Vitamin E Enhances Cancer Immunotherapy by Reinvigorating Dendritic Cells via Targeting Checkpoint SHP1

Dietary supplements are commonly used in general populations (1) and patients with cancer (2). The use of dietary supplements during conventional cancer therapy has brought growing attention as an eminently practical approach (3). However, little is known regarding the effects of dietary components and nutritional supplements on current prevalent immunotherapies. The link between nutritional status and immune homeostasis (4) suggests that dietary interventions may affect antitumor immunity and the efficacy of cancer immunotherapies. Thus, identifying beneficial dietary supplements and clarifying the specific mechanism of functions in enhancing cancer immunotherapy can guide clinical application in patients with cancer.

T cell–based immune-checkpoint therapies (ICT), such as anti-CTLA4, anti–PD-1, and anti–PD-L1, have demonstrated striking clinical success in reviving dysfunctional tumor-infiltrating effector T cells and providing long-lasting protection in a subset of patients with cancer (5). However, ICT is still limited by low clinical response rates and substantial side effects (6). Thus, maximizing anticancer therapeutic effects and minimizing the toxicities of ICT is an unmet clinical challenge (6). A successful anticancer immune response requires a series of stepwise events (7), of which immunogenic antigen presentation in tumors is a key determinant. Dendritic cells (DC) take up and cross-present tumor antigens to prime and activate cytotoxic T lymphocytes for antigen-specific T-cell response (8). However, tumor-infiltrating DCs often become profoundly dysfunctional or tolerogenic for inducing a potent immune response in the suppressive tumor microenvironment (TME; refs. 8, 9), limiting the efficacy of T cell–based therapies. Activating DCs or overcoming DC-suppressive signals can potentiate T-cell antitumor response and enhance the therapeutic effects of ICT and other immunotherapies. Unfortunately, there is no clinically effective strategy to reinvigorate dysfunctional DCs at present. Improved understanding of regulatory mechanisms of DC function and maneuvering DCs accordingly may enable therapeutic development in various clinical settings.

In this study, combining analysis of electronic health records (EHR) of ICT-treated patients with cancer and experimental data from various mouse cancer models, we uncovered that supplementary vitamin E (VitE) increased ICT response. VitE is an abundantly used dietary supplement among adults (1, 10) including patients with cancer (11). As antioxidants, VitE supplements have been implicated in strengthening immunity with anti-inflammatory, anti-atherosclerotic, and antitumor effects in animal models (12). However, the benefits of VitE supplements on patients with cancer are unclear, because its clinical studies showed mixed efficacies (13, 14). Importantly, the mechanisms underlying the disease-modifying activities of VitE remain obscure. Here, we identified that VitE boosted antitumor immunity by mainly enhancing the tumor antigen presentation of DCs and DC-derived extracellular vesicles (DC-EV) via inhibiting SHP1, a DC-intrinsic checkpoint. Compellingly, unleashing DCs and DC-EVs by VitE is an immediately applicable and advantageous strategy for inducing efficient antitumor immunity and enhancing immunotherapy response.

To assess the impact of various dietary supplements on immunotherapy responses, we retrospectively analyzed EHR for the clinical outcomes of patients with cancer receiving immunotherapies (anti–PD-1/PD-L1) at The University of Texas MD Anderson Cancer Center (MDACC) following a rational patient selection process (Supplementary Fig. S1). We found that patients with melanoma who took VitE during anti–PD-1/PD-L1 treatments had a reduced death rate compared with patients taking any of the 14 other common dietary supplements (Supplementary Fig. S2A). Strikingly, patients with melanoma who took VitE while receiving ICT showed a significantly (HR, 0.7; 95% CI, 0.53–0.92; P < 0.05) improved overall survival (OS) compared with those who did not take vitamins (Fig. 1A; Supplementary Table S1) or took multivitamins (Supplementary Fig. S2B) from univariate survival analyses. Also, patients who took VitE showed a trend of better responses to ICT treatment than those who did not take vitamins (Supplementary Fig. S2C). Multivariate analyses confirmed that VitE uptake was significantly associated with OS independent of age, gender, and race (Supplementary Table S2). Likewise, significantly increased OS was found in a mixed cohort of ICT-treated patients with cancer (breast, colon, and kidney cancers) who took VitE (HR, 0.74; 95% CI, 0.57–0.95; P < 0.05; Fig. 1B; Supplementary Tables S3–S6). Furthermore, in another independent mixed cohort of patients with breast, lung, kidney, and bladder cancers from a different hospital, taking VitE while receiving ICT treatment was also associated with a reduced death rate compared with patients not taking VitE (Supplementary Table S7). In contrast, VitE had minimal or no effect on reducing the death rate of chemotherapy-treated (taxol or doxorubicin) patients with cancer (Supplementary Fig. S2D). These clinical data suggest that VitE supplementation improved the clinical outcomes in patients with cancer with ICT treatment.

These clinical findings prompted us to test whether dietary VitE may potentiate ICT efficacy in animal models. Consistent with the clinical data, VitE and ICT combination treatment achieved significantly improved ICT response in mice bearing the EMT6 mammary tumors (Fig. 1C; Supplementary Fig. S2E). Similarly, VitE and ICT combination treatment was significantly more effective in inhibiting B16-GMCSF melanoma growth, reducing lung metastasis, and increasing tumor infiltration of granzyme B (GZMB)⁺ and IFNγ⁺ CD8⁺ cytolytic T cells compared with ICT alone (Fig. 1D; Supplementary Fig. S2F–S2J). Together, VitE supplementation correlated with better ICT response in patients with cancer and enhanced ICT response in mouse tumor models.

When testing in mice bearing the B16-F10 melanoma, we were perplexed that VitE and ICT combination treatments showed no therapeutic effect (Fig. 1E), unlike in the B16-GMCSF, a B16-F10 subline overexpressing granulocyte–macrophage colony-stimulating factor (GM-CSF) that is positively associated with DC infiltration in tumors (ref. 15; Supplementary Fig. S3A–S3C). Likewise, growth factor FMS-like tyrosine kinase 3 ligand (FLT3L) promotes hematopoietic progenitor commitment to the DC lineage, thus increasing DC survival and proliferation (16). In mice bearing the B16-F10 subline overexpressing FLT3L (B16-FLT3L), VitE effectively enhanced their ICT responses (Fig. 1F). The futility in DC-desert tumors (B16-F10) versus the strong activity in DC-enriched tumors (B16-GMCSF and B16-FLT3L; Supplementary Fig. S3A and S3B) suggested that the enhanced ICT response by VitE depends on DC infiltration. Indeed, depletion of DCs by cytochrome c (Cyt c; ref. 17; Supplementary Fig. S3D and S3E) abolished the VitE-augmented ICT response in the B16-GMCSF model (Fig. 1G), but had no effect in the B16-F10 model (Supplementary Fig. S3F). Furthermore, DC depletion by Cyt c also abolished the inhibition of EMT6 tumors by VitE (Supplementary Fig. S3G and S3H), confirming the DC dependency of VitE-mediated tumor inhibition effect across multiple immunogenic tumor models.

