MCT1 relieves osimertinib-induced CRC suppression by promoting autophagy through the LKB1/AMPK signaling

Colorectal cancer (CRC) is one of the most frequently diagnosed cancers worldwide. Development of novel chemotherapeutics is still required to enable successful treatment and improve survival for CRC patients. Here, we found that osimertinib (OSI) exhibits potent anti-CRC effects by inducing apoptosis, independent of its selective inhibitory activity targeting the EGFR T790M mutation. Intriguingly, OSI treatment triggers autophagic flux in CRC cells. Inhibition of autophagy markedly augments OSI-induced apoptosis and growth inhibition in CRC cells, suggesting a protective role of autophagy in response to OSI treatment. Mechanistically, OSI upregulates the expression of monocarboxylate transporter 1 (MCT1) and subsequently activates LKB1/AMPK signaling, leading to autophagy induction in CRC cells. Notably, OSI significantly exaggerates the sensitivity of CRC cells to the first-line drugs 5-fluorouracil or oxaliplatin. Taken together, our study unravels a novel mechanism of OSI-mediated protective autophagy involving MCT1/LKB1/AMPK signaling, and suggests the use of OSI as a potential agent for clinical CRC treatment.


Introduction
Colorectal cancer (CRC) is the third most frequently diagnosed cancer and one of the leading causes of cancerrelated deaths worldwide 1 . The standard treatment for CRC is surgery plus systematic chemotherapy regimens with or without radiation, which has shown favorable therapeutic efficacy and significantly improved the survival of CRC patients 2,3 . However, the prognosis for patients with advanced CRC still remains poor. In addition, a substantial proportion of advanced CRC patients treated with the first-line chemotherapeutic drugs (including 5-FU and oxaliplatin) often relapse due to drug resistance 4 . Thus, there is an urgent need to develop novel potential therapeutic agents for CRC treatment.
Autophagy is a highly conserved process during which de novo-formed membrane-enclosed vesicles engulf aggregated proteins or damaged organelles, and subsequently fuse with lysosomes for degradation 5 . This degradation process enables tumor cells to adapt to multiple cellular stimuli such as nutrient deprivation, oxidative stress and genotoxic stress 6 . Notably, cancer cells express high levels of transcriptional factor such as Mit/TFE and c-Myc in the nucleus to maintain persistent activation of autophagy for cell survival [7][8][9] . Thus, inhibition of autophagy could be beneficial for cancer therapy as it has already been documented in CRC treatment 10 . However, it has also been demonstrated that autophagy can promote cell death either alone or in association with apoptosis. For example, apoptosis and autophagy induced by dihydrotanshinone could synchronously contribute to the anti-CRC activity through enhanced generation of ROS and caspase-dependent signaling pathways 10,11 . Therefore, determining the functional roles of autophagy in a distinct context is critical for the development of rational autophagy manipulation strategies for successful chemotherapy.
Monocarboxylate transporter 1 (MCT1) is one of the most well studied members of monocarboxylate transporter family that promotes the passive transport of lactate. It plays an important role in metabolic reprogramming and has been observed to be overexpressed in various types of tumors that preferentially utilize lactate for oxidative metabolism 12 , including hepatocellular carcinoma (HCC) 13 , and gastric cancer 14 .
In addition, MCT1 was reported to be involved in the resistance of CRC cells to 5-FU, but the underlying mechanism remains largely unknown 15 . Although a link between autophagy and MCT1 upregulation has been observed in HCC cells 16 , whether autophagy, emerging as a mechanism of drug resistance in cancer chemotherapy, plays a role in MCT1-mediated drug resistance remains unclear.
Osimertinib (OSI) is a mutant-selective EGFR inhibitor for the treatment of non-small cell lung cancer (NSCLC) harboring the T790M mutation, exhibiting effective antitumor activity 17 . However, the anti-cancer efficacy of OSI through an EGFR-independent mechanism has rarely been explored. Here, we found that OSI significantly inhibits the growth of CRC cells by stimulating apoptosis irrelevant of its activity targeting the EGFR T790M mutation. Interestingly, OSI upregulates MCT1 and sequentially induces autophagy flux, which in turn ameliorates OSI-induced apoptosis and growth inhibition of CRC cells. Our findings demonstrate a novel role of MCT1 in autophagy modulation, and provide the molecular basis for the use of OSI as a potential therapeutic agent for CRC treatment.

