Mcl-1 Degradation Is Required for Targeted Therapeutics to Eradicate Colon Cancer Cells

Stress-induced apoptosis in mammalian cells is regulated by the Bcl-2 family proteins through a series of orderly events, including mitochondrial outer membrane permeabilization (MOMP), cytosolic release of mitochondrial proteins such as cytochrome c, and activation of caspases (1, 2). Myeloid cell leukemia 1 (Mcl-1) is a prosurvival Bcl-2 family protein frequently overexpressed or amplified in human cancers (3). Mcl-1 inhibits cell death by binding to proapoptotic Bcl-2 family proteins to suppress MOMP and caspase activation (4). A number of recent studies suggest a critical role of Mcl-1 in tumor cell survival and therapeutic resistance (4).

Distinctive from other Bcl-2 family members, Mcl-1 is a very unstable protein, and degradation of Mcl-1 can be triggered by various stresses including anticancer agents (5). Mcl-1 degradation is regulated by its phosphorylation at several sites, leading to subsequent ubiquitination by E3 ligases such as F-box and WD repeat domain-containing 7 (FBW7), Mule, and β-TrCP (6–9). Mcl-1 depletion clearly contributes to apoptosis in hematopoietic cells (10, 11). However, the role of Mcl-1 degradation in solid tumor cells is unclear, as Mcl-1 depletion typically occurs prior to the onset of apoptosis in response to therapeutic treatment, and by itself, is insufficient to trigger cell death (12).

Incorporation of targeted therapy has substantially improved cancer treatment. Most of clinically used targeted drugs are inhibitors of aberrantly activated oncogenic kinases. For example, the multikinase inhibitor regorafenib (Stivarga) and its analogue sorafenib (Nexavar) are used to treat chemotherapy-resistant and metastatic colorectal cancer and other gastrointestinal malignancies (13, 14). Regorafenib and sorafenib inhibit kinases involved in mitogenic signaling and tumor angiogenesis, including CRAF, BRAF, VEGFRs, PDGFR, and c-KIT (15, 16). Our recent study showed that intrinsic and acquired resistance of colorectal cancer cells to regorafenib is associated with blocked Mcl-1 degradation due to inactivating mutations in the tumor suppressor FBW7 (17), suggesting a critical role of Mcl-1 degradation in mediating response to targeted therapy in colorectal cancer cells.

In this study, we used a genetic knock-in (KI) approach to determine the role of Mcl-1 degradation in killing of cancer cells by targeted therapy. Our results indicate that Mcl-1 maintains colorectal cancer cell survival by sequestering the BH3-only Bcl-2 family protein PUMA, which explains why Mcl-1 degradation is necessary, but insufficient for cell death.

Human colorectal cancer cell lines were obtained from the ATCC. FBW7-KO HCT116 cells were obtained from Horizon Discovery. PUMA-KO HCT116 cells were previously described (18). Cell lines were authenticated in 2015 by genotyping, Western blotting of protein expression, and routine detection of mycoplasma contamination by PCR. All cell lines were maintained at 37°C and 5% CO₂ in McCoy 5A modified media (Invitrogen) supplemented with 10% defined FBS (HyClone), 100 U/mL penicillin, and 100 μg/mL streptomycin (Invitrogen). For drug treatment, cells were plated in 12-well plates at 20%–30% density 24 hours before treatment. Chemicals, including regorafenib, sorafenib, UCN-01, sunitinib, roscovitine, sulindac sulfide, etoposide, MG132, ABT-737, ABT-263, SB216763, TW37 (Selleck Chemicals), cycloheximide (Sigma), and UMI-77 (19, 20), were solubilized in DMSO and diluted to appropriate concentrations with the cell culture medium. TRAIL (XcessBio) was diluted with distilled water.

Cells were seeded in 96-well plates at a density of 1 × 10⁴ cells/well. After overnight incubation, various concentrations of agents were added into wells and incubated for additional 72 hours. 3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium (MTS) assay was performed using the MTS assay kit (Promega) according to the manufacturer's instructions. Chemiluminescence was measured with a Wallac Victor 1420 Multilabel Counter (Perkin Elmer). Each assay was conducted in triplicate and repeated three times.