To directly assess VitE's effect on DCs, we generated human monocyte-derived DCs (MoDC) from healthy donors and treated MoDCs with VitE, other nutritional supplements, or vehicle. Compared with other common nutritional supplements, VitE most effectively activated DCs (Fig. 2A; Supplementary Fig. S4A), as indicated by the significantly increased expression of DC activation markers CD40, CD86, and HLA-DR (Fig. 2B). Notably, only natural VitE (DL-α-tocopherol) further enhanced DC activation upon lipopolysaccharide (LPS) stimulation, but not synthesized VitE acetate (VEA) nor antioxidant N-acetylcysteine (NAC; Supplementary Fig. S4B), suggesting that general antioxidant function probably does not contribute to VitE-induced DC activation. VitE was also the most effective supplement that activated mouse bone marrow–derived DCs (BMDC; Supplementary Fig. S4C). Functionally, VitE-treated human MoDCs increased autologous IFNγ⁺CD8⁺ cytotoxic T cells (Fig. 2C) and VitE-treated mouse BMDCs promoted the autologous T-cell proliferation in cocultures (Supplementary Fig. S4D). Furthermore, VitE-treated and ovalbumin (OVA)-loaded BMDCs increased the OVA-derived SIINFEKL antigen-specific OT-1 T cells in coculture (Fig. 2D).

To decipher VitE's effect on DCs and antitumor immunity in vivo, we used the EMT6 orthotopic mouse model. VitE treatment reduced tumor growth compared with the vehicle-treated group in vivo (Supplementary Fig. S4E and S4F) without a direct effect on tumor cell proliferation in culture (Supplementary Fig. S4G). Cytometry by time-of-flight (CyTOF) analysis of CD45⁺ immune cells isolated from EMT6 tumors showed that VitE treatment increased the tumor infiltrations of CD11c⁺ DCs and CD8⁺ T cells (Fig. 2E). Further analyses of CD11c⁺ DCs from EMT6 tumors showed that VitE treatment induced a striking component shift of tumor-associated DCs (TADC) from naïve MoDCs (CD14⁺CD206⁺) and plasmacytoid DCs (pDC; CD11c^loB220⁺) to conventional DCs (CD103⁺ cDC1 and CD11b⁺ cDC2; Fig. 2F; Supplementary Fig. S4H–S4J), which are critical for tumor-reactive T-cell responses (8), indicating that VitE treatment instigated a functional differentiation of the DC compartments and induced polarization to cross-presenting DCs. Consistently, the majority of VitE-induced DCs exhibited the characteristics of activated DCs (CD80⁺CD86⁺MHC-II⁺Ki-67⁺; Fig. 2G). Furthermore, expression of genes associated with DC activation and effector function (Cd40, Ccr7, and Ifnb1) was increased in TADCs isolated from VitE-treated EMT6 tumors versus controls, along with increased cytotoxic IFNγ⁺CD8⁺ T-cell infiltration (Supplementary Fig. S4K and S4L). Significantly, TADCs of VitE-treated tumors promoted T-cell proliferation more than TADCs from vehicle-treated tumors (Fig. 2H). Therefore, VitE-induced enhancement of DC function could adequately trigger antitumor T-cell immunity in vivo.

To examine the association between DC activation and antitumor T-cell immunity in patients, we performed bioinformatics analysis of human breast cancer data in The Cancer Genome Atlas. We discovered that activated DCs (aDC; gene signature defined by CD83, CCR7, LAMP3, CCL19, MARCKSL1, and IDO1), but not total DCs or other immune cells in the TME, were most tightly correlated with CD8⁺ T-cell infiltration and intratumoral immune cytolytic activity (CYT; Supplementary Fig. S5A and S5B). Higher aDCs, but not total DCs, also significantly correlated with longer patient survival in multiple cancer types (Supplementary Fig. S5C). Moreover, pembrolizumab (anti–PD-1)-treated patients with melanoma with higher tumor-infiltrating aDCs had markedly increased OS compared with patients with lower aDC infiltration (PRJEB23709; ref. 18; Fig. 2I). These findings were corroborated in separate cohorts of patients with melanoma (19) and bladder cancer (20) treated with anti-CTLA4 or anti–PD-L1 antibodies, respectively (Supplementary Fig. S5D and S5E). Together, the effect of VitE on DC activation and the profound association of aDCs with patients’ ICT responses indicate that VitE-induced DC activation likely contributes to enhanced ICT efficacy.

In investigating how VitE activates DCs, our reverse-phase protein array (RPPA) analyses on VitE-treated DCs without or with tumor-conditioned medium (TCM) revealed that VitE induced major changes in DCs: (i) increased phosphorylation of several kinases, for example, p-P38(T180/Y182), pERK(T202/Y204), and p-c-JUN(S73); and (ii) decreased phosphorylation at tyrosine 564 (Y564) of SHP1 (Fig. 3A; Supplementary Fig. S6A). SHP1 is a cellular tyrosine phosphatase encoded by the Ptpn6 gene and can reduce phosphorylation of SYK and its downstream MAPKs, for example, pERK1/2 and p-c-JUN (21, 22). SHP1 is a DC-intrinsic checkpoint (22) that regulates multiple Toll-like receptors (TLR) and cytokine sig­naling responses (23). The Y564 phosphorylation on SHP1 is critical for its phosphatase activity (24). We validated the decrease of Y564-phosphorylated SHP1 (pY564-SHP1) in VitE-treated mouse BMDCs (without or with TCM; Supplementary Fig. S6B and S6C) and human MoDCs (Supplementary Fig. S6D). The pY564-SHP1 levels were markedly higher in TADCs than in splenic DCs of EMT6 tumor–bearing mice (Fig. 3B), not in intratumoral myeloid cells and T cells (Supplementary Fig. S6E). There was no significant change for the pY542-SHP2 level between TADCs and splenic DCs (Supplementary Fig. S6E). Conversely, VitE treatment reduced the pY564-SHP1 level in TADCs (Fig. 3C) and increased MAPK pathway activation, correspondingly (Supplementary Fig. S6F). SHP1 expression was undetectable in tested tumor cells (Supplementary Fig. S6G). Silencing SHP1 expression in human MoDCs by PTPN6 siRNA (siPTPN6) led to increased DC activation compared with control siRNA (siCtrl)–transfected MoDCs (Fig. 3D). Mouse siPtpn6-BMDCs induced cocultured OT-1 T-cell proliferation significantly more than siCtrl-BMDCs (Fig. 3E), similar to VitE's effect on DC-primed T-cell proliferation (Fig. 2D). Indeed, VitE-treated siCtrl-BMDCs significantly induced cocultured OT-1 cell proliferation compared with vehicle treatment, while siPtpn6-BMDCs did not further respond to VitE treatment (Fig. 3F), indicating that VitE treatment enhanced DC function by inhibiting SHP1.