Cell culture and reagents
Human cancer cell lines DLD-1, HT29, HCT116, SW620, LoVo, RKO, SW480 and noncancerous colorectal cell line NCM460 were purchased from the ATCC and maintained in DMEM supplemented with 10% fetal bovine serum (Gibco), in a humidified incubator at 37°C under 5% CO 2 atmosphere.

Colony formation assay
Cells were seeded in 24-well plates for about 3 days, treated with indicated concentrations of OSI for 48 h, and maintained in DMEM for another 10 days. Cells were then fixed with 4% paraformaldehyde in PBS for 30 min and stained with crystal violet. After washing with ddH 2 O for three times, samples were dissolved in 0.1% SDS. The absorbance was then detected at 490 nm.

EdU labeling assay
The EdU labeling assay was performed in 96-well plate using the EdU Cell Proliferation Assay Kit (Ribobio). After 24 h treatment, 10 μmol/L EdU was added to each well, and the cells were incubated for another 24 h at 37°C. Cells were then fixed with 4% paraformaldehyde in PBS and stained with reaction cocktail. DAPI was subsequently used for nuclear staining, followed by imaging with DM2500 fluorescence microscope (Leica).

TUNEL assays and flow cytometry
Cells were plated on glass coverslips in 24-well plates, then treated as indicated for 24 h. DeadEnd TM Fluorometric TUNEL system was used to stain cells after fixation in 4% paraformaldehyde according to the manufacturer's instructions (Promega, G3250). The apoptotic and nonapoptotic signals were photographed using DM2500 fluorescence microscope (Leica) and then the percentage of cells with DNA nick end-labeling was evaluated.
Annexin V-FITC/PI Detection Kit (KeyGEN BioTECH, KGA108) was used to measure the ratio of apoptotic cells according to the manufacturer's protocol. Cells were harvested and washed twice with PBS, and then resuspended in 500 μl binding buffer. Annexin V-FITC and PI were added into the cell suspension, respectively, for apoptosis analysis using FACSCalibur flow cytometer (BD FACScan Flow cytometer, United States) and data were analyzed using FlowJo software.

Caspase 3/7 assay
A caspase3/7 test kit (ATT Bioquest) was used to evaluate apoptosis induced by OSI. Cells were seeded in 96-well plates (20,000 cells/well/90 µl, medium alone was used as the control). The cells were then treated with indicated concentrations of OSI for 24 h. Caspase 3/7 assay loading solution was then added and incubated at room temperature for at least 2 h. The fluorescence intensity was detected at Ex/Em = 350/450 nm using a microplate reader.

Immunoblotting and immunoprecipitation
Immunoblotting analysis was performed as described previously 18 . For immunoprecipitation, cells were lysed with IP lysis buffer (20 mM Tris, 137 mM NaCl, 10% glycerol, 1% NP-40 and 2 mM EDTA, pH = 8) and subjected to rotation overnight at 4°C with indicated antibodies (1 μg). Protein A-Sepharose beads (40 μl, GE Healthcare) were then added and incubated with the lysates for 3 h. After centrifugation at 800 × g, the beads were washed four times, boiled with loading buffer and analyzed by immunoblotting with the indicated antibodies.

