following a 4-day coculture with non-transduced (control) T cells or indicated ratios of HLA-A2.1- / NY-ESO-specific T cells. (10B-10F) As in Figure 3C-3G, coculture data at an E:T ratio of 1 :2. (10G and 10H) Immunofluorescent staining of Bad (10G) and Fas (10H) in isogenic YER P (vehicle) and AR3 tumors. (Left) Representative images. (Right) Quantifications showing the mean fluorescent intensities from 5 or 6 fields. Ruler, 50 pm. Data are represented as mean ± SEM. Data are analyzed by the two-way ANOVA test in Figure 10B-1 OF and the one-way ANOVA followed by Tukey’s multiple comparison test in Figure 10G and 10H.

[0021] FIGS. 11 A-11 F illustrate the workflow and data quality of scWGS derived from M486 and YER models. (11 A) Schematic showing the workflow of scWGS library preparation from cultured cells. (11 B) Overdispersion of bin counts computed from scWGS of the isogenicM486 P cell line as well as AR3 and AR5 sublines. (11C) Bar plots for each genetic subclone showing the proportions of M486 P, AR3, and AR5 cells contributing to the indicated subclones. (11 D) Schematic showing the workflow of scWGS library preparation from frozen tumors. (11 E) As in Figure 11 B, except for samples derived from the isogenic YER P (vehicle) and AR3 tumors. (11F) As in Figure 11 C, except for samples derived from the isogenic YER P (vehicle) and AR3 tumors.

[0022] FIGS. 12A-12G show restoring apoptotic sensitivity re-sensitizes AR melanoma to cytotoxic T cells and ICI therapy. (12A) Quantification of endpoint data (Figure 5D) showing the ratios of live cells of indicated M486 P and AR sublines (latter ± overexpression of indicated genes) in cocultures with HLA-A2.1- / NYESO-specific T cells vs. non-transduced (control) T cells. (12B and 12C) As in Figure 5D and 12A, except with a 1:1 E:T ratio. (12D and 12E) Body weight of the mice in Figure 5F (12D) and 5G (12E). (12F and 12G) As in Figure 5G and 12E, except data representative of an independent replicate. In Figures 12A- 12C, data are represented as mean ± SEM and analyzed by the one-way ANOVA followed by Tukey’s multiple comparison test (Figure 12A and 12C) or the two-way ANOVA test (Figure 12B).DETAILED DESCRIPTION

[0023] Clinical relapses on immune checkpoint inhibitors (I Cl s) occur across tumor types, but how genomic evolution results in acquired-resistant (AR) tumor cell-intrinsic phenotypes remains unclear. By integrating analyses of longitudinal clinical melanomas, functional CRISPR-Cas9 screens, human and murine melanoma AR models, the data described herein show that copy-number variant (CNV)-driven subclonal evolution underlies clinical relapses. Informed by screen hits, clinically recurrent and cooccurring resistance-driver genes were functionalized as targets of CN s, including pro-apoptotic gene deletions and anti-apoptotic gene amplifications. Cultured melanoma under the selective pressure of killer CD8 T cells and syngeneic melanoma under CD8 T cell-dependent ICI therapy engendered AR clones that re-capitulated CNVs convergent on apoptosis-regulatory genes. Single-cell whole-genome sequencing resolved distinct AR subclones that can preexist selective pressure. Experimentally dialing down the apoptotic threshold re-sensitized ICI / T cellresistant melanoma. These findings undergird CNV evolution as a driver of acquired ICI resistance and imply apoptotic threshold as a target to thwart resistance evolution.

[0024] The discoveries described herein thus provide new strategies for blocking anti- apoptotic proteins or activating proapoptotic proteins in combination with ICB therapy to prevent or reverse acquired resistance. Provided are methods to detect, prevent, and / orreverse acquired resistance to ICB therapy by blocking anti-apoptotic proteins or activating proapoptotic proteins.Definitions

[0025] All scientific and technical terms used in this application have meanings commonly used in the art unless otherwise specified. As used in this application, the following words or phrases have the meanings specified.

[0026] As used herein, a “control” or “reference” sample means a sample that is representative of normal measures of the respective marker, such as would be obtained from normal, healthy control subjects, or a baseline amount of marker to be used for comparison. Typically, a baseline will be a measurement taken from the same subject or patient. The sample can be an actual sample used for testing, or a reference level or range, based on known normal measurements of the corresponding marker.

[0027] As used herein, a “significant difference” means a difference that can be detected in a manner that is considered reliable by one skilled in the art, such as a statistically significant difference, or a difference that is of sufficient magnitude that, under the circumstances, can be detected with a reasonable level of reliability. In one example, an increase or decrease of 10% relative to a reference sample is a significant difference. In other examples, an increase or decrease of 20%, 30%, 40%, or 50% relative to the reference sample is considered a significant difference. In yet another example, an increase of two-fold relative to a reference sample is considered significant.

[0028] As used herein, "pharmaceutically acceptable carrier" or “excipient” includes any material which, when combined with an active ingredient, allows the ingredient to retain biological activity and is non-reactive with the subject's immune system. Examples include, but are not limited to, any of the standard pharmaceutical carriers such as a phosphate buffered saline solution, water, emulsions such as oil / water emulsion, and various types of wetting agents. Preferred diluents for aerosol or parenteral administration are phosphate buffered saline or normal (0.9%) saline.

[0029] Compositions comprising such carriers are formulated by well-known conventional methods (see, for example, Remington's Pharmaceutical Sciences, 18th edition, A.Gennaro, ed., Mack Publishing Co., Easton, PA, 1990).

[0030] As used herein, the term "subject" includes any human or non-human animal. The term "non-human animal" includes all vertebrates, e.g., mammals and non-mammals, such as non-human primates, horses, sheep, dogs, cows, pigs, chickens, and other veterinary subjects. In a typical embodiment, the subject is a human.

[0031] As used herein, “a” or “an” means at least one, unless clearly indicated otherwise.

[0032] As used herein, “anti-PD-1 therapy” means treatment with an anti-PD-1 antibody (nivolumab / BMS-936558 / MDX-1106, pembrolizumab / MK-3475, Pidilizumab), and / or an anti- PD-L1 antibody (BMS-986559, MPDL3280A, and MEDI4736).Methods

[0033] Described herein are methods for blocking anti-apoptotic proteins or activating proapoptotic proteins in combination with ICB therapy to prevent or reverse acquired resistance. Also described are methods to detect, prevent, and / or reverse acquired resistance to ICB therapy by blocking anti-apoptotic proteins or activating proapoptotic proteins.

[0034] In some embodiments, described herein is a method of enhancing anti-melanoma therapy in a subject in need thereof, the method comprising administering to the subject an effective amount of an activator of a pro-apoptotic protein and / or a downregulator of an anti- apoptotic protein. In some embodiments, the activator of a pro-apoptotic protein is a BAX activator. In some embodiments, the BAX activator is BTSA1.2. In some embodiments, the downregulator of an anti-apoptotic protein is a BH3 mimetic. BH3 mimetics are small molecule antagonists of the anti-apoptotic BCL-2 members that function as competitive inhibitors by binding to the hydrophobic cleft. In some embodiments, the BH3 mimetic is Venetoclax, Navitoclax, S63845, and / or AMG176. Myeloid leukemia 1 (MCL-1) is an antiapoptotic protein of the BCL-2 family that prevents apoptosis by binding to the proapoptotic BCL-2 proteins such as BAK and NOXA. The BCL-2 family includes anti-apoptotic proteins (BCL-2, BCL-XL, BCL-W, MCL-1, BFL-1 / A1), pro-apoptotic pore-formers (BAX, BAK, BOK) and pro-apoptotic BH3-only proteins (BAD, BID, BIK, BIM, BMF, HRK, NOXA, PUMA, etc.). Inhibitors of MCL-1 include, for example, ANG176, AMG397, ABBV-467, and PRT1419. In some embodiments, the administering comprises administering both a BAX activator and a BH3 mimetic.

[0035] In some embodiments, the subject is treated with immune checkpoint blockade (ICB) or immune checkpoint therapy (ICT) as anti-melanoma therapy. Immune checkpoint blockade (ICB) for the treatment of cancer typically comprises an anti-PD-1 or anti-PD-L1 agent. Optionally, the ICB treatment further comprises additional agents, such as anti- CTLA4 and / or anti-LAG3. Representative ICT agents include, but are not limited to, pembrolizumab, nivolumab, and ipilimumab. In some embodiments, the activator of a proapoptotic protein and / or a downregulator of an anti-apoptotic protein is administered concomitantly with, prior to, and / or subsequent to the administering of the one or more anti- PD-1 / L1 antibodies, and / or an anti-CTLA-4 agent.

[0036] Also described is a method of preventing or inhibiting acquired resistance to antimelanoma therapy in a subject in need thereof. In some embodiments, described herein is a method of ameliorating acquired resistance to immune checkpoint blockade therapy in a subject in need thereof, the method comprising administering to the subject an effective amount of an activator of a pro-apoptotic protein and / or a downregulator of an anti-apoptotic protein. In some embodiments, the method comprises administering to the subject an effective amount of an activator of a pro-apoptotic protein and / or a downregulator of an anti- apoptotic protein.

[0037] A method of detecting acquired resistance to anti-melanoma therapy in a subject is also provided. In some embodiments, the method comprises obtaining a biological sample from the subject, wherein the sample comprises tumoral DNA. The method further comprises assaying the sample for deletion or amplification of genes listed in Table S4 and / or Table S5; and detecting acquired resistance when the assaying indicates deletion or amplification of one or more genes listed in Table S4 and or Table S5. Tables S4 and S5 can be found in United States patent application number 63 / 610,823, filed December 15, 2023, the entire contents of which are incorporated herein by reference.

[0038] In some embodiments, the deletion and / or amplification is measured relative to a sample obtained from the subject prior to, or at an earlier time point in, ICB therapy. In some embodiments, the sample comprises a tumor biopsy or liquid biopsy. A liquid biopsy comprises, for example, cell free tumor DNA or circulating tumor cells.Kits and Compositions

[0039] The invention provides kits comprising a set of reagents as described herein, such as antibodies that specifically bind one or more markers of the invention (including genes and their expression products), and optionally, one or more suitable containers containing reagents of the invention. Reagents include molecules that specifically bind and / or amplify and / or detect one or more markers of the invention. Such molecules can be provided in the form of a microarray or other article of manufacture for use in an assay described herein. One example of a reagent is an antibody or nucleic acid probe that is specific for the marker(s). Another example includes probes (or primers) that selectively identify one or more genotypes described herein. Reagents can optionally include a detectable label.Labels can be fluorescent, luminescent, enzymatic, chromogenic, or radioactive.