Deleting the SHP1-encoding Ptpn6 gene by CRISPR/Cas9 in the murine DC cell line DC2.4 (sgPtpn6-DC2.4) also activated DC2.4 cells (Supplementary Fig. S7A and S7B) and increased phosphorylation of ERK and p38 MAPK following LPS stimulation compared with the control sgCtrl-DC2.4 cells (Supplementary Fig. S7C). Functionally, OVA antigen-loaded sgPtpn6-DC2.4 cells increased antigen-specific OT-1 T-cell proliferation (without or with TCM) compared with sgCtrl-DC2.4 cells (Fig. 3G and H; Supplementary Fig. S7D). Analogous to VitE's effect on BMDCs (Fig. 3F), VitE-treated sgCtrl-DC2.4 cells, but not sgPtpn6-DC2.4 cells, significantly induced cocultured OT-1 T-cell proliferation compared with vehicle-treated DCs (Fig. 3G and H), revalidating that SHP1 inhibition was critical for VitE-mediated DC2.4 cell functional activation.

When the sgCtrl-DC2.4 or sgPtpn6-DC2.4 cells were subcutaneously cotransplanted with B16-GMCSF cells into mice, the sgCtrl-DC2.4 cells showed mild antitumor effects, whereas the sgPtpn6-DC2.4 cells highly significantly inhibited tumor growth (Fig. 3I; Supplementary Fig. S7E). Furthermore, B16-GMCSF cells were coimplanted with sgCtrl-DC2.4 into the right flank and with sgPtpn6-DC2.4 cells into the left flank of each mouse. Strikingly, tumors at both flanks regressed similarly (Supplementary Fig. S7F), indicating a systemic antitumor immune response triggered by Ptpn6-deleted DCs. Notably, VitE treatment further reduced the growth of sgCtrl-DC2.4 coimplanted B16-GMCSF tumors in mice but did not further inhibit sgPtpn6-DC2.4 coimplanted B16-GMCSF tumors (Fig. 3J; Supplementary Fig. S7G), indicating that VitE activated DCs by blocking SHP1 to enhance antitumor capacity.

Next, we examined whether VitE may interact with SHP1 in DCs by computational structure analyses. Indeed, molecular docking analysis showed that VitE preferentially docked in the central cavity, at an interface formed by SHP1's N-terminal SH2 (N-SH2) and C-terminal SH2 (C-SH2) domains with the protein tyrosine phosphatase (PTP) domain (Fig. 4A). This region is critical for stabilizing SHP1 autoinhibitory conformation through an N-SH2 domain steric blockade of the phosphatase activation site (25), and docking of VitE in this region indicates that VitE functions as an allosteric inhibitor of SHP1. Specifically, three types of molecular forces are implied in the interaction between folded VitE and all three domains of inactive SHP1: (i) hydrogen bonds with His112 (C-SH2) and Ser250 (PTP); (ii) hydrophobic interactions with the sidechains of His112 (C-SH2), Tyr214 (PTP), and Glu247 (PTP); and (iii) SHP1 intramolecular hydrogen bonds: etheric oxygen from the VitE chroman moiety receives a hydrogen bond from Ser 250 (PTP), which is engaged with Ser107 (N-SH2; Fig. 4B).

Biochemically, our in vitro binding assay demonstrated a direct interaction between purified SHP1 protein and VitE with a dissociation constant of 39 µmol/L (Fig. 4C). Furthermore, we tested the inhibitory function of VitE on the tyro­sine phosphatase activities of SHP1. Clearly, adding different doses of VitE to SHP1 that were immunoprecipitated from LPS-activated DC2.4 cells inhibited the SHP1 phosphatase function (Fig. 4D). Moreover, SHP1 tyrosine phosphatase activities in LPS-activated BMDCs were significantly inhibited by VitE treatment (Fig. 4E). Cumulatively, VitE directly binds SHP1 and inhibits SHP1 phosphatase activities.

VitE can enter enterocytes via the SCARB1 receptor (26). SCARB1 facilitates the uptake of cholesteryl esters from high-density lipoproteins (27). Notably, DCs have the highest SCARB1 expression among all major human immune cell populations (Fig. 4F; Supplementary Fig. S8A). Similarly, mouse DCs, particularly intratumoral conventional DCs (CD103⁺ cDC1 and CD11b⁺ cDC2), which are critical for priming antigen-specific antitumor T-cell immune response, have the highest SCARB1 expression (Supplementary Fig. S8B and S8C). To determine whether SCARB1 is required for VitE-induced DC activation, we generated BMDCs from Scarb1^−/− versus wild-type mice. VitE treatment failed to inhibit pY564-SHP1 in Scarb1^−/− DCs unlike in wild-type DCs (Fig. 4G; Supplementary Fig. S8D and S8E). Also, VitE did not enhance the DC activation in Scarb1^−/− DCs upon TCM stimulation, distinct from VitE-induced activation of wild-type DCs (Supplementary Fig. S8F). Additionally, VitE-treated Scarb1^−/− DCs did not induce proliferation of cocultured T cells, in contrast to VitE-treated wild-type DCs (Fig. 4H). Likewise, pretreatment of DCs with SCARB1-neutralizing antibody (anti-CD36L1) eliminated VitE-induced DC activation (Supplementary Fig. S8G). Taken together, VitE enters DCs via SCARB1 and binds SHP1 to inhibit pSHP1, leading to DC activation, which increases antitumor T-cell immunity (Fig. 4I).