Western blotting was performed as previously described (21). The following antibodies were used: PUMA (18), phospho-Mcl-1 (Ser159/Thr163), cleaved caspases 3, 8, and 9, ERK, phospho-ERK (Thr202/Tyr204), GSK3β, phospho-GSK3β (Ser9), Bad (Cell Signaling Technology), V5, cytochrome oxidase subunit IV (Invitrogen), cytochrome c (Sigma), Bak (Millipore), Bax, HA, Mcl-1 (Santa Cruz Biotechnology), Bcl-2 (Dako), Bim, Bid, Noxa, β-actin (EMD Biosciences), Bcl-X_L (BD Biosciences), and FBW7 (Bethyl).

Adenoviruses expressing PUMA and expression construct of V5-tagged Bcl-X_L are previously described (22, 23). Mcl-1 expression construct was generated by cloning a PCR-amplified full-length human Mcl-1 cDNA fragment into pcDNA3.1/V5-His vector (Invitrogen). Mutations were introduced into Mcl-1 using QuickChange XL Site-Directed Mutagenesis Kit (Agilent Technologies).

Transfection was performed using Lipofectamine 2000 (Invitrogen) according to the manufacturer's instructions. Knockdown was performed 24 hours before regorafenib treatment by transfecting 200 pmol of siRNA for human GSK3β (sc-35527; Santa Cruz Biotechnology), CRAF (AAGCACGCTTAGATTGGAATA-dTdT), Mcl-1 (CGCCGAATTCATTAATTTATT-dTdT), or control scrambled siRNA (GE Dharmacon).

Cells were harvested and suspended in 1 mL of EBC buffer (50 mmol/L Tris-HCl, pH 7.5, 100 mmol/L NaCl, 0.5% Nonidet P-40) supplemented with a protease inhibitor cocktail (Roche Applied Sciences). After disruption of cells by sonication, cell lysates were collected by centrifugation at 10,000 × g for 10 minutes. For immunoprecipitation (IP), 1–2 μg of IP antibodies were mixed with protein G/A-agarose beads (Invitrogen) for 20 minutes at room temperature. The beads were washed twice with PBS containing 0.02% Tween 20 (pH 7.4), incubated with cell lysates on a rocker for 6 hours at room temperature, and then washed thrice with PBS (pH 7.4). Beads were then boiled in 2× Laemmli buffer and subjected to SDS-PAGE and Western blotting.

The Mcl-1–targeting vector was constructed using the pUSER-rAAV (recombinant adeno-associated virus) System. Briefly, two homologous arms of approximately 1 kb each flanking the first intron of Mcl-1 were inserted between 2 USER sites in the AAV shuttle vector pTK-Neo-USER. The coding sequence for the targeted Mcl-1 mutant (S121A/E125A/S159A/T163A) was introduced into the left arm using the QuickChange XL Site-Directed Mutagenesis Kit (Agilent Technologies). For gene targeting, HCT116 cells were infected with the targeting rAAV and selected by G418 (0.5 mg/mL; Corning Mediatech) for 3 weeks. G418-resistant clones were pooled and screened by PCR for targeting events. To target the second allele, Neo flanked by 2 LoxP sites was excised from a heterozygous clone by infection with an adenovirus expressing Cre recombinase (Ad-Cre). The same targeting construct was used in the second round of gene targeting. After the second round, Neo was excised by Ad-Cre infection and targeting was verified by sequencing of genomic DNA and Western blotting.

To identify KI cell lines, genomic DNA was isolated from 5–10 × 10⁴ cells by using ZR-96 Quick-gDNA Kit (ZYMO Research) according to the manufacturer's instructions. One microliter out of 50 μL genomic DNA preparation was used for PCR using previously described conditions (24) and primers listed in Supplementary Table S1. Cycle conditions are available upon request.

For analysis of Mcl-1 mRNA expression, total RNA was isolated from cells using the Mini RNA Isolation II Kit (ZYMO Research) according to the manufacturer's protocol. One microgram of total RNA was used to generate cDNA with SuperScript II reverse transcriptase (Invitrogen). Real-time PCR was performed for Mcl-1 and β-actin using the primer pairs listed in Supplementary Table S1.