[0040] Kits of the invention optionally comprise an assay standard or a set of assay standards, either separately or together with other reagents. An assay standard can serve as a normal control by providing a reference level of normal expression for a given marker that is representative of a healthy individual.

[0041] Kits can include probes for detection of alternative gene expression products in addition to antibodies for protein detection. The kit can optionally include a buffer. Reagents and standards can be provided in combinations reflecting the combinations of markers described herein as useful for detection.

[0042] Also contemplated are kits and compositions comprising the immunotherapeutic and / or supplemental agents and / or additional agents described herein. Such compositions and kits can be packaged to provide a desired combination of therapeutic agents tailored to a patient’s condition. The kit provides therapeutic agents as compositions, or unit dosage forms and / or articles of manufacture, in some embodiments, the kit further comprises instructions for use in accordance with any of the methods described herein. The kit may further comprise a description of an individual suitable for treatment. Instructions supplied in the kits are typically written instructions on a label or package insert (e.g., a paper sheet included in the kit), but machine-readable instructions are also acceptable. The kits of the invention are in suitable packaging. Suitable packaging includes, but is not limited to, vials, bottles, jars, flexible packaging (e.g., sealed Mylar or plastic bags), and the like. Kits may optionally provide additional components such as buffers and interpretative information.EXAMPLES

[0043] The following examples are presented to illustrate the present invention and to assist one of ordinary skill in making and using the same. The examples are not intended in any way to otherwise limit the scope of the invention.Example 1: Genomic copy-number variants drive apoptotic resistance and relapses on immune checkpoint inhibitors

[0044] This Example demonstrates, by integrating analyses of longitudinal clinical melanomas, functional CRISPR-Cas9 screens, human and murine melanoma AR models, that copy-number variant (CNV)-driven subclonal evolution underlies clinical relapses on immune checkpoint inhibitors (ICIs). Informed by screen hits, clinically recurrent and cooccurring resistance-driver genes were functionalized as targets of CNVs, including pro- apoptotic gene deletions and anti-apoptotic gene amplifications. Cultured melanoma under the selective pressure of killer CD8 T cells and syngeneic melanoma under CD8 T celldependent ICI therapy engendered AR clones that re-capitulated CNVs convergent on apoptosis-regulatory genes. Single-cell whole-genome sequencing resolved distinct AR subclones that can preexist selective pressure. Experimentally dialing down the apoptotic threshold re-sensitized ICI / T cell-resistant melanoma. These findings undergird CNV evolution as a driver of acquired ICI resistance and imply apoptotic threshold as a target to thwart resistance evolution.

[0045] Immune checkpoint inhibitors (ICIs) provide the foundation for curative therapy of a large number of malignancies,1-5but resistance limits this potential. Relapses (indicative of tumors acquiring resistance) can follow initial responses, sometimes by several years. In cutaneous melanoma, acquired resistance (AR) or disease progression (DP) to ICI therapy develops in -20-60% of responsive patients.6Knowledge of the clinical mechanisms of AR has not followed pace with the number of ICI-treated patients.

[0046] Earlier analysis identified somatic non-synonymous mutations in interferon receptor signaling and antigen-presentation genes as drivers of acquired ICI resistance in melanoma7and other cancers.6More recently, transcriptomic analysis of pre-and-post tumor tissues from patients with non-small cell lung cancer showed that AR to anti-PD-(L)1 associated with persistent CD8 T-cell inflammation but dysfunctional interferon response.8In the preclinical literature, a recent series of functional CRSPR-Cas9 genomic screens has identified genes that modulate T-cell killing of tumor cells.9-31However, the reductionist and experimental approaches, which often involved a single human or murine tumor cell line, an artificial antigen, and / or a short duration of tumor-T cell interaction, raise questions on the ability of these experimental systems to model clinical relapses on ICI therapy.

[0047] This Example, using whole-exome sequencing data from patient-matched normal as well as pre-and-post melanoma tissues, sought to identify and functionalize somatic, recurrent genetic alterations specifically detected in melanomas with AR to anti-PD-1 ± anti- CTLA-4 therapy. Because of the high copy-number variant (CNV) burden of melanoma, the focus was on AR-specific, recurrently deleted or amplified genes in clinical melanomas that have been identified functionally via CRISPR-Cas9 screens as T cell sensitizer or resister hits / genes, respectively, from independent studies. From this integrative analysis, it was found that putatively functional and recurrent AR-specific CNV events co-occurred frequently to result in intra-tumoral and -patient heterogeneity and loss of tumor cell-death sensitivity in nearly all AR / DP tumors. Experimentally coculturing human melanoma cells to HLA- / antigen- specific killer T cells and treating murine melanoma tumors with ICI therapy generated AR clones that displayed AR-specific co-deletions of pro-apoptotic genes and / or coamplifications of anti-apoptotic genes. Single-cell CNV analysis via single-cell whole-genome sequencing revealed both preexisting and subclonal AR evolution as well as subclonal heterogeneity of driver CNVs, including distinct combinations of pro-apoptotic gene deletions and anti-apoptotic gene amplifications. Finally, AR melanoma models show that restoring (by lowering) the apoptotic threshold rescued the AR phenotype. These findings represent an evolutionary model in which CNV-related genomic instability provides both root and branch mechanisms of clinical relapses on ICI therapy, which may be overcome by co-treatment with agents that enhance tumor cell-death sensitivity.

[0048] EXPERIMENTAL MODEL AND SUBJECT DETAILS

[0049] Patient-derived tissues

[0050] All patients with metastatic cutaneous melanoma were treated at the University of California, Los Angeles (UCLA), Vanderbilt-Ingram Cancer Center, and California Pacific Medical Center. All (except two patients on adjuvant therapy) had an objective (partial or complete) response to anti-PD-1 or anti-PD-1+anti-CTLA-4 blocking antibodies, based on Response Evaluation Criteria in Solid Tumors58and immune-related response criteria.59After > six months of objective responses, in situ or de novo recurrence (on or off active therapy) was defined as acquired resistance or disease progression. Patient-matched longitudinal tumor biopsies (baseline, n = 17; disease progression, n = 20; patients, n = 17) were from the same or distinct anatomic sites. Frozen whole blood or peripheral blood mononuclear cells (PBMCs) was used as the source of normal tissues. Whole-exome sequence (WES) data from patients one to four were generated previously.7Tissue collection and genomic analysis were approved by local institutional review boards with informed consent from each patient.

[0051] Mice

[0052] Male C57BL / 6ROC mice (UCLA Radiation Oncology breeding colony) were used at six to eight weeks of age. All animal experiments were conducted according to the guidelines approved by the UCLA Animal Research Committee.

[0053] Cell lines

[0054] All human and murine cell lines were negative for mycoplasma and identified by RNA-seq and / or GenePrint 10 system (Promega) at periodic intervals during the course of this study for banking and experimental studies. All human cell lines were cultured in either RPMI (A375, M407 and M257) or DM EM (M486, HEK293T) supplemented with penicillin / streptomycin, 2 mM glutamine, 10% heat-inactivated FBS (Omega Scientific or Gibco) in humidified 37°C incubator with 5% CO2. The murine cell line, YUMM1.7ER, its subline AR3cl and engineered AR3cl, were cultures in DMEM / F12 supplemented with non- essential amino acids (Gibco), penicillin / streptomycin, amphotericin B (Gibco), and 10% heat-inactivated FBS (Omega Scientific or Gibco) in humidified 37°C incubator with 5% CO2.

[0055] METHOD DETAILS

[0056] Whole-exome sequencing and analyses

[0057] Genomic DNAs (gDNAs) were extracted from snap-frozen tumors and frozen mouse blood using the QIAGEN AHPrep DNA / RNA Mini Kit, formalin-fixed paraffin-embedded (FFPE) tumors using the QIAGEN QIAamp DNA FFPE Tissue Kit, and patient-matchedPBMCs or frozen blood and cultured cells using the QIAGEN FlexiGene DNA Kit. Whole- exome libraries were prepared using the Roche NimbleGen Exon-Seq Kit, the Roche NimbleGen Seqcap Kit, or the Roche KAPA HyperPlus Library Preparation Kit with the KAPA HyperCap Workflow v3.0 for exome hybridization. Pooled libraries were paired-end sequenced with a read length of 2 x 150 bp on the Illumina HiSeq 3000 or Illumina NovaSeq 6000 S4 or X Plus platforms.

[0058] For somatic mutation calling, BWA-MEM was used for mapping and Picard for the removal of duplications. Somatic single-nucleotide variants (SNVs) and small insertiondeletions (INDELs) of tumors were prepared as previously reported.35Specifically, SNVs were called by using a combination of the Unified Genotyper tool of GATK, MuTect and VarScan2 and INDELs using the combined calls of GATK-UGF, SomaticIndelDetector of GATK (I ndel Locator) and VarScan2. For each detection tool, SNVs / INDELs were supported by at least five reads in the tumor samples and none in the patient-matched normal tissues. Oncotator was then used to annotate somatic SNV / INDELs. Sequenza was used to detect tumor purities and ploidies using default parameters. CNVs were called using the union of the copy number calls derived from Sequenza and VarScan2.

[0059] The CNV profiles of baseline and DP tumors from each patient were analyzed using SigProfilerAssignment with default parameters to decipher CN signature compositions, with COSMIC CN signatures version 3.3 as reference. The associations of gene deletions / co- mutations with signature CN 10 were evaluated using two-sided Fisher’s exact test of CN10 non-attributed or attributed (= 0, > 0) against gene deletions / co-mutations presence or absence. DP-specific recurrent gene amplifications or deletions were defined as detection in > three patients. Functional enrichment analyses were performed on DP-specific recurrent amplified or deleted genes (n > 4 patients) using Metascape,60with REACTOME gene sets as the reference.

[0060] DP-related SMGs were identified using MutSig2CV with each patient’s mutational profile of a single DP tumor or a combination of multiple DPs (if applicable). For a patient with multiple DP tumors, a mutation was considered to exist in this patient if it was detected in > one tumor. Each baseline tumor’s mutational profile was used as input to identify baseline-related SMGs. To circumvent the limitation of small cohorts, type I error was inflated by not performing multiple testing and identified MutSig2CV genes at a P value cutoff of < 0.05. To reduce false-positive SMGs, SMGs were nominated by requiring positive RNA expression level (mean log2CPM > 0) of the given genes in the Cancer Cell Line Encyclopedia RNA expression database.