As VitE treatment of OVA antigen–loaded DCs significantly increased cocultured antigen-specific OT-1 T-cell proliferation compared with vehicle treatment in vitro (Fig. 3F–H), we further assessed the effect of VitE on antigen-specific T-cell response in vivo. VitE treatment alone inhibited the growth of B16-GMCSF-OVA melanoma (Fig. 5A), decreased pSHP1 expression in TADCs (Supplementary Fig. S9A), and doubled tumor infiltration of cross-presenting CD103⁺CD11c⁺ DCs (Fig. 5B), which indeed led to increased tumor infiltration of OVA antigen-specific CD8⁺ T cells (Fig. 5C) compared with vehicle treatment. Given the strong antitumor immunity of DCs triggered by inhibiting SHP1 with VitE and by blocking SHP1 genetically (Fig. 3I and J), we reasoned that it could be caused by an increased ability of DC cross-presentation of antigens, leading to better cross-priming of CD8⁺ T cells, as antigen cross-presentation is the most critical function of DCs (28). To investigate whether VitE treatment or SHP1 knockout directly affects DC cross-presentation, we measured the membrane expression level of OVA-derived peptide (SIINFEKL) bound to the H-2K^b of MHC-I (H-2K^b-SIINFEKL) on OVA-loaded DCs after SHP1 knockout or VitE treatment. H-2K^b-SIINFEKL molecules were significantly increased on membranes of sgPtpn6-DC2.4 cells compared with sgCtrl-DC2.4 cells (Fig. 5D). VitE treatment of BMDCs significantly restored their H-2K^b-SIINFEKL level that was reduced by TCM (Fig. 5E; Supplementary Fig. S9B and S9C). Of note, both VitE treatment and SHP1 knockout significantly increased H-2K^b-SIINFEKL molecules on DC2.4 cells compared with vehicle-treated sgCtrl-DC2.4 cells, but VitE treatment did not further increase H-2K^b-SIINFEKL molecules on sgPtpn6-DC2.4 cells (Fig. 5F). Thus, SHP1 inhibition in DCs by VitE treatment or Ptpn6 knockout enhances antigen cross-presentation that boosts CD8⁺ T-cell antitumor immunity.

When exploring the potential mechanism of enhanced cross-presentation in SHP1-inhibited DCs, we observed a remarkable reduction of phagolysosome fusion after silencing SHP1 in BMDCs, as indicated by reduced staining of lysotracker and Lamp1 (Fig. 5G). Reduced phagolysosome fusion indicates a less degradative phagosome that has been tightly linked with an enhanced cross-presentation efficiency (29). Multiple Rab GTPase proteins are reported to regulate phagolysosome trafficking and fusion (30). To examine the SHP1 downstream molecular pathways that are involved in regulating phagolysosome fusion and subsequent cross-presentation, we performed Rab GTPase library mini-screen and identified that overexpression of SHP1 most dramatically decreased Rab34 expression (Fig. 5H), whereas knockdown of SHP1 increased Rab34 expression (Fig. 5I). Furthermore, knockdown SHP1 in mouse BMDCs reduced cytosolic Rab34 but increased membrane accumulation of Rab34 (Fig. 5J), which was reported to decrease phagolysosome fusion resulting in increased cross-presentation efficiency of DCs (29, 31). Consistently, VitE-treated BMDCs also had increased total Rab34 expression (Fig. 5K), as well as increased membrane Rab34 and decreased cytosolic Rab34 proteins (Fig. 5L). Moreover, compared with control BMDCs, silencing Rab34 in BMDCs (siRab34-DCs) reduced the proliferation of antigen-specific OT-1 T cells in coculture (Fig. 5M), signifying the decreased cross-presentation efficiencies of siRab34 DCs. These data indicate that inhibiting SHP1 by VitE or genetic knockdown increases Rab34 membrane accumulation and suppresses phagolysosomal fusion, thereby promoting antigen cross-presentation and T-cell activation.

An efficient antitumor response by DCs requires multifaceted communication modes (32) between DCs and T cells, including direct local intercellular contact and indirect distant communication by extracellular vehicles (EV). DC-EVs carry biomolecules from DCs including membrane peptide–MHC-I (pMHC-I) complexes, which also function in antigen cross-presentation (32). Remarkably, VitE treatment of DCs also increased the level of H-2K^b–SIINFEKL complexes in DC-EVs compared with vehicle treatment (Fig. 6A). VitE-treated DC-EVs had increased binding to CD8⁺ T cells compared with vehicle-treated DC-EVs (Fig. 6B), and the pMHC-I complex was more detected in them (Fig. 6C). Moreover, compared with vehicle-treated DC-EVs, VitE-treated DC-EVs more effectively enhanced T-cell proliferation and IFNγ production (Fig. 6D). Similarly, DC-EVs from sgPtpn6-DC2.4 also enhanced T-cell proliferation (Supplementary Fig. S9D). Thus, DC-EVs from VitE-treated DCs or sgPtpn6-DCs have a greater capacity of antigen presentation relative to the control DC-EVs (Fig. 6E).

To test the effect of DC-EVs from VitE-treated DCs or sgPtpn6 DCs in antitumor immunity and tumor inhibition in vivo, we induced mammary tumors in mice by mammary fat pad injection of OVA-expressing EO771 cells (EO771-OVA) and treated mice with DC-EVs from vehicle- versus VitE-treated BMDCs or from sgCtrl-DC2.4 versus sgPtpn6-DC2.4 cells loaded with OVA. VitE-treated DC-EVs significantly exceeded vehicle-treated DC-EVs in tumor inhibition and increased mouse survival time (Fig. 6F; Supplementary Fig. S9E) and induced a higher CD8⁺/regulatory T-cell ratio and higher tumor infiltration of OVA-specific CD8⁺ T cells (Supplementary Fig. S9F), indicating that VitE-treated DC-EVs elicit a specific antitumor immune response. Similarly, sgPtpn6-DC-EVs significantly inhibited EO771-OVA tumor growth whereas sgCtrl-DC-EVs did not (Supplementary Fig. S9G). Moreover, combining sgPtpn6-DC-EVs with ICT yielded a strong therapeutic response in mice bearing the ICT-resistant EO771-OVA tumors, as indicated by delayed tumor growth, prolonged survival, and reduced lung metastasis (Fig. 6G; Supplementary Fig. S9H–S9K). Consistently, sgPtpn6-DC-EVs alone (Fig. 6H) or in combination with ICT (Fig. 6I) markedly inhibited tumor growth in mice bearing the B16-GMCSF-OVA melanoma (Supplementary Fig. S9L–S9N) and increased tumor infiltration of OVA antigen–specific CD8⁺ T cells (Fig. 6J). Together, EVs from SHP1-inhibited (by VitE treatment or Ptpn6 knockout) DCs offer a promising option to generate an antigen-specific immune response and empower immunotherapy.