Immunofluorescence
Cells were plated on the glass coverslips in 24-well plates. After treatment, cells were fixed with 4% paraformaldehyde in PBS for 30 min. After washing with PBS, cells were permeabilized with 0.4% Triton X-100 and blocked with 5% goat serum for 30 min. Indicated primary antibodies were used to incubate with the cells overnight at 4°C, followed by secondary antibody (DyLight 488-conjugated goat anti-rabbit IgG or DyLight 594-conjugated goat anti-mouse IgG) at 37°C for 1 h. After staining nuclei with DAPI for 10 min, images were viewed with a confocal laser scanning microscopy (Zeiss). For DQ-BSA assay, cells were pre-probed with DQ-BSA Red (10 μg/ml) for 1 h and then treated with OSI for 24 h. After fixing with 4% paraformaldehyde in PBS for 30 min, images were visualized using a confocal laser scanning microscopy (Carl Zeiss Microimaging).

Tumor xenograft model
All animal experiments were approved by the Institutional Animal Care and Treatment Committee of Sichuan University. Six-week-old nude mice (BALB/c, 18-20 g each) were purchased from HFK Bioscience Co., Ltd (Beijing). The mice were housed under standard conditions. For the subcutaneous xenograft model, DLD-1 cells (1 × 10 7 cells/mouse) were suspended in PBS and subcutaneously implanted into flanks of mice. When the tumor volume reached~200 mm 3 (14 days post-injection), the mice were divided into a control group (daily intraperitoneal injection of vehicle) and OSI group (daily intraperitoneal injection of 15 mg/kg OSI). The tumor volumes were measured every day and evaluated according to the following formula: tumor volume (mm 3 ) = (length × width 2 )/2. The mice were euthanized after two weeks and tumors were harvested.

Statistical analysis
All statistical analysis and graphics were performed using GraphPad 6 software. One-way ANOVA or Students t-test was used to analyze statistical differences. All data are presented as the mean ± SD from at least three individual experiments. A value of P < 0.05 was considered as statistically significant.

OSI inhibits CRC cells growth by stimulating apoptosis in vitro and in vivo
Aberrant amplification of epidermal growth factor receptor (EGFR) plays a key role in the tumorigenesis and progression of CRC 19 , thus represents a potential target for CRC therapy. In this regard, we screened a smallmolecule library of 15 EGFR inhibitors and identified OSI as the most effective candidate in suppressing CRC cell growth ( Supplementary Fig. 1A, B). Notably, the anti-CRC effect of OSI is independent of the KRAS status in CRC cells ( Supplementary Fig. 1C, D). In addition, OSI exhibits greater efficacy than olmutinib, another clinical used third-generation EGFR inhibitor, suggesting an EGFRindependent mechanism of OSI in CRC suppression.
To validate the anti-CRC activity of OSI, we detected the relative absorbance of various CRC cell lines (including DLD-1, HT29, LoVo, HCT116, SW480, RKO, and SW620) and noncancerous colorectal cell line (NCM460) in MTT assays following OSI treatment. Indeed, OSI treatment for 24 h markedly decreased the relative absorbance of CRC cell lines, but exhibited less toxicity to noncancerous cells (Fig. 1a). In agreement with these observations, OSI significantly inhibited the proliferation of CRC cells as evidenced by decreased EdU incorporation (Fig. 1b) and colony formation (Fig. 1c). These observations suggest that OSI shows significant anticancer activity against CRC. Further studies described in this work were performed using DLD-1 and HT29 cell lines as they showed the strongest sensitivity to OSI treatment. Since apoptosis is a major form of cell death induced by chemotherapeutic agents 20,21 , we next examined whether OSI could induce apoptosis in CRC cells by analyzing TUNEL-positive cells and caspase3/7 activity. As showed in Fig. 1d, e, OSItreated CRC cells exhibited obvious induction of apoptosis compared to cells treated with vehicle. This was further confirmed by the increased level of cleaved-PARP and cleaved-caspase 3 in OSI-treated CRC cells (Fig. 1f).
To evaluate the anti-CRC effect of OSI in vivo, a CRC xenograft model was generated by subcutaneously inoculating DLD-1 cells into nude mice. As shown in Fig. 1g, and supplementary Fig. 2A, B, the size, weight and growth rate of tumors were remarkably decreased in OSI-treated mice compared with those of the control group. Furthermore, OSI treatment resulted in relatively weaker Ki67 staining compared with the control group (Fig. 1h). In addition, obvious apoptosis was observed in tumors from OSI-treated mice as evidenced by increased cleaved-caspase 3 intensity and protein levels, as well as upregulated cleaved-PARP levels ( Fig. 1j-l). H&E staining of major organs and analysis of body weight changes showed no obvious toxic effect in response to OSI treatment in mice ( Supplementary Fig. 2C, D). Collectively, these results demonstrate that OSI exhibits significant antitumor effect in CRC cells by triggering apoptosis both in vitro and in vivo.