[0061] The co-occurrence of mutated / deleted genes was analyzed using the somatic / nteractions function in the R package ‘maftools,’ which performs pair-wise Fisher’s exact test to identify such significantly correlated pairs of gene alterations. Comutated genes were defined as gene pairs with an odds ratio > 1 and P value < 0.05. The patient-level frequencies of co-mutations in defined tumor cohorts was also calculated. For the TCGA-SKCM therapy-naive cohort, the copy number alteration data were downloaded from cBioPortal. The WES data of a pre-ICI melanoma cohort have been reported.45When multiple tumors were available from a given patient, they were considered as an entirety, e.g., a deleted gene was counted in a given patient if it was identified from at least one of the patient-matched tumors.

[0062] For phylogenetic analysis, the PHYLIP program with the parsimony algorithm was used, as previously reported33’35-37. Each tree was annotated with truncal driver mutations and with putative driver genetic alterations of acquired ICI resistance.

[0063] For the syngeneic murine model YER, paired-end reads were aligned to the mouse reference genome (mm10) using the BWA-MEM algorithm. The resulting alignments were processed with SAMtools and Picard tools to sort the alignments and remove PCR duplicates. Somatic SNVs and indels were detected using Strelka2 with default parameters and subsequently annotated with ANNOVAR. CNVs were called using CNVkit with default settings. Using CNV profiles as input, MEDICC247was employed to construct a phylogenetic tree based on minimum event distances.

[0064] STRING analysis

[0065] Protein-protein interaction network analysis was conducted using the STRING database (version 12.0) to visualize potential interactions among proteins encoded by recurrent DP-specific amplified or deleted cell-death genes. A median confidence score of 0.4 was used as the threshold. Disconnected nodes were excluded. The k-means algorithm defined functional clusters.

[0066] Cataloging resister / sensitizer genes from CRISPR-cas9 screens

[0067] Data were combined from 23 published studies of functional CRISPR-Cas9 screens to identify genes that modulate tumor cell killing by T cells (excluding CAR T cells) or ICI treatments, in vitro or in vivo, in human or mouse models. These studies were identified by a PubMed search using the keywords ‘CRISPR screening’, ‘tumor-intrinsic’, and ‘T cell killing’, ‘immune evasion’, and ‘immunotherapy target’. Functional hits were categorized into resister genes (whose loss-of-function / knockout hits promoted resistance or gain-of- function / activation hits enhanced sensitivity) and sensitizer genes (whose loss-of-function hits enhanced sensitivity or gain-of-function hits promoted resistance), according to thespecific cut-offs defined by each study. Fensitizer or resister hits were further defined as those genes with concordant evidence in > two independent measurements within the context of CRISPR screens.

[0068] Immunofluorescence

[0069] FFPE tissue sections were heated at 60 °C for 30 minutes and immersed in xylene followed by a gradient of ethanol solutions for deparaffinization and re-hydration. For antigen retrieval, tissue sections were heated in Tris-EDTA buffer (pH 9.0) (ab93684, Abeam) at 95 °C for 15 minutes. Tissues were permeabilized in 0.4% Triton X-100 in PBS for 15 minutes followed by blocking with 10% normal goat serum in PBS for 1 hour. Tissues were incubated with validated primary antibodies, including anti-B2M (MA1-19413, Invitrogen, 1:200), anti-GSDMC (PA5- 116594, Invitrogen, 1:100), anti-FAS (MA5-32489, Invitrogen, 1:100), anti-DFFB (PA5-115114, Invitrogen, 1 :100), anti-CASP9 (MA1- 16842, Invitrogen, 1:100), and anti-Bad (9292S, Cell Signaling Technology, 1 :100) at 4 °C overnight. Tissues were then incubated with goat anti-mouse IgG (highly cross-absorbed secondary antibody, Alexa Fluor Plus 488, A32723, Invitrogen, 1 :400) or goat anti-rabbit IgG (highly crossabsorbed secondary antibody, Alexa Fluor 555, A21429, Invitrogen, 1 :400). Nuclei were counterstained by DAPI (D9542, Sigma-Aldrich). Staining was imaged with a Leica confocal SP8 Light-Sheet microscope and quantified fluorescence intensity by Imaged (version, 1.54d).

[0070] Peripheral T cell transduction

[0071] For peripheral T cell transduction, the NY-ESO-1 TCR retroviral vector (Antoni Ribas, UCLA) was used and retrovirus was generated by co-transfection of HEK-293T with pMSGV, pRD114 and pHIT using BioT (B01-00, Bioland Scientific). Fourteen hours after transfection, cells were treated with 10 mM sodium butyrate (A11079.36, ThermoFisher) and 20 mM HEPES (15630080, Gibco) in fresh media for 6-8 hours and harvested retroviral supernatants 48 and 72 hours after transfection. Human primary T cells were isolated from PBMCs from healthy donors (UCLA Virology Core) by negative selection using the Pan T Cell Isolation Kit (130-096-535, Miltenyi Biotec) and magnetic LS columns (130-122-729, Miltenyi Biotec). Isolated T cells were cultered in RPMI 1640 supplemented with 10% fetal bovine serum (Omega Scientific or Gibco) and 50 lU / ml human IL2 (130-097-748, Miltenyi Biotec) and activated the primary T cells on the same day with 25 pL CD3 / CD28 dynabeads (11161 D, Gibco) per million T cells. Two days after activation, cells were transduced with retrovirus harboring NY-ESO-1 TCR in 6-well plates pre-coated with RetroNectin (T110A, Takara). Media was replenished every 2-3 days. Five days after transduction, the proportion of NY-ESO-1 TCR-positive T cells was measured by flow cytometry using anti-human CD3(317306, BioLegend) and anti-human TCR VP13.1 (362410, BioLegend) antibodies. CD3 / CD28 dynabeads were removed using magnetic isolation 7 days after activation. Engineered T cells were freshly used or cryopreserved using Cryostor CS10 cryopreservation medium (210102, Biolife Solutions) 12-14 days after activation.

[0072] Tumor cell-T cell coculture

[0073] For tumor cell-T cell coculture, mCherry-NLS-labeled or H2B-GFP-labeled melanoma cells were seeded at a concentration of 4 x 104per well in 96-well plates and incubated overnight. NY-ESO-1 TCR-transduced T cells were added on the following day at the indicated E:T ratios with triplicates per group. Plates were imaged using the IncuCyte SX5 real-time live-cell imaging system (Sartorius) at 10x zoom with 2-hour intervals for 4 days. Non-transduced T cells or cultures without T cells served as negative controls. The integrated mCherry or GFP count per well was analyzed as quantitative measurements of live melanoma cells. All measurements were normalized to the signals at the starting time point.

[0074] Human and murine melanoma cell line transduction

[0075] The mCherry-NLS (39319, Addgene) was subcloned into the pLV-neo lentiviral vector, H2B-GFP (11680, Addgene) into a pLV-blast vector, and generated lentiviruses. Melanoma cell lines were then transduced to express mCherry-NLS or H2B-GFP after neomycin (A1720-1G, Sigma) or blasticidin (R21001 , Gibco) selection, respectively, for one week. The FLAG-tagged caspase 3 cleavage-resistant DFFA (DFFA-CR)43in a lentiviral vector (pHAGE_EF1a) was obtained from Stephen Elledge via a material transfer agreement (Dana Farber Cancer Center). The cDNA of human BAX, BAD, P53 and TNFR1 were from UCLA Molecular Screening Shared Resource and subcloned into pLV-blast, pLV-hygro, pLV-puro, and pLV-neo vectors, respectively. Mouse Bad and Fas genes were PCR amplified from the cDNA of YUMM1.7ER cells and cloned into pLV-neo and pLV-blast vectors, respectively. shRNA sequences for human FAS and BAX (UCLA Molecular Screening Shared Resource) are: ACAAACTTCATCAAGAGTA (shFAS-1 ; SEQ ID NO: 1), GGCTTAGAAGTGGAAATAA (shFAS-2; SEQ ID NO: 2), CGGAACTGATCAGAACCATCA (shBAX-1 ; SEQ ID NO: 3), and GCCAGCAAACTGGTGCTCA (shBAX-2; SEQ ID NO: 4). Vector control, shRNA or cDNA-containing lentiviruses were generated by co-transfection of the above constructs with pMD2.G, pRSV-Rev and pMDLg / pRRE into HEK-293T cells using calcium phosphate. Human and mouse cell lines were transduced with lentiviruses using 10 pg / mL polybrene (TR-1003-G, Millipore) for 48 hours and performed antibiotic selection for one week. Synonymous mutations were introduced into the shBAX target sequence within the human BAX cDNA using PCR mutagenesis (A393G, C394T, G396A and C399T) and subcloned this mutant human BAX cDNA into the pLV-blast lentiviral vector. The BAXknocked-down cell line was then transduced with the knockdown-resistant, mutant BAX virus to rescue BAX expression.

[0076] Murine tumor studies

[0077] One million YUMM1.7ER cells or its subline, AR3cl were introduced and AR3cl was engineered (with Bad + Fas overexpression) subcutaneously on both flanks and tumors measured with a caliper every two to four days to calculate the tumor volumes based on the formula (length x width2) / 2. Once tumors reached volumes of 100 to 250 mm3, mice were assigned randomly into experimental groups. Tumor-bearing mice were then treated by injecting anti-PD-1 (P362, Leinco) and anti-CTLA4 (BE0131, Bio X Cell) intraperitoneally twice a week (300 pg / mouse for each antibody for the first two weeks and then 200 Ag / mouse for the remainder time). Venetoclax was formulated in 10% DMSO + 40% PEG 300 + 5% Tween 80 + 45% saline and administered orally via gavage once a day at 25 mg / kg.

[0078] Generation of acquired resistant sublines in vitro and in vivo

[0079] To generate human melanoma sublines with acquired resistance to T cells, M486 cells were seeded at 2.5 x 105per well in 12-well plates and, after overnight cultures, M486 cells were treated continuously with fresh NY-ESO-1 TCR transduced T cells every 6 days for 8 rounds until the formation of visible colonies. Monoclonal sublines (AR1, AR2, AR3) were derived using an E:T ratio of 1:1 (based on the seeding melanoma cell numbers) followed by ring cloning, and polyclonal sublines (AR4, AR5) were derived using an E:T ratio of 1:2 and comprised resistant colonies from multiple wells. The initial acquired resistant sublines were expanded without T cell treatment for three weeks followed by continuous rounds of T cell treatments until cryo-preservation. To verify resistance, M486P and AR sublines were subjected to coculture with NY-ESO-1 TCR-transduced T cells in 12-well plates at the indicated E:T ratios for 4 days. After coculture, plates were washed with PBS and cell confluence was quantified with brightfield imaging using the IncuCyte SX5 at 10x zoom. Plates were fixed with 4% paraformaldehyde (J61899.AK, Thermo) and stained with 0.1% crystal violet solution (V5256, Sigma).