Next, we further tested whether VitE may enhance the efficacy of anticancer therapies, especially those that release tumor antigens and enrich DC infiltration, that is, tumor vaccines (33) or immunogenic chemotherapies (34). Mice were inoculated with the GM-CSF–secreting tumor cell vaccine (GVAX) after injection of B16-GMCSF tumor cells and treated with vehicle or VitE (Fig. 7A). Although GVAX + vehicle showed no therapeutic effect, GVAX + VitE led to a significant tumor inhibition (Fig. 7A) and increased tumor-infiltrating cytolytic GZMB⁺CD8⁺ T cells (Supplementary Fig. S10A). Similarly, aggressive 4T1 tumors of a mouse triple-negative breast cancer (TNBC) model are resistant to either GVAX or ICT, and only partly respond to the GVAX + ICT combination; still, VitE treatment combined with GVAX + ICT significantly enhanced the antitumor effect (Fig. 7B; Supplementary Fig. S10B), increased tumor infiltration of CD11c⁺CD40⁺ activated DCs with reduced pSHP1 expression (Supplementary Fig. S10C), along with the increased tumor-infiltrating antigen-experienced CD44⁺CD62L⁻CD8⁺ T cells and tumor-reactive gp70 tetramer–specific CD8⁺ T cells that recognize the 4T1 tumor–associated gp70 antigen (Fig. 7C; Supplementary Fig. S10D).

Clinically, cancer types with low mutation burden are resistant to ICT treatment, and combination with certain immunogenic chemotherapeutics improves ICT efficacy in patients with late-stage cancer (35). Thus, we assessed whether VitE further enhances chemotherapy + ICT antitumor effects in 4T1 tumor–bearing mice treated with low-dose doxorubicin + ICT (Supplementary Fig. S11A). Strikingly, compared with the doxorubicin + ICT group, doxorubicin + ICT + VitE treatment substantially prolonged survival of mice bearing the ICT-resistant 4T1 tumors (Fig. 7D) and inhibited tumor growth and lung metastasis (Supplementary Fig. S11B and S11C). Moreover, adding VitE into the combinatorial treatment (doxorubicin + ICT) increased the infiltration of CD86⁺-activated DCs with reduced pSHP1 expression (Supplementary Fig. S11D and S11E), along with increased tumor-infiltrating gp70 tetramer–specific CD8⁺ T cells (Supplementary Fig. S11F).

Pancreatic ductal adenocarcinoma (PDAC) is one of the most aggressive and lethal types of cancer that is resistant to all standard treatments such as gemcitabine (GEM) chemotherapy and immunotherapies. The HY19636KPC cells, which were derived from the P48-Cre; LSL-Kras^G12D; Trp53^fl/+ (KPC) mice, develop PDAC in mice (36) that recapitulates the extremely aggressive human PDAC phenotype with low immunogenicity and poor response to therapies, for example, GEM or/and ICT (37). Notably, supplementing VitE to the GEM + ICT combinatorial regimen effectively inhibited the HY19636KPC PDAC tumor growth, reduced tumor-induced ascites, and prolonged the survival time of tumor-bearing mice compared with the GEM + ICT + vehicle group (Fig. 7E–G; Supplementary Fig. S12A–S12C). Moreover, GEM + ICT + VitE treatment increased tumor infiltration of CD86⁺-activated DCs with low pSHP1 expression (Supplementary Fig. S12D and S12E). Collectively, VitE greatly enhanced the antitumor effect of tumor vaccines or immunogenic chemotherapies that release tumor antigens and enrich DC infiltration, and ultimately sensitized various highly ICT-resistant tumors to the ICT antitumor effect.

A determining step in the multistep cancer–immunity cycle is eliciting immunogenic antigen presentation in tumors. DCs are critical for the cross-presentation of tumor antigens to activate the T cells. Here, we discovered that natural VitE, a commonly used dietary supplement, activates antigen-specific immune responses and enhances immunotherapy responses by increasing antigen presentation of DCs and DC-EVs via targeting a DC checkpoint, SHP1 (Fig. 7H).

VitE, a fat-soluble vitamin, is required for the proper function of many organs in the body (38). But the potential general anticancer effect of VitE has been investigated for decades with controversial findings (39, 40). Although some early preclinical studies indicated that VitE had antitumor activities (38, 41), large-scale clinical intervention studies have not shown a significant effect (13, 42, 43). Our findings here reveal one possible reason why VitE lacked efficacy in these clinical intervention studies. We found that VitE increases antigen presentation of DCs, and VitE-enhanced anticancer immunity is dependent on DC-induced antigen-specific immune responses. Therefore, VitE supplement may only produce a significant effect in patients with highly immunogenic tumors that are already DC-enriched but would typically be insufficient to induce antitumor immune responses without much tumor antigen release as in these clinical intervention studies. However, when combined with cancer vaccines or immunogenic chemotherapies that induce antigen release and enrich tumor-infiltrating DCs in nonimmunogenic tumors, VitE can indeed potentiate anticancer immune responses and further enhance ICT response. Here, we unveiled the anticancer efficacy of therapies fostering cooperation among nonredundant mechanisms of anticancer immunity: (i) releasing tumor antigens, (ii) increasing antigen presentation of DCs, and (iii) activating T cell–mediated adaptive immune response. It will be important to conduct randomized clinical trials with carefully controlled VitE dosages and other extraneous confounding factors to prospectively test the efficacy of VitE in combination with ICT and/or other immune induction therapies.

The key downstream target of VitE in DCs is SHP1, a DC-intrinsic checkpoint and master regulator of DCs (22). TADCs have markedly elevated SHP1 activity that limits tumor antigen cross-presentation, whereas VitE inhibits SHP1 activity and thus rescues the dysfunctional TADCs. Commonly, VitE is regarded as an antioxidant with anti-inflammatory properties (38). In this study, our computational structural analysis and biochemical assays revealed that VitE, after entering DCs via the SCARB1 receptor, directly binds to SHP1 and functions as an allosteric inhibitor of SHP1 to reinvigorate TADC function. Targeting SHP1 is the key mechanism of VitE-induced DC activation, as knockout of the SHP1 encoding gene Ptpn6 led to DC activation and blunted the effect of VitE on DCs. Thus, SHP1 is an attractive target to effectively activate DCs for the development of potent immunotherapy. The structural insight on the interaction between VitE and SHP1 will guide us to develop more specific allosteric SHP1 inhibitors to effectively target SHP1 for enhancing the functions of DCs.