Adherent and floating cells were harvested, stained with Hoechst 33258 (Invitrogen), and analyzed for apoptosis by nuclear staining and counting cells with condensed and fragmented nuclei. At least 300 cells were analyzed for each treatment. Annexin V/propidium iodide (PI) staining was performed using Annexin-Alexa Fluor 488 (Invitrogen) and PI as described (24). For colony formation assays, equal numbers of cells were subjected to various treatments and plated into 12-well plates at different dilutions. Colonies were visualized by crystal violet staining 14 days after plating. Each experiment was performed in triplicate and repeated at least twice. For analysis of cytochrome c release, cytoplasmic and mitochondrial fractions were separated by Mitochondrial Fractionation Kit (Active Motif) according to the manufacturer's instructions, followed by Western blotting of cytochrome c in the cytoplasmic and mitochondrial fractions.

All animal experiments were approved by the University of Pittsburgh Institutional Animal Care and Use Committee. Female 5- to 6-week-old Nu/Nu mice (Charles River) were housed in a sterile environment with micro isolator cages and allowed access to water and chow ad libitum. Mice were injected subcutaneously in both flanks with 4 × 10⁶ wild-type (WT) or Mcl-1-KI HCT116 cells. After tumor growth for 7 days, mice were treated daily with regorafenib at 30 mg/kg by oral gavage for 10 consecutive days. Regorafenib was dissolved in Cremephor EL/95% ethanol (50:50) as a 4× stock solution. Tumor size was monitored by calipers, and tumor volumes were calculated according to the formula 0.5 × length × width². Mice were euthanized when tumors reached 1.0 cm³ in size. Tumors were dissected and fixed in 10% formalin and embedded in paraffin. Terminal deoxynucleotidyl transferase–mediated dUTP Nick End Labeling (TUNEL; Millipore), active caspase-3 (Cell Signaling Technology), CD31 (Spring Bioscience), and Carbonic Anhydrase 9 (CA9; Santa Cruz Biotechnology) immunostaining was performed on 5-mm paraffin-embedded tumor sections as previously described (25), with an Alexa Fluor 488–conjugated secondary antibody (Invitrogen) for detection.

Interactions of FBW7/Mcl-1 and PUMA/Mcl-1 were detected by in situ proximity ligation assay (PLA) in paraffin-embedded sections of WT or Mcl-1-KI HCT116 tumor tissues from nude mice treated with or without regorafenib for 1 day. PLA was performed using the Duolink In Situ Redstarter Kit (Sigma) according to the manufacturers' instructions. Incubation with primary antibodies for FBW7 (rabbit, Abcam, 1:100), PUMA (rabbit, Abcam, 1:50), and Mcl-1 (mouse, BD Biosciences, 1:100) was performed at 4°C overnight. The sections were mounted with Vectashield Mounting Medium (Vector Laboratories) with DAPI (Sigma) for nuclear counter staining. PLA signals were visualized by fluorescence microscopy (Olympus) and quantified by counting.

Statistical analyses were performed using GraphPad Prism V software. P values were calculated by the Student t test, and considered significant if P < 0.05. The means + 1 SD was displayed in the figures.

After analyzing over 50 different anticancer agents, we identified several kinase inhibitors that induced depletion of Mcl-1 protein at a concentration that triggered apoptosis in HCT116 colon cancer cells, including FDA-approved drugs regorafenib, sorafenib, and sunitinib, the broad-range kinase inhibitor UCN-01, and the CDK inhibitor roscovitine (Fig. 1A). In contrast, Mcl-1 stability was not affected by other agents such as the death receptor ligand TRAIL, the DNA-damaging agent etoposide, and the nonsteroidal anti-inflammatory drug sulindac sulfide. We found that regorafenib and sorafenib induced Mcl-1 degradation in a dose- and time-dependent manner (Fig. 1B; ref. 17), and in different colorectal cancer cells regardless of their KRAS, BRAF, PIK3CA, or p53 status (Fig. 1C). The half-life of Mcl-1 protein, but not its mRNA expression, was reduced following regorafenib or sorafenib treatment (Fig. 1D; Supplementary Fig. S1A and S1B). Mcl-1 depletion was blocked by the proteasome inhibitor MG 132 (Fig. 1E), suggesting ubiquitin/proteasome–dependent protein degradation.