[0080] To generate syngeneic murine melanomas with AR to ICB, one million YUMM1.7ER cells were injected subcutaneously on both flanks of male C57BL / 6ROC mice. Once tumors reached volumes of 100 to 250 mm3, the ICB treatment was started. Anti-PD-1 (P362, Leinco) and anti-CTLA4 (BE0131, Bio X Cell) were intraperitoneally injected twice a week (300 pg / mouse for each antibody for the first two weeks and then 200 pg / mouse). Tumors that regressed after initial treatments but later re-grew as AR tumors (AR1 , AR2, AR3) were then selected and processed for further analysis. A portion of the AR3 tumor, excised on day148 on anti-PD-1 + anti-CTLA4, was used for dissociation (kit and gentleMACS Octo Dissociator from Miltenyi Biotec) and derivation of a cell line (AR3cl). Within eight weeks of in vitro culture, the AR3cl was used for experiments.

[0081] Single-cell whole-genome sequencing and analysis

[0082] From flash-frozen YER tumors, 3x3x3 mm tumor pieces were dissociated in 4 ml NST-DAPI buffer using the gentleMACS Octo Dissociator (Miltenyi Biotec) with protocol ‘4c_nuclei’. Nuclei suspensions were flow-sorted with the BD FACSAria II where DAPI intensity was used to gate and harvest the diploid or aneuploid nuclei populations. Cultured M486 and AR sublines were digested by trypsin-EDTA and stained by 10 pg / ml Hoechst33342 (62249, ThermoFisher) and 5 pg / ml propidium iodide (556463, BD) for 30 mins on ice. Then cells were diluted to 28,000 cells per ml and dispensed into ICELL8 350v nanowell chip (640019, Takara) using the ICELL8 ex system (Takara Bio). For frozen tumors, the sorted nuclei were diluted to 28,000 nuclei per ml and dispensed. The wells with single live cells or nuclei were selected using the ICELL8 CellSelect Software (Takara Bio). Single-cell WGS libraries were prepared following a published protocol.61Briefly, the selected single cells or nuclei were lysed using a lysis buffer (30 mM Tris-HCI PH8.0, 5% Tween 20, 0.5% TritonX-100, 1.36 All / rnl protease) for 30 mins at 54.5°C. Tagmentation was then performed using the Illumina Tagment DNA enzyme kit (20034198, Illumina) at 54.5°C for 12 mins. The reactions were stopped by dispensing neutralization buffer into the sample wells and incubating at 49.4°C for 30 mins. 72 indexed PCR forward primers and 72 indexed reverse primers were used to barcode all the sample wells. PCR master mix was dispensed into each sample well and single-cell whole genomes were amplified for 12 cycles. The PCR products were collected by the Collection Kit (640212, Takara) by centrifuging the chip facing downward at 3,000g for 10 mins. The libraries were purified by 1.8x Ampure XP beads (A63881, Beckman), followed by Qubit and TapeStation QC. The libraries were sequenced using Novaseq X Plus targeting 1 million reads per cell or nucleus.

[0083] FASTQ reads from M486 and YER samples were aligned to the human reference genome hg19 and the mouse reference genome mm10, respectively, using bowtie2.62PCR duplicates were subsequently removed using sambamba.63Cells exhibiting excessive noise were excluded based on the following criteria: (1) cells with low-quality mapping libraries (Q < 1), (2) cells with read counts below 100K when the average read number per cell is around 1M, and (3) cells with more than 10% bins containing no mapped reads. Aligned reads were then counted in variable bins averaging 220 kb and normalized using GC content via lowess regression. Bin-wise ratios were calculated by dividing the bin counts by the mean bin count of each cell. Circular binary segmentation (CBS) function of the R package DNACopy64was employed for segmentation. Low-quality CN profiles (average correlation < 0.8) were filteredout by calculating the average cell correlation with its five nearest neighbors using the k- nearest neighbor (KNN) filtering function in CopyKit (doi.org / 10.1101 / 2022.03.09.483497).

[0084] To identify diploid cells, the coefficient of variation was calculated from the CBS segment ratio values for each cell. The expected coefficient of variation was initially simulated for 1000 diploid cells using a normal distribution / V (0,0.01). By applying an expectation-maximization algorithm to the combined dataset of single-cell segment ratios and the simulated diploid coefficients of variation, a mixture of normal distributions was fitted using the normalmixEM function from the R package mixtools (v1.2.0). Cells were classified as diploid and excluded from further analysis if their coefficient of variation exceeded 5 standard deviations from the mean of the distribution that included the simulated diploid dataset.

[0085] Segment ratios of individual cells were transformed using Iog2 and subjected to dimension reduction via LIMAP using the R package uwot (vO.1.16) with specified parameters (seed = 31, min dist = 0.1, n_neighbors = 20, distance = 'manhattan', spread = 3). Subsequently, subclones were identified using a density-based clustering algorithm (dbscan) in the R package dbscan (v1.1.12). Cells identified as noise by hdbscan were filtered out, and those failing to meet the minimum cell count criterion (n = 6) in each subclone were also excluded to mitigate clustering inaccuracies. Heatmaps were generated using the R package ComplexHeatmap (v2.10.0).

[0086] To reconstruct the phylogenetic relationship, the total number of CNV events was first determined, computed as the count of CNV breakpoints in the consensus CNV profile of each subclone. The length of CNV events was estimated based on the number of bins between each pair of breakpoints, excluding events with a length of < 1 bin. Ploidy estimation for each sample was conducted using DAPI signals. Specifically, the formula 2 x (median DAPI intensity of A peak / median DAPI intensity of D peak) was employed if the majority of sequenced cells originated from the aneuploid peak. Subclonal consensus integer profiles were then derived by computing the median of every integer CN across all single cells assigned to the same subclone, rounded to the nearest integer. The phylogenetic tree of subclones was generated using MEDICC2,47utilizing minimum event distances. To root the tree, a diploid cell with a copy number of 2 was introduced as the root node. Trees were visualized using the R package ggtree (v3.11.0).65

[0087] Western blots

[0088] Protein levels were measured using Western blots after lysis of cells in RIPA buffer (PI89901 , Thermo Fisher Scientific) with protease inhibitor (PI78430, Thermo Fisher Scientific) and phosphatase inhibitor (PI78428, Thermo Fisher Scientific) cocktails. Proteinconcentrations were determined by the BCA protein assay (PI23227, Thermo Fisher Scientific). Primary antibodies used are as follows: TUBULIN (2144S, CST), BAX (41162, CST), FAS (MA5-32489, Thermo Fisher), FLAG (14793S, CST), P53 (2527S, CST), TNFR1 (3736T, CST), BAD (9292S, CST), TAP1 (49671 S, CST), TAP2 (25657S, CST), and INFGR1 (34808S, CST). All antibodies were diluted 1 :1 ,000.

[0089] QUANTIFICATION AND STATISTICAL ANALYSIS

[0090] Student’s t-tests were used to compare two independent groups. One-way ANOVA tests were used for multiple comparisons. Two-way ANOVA tests were used for comparing cell and tumor growth curves. Fisher's exact test was utilized to assess the co-occurrence or association between two events. All details related to statistical analysis can be found in figure legends and in the corresponding methods. All analyses were done using R.4.02, Python 3.8.0, Python 2.7.17 and GraphPad Prism 10. The P values were represented as follows: ns, not significant. *p < 0.05, **p < 0.01 , ***p < 0.001. The sample size was determined on the basis of animal experimental trials and in consideration of previous publications on similar experiments to allow for confident statistical analysis.

[0091] RESULTS

[0092] Recurrent, AR-specific CNVs attenuate tumor cell death

[0093] WES data generated from patient-matched normal tissues (n = 17) was analyzed along with baseline (n = 17) and AR tumors (n = 20) (Fig. 6). The patient cohort consists of 13 males and four females with advanced BRAFMUT(n = 8), / VRASMUT(n = 7), and / VF7MUT(n = 2) cutaneous melanoma. Whole-exome sequencing achieved an average of 238x coverage. The average tumor mutational burden per tumor was 2,864 somatic mutations (range, 20-14,172) or 436 nonsynonymous somatic mutations (range, 1-2,099). No significant difference was observed in the tumor mutational burdens between baseline and AR tumors.

[0094] Copy number variations (CNVs) drive therapeutic resistance in melanoma, as previously reported by the present inventor and colleagues.32'38To evaluate patterns of CNVs before and after clonal evolution on ICI, 22 CN variation signatures derived from a recent pan-cancer analysis39were referenced and 14 CN signatures detected among baseline and AR CNVs (Fig. 1A). CN signatures 10 (CN10) and 12 (CN12)- focal loss-of- heterozygosity signatures recurrent around tumor suppressor genes- were observed only in AR (6 of 20, 30%) but not in baseline tumors. Conversely, CN5 and CN7 (signatures of chromothripsis / amplicon events) were observed only in baseline (3 of 17, 17.6%) but not AR tumors. Additionally, baseline (vs. AR) tumors preferentially displayed CN21 (a signature negatively associated with loss-of-heterozygosity genomic regions) (3 of 17 or 17.6% vs. 1 of20 or 5%). These data suggest loss-of-heterozygosity (LOH) contributes to tumor-cell fitness during ICI therapy.

[0095] CNVs were then analyzed at the gene level. The somatic gene deletion and amplification patterns of baseline and AR tumors were determined in each patient and genes that were specifically and recurrently (defined as > three patients) deleted (n = 5,035) or amplified (n = 2,372) in the AR tumors were extracted. Recurrent AR-specific deleted genes enriched for cell death regulation (e.g., transcriptional regulation by TP53, death receptor signaling) and interferon signaling, among others (Fig. 1 B). Recurrent AR-specific amplified genes enriched for the cell cycle, Rho GTPases, RNA metabolism, and, again, cell death regulation (Fig. 1C). The large numbers of AR-specific and recurrent gene deletions and amplifications likely included passenger events. To functionalize CNV-affected genes as potential drivers of AR to ICI, data were synthesized from the emerging literature, which collectively identifies via CRISPR-Cas9 knockout and activation screens9'31the universe of cancer genes that modulate tumor cell killing by cytotoxic CD8 T cells or ICI therapy, in vitro and / or in vivo (Fig. 7A-7D). By gleaning data derived from all cancer types and including only hits observed in > two independent screens, 519 resister genes (whose loss-of- function / knockout hits promoted resistance or gain-of-function / activation hits promoted sensitivity) (Fig. 7A) and 877 sensitizer genes (whose loss-of-function hits promoted sensitivity or gain-of-function hits promoted resistance) (Fig. 7B) were identified. Resister genes enriched for known resistance pathways (interferon, PI3K-PTEN-AKT-mTOR and MHC class I antigen presentation)7’3540'42and, notably, transcriptional regulation by TP53 (Fig. 7C). Sensitizer genes enriched for previously reported mechanisms of immunotherapy sensitization, including autophagy, toll-like receptor and TGFp signaling, as well as, curiously, adipocyte differentiation (Fig. 7D).