OSI stimulates autophagic flux in CRC cells
As autophagy plays an important role in chemotherapy 22 , it is of particular interest to investigate whether autophagy was involved in the anti-CRC effect of OSI. We first evaluated the formation of autophagosomes in OSItreated cells by analyzing the turnover of LC3-I to lipidated LC3-II and the formation of LC3 puncta, two hallmarks of autophagy 23,24 . OSI treatment resulted in increased LC3-II conversion (Fig. 2a, Supplementary Fig.  3A, B) and LC3 puncta accumulation (Fig. 2b, c, Supplementary Fig. 3C, D). In addition, OSI treatment upregulated the protein levels of Atg5, a key component in the ubiquitin-like conjugating system, in a dosedependent manner (Fig. 2a). Moreover, stronger LC3 staining (Fig. 2d, e) and increased LC3-II protein levels (Fig. 2f, g) were observed in OSI-treated tumor xenografts. These results indicate that OSI promotes autophagosome accumulation in vitro and in vivo.
The accumulation of autophagosomes may be attributed to either autophagy initiation or blockage of autophagic flux 25 . We thus set out to examine the interaction of Beclin1 with its negative regulator Bcl-2 to validate whether OSI treatment could promotes autophagy initiation 5,23 . As shown in Fig. 2h, OSI treatment disrupted the interaction between Beclin1 and Bcl-2. Furthermore, inhibition of class III PI3K by 3-MA 23 , or siRNAmediated ATG5 or BECN1 silencing, prominently inhibited the conversion of LC3-II and the formation of LC3 puncta in OSI-treated CRC cells ( Fig. 2i-m, Supplementary Fig. 3E, F). Taken together, these data indicate that OSI induces autophagosome formation by promoting autophagy initiation in CRC cells. Interestingly, we found that olmutinib, another third-generation EGFR inhibitor, shows no obvious autophagy induction in CRC cells ( Supplementary Fig. 3G), again suggesting that OSIinduced autophagy induction might represent an EGFRindependent mechanism.
To further determine whether OSI promotes autophagic flux, we used a tandem mRFP-GFP tagged LC3 construct and found that OSI-treated CRC cells displayed increased formation of red fluorescent autolysosomes (GFP − RFP + signal), while combinatorial treatment of CQ (chloroquine) with OSI resulted in accumulation of yellow fluorescent autophagosomes (GFP + RFP + signal) (Fig. 3a,  b). We also detected the colocalization of LC3 (autophagosome marker) with LAMP1 (lysosome marker) in OSItreated CRC cells. As shown in Supplementary Fig. 4A, B, OSI treatment resulted in obvious colocalization of LC3 and LAMP1, suggesting that OSI promotes the fusion of the autophagosome with lysosome. In addition, combination treatment OSI with CQ led to increased accumulation of LC3-II and endogenous LC3 puncta (Fig. 3c,  Supplementary Fig. 4C, D). In line with these data, using a highly self-quenched BODIPY conjugated bovine serum albumin (DQ-BSA), we found that OSI treatment increased the DQ-BSA fluorescent signal (Fig. 3d, e), suggesting increased proteolytic degradation in response  to OSI treatment. Consistently, ubiquitinated protein conjugates were decreased following OSI treatment ( Supplementary Fig. 4E, F). Taken together, these findings demonstrate that OSI promotes autophagic flux in CRC cells. We also analyzed the protein levels of p62/SQSTM1, which is supposed to be downregulated during a complete autophagy flux due to its degradation in autolysosome 26 . Unexpectedly, the protein levels of p62/SQSTM1 were upregulated after OSI treatment (Fig. 2a), which might contribute to the enhanced transcription of p62/SQSTM1 induced by OSI ( Supplementary Fig. 5).

