See article, p. 935

  • Adoptive transfer of MART1-specific T cells induced tumor regression in a patient with melanoma.

  • Inflammation-induced tumor cell dedifferentiation resulted in loss of the tumor antigen and relapse.

  • Inflammation-induced differentiation may be a mechanism of resistance to adoptive T-cell therapies.


Immunotherapy with adoptive T-cell transfer (ACT) achieves responses in patients with melanoma, but most patients develop resistance and eventually relapse. Proposed resistance mechanisms include T-cell exhaustion or immune tolerance, and selection for T cells with genetic alterations that increase proliferation and result in loss of the target antigen. Recently an alternative resistance mechanism has been observed in mice: inflammation-induced dedifferentiation of tumor cells to precursor cells of neural crest origin results in loss of the tumor antigen. Mehta, Kim, and colleagues observed this resistance mechanism in a patient with metastatic melanoma who received MART1-specific ACT. After initial tumor regression and 3 months of progression-free survival, the patient relapsed. At tumor progression, a biopsy revealed loss of MART1 expression as well loss of the melanocytic antigens gp100 and tyrosinase, indicative of a phenotypic switch instead of selective loss of MART1. The relapsed tumor exhibited increased CD8+ cytotoxic T-lymphocyte infiltration, consistent with increased inflammatory signals in the tumor microenvironment, and expression of inflammation-induced dedifferentiation markers including NGFR was increased. Further, coculturing F5 TCR transduced T cells with human melanoma cell lines was sufficient to induce their dedifferentiation in vitro, inducing downregulation of MART1 and upregulation of NGFR. This effect was mediated by TNFα released by T cells, and TNFα alone was sufficient to induce reversible dedifferentiation of melanoma cells. This case study demonstrates that inflammation-induced dedifferentiation of tumor cells may promote resistance to ACT in patients with melanoma.

See article, p. 944

  • Adoptively transferred NY-ESO-1c259 cells showed antitumor activity in patients with synovial sarcoma.

  • NY-ESO-1c259 cells were identified in the peripheral blood 6 months post-infusion.

  • Persistent NY-ESO-1c259 cells were largely functional central memory and stem cell memory T cells.


NY-ESO-1 is an immunogenic tumor antigen that is highly expressed in cancers such as melanoma and synovial sarcoma, and NY-ESO-1+ autologous T-cell therapy has previously been shown to produce objective responses in 11 of 18 patients with synovial sarcoma. D'Angelo and colleagues sought to identify biological correlates and further evaluate the safety and feasibility of NY-ESO-1+ autologous T-cell therapy in the initial cohort of a pilot trial of patients with synovial sarcoma. In a total of 12 patients, the overall response rate was 50% (1 confirmed complete response, 5 confirmed partial responses) with a median duration of response of 34 weeks, median progression-free survival of 15 weeks, estimated median overall survival of 120 weeks, and no fatal adverse events (low-grade cytokine release storm occurred in 5/12 patients). Interrogation of the T-cell subsets demonstrated the expansion, long-term persistence, and functionality of NY-ESO1+ T cells in responding patients; further, long-term persistent T cells were derived from all T-cell subsets (from naïve to differentiated memory T cells) and enriched for memory T cells. TCRBV sequencing of the apheresis CD8+ T cells and NY-ESO1+ CD8+ T cells in responding patients further showed that persistent clones were derived from a preexisting pool of NY-ESO1–experienced stem cell/central T cells. These results confirm the efficacy of autologous T-cell transfer therapy in a solid tumor and provide in-depth characterization of persistent T cells in a clinical context.

See article, p. 958

  • CNS-produced cytokines may drive CAR T cell–associated neurotoxicity independently of CAR T cells.

  • CAR T cell–associated toxicities are CAR T cell construct–indepen-dent.

  • CAR T cell–induced neurotoxicity is associated with elevated inflammatory cytokines and BBB disruption.