[0096] Recurrent AR-specific CNV-affected genes were then functionalized by integrative analysis of clinical tumor-derived data and preclinical hits from knockout screens. Of 5,053 recurrent, AR-specific, deleted genes, 108 overlapped with 519 resister hits (Fig. 1 D and Table 1). The top overlapping genes included interferon genes (JFNA16, IFNGR2), the epigenetic regulators PRDM4 and the DNA repair genes KDM4A and RECQL4. Importantly, among the top overlapping genes were cell death-promoting genes (TNFRSF1A, DNASE1L2, BAX, PPP6C, FAS, DAPK3, TRAF3). Inspection of the 4,927 non-overlapping genes recurrently and specifically deleted in AR tumors uncovered additional cell-death promoting genes (BAD, GSDMC, DFFB, BNIP3L, CAPS9, APAF1 and TP53). Of the 2,372 recurrent, DP-specific, amplified genes, 90 overlapped with 877 sensitizer genes (Fig. 1E and Table 2). These 90 genes included genes that function in PI3K-AKT-mTOR (YWHAZ), chromatin remodeling (ARID2), autophagy (VMP1, CALC0C02), and, somewhatparadoxically, the immunoproteasome (PSMB8) as well as antigen presentation (TAP1, TAP2 and TAPBP). Notably, highly recurrent amplifications of genes that inhibit cancer cell death (BIRC2, DAXX, MCL1, STAT3, TBK1) occurred specifically in AR tumors.

[0097] Table 1 : Overlapped genes between DP-specific recurrently deleted genes and resister genesRECQL4 APC EML1 SLC25A3 CIC B0LA3 EXTL3 SWI5 DNLZ CHRAC1 FAS SYVN1 FAM49B EIF2AK2 FBXW8 TGFB1 IFNGR2 GTPBP4 FIBP TMED2 ITGB1 KDM4A GNPTAB TMEM181 RPL8 MATR3 HDAC3 TRAF3 TNFRSF1A MLANA HMBS TRIM28 WDR55 NDUFAF6 HNRNPAB WDR20 EHMT1 REST IFNA16 ZNF146 FPGS SLC25A32 INTS5 UBA52 GPI UBR5 KLHL9 E4F1 KMT2B UPF2 LILRA1 NHLRC4 LAIR1 WNK1 MAPKAP1 S0X8 LENG8 ABCC2 MEN1 BSG LMX1B ACY1 NDUFA8 CARM1 LRRC8A ALAS1 NDUFV1 CCDC17 PPP6C AP2S1 NUP188 CDT1 PRDM4 ARHGAP35 NUP214 DAPK3 SUPT16H ATF1 PDIA3 D0T1L SUPV3L1 CCNC PPP2R4 F10 TRAFD1 CDK7 PSME1 NDUFA13 TYROBP CDKN1A RBM22 PLIN4 DNASE1L2 CHD8 RELB ZBTB7A DOHH DNAJC9 SEL1L NDUFB7 MLLT1 DUSP5 SIRT1 ZNF561 BAX EIF3K SLC25A20 OR10H3

[0098] Table 2: Overlapped genes between DP-specific recurrently amplified genes and sensitizer genesPSENEN ACTB PRDX1 HIST1H2AC B / RC2 T0MM7 TAP2 MCL1 CCZ1B CDK2 TAPBP YWHAZ DAXX HEXIM 1 TM2D1 ZNRD1 PSMB8 NACA VMP1 BRAT1 TAP1 PIGK VPS52 C1GALT1 EMC2 PPP1R8 YEATS4 FBXL18PITRM 1 G PATCH 8 PUM1 ZZZ3ACLY JUN RAB5C ANKRD34AACTR2 KAT2A RBBP4 C0X6B1ACTR6 KAT7 RNF19B NDUFB9ANKRD52 KCMF1 S100PBP PABPC1ARID2 LSM10 SARNP PDE7AATF7IP MED18 SLC35B1 PPP1R15ACALC0C02 MED24 SMARCE1 RB1CC1CAND1 MOGS SPOP TADA1CHMP3 M0N2 STAT3 UBE2HCS NSUN4 TBK1 UNC50DPH5 PAN2 TMEM50A ZBTB10DR1 PCBP1 TSFM BAG6EFTUD2 PCBP2 USP39 EHMT2EL0VL1 PPCS YBX1GADD45A PPHLN1 YTHDF2

[0099] Although the tumor cohort size is not powered for the analysis of significantly mutated genes (SMGs), provisional SMGs were identified as those that display RNA expression in the Cancer Cell Line Encyclopedia. There were 111 SMGs among baseline tumors and 94 SMGs among AR tumors (Fig. 1F), with -50% overlapping. Among the baseline-specific provisional SMGs, FBXO43 and IPO11, upon knockout, sensitized tumor cells to T-cell- mediated killing. Conversely, among the AR-specific provisional SMGs, B2M, JAK2 and TCF23 , upon knockout, de-sensitized tumor cells to T-cell-mediated killing. Additionally, in all AR tumors with predicted loss-of-function B2M and JAK2 non-synonymous mutations, deletion of the other copy or copy-neutral loss-of-heterozygosity were observed respectively, both of which resulting in bi-allelic loss-of-function somatic alterations (Fig. 1G). Furthermore, one of the two TCF23-mutant AR tumors displayed bi-allelic loss-of-function alterations (Fig. 1G).

[0100] Thus, from the above integrative analysis, results from CRISPR screens undergird the functional confluence of AR-specific deletions and amplifications on reducing tumor cell death susceptibility, as shown by STRING network analysis (Fig. 1H). From a closer inspection, despite the highly variable genomic spans of AR-specific amplification or deletions, they overlapped in regions harboring pro- or anti-cell death genes, respectively (Fig. 7E). To further validate a T cell-protective role of pro-apoptotic gene deletions specifically in melanoma, a coculture assay was set up to measure human melanoma cell line killing by primary T cells transduced with a HLA- and antigen-specific T cell receptor (TCR). Three nuclear-mCherry+ / HLA-A2.1+ / NY-ESO+human melanoma cell lines (A375, M407 and M257) were used for cocultures with non-transduced or TCR-transduced T cellsfrom multiple distinct healthy donor PBMCs. Melanoma cell lines were engineered with loss- of-function alterations of individual cell death-promoting genes which displayed AR-specific deletions. AR-specific BAX or FAS deletions were mimicked by partial BAX knockdown using shRNAs and AR-specific DFFB deletion by expression of a dominant-negative DFFA (DFFA-CR)43. At the same effector to target (E:T) ratio, partial BAX knockdown significantly reduced A375 and M407 cytotoxicity by TCR-transduced (but not non-TCR-transduced) T cells from three donors (Fig. 11— J and Fig. 8A and 8B). Of the two shRNAs targeting BAX, only the more effective shBAX-2 reduced M407 killing by TCR-transduced (but not control) T cells from three donors (Fig. 1J and Fig. 8B). BAX knockdown was on-target, as expression of a knockdown-resistant BAX rescued A375’s resistance to T cells caused by shBAX-1 and sensitized A375 (without shBAX perturbation) to killing by T cells (Fig. 1K and Fig. 8C). In M257 (but not in A375), DFFA-CR significantly reduced killing by T cells from three donors (Fig. 1 L and Fig. 8D). In M407, but not in A375 or M257, partial FAS knockdown attenuated killing by TCR-transduced (but not non-TCR-transduced) T cells from three donors (Fig. 1M and Fig. 8E). These data support distinct tumor cell context-dependent thresholds for killing by T cells, which is consistent with widely variable apoptotic priming states across solid and blood cancers44.

[0101] ICI selects for heterogeneous genetic co-alterations

[0102] Introducing individual loss-of-function alterations in cell death-promoting genes led to cell line-specific and significant loss of sensitivity to melanoma killing by CD8 T cells, although the biologic effect size from perturbing a single gene dosage was small (Fig. 11- 1M). Considering deletions of cell death-promoting genes, recurrent DP-specific co-deletions of cell death-promoting genes (e.g., co-deletions in DFFB and CAPS9 in three DP tumors) were noted. Also observed were recurrent DP-specific co-amplifications of cell death- protective genes (BIRC2 and STAT3 in Pt04, MCL1 and STAT3 in Pt12, TBK1 and STAT3 in Pt15, BIRC2 and MCL1 in Pt16). Such concomitant (usually) single-allele CN alterations of multiple cell death-regulatory genes suggested functional cooperativity. Therefore, a statistical co-occurrence analysis was performed (Fig. 2A) of nominated ICI resistance-driver gene alterations (Fig. 1 D-1F). Several significant co-occurring loss-of-function mutations / CN losses were identified in several pairs of genes, e.g., B2M and GSDMC or FAS; APAF1 and BNIP3L, BAD and TRAF3 as well as DFFB and CASP9 (Fig. 2A). Interestingly, the codeletion of DFFB and CASP9 was associated with CN10 and CN12 attributions (Fig. 1A) in DP tumors (P = 0.0175, two-sided Fisher’s exact test), which implied that the unknown genomic instability process(es) underlying CN10 / 12 as important resistance-evolutionary mechanism(s). Also detected were significant co-occurring amplifications of STAT3 and TBK1 (Fig. 2A).

[0103] To visualize co-mutations of resistance-driver alterations at the level of individual patients, phylogenies were constructed (Fig. 2B-2D). Baseline and DP tumors, with respect to normal tissues, uniformly followed a branched evolutionary pattern. DP (vs. baseline) tumors were more, equally or less closely related to their most recent common ancestral tumor clone. Importantly, in twelve DP tumors from eleven patients, single-copy (sometimes double-copy) co-deletions were observed in two to ten death-promoting genes (Fig. 2B-2D). Co-amplification of at least two cell death-protective genes in six DP tumors from six patients. Notably, DP tumors from five patients displayed both amplification of cell death- protective and deletions of cell death-promoting genes, relative to their patient-matched baseline tumors (Fig. 2B-2D). As validation, co-deletion-associated reductions in caspase 9- DFFB (Fig. 2E), B2M-FAS (Fig. 2F), and B2M-GSDMC (Fig. 2G) protein levels were confirmed upon relapse.

[0104] The hypothesis that the frequencies of resistance-driving loss-of-function or gain-of- function mutations and their co-mutations are higher in cutaneous melanoma with acquired ICI resistance, compared to ICI therapy-naive cutaneous melanomas was further tested. Three melanoma cohorts were analyzed: (i) TCGA-SKCM tumors (367 tumors from 367 patients), (ii) pre-ICI melanomas from prior45and current studies (n = 56 tumors from 55 patients), and (iii) post-ICI melanomas from the current study (n = 20 tumors from 17 patients). Consistently, melanomas that relapsed on ICI displayed significantly higher frequencies of deletions in death-promoting genes, amplifications in death-inhibitory genes and their co-mutations across most analyses (Fig. 9).