Inhibition of autophagy augments the antitumor effect of OSI in CRC cells
To evaluate whether autophagy was involved in the anti-CRC effect of OSI, CRC cells were treated with OSI combined with 3-MA, siATG5, siBECN1, or CQ, respectively. As shown, combinational use of autophagy inhibitors (3-MA or CQ), siATG5 or siBECN1 with OSI significantly potentiated OSI-induced growth suppression in CRC cells (Fig. 4a-d). In line with these observations, OSI-induced apoptotic cell death was augmented in the presence of autophagy inhibitors, as indicated by increased levels of cleaved-PARP and cleaved-caspase 3 (Fig. 4e). Collectively, these data demonstrate that inhibition of autophagy significantly promotes OSI-induced CRC suppression, partially by enhancing apoptosis induction, suggesting a protective role of autophagy in OSI-treated CRC cells.

LKB1/AMPK signaling plays a major role in OSI-induced autophagy
It has been previously reported that AMPK is a canonical upstream signaling molecule for autophagy induction 27 . Therefore, we investigated the phosphorylation status of AMPK to validate whether AMPK was involved in OSI-induced autophagy 28,29 . As shown in Fig. 5a, the phosphorylation levels of AMPK were enhanced after OSI treatment. In addition, OSI treatment resulted in relatively stronger staining of phosphorylated AMPK (Thr172) in CRC xenografts treated with OSI compared with the control group (Fig. 5b, c). Inactivation of AMPK, either by siRNA or a dominant-negative mutant of AMPK (DN-AMPK), significantly inhibited OSI-induced LC3-II conversion (Fig. 5d, e) and LC3 puncta accumulation ( Fig. 5f-i). We then examined whether LKB1 or CaMKKβ, two well-known AMPK upstream kinases, were responsible for the activation of AMPK in OSI-treated CRC cells. As shown in Fig. 5j, the phosphorylation level of LKB1, but not CaMKKβ, was enhanced following OSI treatment. Consistently, siRNA-mediated LKB1 silencing markedly inhibited OSI-induced LC3-II conversion ( Supplementary  Fig. 6A). These results suggest that OSI induces autophagy by activating LKB/AMPK signaling pathway in CRC cells.