Stress-induced Mcl-1 degradation is mediated by its phosphorylation (26). Regorafenib treatment promoted Ser159/Thr163 phosphorylation of Mcl-1, coinciding with a sharp decline in Mcl-1 protein (Fig. 2A). Consistent with the requirement of FBW7 for Mcl-1 degradation (17), regorafenib treatment promoted binding of FBW7 with phosphorylated Mcl-1 (p-Mcl-1; Fig. 2B), and ubiquitination of Mcl-1, which was absent in FBW7-knockout (KO) HCT116 cells (Fig. 2C). GSK3β, which is activated by suppression of its inhibitory Ser9 phosphorylation (Fig. 2D), is responsible for the effect of regorafenib on Mcl-1, as regorafenib-induced Mcl-1 phosphorylation, ubiquitination, and degradation was abrogated by the GSK3β inhibitor SB216763 (Fig. 2E–G), or GSK3β knockdown by siRNA (Supplementary Fig. S1C). We then dissected the upstream kinases responsible for GSK3β activation and Mcl-1 degradation. Knockdown of CRAF or inhibition of ERK alone was sufficient to induce GSK3β dephosphorylation and Mcl-1 degradation (Fig. 2H and I), while knockdown of other regorafenib targets, including c-KIT, PDGFR, VEGFR, AKT, and BRAF, did not affect Mcl-1 stability (data not shown). These results suggest that inhibition of CRAF/ERK signaling by regorafenib promotes GSK3β-mediated Mcl-1 phosphorylation, which facilitates its binding to FBW7 and subsequent ubiquitination and proteasomal degradation.

To determine the role and mechanism of Mcl-1 degradation, we sought to specifically inhibit Mcl-1 degradation by blocking its phosphorylation. Previous studies identified 4 phosphorylation sites on Mcl-1, including S121, E125, S159, and T163, which can be phosphorylated by various kinases to promote Mcl-1 degradation in response to different stimuli (8, 9). A transfected quadruple mutant in these 4 sites (S121A/E125A/S159A/T163A) was resistant to regorafenib-induced Mcl-1 ubiquitination and degradation (Supplementary Fig. S1D and S1E), which prompted us to design a targeting strategy to KI this mutant in HCT116 cells (Fig. 3A). After two rounds of homologous recombination, we identified HCT116 cell lines with KI of the Mcl-1 phosphorylation site mutant (Mcl-1-KI; Fig. 3B). KI of the Mcl-1 mutant did not affect its steady-state expression and mitochondrial localization (Fig. 3C and D), but substantially increased its half-life (Fig. 3E). We then compared Mcl-1-KI and wild-type (WT) HCT116 cells for their response to regorafenib and sorafenib. In stark contrast to WT cells, Mcl-1-KI cells almost completely lacked regorafenib-induced Mcl-1 degradation, FBW7 binding, and ubiquitination (Fig. 3F–H), indicating complete dependence of Mcl-1 phosphorylation and ubiquitination on these 4 sites.

Mcl-1-KI cells were found to be highly resistant to regorafenib and sorafenib, showing markedly increased viability (Fig. 4A), enhanced clonogenic survival (Fig. 4B), and reduced apoptosis measured by nuclear fragmentation (Supplementary Fig. S2A), Annexin V/propidium iodide (PI) staining (Supplementary Fig. S2B), and activation of caspases 3, 8, and 9 (Fig. 4C). The Mcl-1 dependence was verified by knockdown of Mcl-1, which restored regorafenib sensitivity in Mcl-1-KI cells (Fig. 4A; Supplementary Fig. S2A). Mcl-1-KI cells were defective in regorafenib-induced mitochondrial events (1, 2), including release of cytochrome c and collapse of MOMP (Fig. 4D; data not shown), and Bax conformational changes and oligomerization (Fig. 4E and F). Inhibiting Mcl-1 upstream GSK3β or knockout of its downstream BAX also suppressed apoptosis induced by regorafenib or sorafenib (Supplementary Fig. S2C–S2E).

Mcl-1-KI cells were defective in Mcl-1 degradation and apoptosis induced by other kinase inhibitors, including UCN-01, sunitinib, and roscovitine, but remained sensitive to the agents that did not induce Mcl-1 degradation (Fig. 5A and B). Interestingly, all of the kinase inhibitors that relied on Mcl-1 degradation were found to induce the expression of the BH3-only Bcl-2 family protein PUMA (Fig. 5C). These kinase inhibitors, including regorafenib, sorafenib, UCN-01, and sunitinib, have been shown to be dependent on PUMA to induce apoptosis in colorectal cancer cells (24, 25, 27, 28). In contrast, regorafenib treatment did not affect the expression of other Bcl-2 family members, except for a slight induction of Bim (Supplementary Fig. S3A).