We found that VitE promotes antitumor T-cell immunity through both direct intercellular contact with DCs and via DC-EVs. DC-EVs also mediate the VitE effect of increasing cross-presentation of tumor antigens. DC-EVs are attractive candidates for drug development and delivery because of their safety, targeting capabilities, and clinical practicality (44). DC-EV–based clinical trials have been conducted in advanced melanomas, yet showed limited efficacy (44). Realistically, the weak antigen presentation of in vitro–generated MoDCs cannot effectively license adaptive T cells for antitumor immunity, which could result in the DC-EVs’ lack of efficacy. Our data showed that DC-EVs derived from SHP1-inhibited (i.e., VitE-treated or Ptpn6 knockout) DCs significantly enhanced their cross-presentation of tumor antigens, triggering potent systemic antigen-specific T-cell antitumor immunity, and facilitated immunotherapy compared with corresponding control DC-EVs, highlighting that SHP1 inhibition fundamentally alters the biological functions of DCs and DC-EVs. Given their unique effector function, adoptive transfer of SHP1-silenced DCs or DC-EVs can be developed as therapeutics for various diseases that demand boosted antigen cross-presentation and activate antigen-specific T-cell immunity for effective treatment.

Based on our findings in this study, we envision that combining VitE with DC-enrichment therapies could be a new strategy to potentially enhance various anticancer immunotherapies. Certain immunogenic chemotherapeutics, radiotherapy, or targeted therapies for cancer treatment trigger immunogenic cell death that promotes antitumor immune responses (45), which depend on DC functions to license cytotoxic T cells. Cancer vaccine therapies release tumor antigens to induce antitumor immune responses, which also depend on DCs to elicit potent T-cell immunity (33). Although these cancer therapies are the major pillars of the cancer treatment powerhouse, they frequently are ineffective. It is conceivable that activating DCs by VitE treatment or SHP1 inhibition and increasing DC-mediated antigen processing and presentation to T cells can yield more effective immune responses and might enhance the efficacies of these anticancer therapies.

In summary, we discovered that dietary VitE supplement enhances antitumor immunity and immunotherapy efficacy largely by a DC-dependent function via inhibiting SHP1, despite that VitE also has other possible antitumor functions (46, 47). Our preclinical findings pave the way for clinical trials to test combinatorial therapies of VitE, or SHP1-depleted DCs/DC-EVs, with antigen-releasing therapies (vaccines, immunogenic chemotherapies) and ICT as promising and efficacious anticancer immunotherapies.

All procedures in studies involving human subjects were performed under the ethical standards of the institutional research committee and the Helsinki declaration or comparable ethical standards following the guidelines approved by the Institutional Review Board (IRB) at MDACC and the Houston Methodist Hospital. In this study, we performed retrospective reviews of the institutional medical databases fed from the electronic health record. To protect patient confidentiality, deidentified patient data were included, and no Protected Health Information was collected. No treatments were given, and no subjects were contacted. It did not involve more than minimal risk to the subjects. The informed consent from the patients was therefore waived, and the request for Waiver of Informed Consent has been approved by the IRB committee. All patient-related data and unique identifiers were removed so that human information was anonymized before any further processing.

Our clinical data mining study was conducted on patients with cancer encountered during 2016 to 2021 at MDACC. The study population consisted of patients with cancer with the International Classification of Diseases code (10th version). Anonymized aggregate-level data were collected using the SlicerDicer function within MD Anderson Epic electronic medical records. Using the Epic SlicerDicer, we identified all the patients with cancer with a visit diagnosis, billing diagnosis, or active problem list with malignant cancers. Epic SlicerDicer was used to further identify patients who received anti–PD-1 antibody (pembrolizumab, nivolumab, or cemiplimab) or anti–PD-L1 antibody (durvalumab, atezolizumab, or avelumab) treatment during the study period. To analyze the impacts of common dietary components and nutritional supplements on patients’ outcomes, the 2,715 patients with melanoma receiving anti–PD-1/PD-L1 treatment during 2016 to 2019 were further divided into 15 patient subgroups based on the supplements they took during ICT immunotherapeutic treatment. The deceased patient percentage of each subgroup was calculated based on the patient number of each subgroup and the average patient deceased rate. Patient information, including age, sex, disease stage, and survival (alive or dead), was extracted. No major selection bias is envisioned in these clinical data collections and analyses. Simultaneously, a similar analysis was performed in 3,967 patients with breast cancer, 475 patients with melanoma, and 323 patients with prostate cancer receiving taxol treatment and in 3,489 patients with breast cancer, 460 patients with melanoma, and 416 patients with prostate cancer with doxorubicin treatment encountered at MDACC during 2016 to 2019. Patient data from the Houston Methodist Hospital during 2016 to 2021 were collected, processed, and analyzed following the same standard. Due to limited patient numbers with the record of dietary supplement administration, only death rate data were collected and analyzed.

For OS analysis and anti–PD-1/PD-L1 therapeutic response analysis, patients encountered during 2016 to 2021 at MDACC were separated into two subgroups: patients who took VitE and patients who took neither VitE nor multivitamins. The survival data were updated to August 31, 2021. In patients with melanoma, patients who only took multivitamins were also separated as a subgroup for survival analysis. Patients who were treated with more than one round of ICTs were excluded from the analysis. Additionally, patients without the date information of starting ICT treatment were also excluded from the survival analysis. To perform patient OS analysis, patient death date and the date starting the ICT treatment were pulled out from Epic. The survival time was calculated as the days between the two dates. The univariate survival analysis and multivariate survival analysis (Cox proportional hazards regression analysis) were performed using the IBM SPSS Statistics version 23 (RRID: SCR_019096). As patients with breast cancer, kidney cancer, or colon cancer shared similar median survival time with limited patient numbers separately, they were combined together as a mixed cohort for OS analysis. Therapeutic response to ICTs was defined as the event/body site (Onc Hx) recorded within 6 months after receiving the ICTs, including progressed disease (PD), complete remission/response (CR), partial remission/mixed response (PR), and stable disease (SD). For therapeutic response analysis, the disease control rate was calculated as the ratio of patients with CR, PR, or SD compared with total patients with available therapeutic response data (patients of CR, PR, SD, plus PD).