OSI induces autophagy through MCT1-mediated activation of AMPK in CRC cells
The lactate transporter MCT1 has been previously reported to be associated with AMPK activation 30 . We thus speculated that MCT1 might be involved in AMPKmediated autophagy induction in OSI-treated CRC cells. We found that OSI treatment led to upregulation of MCT1 in CRC cells (Fig. 6a), but olmutinib failed to elevate the protein level of MCT1 ( Supplementary Fig.  3G). The upregulation of MCT1 in response to OSI treatment was further confirmed in xenograft tumors (Fig.  6b-d). Next, we investigated whether MCT1 was required for OSI-induced AMPK activation and autophagy induction in CRC cells by knockdown or enforced expression of MCT1. As shown in Fig. 6e-g, siRNA-mediated MCT1 silencing markedly repressed the OSI-induced phosphorylation levels of LKB1 and AMPK, LC3-II conversion and LC3 puncta accumulation. In addition, exogenous MCT1 expression resulted in elevated LC3-II conversion (Fig. 6h), LC3 puncta accumulation (Fig. 6i, j) and AMPK phosphorylation, to a similar level to that observed in cells treated by OSI alone. Notably, the increased LC3 lipidation induced by MCT1 overexpression could be counteracted by DN-AMPK-mediated AMPK inactivation (Fig.  6k-m) or LKB1 knockdown (Supplementary Fig. 6B). Taken together, these results demonstrate that OSI (see figure on previous page) Fig. 2 OSI induces autophagy in CRC cells in vitro and in vivo. a Immunoblotting analysis of LC3, Atg5, and p62/SQSTM1 expression in CRC cells treated with indicated concentrations of OSI for 24 h. b The formation of endogenous LC3 puncta in cells treated with DMSO or 5 μM OSI for 24 h. c Total number of endogenous LC3 puncta per cell in (b). d, e LC3 expression in xenograft tissues was examined by IHC. Representative images were provided as indicated in (d) and relative intensity of LC3 staining was quantified in (e). f Immunoblotting analysis of LC3 and p62/SQSTM1 expression in tumor xenografts (Each protein of interest from each group was electrophoretically transferred onto a PVDF membrane, incubated with indicated primary and secondary antibodies, and developed as a digital image.) g Relative intensity of LC3 in (f). h Co-immunoprecipitation analysis of the interaction between Beclin 1 and Bcl-2 in CRC cells treated with or without 5 μM OSI for 24 h. i Immunoblotting analysis of LC3 expression in CRC cells treated with or without 5 μM OSI in the presence or absence of 5 mM 3-MA for 24 h. j CRC cells were treated as in (i), the LC3 puncta were analyzed by immunofluorescence. Scale bar, 10 μm. k, l Immunoblotting analysis of LC3 expression in CRC cells transfected with siScramble, siATG5 (k), or siBECN1 (l) for 24 h, followed by treatment with or without 5 μM OSI for another 24 h. m CRC cells were treated as in (k, l). The LC3 puncta were analyzed by immunofluorescence. Scale bar, 10 μm. Data are presented as mean SEM, Student's t-test, and are representative of 3 independent experiments. **P < 0.01; ***P < 0.001

MCT1 antagonizes the antitumor efficacy of OSI and is overexpressed in human CRC tissues