The concordance of Mcl-1 degradation and PUMA induction prompted us to determine the functional relationship between Mcl-1 and PUMA. Mcl-1 degradation and PUMA induction did not affect each other and had similar impact on regorafenib-induced apoptosis (Supplementary Fig. S3A–S3C), indicating they are two independent events in response to regorafenib. In Mcl-1-KI cells, the degradation-resistant Mcl-1 had markedly enhanced binding to PUMA (Fig. 5D), which led to reduced interaction of Bcl-X_L with PUMA, but increased interaction of Bcl-X_L with Bax (Fig. 5E). Furthermore, treating cells with the Mcl-1 inhibitor UMI-77 (19, 20), or the pan-Bcl-2 inhibitor TW37 (29), almost completely restored regorafenib-induced apoptosis and caspase activation in Mcl-1-KI cells, as well as apoptosis induced by other kinase inhibitors that required Mcl-1 degradation (Fig. 5F and G). In contrast, the Bcl-2/Bcl-X_L inhibitor ABT-737 or ABT-263, which mimics the BH3 domain of Bad that is unable to bind to Mcl-1 (30), was unable to restore apoptosis induced by kinase inhibitors in Mcl-1-KI cells (Supplementary Fig. S3D). Similarly, infecting Mcl-1-KI cells with a low, nonapoptotic dose of PUMA-expressing adenovirus (Ad-PUMA) restored regorafenib-induced apoptosis and caspase activation, while a control adenovirus lacking the BH3 domain of PUMA (Ad-ΔBH3) was unable to do so (Fig. 5H and I; Supplementary Fig. S3E). These results suggest that Mcl-1 degradation is a key event that allows PUMA, upon its induction, to bind to other prosurvival Bcl-2 family proteins such as Bcl-X_L to promote apoptosis, explaining why Mcl-1 degradation is necessary but insufficient for apoptosis induction.

We used xenograft tumors to analyze the in vivo effects of blocking Mcl-1 degradation. The growth inhibitory effect of regorafenib was statistically significant (P < 0.05) in WT tumors, but not (P > 0.05) in Mcl-1-KI tumors (Fig. 6A). Compared with WT tumors, Mcl-1-KI tumors were significantly more resistant to regorafenib (Fig. 6A and B, P < 0.05). Apoptosis analyzed by TUNEL and active caspase-3 staining was significantly reduced in Mcl-1-KI tumors relative to WT tumors (Fig. 6C and D). Staining of tumor vasculature by CD31 revealed an attenuated antiangiogenic effect of regorafenib in Mcl-1-KI tumors (Fig. 6E). The antihypoxic effect of regorafenib analyzed by Carbonic Anhydrase 9 (CA9) staining was also reduced in Mcl-1-KI tumors relative to WT tumors (Fig. 6F). Consistent with the in vitro findings, Mcl-1-KI tumors lacked regorafenib-induced Mcl-1 degradation, but had intact GSK3β dephosphorylation (Fig. 6G). Probing in situ protein–protein interactions by PLA (31) detected enhanced interaction between WT Mcl-1 to FBW7 in WT tumors, and between mutant Mcl-1 and PUMA in Mcl-1-KI tumors (Fig. 6H and I). These results demonstrate a pivotal role of Mcl-1 phosphorylation and degradation in mediating the in vivo antitumor effects of regorafenib.

One of the hallmarks of cancer is resistance to apoptosis, which maintains survival of cells en route to oncogenic transformation (32). Overexpression or amplification of Mcl-1 is one of the most frequent alterations in human cancers (33). Defective Mcl-1 degradation allows tumor cells to evade the fate of death and underlies development of therapeutic resistance. Distinguished from other Bcl-2 family members, Mcl-1 contains two PEST (proline/glutamic acid/serine/threonine) sequences, which are enriched in putative phosphorylation sites that mediate its degradation. Our results from genetic knock-in experiments demonstrate that the previously described 4 phosphorylation sites (S121/E125/S159/T163) are essential for Mcl-1 ubiquitination and degradation induced by inhibition of oncogenic kinases, accounting for most, if not all, of phosphorylation that engages Mcl-1 to bind to the E3 ubiquitin ligase FBW7. Upon GSK3β-mediated phosphorylation, Mcl-1 is likely recruited by FBW7 to the SCF ubiquitin ligase complex composing of FBW7, CUL1, SKP1, and ROC1 (9). This complex then covalently links ubiquitin chains to Mcl-1, leading to its degradation in the 26S proteasome. In addition to GSK3β, several other kinases have also been implicated in regulating Mcl-1 turnover, including p38, JNK, CDK1, and casein kinase II (8, 9, 26). Mcl-1 stability can be regulated by other E3 ubiquitin ligases including Mule and β-TrCP (6–9), and by the deubiquitinase USP9X (34). Depending on apoptotic stimuli and cell types, these proteins may also be involved in therapeutic resistance caused by Mcl-1 stabilization.