[0105] Experimental antitumor T-cell Immunity selects for combinatorial deletions of pro- apoptotic genes

[0106] To test whether tumor cell-intrinsic combinatorial deletion of specific genes underlie acquired ICI resistance, in vitro and in vivo models were developed to enable somatic genomic analysis. In vitro, a HLA-A2.1+ / NY-ESO+human melanoma cell line (M486) with patient-matched blood was identified. M486 was then exposed to eight rounds (once every six days) of fresh primary human T cells (transduced with a HLA- / antigen-specific T cell receptor) continuously (at a fixed E:T ratio determined by the initial seeding density of melanoma cells) over a period of 48 days (Fig. 3A). This coculture resulted in tumor cell killing during early rounds, cytostasis or slow cycling later, and finally resumption of clonogenic proliferations. Monoclonal (acquired resistant or AR) sub-populations (AR1, AR2, and AR3) and polyclonal (mixed) sub-populations (AR4 and AR5) (Fig. 3A) were expanded. In vivo, the Bra / V600MUTsyngeneic murine melanoma model with high mutational burden (YUMM1.7ER or YER) was used, which had been characterized as tumors capable of responding on average to anti-PD-1+anti-CTLA-4 therapy in a CD8 T cell-dependentmanner.46A cohort of YER parental (P) tumors were treated with combined ICI therapy twice a week. Tumors that relapsed were then selected, after initial partial responses lasting 25-35 days (AR1, AR2), for banking plus analysis and a tumor that relapsed, after an initial complete response lasting > 110 days (AR3), for banking, analysis plus establishment of a syngeneic subline (Fig. 3B).

[0107] Next, T-cell resistance of M486 AR1-5 sublines was validated by brightfield imaging and quantification followed by crystal violet staining (Fig. 10A). Compared to the M486 parental (P) cell line, M486 AR1 to AR5 displayed clear resistance to growth suppression by four-day cocultures with fresh primary HLA- / antigen-specific T cells (Fig. 10A). This was consistent with both brightfield image quantification and crystal violet staining (Fig. 10A). To corroborate these findings, the P and AR lines were engineered with nuclear-localized green fluorescent protein and measured melanoma cell confluence after exposure to HLA- / antigen- specific vs. control T cells (Fig. 3C-3G). Continuous selection with killer T cells and expansion of subclonal variant populations had resulted in a stable T-cell resistant phenotype, at two E:T ratios tested (Fig. 3C-3G; Fig. 10B-10F). To validate stable T-cell resistance after in vitro expansion of the YER AR3-derived cell line (AR3cl), AR3cl were reimplanted into the syngeneic host (C57BL / 6 mice) and the YER AR3cl tumors re-challenged (compared to the YER P tumors) with combined ICI therapy. Indeed, YER AR3cl tumors retained resistance to ICI therapy (Fig. 3H).

[0108] WES data was then generated from isogenic M486 lines and the patient-matched blood to identify somatic genetic alterations and phylogenetic relationships. Truncal alterations included a putative driver point mutation (PIK3CA E542K) and amplification of the antigen-presentation gene, TAP1 / 2 (CN = 4) (Fig. 3I). In contrast to M486 P, all M486 AR monoclonal and polyclonal populations shared single-copy deletions of pro-apoptotic genes (TNFRSF1A, BAD, and TP53), interferon genes (IFNGR1 / 2), and TAP1 / 2. This set of deleted genes re-capitulated DP-specific co-deletions in patient #3 (Pt03) (TNFRSF1A, BAD, and TP53), Pt05 (BAD, and TP53), Pt07 (TNFRSF1A, TP53), and Pt09 (TNFRSF1A, BAD). Additionally, M486 AR1 uniquely harbored a single-copy deletion of IFNA16. Western blot analysis of cell lysates confirmed that the reduced gene dosage of AR sublines resulted in downregulated levels of proteins (Fig. 3J). WES data was also generated from YER P and AR1-3 tumors, along with surrogate normal tissue from the blood of the inbred C57BL / 6 colony. From phylogenetic analysis, truncal CN gains in pro-apoptotic genes (Bad, Casp7, Fas) and deletions in anti-apoptotic genes (Bcl3, Bcl2l12) were observed, in addition to the expected truncal Braf mutation (V637E) (Fig. 3K). This set of deleted genes re-capitulated DP-specific co-deletions in Pt03 (BAD, FAS). Importantly, evolution past the most recent ancestral node shared by YER AR1-AR3 (but not shared by the two vehicle-treated tumors)was significant for reversion to CN neutrality (from one-copy deletion) in both anti-apoptotic genes (Bcl3, Bcl2l12), denoting a relative gene dosage gain specific to the evolution of acquired resistance. Moreover, resistance-specific CN alterations unique to YER AR3 were observed, namely, reversion to CN neutrality in all three pro-apoptotic genes (Bad, Casp7, Fas), denoting a relative gene dosage loss specific to resistance evolution in YER AR3. Using the YER P cell line and AR3cl, Western blots confirmed the predicted decrease in the protein expression of Bad, Fas and Casp7 in YER AR3cl (vs. YER P) (Fig. 3L). Using the corresponding tumors, these findings were corroborated by performing the Bad and Fas immunofluorescence, showing unique loss of protein expression in the YER AR3 tumor (Fig. 10G and 10H).

[0109] Resistance evolves by subclone-private and preexisting CNAs

[0110] To resolve subclonal evolutionary patterns and mutations with and without the selective pressure of tumor cell killing by T cells, single-cell WGS (scWGS) of the M486 P, AR3 and AR5 cell lines was conducted (Fig. 11 A). From a total of 3,384 single cells sequenced, 3,084 (91.1%) of high-quality aneuploid nuclei were retained for CN analysis and achieved an average read count of 3.76M per cell (range, 0.11M-11.43M), an average of 204 reads / 220 Kb bin (range, 7-596 reads), and a lower level of overdispersion (i.e. , variance of read count distribution across 220 kb genomic bins) (median values < 0.05) (Fig. 11 B). Unbiased clustering of CN profiles from 3,084 cells identified five subclones (c1-c5) (Fig. 4A and Fig. 11C), which comprised three M486 resistance-specific subclones (c1, c2, and c3) and two M486 P-specific subclones (c4 and c5) (Fig. 4B). The phylogeny of these subclones was then reconstructed by computing the consensus integer CN profiles for each subclone and then determining the lineage relationships using minimum event distance with intra-tumoral CN comparisons 2.47As expected, all subclones originated from the same genetic lineage and evolved in a branched pattern (Fig. 4C). Also observed was that the P- specific subclones (c4 and c5) evolved prior to the emergence of a resistance-specific clade (c1-c3). The monoclonal AR subline (AR3) mostly comprised c1 , whereas the polyclonal subline (AR5) comprised the two more phylogenetically related subclones, c2 and c3 (Fig. 4C). Consistent with the phylogenetic analysis based on bulk WES (Fig. 3I), deletions of chr6p (TAP1 / 2), chr6q (IFNGR1), chr17p (TP53), and chr21 (IFNGR2) were detected in M486 resistance-specific subclones (c1-c3) relative to M486 P-specific subclones (c4, c5) (Fig. 4D). However, bulk WES-based analysis could not resolve the “intra-tumoral” heterogeneity in resistance-associated CN losses, e.g., chr6q (IFNGR1) CN loss was more profound in c1 and c3 relative to c2 (Fig. 4D). Additionally, the resistance-specific CN alterations based on bulk WES, deletions in chr11q (BAD) and chr12p (TNFRSF1A) (Fig. 3I), were in fact preexisting in the M486 P-specific subclone (c4) (Fig. 4D). Importantly, bulkWES-based CN analysis failed to detect a resistance-specific deletion in chr15 (B2M), but scWGS clearly detected chr15 (B2M) deletions in numerous cells within the resistancespecific subclone, c3 (Fig. 4D). Again, this resistance-specific deletion in B2M (c3) preexisted in a M486 P-specific subclone (c4) (Fig. 4D).

[0111] Next, to resolve the in vivo subclonal evolutionary patterns and mutations using the pair of flash-frozen YER vehicle-treated and AR3 tumors, 2,358 nuclei were profiled by scWGS (Fig. 11D) and 1 ,578 (66.9%) of high-quality aneuploid nuclei retained for CN analysis. Data from these nuclei yielded an average read count of 1.14M / nucleus (range, 0.11M-4.11M), an average of 74 reads / 220 Kb bin (range, 7-255 reads), and low levels of overdispersion (median values < 0.05) (Fig. 11 E). Unbiased clustering of CN profiles from these YER 1 ,578 nuclei identified five subclones (c1-c5) (Fig. 4E), which comprised two resistance-specific or -enriched subclones (c1 and c2) and three vehicle-specific or - enriched subclones (c3-c5) (Fig. 4F and Fig. 11 F). Per scWGS-derived CN-based phylogenetic analysis, YER tumors on vehicle and ICI treatment displayed parallel and branched evolution (Fig. 4G). Consistent with the pattern observed in the M486-T cell coculture model, subclones without killer T cell-imposed selective pressure emerged prior to the evolution of a resistance-enriched clade composed of c1 and c2 (Fig. 4G). Consistent with bulk WES-based CN analysis (Fig. 3K), scWGS-based CN analysis revealed that the vehicle treatment-associated YER tumor subclones, c3-c5, harbored baseline amplification of chr19 (containing the anti-apoptotic genes Bad, Fas and Casp7) and deletion of chr7 (containing the pro-apoptotic genes Bcl3 and Bcl2l12) (Fig. 4H). In contrast to bulk WESbased CN analysis of the YER AR3 tumor, which detected CN neutrality in the corresponding regions of chr19 and chr7, scWGS-based CN analysis of the YER AR3- specific or -enriched clones (c1 and c2) uncovered intra-tumoral heterogeneity in these regions. Specifically, with respect to the region of chr19 encompassing Bad, Fas and Casp7, while the majority of resistant c2 cells reverted to CN neutrality, resistant c1 cells displayed deletion (Fig. 4H). Moreover, with respect to the region of chr7 encompassing Bcl3 and Bcl2l12, while resistant c2 cells reverted to CN neutrality, resistant c1 cells displayed attenuated deletion relative to sensitive c3-c5 cells. Thus, the two resistant subclones, c1 and c2, appeared to have achieved resistance by differentially dialing down the pro-apoptotic (c1) or up the anti-apoptotic (c2) thresholds (Fig. 4H). Furthermore, the CN heterogeneity of chr7 and chr19 evident in minute subpopulations of YER vehicle treatment-associated subclones (c3-c5) supports preexisting resistant clones. Interestingly, scWGS-based CN analysis detected, while bulk WES-based CN analysis missed, amplification of chr9 (Cxcr5) in the ICI-sensitive clones (c3-c5) and amplification of chr11 (Egfr and Mapk9) in the ICI- resistant clone, c1.