Given the role of MCT1 in mediating OSI-induced autophagy, it was necessary to explore whether MCT1 interfered with the OSI-induced CRC suppression. As shown in Fig. 7a, MCT1 was overexpressed in various CRC cells compared with noncancerous NCM460 cells. siRNA-mediated MCT1 silencing remarkably aggravated the OSI-induced growth suppression in CRC cells (Fig. 7b, c), at least in part, by promoting apoptosis as evidenced by the enhanced levels of cleaved-PARP (Fig. 7d). Interestingly, there was no synergistic effect on CRC suppression and LC3-II conversion of OSI with AZD3965 (a MCT1 inhibitor that blocks its lactate Data are presented as mean SEM, Student's t-test, *P < 0.05; **P < 0.01; ***P < 0.001 transporting function) ( Supplementary Fig. 7A, C), implying that the mechanism of MCT1-mediated autophagy might be independent of the monocarboxylatetransporting function of MCT1.
As OSI has been reported to induce autophagy in NSCLC cells 31 , we analyzed whether MCT1 and protective autophagy were involved in the resistance of NSCLC cells to OSI. We found that the protein level of MCT1 was upregulated after OSI treatment both in T790M mutation harboring NCI-1975 cells and EGFR WT A549 cells ( Supplementary Fig. 8A, C). Moreover, silencing MCT1 augmented OSI-induced growth suppression in NSCLC cells ( Supplementary Fig. 8B, D). These results indicate that the upregulation of MCT1 and subsequent autophagy . d CRC cells were transfected with siScramble or siAMPK for 24 h and followed by treatment with or without 5 μM for another 24 h. LC3 and phosphorylated AMPK (Thr172) levels were detected by immunoblotting. e CRC cells were transfected with empty vector or DN-AMPK plasmid for 24 h, followed by treatment with or without 5 μM OSI for another 24 h. LC3 and AMPK phosphorylation levels were determined by immunoblotting. f CRC cells were treated as in (d), the endogenous LC3 puncta in CRC cells were assessed by immunofluorescence. Scale bar, 10 μm. g The number of LC3 puncta per cell in (f). h CRC cells were treated as in (e), the endogenous LC3 puncta in CRC cells were assessed by immunofluorescence, Scale bar, 10 μm. i The number of LC3 puncta per cell in (h). j Immunoblotting analysis of phosphorylated LKB1 and phosphorylated CaMKKβ levels in CRC cells treated with or without 5 μM OSI. Data are presented as mean SEM, Student's t-test, and are representative of three independent experiments. ***P < 0.001 initiation might play a role in the drug resistance of NSCLC cells to OSI.
It has been reported that MCT1 is overexpressed in many cancers, such as HCC and gastric cancer 16 . Here, our study suggested that the expression of MCT1 was upregulated in CRC specimens compared with normal tissues by immunoblot and immunohistochemical analysis ( Fig. 7e-h), which was consistent with the Oncomine data (Fig. 7i). These data indicate that MCT1 may play important role in CRC cell growth. Together, our study suggests that MCT1 might confer resistance to the antitumor effect of OSI, and may serve as a promising therapeutic target for CRC.
OSI enhances the anti-CRC efficacy of 5-FU and oxaliplatin 5-Fluorouracil and oxaliplatin, two first-line chemotherapeutic drugs for clinical CRC treatment, have been reported to lead to drug resistance 32,33 . A recent study demonstrated that OSI increases sensitivity of NSCLC cells to radiation by delaying DNA damage repair 34 . Thus, we question whether OSI could also enhance the chemosensitivity of CRC cells to 5-FU or oxaliplatin. As indicated, combinational treatment of OSI with 5-FU or oxaliplatin markedly decreased the relative absorbance ( Supplementary Fig. 9A, 9B) and proliferation rate ( Supplementary Fig. 9C, 9D) of CRC cells. Taken together, these results demonstrate that OSI effectively sensitizes CRC cells to 5-FU and oxaliplatin treatment, which was consistent with a previous study suggesting that OSI synergizes with 5-FU 35 , and the underlying molecular mechanism remains rarely unexplored.