The success of autologous chimeric antigen receptor (CAR) T-cell therapy in patients with hematologic malignancies, particularly CD19+ B-cell acute lymphoblastic leukemia (B-ALL), has been partially mitigated by the development of severe side effects, such as neurotoxicity and cytokine release syndrome, in a subset of patients. Although recent studies have identified potential biomarkers for the development of these significant toxicities, particularly for the CD19 CAR 19-41BB T-cell product, currently it is unknown whether these biomarkers and cytotoxicities are CAR construct–specific. To identify biomarkers predictive of CD19 CAR 19-28ζ T cell–associated neurotoxicity, Santomasso, Park, Salloum, and colleagues characterized the neurologic symptoms and blood of 53 adult patients with B-ALL treated with a CD19 CAR 19-28ζ T-cell product. Although no patients developed fatal neurotoxicity, 22 patients developed severe neurotoxicity, which was associated with high pretreatment disease burden, higher peak CAR T-cell expansion, early and elevated systemic inflammation, blood–brain barrier (BBB) disruption, elevated production of central nervous system (CNS)–specific cytokines such as IL6, IL8, and MCP1, and elevated levels of NMDA receptor agonists in cerebrospinal fluid (CSF). Further, radiographic changes were identified in 4 of 14 patients who underwent MRI imaging during acute neurotoxicity symptoms, and low platelet count and temperature ≥38°C on Day 3 post-treatment were predictive of the development of severe neurotoxicity. Together, these findings identify potential biomarkers of severe neurotoxicity in patients with B-ALL who receive CD19 CAR T-cell therapy.

See article, p. 972

  • Engineered SmarT cells recognizedistinct antigens for activation, costimulation, and cytokine support.

  • SmarT cells targeting PSCA, TGFβ, and IL4 induce tumor regression and remain active upon rechallenge.

  • SmarT cells may limit the on-target, off-tumor toxicity of CAR T therapies to broaden their use.


Adoptive transfer of chimeric antigen receptor (CAR)–modified T cells can elicit antitumor responses, but few tumor antigens are absolutely tumor-specific. Thus, CAR T-cell therapy often induces “on-target, off-tumor” toxicities, which can sometimes be life-threatening and limit the clinical use of CAR T-cell therapies. To this end, Sukumaran and colleagues developed tumor-specific molecule-pattern activated and regulated T cells (SmarT cells) to limit on-target, off-tumor toxicity for CAR T cells with enhanced potency, safety, and tumor cell selectivity. These SmarT cells were designed to recognize three independent pancreatic tumor antigens: prostate stem cell antigen (PSCA), transforming growth factor β (TGFβ), and interleukin 4 (IL4). These three antigens each delivered independent intracellular signals to the engineered T cells with PSCA triggering activation (TCRζ chain) for tumor recognition, TGFβ triggering costimulation (41BB) for T-cell survival, and IL4 triggering cytokine support (IL7) for T-cell proliferation. In vivo, in pancreatic cancer xenografts expressing PSCA, TGFβ, and IL4, SmarT cells expanded, induced tumor regression, and remained active upon tumor rechallenge. As expected, SmarT cells failed to expand in tumors expressing only PSCA. These findings indicate that engineered SmarT cells, designed to target multiple tumor antigens, may enhance CAR T-cell selectivity to limit their on-target, off-tumor toxicity and maximize their activity at the tumor site. This strategy may be used with multiple tumor antigens across tumor types, to potentially extend the clinical use of CAR T-cell therapy.

See article, p. 988

  • The mono-ADP-ribosylhydrolase gene MACROD2 is frequently deleted in human colorectal cancer.

  • Deletion of MACROD2 promotes intestinal tumor growth in a haploinsufficient manner.

  • MACROD2 loss impairs DNA repair and drives chromosome instability by repressing PARP1 activity.


MACROD2 encodes a mono-ADP-ribosylhydrolase required for regulation of PARP1 activity in the DNA damage response and has been found to undergo focal deletions in cancer, suggesting a potential role for somatic genomic alterations of MACROD2 in tumorigenesis. Using comprehensive DNA copy-number analysis, Sakthianandeswaren, Parsons, Mouradov, and colleagues identified recurrent focal alterations of the MACROD2 locus, including heterozygous and homozygous deletions often involving the catalytic macrodomain, in approximately 30% of human colorectal cancers, consistent with previous studies. Deletion of MACROD2 enhanced the growth of intestinal tumors both in the ApcMin/+ mouse model and human colorectal cancer xenografts in a gene dosage–dependent manner, in line with a potential tumor-suppressor function. MACROD2 proteins with in-frame deletions of the macrodomain showed loss of recruitment to sites of DNA damage. In addition, heterozygous or homozygous loss of MACROD2 suppressed PARP1 transferase activity, resulting in defective homologous recombination–mediated DNA repair and increased sensitivity to genotoxic stress–induced DNA damage. Furthermore, MACROD2 haploinsufficiency resulted in increased structural and numerical chromosome abnormalities and defects in mitotic chromosome segregation, including anaphase bridges, lagging chromosomes, and centrosome amplification, phenocopying the effects of PARP1 inhibitor treatment and supporting a role for MACROD2 loss in promoting chromosome instability and aneuploidy. These findings identify MACROD2 as a haploinsufficient tumor suppressor required for the maintenance of genomic integrity in human colorectal cancer, and suggest that this function is mediated at least in part via regulation of PARP1 activity.