[0112] Restoring gene dosage re-sensitizes AR melanoma to T cells

[0113] The functional roles of deletions in pro-apoptotic genes to the T cell-resistant phenotype were next tested directly using in vitro and in vivo experimental models. Because single-copy deletions of P53, BAD, and TNFR1 were truncal to the evolution of all M486 AR sublines (AR1-AR5) (Fig. 3I and 3J), all AR sublines were engineered to stably overexpress P53 alone, BAD alone, or TNFRI alone (Fig. 5A), both BAD and TNFR1 (Fig. 5B), as well as P53, BAD, plus TNFRI (Fig. 5C). M486 AR subline killing by HLA-A2.1- / NY-ESO-specific (vs. control / non-transduced) T cells (cocultures at E:T of 2:1 or 2:1) was then measured, with or without restoring gene dosage by overexpression. In AR1 to AR5 sublines cocultured with T cells, single-gene (P53, BAD, or TNFRI) overexpression often caused significant melanoma growth inhibition but failed to confer full sensitivity (i.e. , the sensitivity of the P M486 cell line) to HLA- / antigen-specific killing by T cells (Fig. 5D, Fig. 12A-12C). Compared to single-gene overexpression, double-gene overexpression (BAD and TNFR1) caused significant melanoma growth inhibition in all AR sublines, conferring full sensitivity more frequently at the lower E:T ratio (1 :1) (Fig. 5D, Fig. 12A-12C). Importantly, triple-gene (P53, BAD, and TNFR1) overexpression caused not only significant melanoma growth inhibition consistently but also a net loss of growth (i.e., melanoma death exceeding proliferation), resulting in sensitivity to HLA- / antigen-specific killing by T cells equal to or greater than that observed in the M486 P cell line (Fig. 5D, Fig. 12A-12C). In contrast, in the presence of control T cells, the growth patterns of AR sublines did not differ significantly with or without overexpression, even the overexpression of three genes that are hemizygously deleted in the M486 P line (Fig. 5D). Moreover, because single-copy deletions of Bad and Fas occurred specifically in YER AR3cl but not the YER P cell line (Fig. 3K and 3L), YER AR3cl was engineered to stably overexpress Bad and Fas (Fig. 5E). Importantly, overexpression of Bad and Fas sensitized AR3 tumors to ICI therapy in vivo, which did not affect weight gains in mice (Fig. 5F; Fig. 12D). Lastly, restoring the functional gene dosage pharmacologically using the BCL-2 inhibitor venetoclax was found to also re-sensitize AR3 tumors to ICI therapy without negatively affecting weight gain (Fig. 5G; Fig. 12E-12G).

[0114] DISCUSSION

[0115] Analysis of somatic gene alterations specific to relapsing clinical melanomas after ICI therapy was integrated with analysis of cancer genes that regulate cytotoxicity by T cells, as discovered through in vitro / in vivo CRISPR-Cas9 screens. This comparative genomic study functionalized recurrent relapse-specific CNVs that potentially confer melanoma fitness under the selective pressure of cytotoxic T cells unleashed by ICI therapy. Gene targets of overlapping relapse-specific CNVs converged on cell-death regulatory genes. Notably, compared to baseline / pretreatment melanomas, relapsing melanomas commonly displaymultiple hemizygous deletions of distinct pro-apoptotic genes, which is consistent with a model of cumulative haploinsufficiencies of tumor suppressor genes proposed to explain recurrent hemizygous deletions / LOHs common across cancers (in contrast to Knudson’s “two-hit” hypothesis of recessive tumor suppressor genes).48'50Functionally, these findings support a tumor cell-intrinsic model of clinical relapses on ICI in which escape from ICI- unleashed antitumor immunity selects for multiple, yet often distinct, non-redundant genetic hits to resist T-cell attack51and to avoid T-cell recognition.52Attenuation of the apoptotic threshold, by frequently co-occurring deletions of pro-apoptotic genes and amplifications of anti-apoptotic genes, facilitates resistance to additive T cell-mediated cytotoxicity - a notion consistent with tumor regression by cytotoxic T cells as death by many cuts or multiple sublethal attacks.53Thus, the pretreatment or early on-treatment apoptotic thresholds of immunogenic death as well as the quantitative (how numerous) and qualitative (how cytotoxic) properties of intratumoral CD8 T cells are distinct across individual tumor subclones, tumor sites and patients, which in turn shape the individualized and heterogeneous kinetics of resistance evolution and clinical relapses.

[0116] Although clinical relapses after ICI therapy can follow highly variable kinetics and metastatic patterns, the generally accepted parsimonious definition of AR to ICI refers to relapses in patients who have a period of initial response to ICI therapy followed by clinical and / or radiological DP. From here, variations of this definition center around the depths (stable disease, partial or complete response) of the initial response, durations of the response (where delayed relapses comprise > six months of response duration, in contradistinction to the phenomenon of primary resistance), and treatment continuation / discontinuation status. Clinical relapses in the current cohort of seventeen patients mostly conform to the general definition with partial / complete response, delayed relapses, and DP on treatment. However, two patients were treated with ICI therapy in the post-surgical or adjuvant setting, where an initial response was unevaluable, and one of these two patients had her adjuvant ICI treatment stopped after two months due to toxicities, although her disease recurred or relapsed after 3 years of initiating adjuvant ICI treatment and her recurrent tumor sample was obtained off ICI therapy. The post-treatment melanomas from these two adjuvant ICI-treated patients harbored CNVs affecting multiple apoptosis-regulatory genes that were absent in their patient-matched pretreatment, surgically resectable melanomas. These observations, albeit in only two patients, suggest a CNV-driven resistance mechanism similar to those observed in advanced melanomas not amenable to surgical treatment. Thus, a mechanism-based definition (e.g., tumor cell- intrinsic, apoptosis threshold, etc.) of clinical relapses on / after ICI therapy may inform strategies to overcome AR, regardless of the specific definition of ICI AR.

[0117] Tumor cell killing by T cells requires the pro-apoptotic effects of perforin and granzymes, death receptors, and / or interferons.52 53When tumor cells evolve to resist killing by T cells, they down-regulate both extrinsic and intrinsic programmed cell death mechanisms, which indicates that down-regulating both slow and rapid killing mechanisms contributes to acquired ICI resistance. The present findings indicate that blunting of killing mechanisms occurred via reduced I FNy, TNFa, FAS signaling and / or reduced BAD / BAX, APAF1 / CASP9 signaling. Moreover, reduced DFFB levels in tumor cells may attenuate killing by granzymes from cytolytic T cells, as inactive DFFB is cleaved and activated by granzymes B and M. Reduced expression of BAD / BAX in tumor cells may also blunt the effect of rapid killing by granzyme B, which can activate pro-apoptotic BCL2 family members. The terminal executioner of apoptosis, CASP3, can cleave and activate pore-forming gasdermin proteins, leading to an inflammatory form of cell death (pyroptosis).54Reduced expression of GSDMC may weaken the link between apoptosis and pyroptosis. Additionally, the results suggest that blocking anti-apoptotic proteins (e.g., BCL-2, BIRC2) or activating pro-apoptotic proteins (e.g., BAX ) may reverse or even prevent acquired ICI resistance.

[0118] It is important to note that most of the functional CRISPR-Cas9 screens analyzed here were in vitro and involved only tumor and T cells. Clinically, relapse may be caused by deletions of genes such as CXCL17 that affect cytotoxic T cell homeostasis55'57in the context of an intact tumor immune microenvironment. Notably, investigators of recently published in vivo CRISPR-Cas9 screens analyzed their data against prior results from in vitro CRISPR-Cas9 screens.11They found that tumor cell-intrinsic loss of antigen presentation genes conferred T cell cytotoxicity in vitro but sensitized to in vivo ICI therapy in mice. One in vivo mechanism was based on IFN sensing by tumor cells (in response to ICI therapy) that results in classical HLA class I upregulation and thereby NK cell inhibition. Interestingly, the present study found that TAP1 / 2 and TAPBP were recurrently amplified specifically in melanomas relapsing on / after ICI therapy and that these genes were identified as sensitizer hits only in CRSPR-Cas9 screens carried out in vivo.11’12’21Therefore, evasion of NK cell cytotoxicity may contribute to melanoma relapses on / after ICI therapy.

[0119] REFERENCES

[0120] 1. Chamoto, K., et al. (2023). Nat Rev Immunol. 10.1038 / s41577-023-00867-9.

[0121] 2. Ribas, A., and Wolchok, J.D. (2018). Science 359, 1350-1355.

[0122] 3. Sharma, P., and Allison, J.P. (2015). ICell 161, 205-214. S0092-8674(15)00317-7.

[0123] 4. Sharma, P., and Allison, J.P. (2015). Science 348, 56-61. 10.1126 / science.aaa8172.

[0124] 5. Topalian, S.L., et al. (2016). Nat Rev Cancer 16, 275-287. 10.1038 / nrc.2016.36.

[0125] 6. Schoenfeld, A.J., and Hellmann, M.D. (2020). Cancer Cell 37, 443-455.

[0126] 7. Zaretsky, J.M., et al. (2016). N Engl J Med 375, 819-829. 10.1056 / NEJMoa1604958.

[0127] 8. Memon, D., et al. (2024). Cancer Cell. 10.1016 / j.ccell.2023.12.013.

[0128] 9. Coelho, M.A., et al. (2023). Cancer Cell 41, 288-303 e286.

[0129] 10. Dervovic, D., et al. (2023). at Commun 14, 3150. 10.1038 / s41467-023-38841 -7.

[0130] 11. Dubrot, J., et al. (2022). Nat Immunol 23, 1495-1506. 10.1038 / s41590-022-01315-x.

[0131] 12. Dubrot, J., et al. (2021). Immunity 54, 571-585 e576. 10.1016 / j.immuni.2021 .01 .001 .

[0132] 13. Frangieh, C.J., et al. (2021). Nat Genet 53, 332-341. 10.1038 / s41588-021-00779-1 .

[0133] 14. Griffin, G.K., et al. (2021). Nature 595, 309-314. 10.1038 / s41586-021-03520-4.

[0134] 15. Herzfeldt, A.K., et al. (2023). Elife 12. 10.7554 / eLife.84314.

[0135] 16. Hou, J., et al. (2021). J Immunother Cancer 9. 10.1136 / jitc-2020-001819.