C57BL/6J (RRID:IMSR_JAX:000664), BALB/c (RRID:IMSR_JAX:000651), and B6;129S-Scarb1^tm1Kri/J (Scarb1^−/−, IMSR; cat. #JAX:003379, RRID:IMSR_JAX:003379) mice were purchased from The Jackson Laboratory and allowed to acclimatize for at least one week before experimentation. OT-1 T-cell receptor–transgenic mice (C57BL/6-Tg (TcraTcrb)1100Mjb/J, RRID:IMSR_JAX:003831) have a T-cell receptor that recognizes OVA residues 257 to 264 in the context of H2K^b. For all experiments, 8- to 12-week-old mice were matched by age and sex and randomly assigned to specific treatment groups. Animals were bred and maintained in a specific pathogen-free animal facility, and all mouse protocols and experiments were conducted in accordance with an Institutional Animal Care and Use Committee–approved protocol at MDACC (protocol number 00000883) and in compliance with all relevant ethical regulations.

Mouse melanoma B16-F10 cells (ATCC; cat. #CRL-6475, RRID: CVCL_0159) and mouse breast carcinoma cells: EMT6 (ATCC; cat. #CRL-2755, RRID: CVCL_1923), EO771 (ATCC; cat. #CRL-3461, RRID: CVCL_GR23), 4T1 (ATCC; cat. #CRL-2539, RRID: CVCL_0125) were purchased from the ATCC. The murine melanoma cell lines B16-GMCSF (RCB; cat. #RCB1158, RRID: CVCL_L292; ref. 15) and B16-FLT3L (RRID: CVCL_IJ12; ref. 48) were generated as described previously. The hybridoma RMP1-14 for αPD-1 production was provided by Dr. Hideo Yagita at the Juntendo University School of Medicine. HY19636 cells were established from KPC mice (p48-cre⁺Kras^LSL-G12D/+Trp53^lox/+, C57BL/6 background; ref. 36). EO771-OVA and B16-GMCSF-OVA were derived from EO771 and B16-GMCSF, respectively, following the protocol previously described (49), in which tumor cells were transfected with the plasmid pCI-neo-mOVA (RRID: Addgene_25099). Mouse DC DC2.4 (Millipore; cat. #SCC142, RRID: CVCL_J409) was cultured in DC medium (RPMI 1640 containing 2 mmol/L L-glutamine and 25 mmol/L HEPES, supplemented with 10 mmol/L sodium pyruvate, 1% MEM nonessential amino acids, 100 U/mL penicillin/streptomycin, 50 µmol/L 2-mercaptoethanol, and 10% FBS). Details of cell line generation using CRISPR/Cas9 KO, shRNA, and siRNA knockdowns are included in the Supplementary Methods section, and gRNA and shRNA sequences are listed in Supplementary Table S8. These cell lines were authenticated by the Characterized Cell Line Core Facility at MDACC. All cells were used at low passage numbers and were found to be negative for Mycoplasma upon repeated testing every 2 months using the MycoAlert Mycoplasma Detection Kit (Lonza; cat. #LT07-118).

For the mammary tumor models, 2 × 10⁵ EMT6, 4T1, or EO771 cells were orthotopically injected into the mammary fat pads of female BALB/c mice (4T1, EMT6) or C57BL/6J (EO771) on day 0. For the melanoma tumor challenges, 2 × 10⁵ B16-F10, B16-GMCSF, or B16-FLT3L cells were subcutaneously (s.c.) injected into the right flank of C57BL/6 mice on day 0. Syngeneic orthotopic PDAC tumors were established by surgical implantation. Briefly, 2,000 HY19636 KPC cells in 50 µL PBS were injected into the pancreas of C57BL/6J mice.

Treatments were given as single agents or in combination with the indicated regimens for each group. VitE [(+)-α-Tocopherol, type VI, Sigma-Aldrich; cat. #T1539] was formulated in the vehicle (10% ethanol in sterile water) and administered by oral gavage once a day at a dose of 50 mg/kg body weight (100 µL at 10 mg/mL per day), corresponding to 400 to 600 mg/day in a human weighing 60 kg, which is within the range of doses recommended for supplement and pharmacologic use. VitE treatment was initiated on day 4 lasting until the ending point. For ICT experiments, mice were given by intraperitoneal (i.p.) injection of anti-mouse PD-1 antibody (200 μg/mouse, clone RMP1-14, the hybridoma RMP1-14 for αPD-1 production) and/or anti-mouse CTLA4 antibody (100 μg/mouse, Bio X Cell; cat. #BE0164, RRID: AB_10949609) on days 7, 10, 13, and 16 for the indicated tumor models. Rat IgG2a isotype control (200 μg, Bio X Cell; cat. #BE0089, RRID: AB_1107769) and/or mouse IgG2b isotype control (100 μg, Bio X Cell; cat. #BE0086, RRID: AB_1107791), respectively, was used in control mice corresponding to the ICT treatment group. For in vivo DC adoptive transfer experiments, 2 × 10⁵ B16-GMCSF tumor cells mixed with sgCtrl-DC2.4 or sgPtpn6-DC2.4 cells at a ratio of 1:1 was injected subcutaneously into C57BL/6J mice, respectively. For DC-EV treatment experiments, 30 µg DC-EVs were administered intravenously (i.v.) on days 2, 9, and 16 after tumor transplantation.

Where indicated, mice were vaccinated subcutaneously on their contralateral flank with either PBS or 1.0 × 10⁶ 35 Gy-irradiated GM-CSF–secreting B16 cells or GM-CSF–secreting 4T1 cells (GVAX) on days 1, 4, and 7 after tumor injections. For chemo combination treatment, HY19636 KPC tumor-bearing mice were treated with GEM (LC Laboratories; cat. #G-4177) at 25 mg/kg body weight by i.p. injection every 4 days from day 10, and 4T1 TNBC tumor–bearing mice were given with doxorubicin (5 mg/kg body weight; LC Laboratories; cat. #G-4000) by i.v. once a week from day 10.

Tumor size was measured with a digital caliper, and tumor volume was calculated using the formula V  =  (L  ×  W²)/2 and expressed as mm³, where V is tumor volume, L is the length of the tumor (longer diameter), and W is the width of the tumor (shorter diameter). For HY19636 KPC mice, KPC tumors were measured by gross tumor diameter using twice-weekly palpation and external caliper measurement. For MRI, HY19636 KPC-bearing mice were imaged with Bruker 7T small animal Magnetic Resonance Imager (RRID:SCR_019759).