Our results clarify the functional role of Mcl-1 in apoptosis in solid tumor cells. Unlike hematopoietic cells, solid tumor cells such as colorectal cancer cells are not highly primed to cell death, and do not die in response to Mcl-1 inhibition or depletion alone (10). At least two independent events are necessary for initiating apoptotic response to regorafenib, including early Mcl-1 degradation and later PUMA induction (24, 25). Mcl-1 degradation unleashes the full proapoptotic activity of PUMA, and allows it to bind to and neutralize other prosurvival factors such as Bcl-X_L (23). In Mcl-1-KI and FBW7-mutant colorectal cancer cells that are deficient in Mcl-1 degradation, Mcl-1 sequesters PUMA from Bcl-X_L and other antiapoptotic proteins. Blocking Mcl-1 degradation not only abolishes apoptosis induction, but also attenuates the antiangiogenic and antihypoxic activities of regorafenib. This can probably be explained by suppression of the release of antiangiogenic and antihypoxic signals by apoptotic cells into microenvironment (35). Our findings further indicate that the effects of targeted therapy on cancer cells and their microenvironment are closely related and influence each other via cell death (25, 36).

Inhibiting prosurvival Bcl-2 family proteins in tumor cells has emerged as an attractive therapeutic strategy, leading to recent FDA approval of the Bcl-2-selective inhibitor ABT-199 (Venetoclax) for the treatment of chronic lymphocytic leukemia (37). However, ABT-199 and related agents do not bind to Mcl-1, and resistance to these agents quickly emerges due to overexpression of Mcl-1 (38–40). There is an urgent need for developing small-molecule Mcl-1 inhibitors. Mcl-1 binds to the BH3 domains of proapoptotic proteins through a surface grove containing 4 hydrophobic pockets (p1–p4; refs. 41, 42). Several different chemical classes of Mcl-1 inhibitors have been described (19, 20, 43–49). However, the existing Mcl-1 inhibitors lack sufficient potency and specificity for further development. Targeting Mcl-1, especially for solid tumor cells, has been hampered by lack of a reliable system for evaluating specificity. Our data suggest that Mcl-1–binding BH3 mimetics or enhancing PUMA expression can be used to overcome therapeutic resistance caused by Mcl-1 stabilization in colorectal cancer cells. In this context, the previously identified Mcl-1 inhibitors exhibited on-target activity in the isogenic Mcl-1-KI cells, which are potentially useful for identifying new inhibitors with improved potency, specificity and in vivo activity to overcome therapeutic resistance in colon and other solid tumor cells.

No potential conflicts of interest were disclosed.

Conception and design: J. Tong, L. Zhang

Development of methodology: J. Tong, P. Wang, L. Zhang

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): S. Tan, D. Chen, F. Zou, J. Yu, L. Zhang

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): J. Tong, D. Chen, F. Zou, J. Yu, L. Zhang

Writing, review, and/or revision of the manuscript: J. Tong, Z. Nikolovska-Coleska, J. Yu, L. Zhang

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): P. Wang, D. Chen, J. Yu, L. Zhang

Study supervision: L. Zhang

Other (provided key reagents): Z. Nikolovska-Coleska

We thank our laboratory members for critical reading and discussion.

This work is supported by NIH grants (R01CA106348, R01CA172136, and R01CA203028 to L. Zhang; U01DK085570 and U19AI068021 to J. Yu; R01CA149442 to Z. Nikolovska-Coleska) and National Natural Science Foundation of China (81672942 to F. Zou). This project used the UPCI shared facilities that were supported in part by award P30CA047904.

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.