[0136] 17. Joung, J., et al. (2022). Nat Commun 13, 1606. 10.1038 / s41467-022-29205-8.

[0137] 18. Kearney, C.J., et al. (2018). Sci Immunol 3. 10.1126 / sciimmunol.aar3451 .

[0138] 19. Lawson, K.A., et al. (2020). Nature 586, 120-126. 10.1038 / s41586-020-2746-2.

[0139] 20. Li, F., et al. (2020). Cancer Discov 10, 270-287. 10.1158 / 2159-8290.CD-19-0780.

[0140] 21. Manguso, R.T., et al. (2017). Nature 547, 413-418. 10.1038 / nature23270.

[0141] 22. Pan, D., et al. (2018). Science 359, 770-775. 10.1126 / science.aao1710.

[0142] 23. Patel, S.J., et al. (2017). Nature 548, 537-542. 10.1038 / nature23477.

[0143] 24. Upadhyay, R., et al. (2021). Cancer Discov 77, 599-613.

[0144] 25. Vredevoogd, D.W., et al. (2019). Cell 178, 585-599 e515. 10.1016Zj.cell.2019.06.014.

[0145] 26. Wang, G., et al. (2020). Cancer Discov 10, 1912-1933.

[0146] 27. Wang, X., et al. (2021). Cell 184, 5357-5374 e5322. 10.1016Zj.cell.2021 .09.006.

[0147] 28. Williams, J.B., et al. (2020). Nat Commun 77, 602. 10.1038 / s41467-020-14290-4.

[0148] 29. Wu, L., et al. (2023). Cell Metab 35, 1580-1596 e1589. 10.1016 / j.cmet.2023.07.001 .

[0149] 30. Young, T.M., et al. (2020). Sci Immunol 5. 10.1126 / sciimmunol.abb9561 .

[0150] 31. Zhang, Z., et al. (2022). Cell Rep Med 3, 100655. 10.1016 / j.xcrm.2022.100655.

[0151] 32. Dharanipragada, P., et al. (2023). Cancer Discov 73, 880-909.

[0152] 33. Hong, A., et al. (2021). Cancer Discov 77, 714-735. 10.1158 / 2159-8290.CD-20-0873.

[0153] 34. Hugo, W., et al. (2015). Cell 762, 1271-1285. 10.1016 / j.celL2015.07.061.

[0154] 35. Liu, S., et al. (2023). Nat Med 29, 1123-1134. 10.1038 / s41591 -023-02304-9.

[0155] 36. Moriceau, G., et al. (2015). Cancer Cell 27, 240-256.

[0156] 37. Shi, H., et al. (2014). Cancer Discov 4, 80-93.

[0157] 38. Shi, H., et al. (2012). Nat Commun 3, 724. ncomms1727 [pii]10.1038 / ncomms1727.

[0158] 39. Steele, C.D., et al. (2022). Nature 606, 984-991. 10.1038 / s41586-022-04738-6.

[0159] 40. Garcia-Diaz, A., et al. (2017). Cell Rep 79, 1189-1201. 10.1016 / j.celrep.2O17.04.031 .

[0160] 41. Peng, W., et al. (2016). Cancer Discov.6(2):202-16. Epub 2015 Dec 8.

[0161] 42. Shin, D.S., et al. (2016). Cancer Discov. 2159-8290.

[0162] 43. Kula, T., et al. (2019). Cell 178, 1016-1028 e1013. 10.1016Zj.cell.2019.07.009.

[0163] 44. Lopez, A., et al. (2022). Nat Commun 73, 1199. 10.1038 / s41467-022-28741 -7.

[0164] 45. Hugo, W., et al. (2016). Cell 765, 35-44.

[0165] 46. Wang, Y., et al. (2021). Cancer Cell 39, 1375-1387. 10.1016 / j.ccell.2O21 .07.023.

[0166] 47. Kaufmann, T.L., et al. (2022). Genome Biol 23, 241. 10.1186 / s13059-022-02794-9.

[0167] 48. Beroukhim, R., et al. (2010). Nature 463, 899-905. 10.1038 / nature08822.

[0168] 49. Bignell, G.R., et al. (2010). Nature 463, 893-898. 10.1038 / nature08768.

[0169] 50. Solimini, N.L., et al. (2012). Science 337, 104-109. 10.1126 / science.1219580.

[0170] 51. McKenzie, B., and Valitutti, S. (2023). Trends Cancer 9, 198-211.

[0171] 52. Marincola, F.M., et al. (2000). Adv Immunol 74, 181-273.

[0172] 53. Weigelin, B., and Friedl, P. (2022). Trends Cancer 8, 980-987.

[0173] 54. Wang, Y., et al. (2017). Nature 547, 99-103. 10.1038 / nature22393.

[0174] 55. Hernandez-Ruiz, M., et al. (2019). J Leukoc Biol 105, 1027-1039.

[0175] 56. Hiraoka, N., et al. (2011). Gastroenterology 140, 310-321.

[0176] 57. Li, Y., et al. (2022). Int J Mol Sci 24. 10.3390 / ijms24010615.

[0177] 58. Wolchok, J.D., et al. (2009). Clin Cancer Res 15, 7412-7420.

[0178] 59. Eisenhauer, E.A., et al. (2009). Eur J Cancer 5, 228-247.

[0179] 60. Zhou, Y., et al. (2019). Nat Commun 10, 1523. 10.1038 / s41467-019-09234-6.

[0180] 61. Wang, K., et al. (2023). Cell 186, 3968-3982 e3915. 10.1016 / j.cell.2023.07.024.

[0181] 62. Langmead, B., and Salzberg, S.L. (2012). Nat Methods 9, 357-359.

[0182] 63. Tarasov, A., et al. (2015). Bioinformatics 31, 2032-2034.

[0183] 64. Olshen, A.B., et al. (2004). Biostatistics 5, 557-572.

[0184] 65. Yu, G. (2020). Curr Protoc Bioinformatics 69, e96. 10.1002 / cpbi.96.Resistance

[0185] Assays for detection of acquired resistance can be based on sequencing of tumoral DNA obtained from a patient sample, optionally comparing to a sample obtained from the same patient prior to or at an earlier timepoint during treatment. Samples are typically obtained from tumor biopsy or liquid biopsy (cell free tumor DNA or circulating tumor cells). Examples of genes whose deletion or amplification can be informative include: a) genes which promote tumor cell death (e.g., TNFRSF1A, FAS, BAX) b) loss-of-function co-mutations of cell death-promoting genes, antigen-presenting gene, B2M, and IFNy genes (JFNA, IFNGR2, JAK1 / 2) c) protein expression of cell death-promoting genes (CASP9, DFFB, GSDMC, FAS) and B2M d) genes amplified specifically during relapse enriched for immune synapse regulation and cMET-RAP1-RAC1 signaling.

[0186] Examples include assays directed at detecting deletion of genes (see Table S5) specifically observed in acquired ICB resistant melanoma. Of particular interest are resister genes (Table S4) identified in CRISPR-CAS9 functional screens. Additional examples include assays directed at detecting amplification of genes (see Table S5) specifically observed in acquired ICB resistant melanoma. Of particular interest are sensitizer genes (Table S4) identified in CRISPR-CAS9 functional screens. Tables S4 and S5 can be found in United States patent application number 63 / 610,823, filed December 15, 2023, the entire contents of which are incorporated herein by reference.

[0187] Throughout this application various publications are referenced. The disclosures of these publications in their entireties are hereby incorporated by reference into this application in order to describe more fully the state of the art to which this invention pertains.

[0188] Those skilled in the art will appreciate that the conceptions and specific embodiments disclosed in the foregoing description may be readily utilized as a basis for modifying or designing other embodiments for carrying out the same purposes of the present invention. Those skilled in the art will also appreciate that such equivalent embodiments do not depart from the spirit and scope of the invention as set forth in the appended claims.

Claims

What is claimed is:

1. A method of enhancing anti-melanoma therapy in a subject in need thereof, the method comprising administering to the subject an effective amount of an activator of a pro-apoptotic protein and / or a downregulator of an anti-apoptotic protein.

2. The method of claim 1 , wherein the activator of a pro-apoptotic protein is a BAX activator.

3. The method of claim 2, wherein the BAX activator is BTSA1.2.

4. The method of claim 1 , wherein the downregulator of an anti-apoptotic protein is a BH3 mimetic.

5. The method of claim 4, wherein the BH3 mimetic is Venetoclax, Navitoclax, S63845, and / or AMG 176.

6. The method of claim 1 , wherein the downregulator of an anti-apoptotic protein is an inhibitor of MCL-1.

7. The method of claim 6, wherein the inhibitor of MCL-1 is AMG176, AMG397, ABBV-467, and / or PRT1419.

8. The method of claim 1, wherein the administering comprises administering both a BAX activator and a BH3 mimetic.

9. The method of claim 1 , wherein the subject is treated with one or more anti- PD-1 / L1 antibodies, and / or an anti-CTLA-4 agent, as anti-melanoma therapy.

10. The method of claim 9, wherein the activator of a pro-apoptotic protein and / or a downregulator of an anti-apoptotic protein is administered concomitantly with, prior to, and / or subsequent to the administering of the one or more anti-PD-1 / L1 antibodies, and / or an anti-CTLA-4 agent.

11. A method of preventing or inhibiting acquired resistance to anti-melanoma therapy in a subject in need thereof, the method comprising administering to the subject an effective amount of an activator of a pro-apoptotic protein and / or a downregulator of an anti- apoptotic protein.

12. The method of claim 11 , wherein the activator of a pro-apoptotic protein is a BAX activator.

13. The method of claim 12, wherein the BAX activator is BTSA1.2.

14. The method of claim 11 , wherein the downregulator of an anti-apoptotic protein is a BH3 mimetic.

15. The method of claim 14, wherein the BH3 mimetic is Venetoclax, Navitoclax, S63845, and / or AMG 176.

16. The method of claim 11 , wherein the downregulator of an anti-apoptotic protein is an inhibitor of MCL-1.

17. The method of claim 16, wherein the inhibitor of MCL-1 is AMG176, AMG397, ABBV-467, and / or PRT1419.

18. The method of claim 11 , wherein the administering comprises administering both a BAX activator and a BH3 mimetic.

19. The method of claim 11 , wherein the subject is treated with one or more anti- PD-1 / L1 antibodies, and / or an anti-CTLA-4 agent, as anti-melanoma therapy.

20. The method of claim 19, wherein the activator of a pro-apoptotic protein and / or a downregulator of an anti-apoptotic protein is administered concomitantly with, prior to, and / or subsequent to the administering of the one or more anti-PD-1 / L1 antibodies, and / or an anti-CTLA-4 agent.

Images