A stock solution of VitE [(+)-α-Tocopherol, type VI, Sigma-Aldrich; cat. #T1539] was prepared by dissolving VitE in absolute ethanol to a final concentration of 100 mg/mL. To optimize cellular uptake, the VitE stock solution was then mixed in FBS at 1 mg/mL and incubated in a 37°C incubator for 1 hour protected from the light with periodic vortexing. VitE was added to the culture media of mouse BMDCs, mouse TADCs, or human MoDCs at 50 µg/mL (116 µmol/L) or at the indicated concentrations for 24 to 48 hours, in which vehicle (0.02% ethanol in medium) as control. This dose of VitE (50 µg/mL) is safe in humans and this treatment mimics the effects of in vivo feeding with VitE (50 mg/kg) in mice and taking a daily VitE supplement in humans. Vehicle- or VitE-treated DCs were washed twice with PBS before continuing to perform antigen presentation assay, coculture with T cells, or conduct the indicated staining assay.

The single cells from mouse tumor tissues were prepared as described previously (50). The cells were incubated with metal-conjugated Ab cocktails against surface and intracellular proteins (Supplementary Table S9). To analyze the phosphorylated signal proteins, single cells isolated from tumor tissues of EMT6-bearing mice with indicated vehicle or VitE treatment were stimulated by LPS (100 ng/mL) for 15 minutes in vitro and then stained with distinct isotope-labeling phosphorylated Ab cocktails after permeabilization. Cell samples were then analyzed by Fluidigm Helios Cell Mass Cytometer (RRID:SCR_019917) in the Flow Cytometry and Cellular Imaging Core Facility, which is supported in part by the NIH through MDACC Support Grant (CA016672) and a Shared Instrumentation Award from the Cancer Prevention Research Institution of Texas (CPRIT; RP121010). Detailed CyTOF sample collection and data analysis are described in the Supplementary Methods.

Detailed protocols for flow cytometry and immunoblotting are described in the Supplementary Methods, and antibodies used in these protocols are listed in Supplementary Tables S10 and S11, respectively.

Statistical analyses were performed with GraphPad Prism v.8.0 (RRID:SCR_002798). Unless mentioned otherwise, all data are presented as mean with SEM. Sample size, error bars, and statistical methods are reported in the figure legends. P values are shown in figures or associated legends. The statistical significance of differences between two experimental groups was assessed by unpaired two-tailed Student t test, and one-way analysis of variance (ANOVA) was performed for comparisons between more than two groups unless indicated otherwise. Pearson correlation analysis was performed for correlation analysis. No statistical methods were used to predetermine sample size in the mouse studies, and mice with matched sex and age were randomized into different treatment groups and, where possible, mixed among cages. For the Kaplan–Meier analyses of clinical data, the log-rank test was used to compare groups. The significance of survival outcomes in each cohort was not affected when adjusted for race/ethnicity, age at the start of ICT in a multivariate analysis. Statistics on patient demographics were conducted by a two-sided Fisher exact test, and related analyses were performed on the IBM SPSS Statistics version 23 (RRID: SCR_019096).

The human clinical data generated in this study are not publicly available due to patient privacy requirements of MDACC and the Houston Methodist Hospital. Other data generated in this study are available within the article and its Supplementary Data files.

X. Yuan reports a patent for 63/252,721 pending. Y. Duan reports a patent for 63/252.721 pending. Y. Xiao reports a patent for 63/252.721 pending. Y. Qi reports a patent for 63/252.721 pending. A.E. Yuzhalin reports grants from the U.S. Department of Defense outside the submitted work. D. Yu reports grants from the NIH and METAvivor Research and Support, Inc. during the conduct of the study, as well as a patent for 63/252.721 pending. No disclosures were reported by the other authors.

X. Yuan: Conceptualization, data curation, formal analysis, validation, investigation, methodology, writing–original draft, writing–review and editing. Y. Duan: Data curation, formal analysis, investigation, methodology, writing–review and editing. Y. Xiao: Data curation, formal analysis, validation, investigation, methodology. K. Sun: Data curation, formal analysis, validation. Y. Qi: Data curation, formal analysis, investigation. Y. Zhang: Data curation, formal analysis, investigation. Z. Ahmed: Data curation, formal analysis, investigation. D. Moiani: Formal analysis, investigation. J. Yao: Data curation, formal analysis. H. Li: Formal analysis, investigation. L. Zhang: Formal analysis, investigation. A.E. Yuzhalin: Writing–review and editing. P. Li: Investigation. C. Zhang: Formal analysis. A. Badu-Nkansah: Writing–review and editing. Y. Saito: Investigation. X. Liu: Investigation. W. Kuo: Methodology. H. Ying: Resources, investigation. S. Sun: Methodology. J.C. Chang: Resources, validation, investigation. J.A. Tainer: Resources, validation, investigation, writing–review and editing. D. Yu: Conceptualization, resources, data curation, formal analysis, supervision, validation, investigation, methodology, writing–original draft, project administration, writing–review and editing.

We thank the NIH Tetramer Core Facility for providing gp70 tetramer and OVA tetramer, Dr. Jedd Wolchok for the B16-GMCSF cell line, Dr. Hideo Yagita for the hybridoma RMP1-14, and Drs. Stephanie Watowich, Chunru Lin, Jianjun Gao, and Florencia McAllister for critical comments on the manuscript. This work was supported by NIH grants R01CA184836 (D. Yu), R01CA208213 (D. Yu), and R01CA231149 (D. Yu), METAVivor research grants #56675 and #58284 (D. Yu), MD Anderson Duncan Family Institute for Cancer Prevention and Risk Assessment (D. Yu), and NIH Cancer Center Support Grant P30CA016672 to MDACC (Flow Cytometry and Cellular Imaging Facility, Functional Genomics Core, Functional Proteomics Core, Research Histology Core, Characterized Cell Line Core, and Research Animal Support Facility-Houston). J.A. Tainer is supported by Cancer Prevention Research Institute of Texas (CPRIT) grant RP180813, NIH grants P01CA092584 and R35CA220430, and the Robert A. Welch Chemistry Chair. D. Yu is the Hubert L. & Olive Stringer Distinguished Chair in Basic Science at MDACC.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.