See article, p. 1006

  • FATP1 is overexpressed in melanoma and promotes lipid transfer from adipocytes to melanoma cells.

  • FATP1 accelerates tumorigenesis and enhances lipid uptake in a zebrafish model of BRAFV600E melanoma.

  • Adipocyte–melanoma cell cross-talk supports tumor growth and may be targeted with FATP inhibition.


Advanced metastatic melanomas often occur in adipocyte-rich subcutaneous tissue, but the role of adipocytes in melanoma progression is not well understood, prompting Zhang and colleagues to investigate adipocyte–melanoma cell cross-talk. Histologic analysis of advanced melanomas revealed that tumor-adjacent adipocytes were smaller than those farther from the tumor, indicative of tumor-induced lipolysis. In an adipocyte–melanoma co-culture system, adipocytes enhanced melanoma cell proliferation and invasion. Further, adipocytes increased the lipid content of melanoma cells, with an increase in the number and size of lipid droplets. Consistent with these findings, in a zebrafish model of melanoma, adipocytes increased the lipid content of melanoma cells, and subcutaneous metastases grew next to adipocytes and exhibited dysregulation of lipid gene expression. Mechanistically, long-chain fatty acids were released by adipocytes into the extracellular milieu and were taken up by melanoma cells via FATP/SLC27A lipid transporter proteins. This resulted in an increased dependence on extrinsic lipid uptake and β-oxidation and a decreased dependence on de novo lipogenesis. Specifically, FATP1 was upregulated in primary and metastatic melanoma samples, and was associated with increased long-chain fatty acid uptake and poor survival. In a zebrafish model exhibiting melanocyte-specific BRAFV600E expression, concurrent FATP1 expression accelerated melanoma development, and treatment with a small-molecule FATP inhibitor decreased the tumor lipid content and suppressed tumor growth. Collectively, these findings reveal an adipocyte–melanoma cell cross-talk that supports tumor growth, and suggest the potential for therapeutic targeting of FATP.

See article, p. 1026

  • YAP–TEAD upregulates the activin receptor, enhancing SMAD/TGFβ signaling to promote Treg generation.

  • Inhibiting YAP or activin reduces the frequency of FOXP3+ Tregs to suppress melanoma growth in vivo.

  • Immunotherapeutic targeting of YAP or activin may suppress Tregs to enhance antitumor immunity.


Regulatory T cells (Treg) can suppress the antitumor immune response, suggesting the potential for immunotherapy targeting pathways that support Treg function. The transcription factor FOXP3 is required for Treg function, but the mechanisms regulating FOXP3 expression and Treg activity are not well understood. Ni, Tao, Barbi, and colleagues found that the transcriptional coactivator YAP is highly expressed in Tregs and supports their function. In mice with T cell–specific Yap deletion, challenge with B16 melanoma did not result in tumor growth, in comparison to wild-type mice who exhibited rapid tumor growth and restrained activation of CD4+ and CD8+ T cells, and the YAP-deficient cells exhibited higher levels of IFNγ and TNFα production, indicative of a more robust antitumor immune response. Further, Yap deletion reduced the number of FOXP3+ Tregs in the tumor microenvironment. Treating melanoma-bearing mice with the YAP inhibitor verteporfin reduced tumor growth, and combination therapy with anti–PD-1, the cancer vaccine GM-Vac, and verteporfin resulted in a dramatic suppression of tumor growth. Mechanistically, YAP interacted with the transcription factor TEAD to directly bind the activin receptor (AcVR1C) promoter and increase its expression. Activin then enhanced SMAD/TGFβ signaling to promote Treg differentiation. An anti-activin antibody reduced the frequency of FOXP3+ Tregs and suppressed melanoma growth in vivo, with enhanced activity in combination with GM-Vac. Altogether, these findings demonstrate that YAP-mediated activin signaling supports the generation of Tregs that suppress the antitumor immunity, suggesting the potential for immunotherapies targeting YAP or activin.

Note:In This Issue is written by Cancer Discovery editorial staff. Readers are encouraged to consult the original articles for full details.