Discussion
OSI is a first-line chemotherapeutic agent for NSCLC patients harboring the EGFR T790M mutation 36,37 . Interestingly, Hirano et al. suggested that OSI has potent efficacy against NSCLC harboring EGFR 19 deletion and L858R mutation, as well as those harboring exon 20 insertion mutations, suggesting a wide therapeutic window of OSI in NSCLC 38 . In this study, we found that OSI has potent anticancer effects in CRC, which might interestingly be independent of its selective inhibitory activity targeting EGFR T790M mutation. Our findings may extend the clinical potential of OSI and provide a new paradigm for cancer therapy.
Autophagy was proposed as a potential cell death mechanism during ionizing radiation and chemotherapy 39 . However, accumulating evidence suggests that autophagy could facilitate the resistance of cancer cells to chemotherapeutic drugs, and inhibition of autophagy could sensitize tumor cells to cancer therapies. Thus, it is necessary to determine the role of autophagy in cancer therapy for rational manipulation of autophagy to assist chemotherapy. Our data demonstrate that OSI induces autophagic flux, which counteracts apoptosis induction and growth suppression of CRC cells. Hence, it is reasonable to conclude that autophagy plays a protective role against OSI-induced CRC suppression. It has been reported that OSI could induce acquired drug resistance in NSCLC patients 40 , mainly due to a second mutation of EGFR (such as C797 mutation) 41 , or activation of by-pass pro-survival signaling pathways such as MET/ERK, HER2, IRE1α or RAS [41][42][43][44] . Our data demonstrate autophagy as an additional drug resistance mechanism for OSI treatment in both NSCLC and CRC, suggesting that combinatorial use of OSI with autophagy inhibitors may benefit their therapeutic efficacy for cancer treatment. Further studies are needed to detect the combinatorial efficacy of OSI and autophagy inhibitors in vivo.
AMPK, a well-known energy sensor, could switch on glycolysis and induce autophagy to maintain ATP level 45 . It has been reported that AMPK could phosphorylate ULK1 to form PI3K complex, leading to autophagy induction 46,47 . Our study reveals that LKB1 mediatedactivation of AMPK is crucial for the initiation of autophagy in OSI-treated CRC cells. Further investigation shows that OSI treatment upregulates the protein level of MCT1, which subsequently activates LKB1/AMPK signaling, leading to autophagy induction in CRC cells. The MCT1/LKB1/AMPK is proposed as a new mechanism of (see figure on previous page) Fig. 6 OSI induces autophagy through upregulation of MCT1 in CRC cells. a Immunoblotting analysis of MCT1 expression in CRC cells treated with or without 5 μM OSI for 24 h. b Immunoblotting analysis of MCT1 and phosphorylated AMPK in tumor xenografts obtained from vehicle-or OSItreated mice. (Each protein of interest from each group was electrophoretically transferred onto a PVDF membrane, incubated with indicated primary and secondary antibodies, and developed as a digital image.) c Immunohistochemical analysis of MCT1 expression in tumor xenografts. Scale bar, 50 μm. d Relative intensity of MCT1 staining in (c). e CRC cells were transfected with siScramble or siMCT1 for 24 h, followed by treatment with or without 5 μM OSI for another 24 h. The protein levels of LC3, phosphorylated AMPK, phosphorylated LKB1 and MCT1 were analyzed by immunoblotting. f CRC cells were treated as in (e), the endogenous LC3 puncta were analyzed by immunofluorescence. Scale bar, 10 μm. g The number of LC3 puncta in (f). h CRC cells were transfected with empty vector or Flag-MCT1 plasmid for 48 h, the protein levels of MCT1 and phosphorylated AMPK were analyzed by immunoblotting. i CRC cells were treated as in (h), the endogenous LC3 puncta were analyzed by immunofluorescence. Scale bar, 10 μm. j The number of LC3 puncta in (i). k Immunoblotting analysis of LC3, MCT1 and phosphorylated AMPK levels in CRC cells co-transfected with Flag-MCT1 and DN-AMPK plasmids for 48 h. l CRC cells were treated as in (k), the endogenous LC3 puncta were analyzed by immunofluorescence. Scale bar, 10 μm. m The number of LC3 puncta per cell in (l). Data are presented as mean SEM, Student's t-test, and are representative of three independent experiments. *P < 0.05; ***P < 0.001 Fig. 7 MCT1 antagonizes the antitumor efficacy of OSI and is overexpressed in human CRC tissues. a Immunoblotting analysis of MCT1 protein levels in CRC cell lines and noncancerous colorectal cell line NCM460. b, c Representative images of colony formation assays of CRC cells transfected with siScramble or siMCT1 for 24 h, followed by treatment with or without 5 μM OSI for another 24 h. d Immunoblotting analysis of cleaved-PARP expression in CRC cells transfected with siScramble or siMCT1 for 24 h, followed by treatment with or without 5 μM OSI for another 24 h. e Immunoblotting analysis of MCT1 expression in 10 pairs of tumor and adjacent normal tissues obtained from CRC patients. f Relative quantification of MCT1 levels in (e). g Immunohistochemical staining of MCT1 in 10 pairs of tumor and adjacent normal tissues obtained from CRC patients. Scale bar, 50 μm. h Relative intensity of MCT1 levels in (g). i Data represent the Z-score of MCT1 from Oncomine database. Data are presented as mean SEM, Student's t-test, and are representative of 3 independent experiments. **P < 0.01; ***P < 0.001 autophagy modulation axis. A previous study has shown that AMPK could be activated by MCT1-mediated lactate uptake 30 . Further studies are needed to delineate the mechanism underlying MCT1-mediated activation of LKB1/AMPK signaling and subsequent autophagy induction.
MCT1 plays an important role in metabolic reprogramming by transporting of monocarboxylates such as lactate, pyruvate, and ketone bodies from hypoxic cancer cells or cancer-associated fibroblasts (CAFs) to aerobic cancer cells 48 . Robust lactate metabolism in MCT1positive cancer cells produces sufficient energy for selfrenewal and distant metastasis, indicating a predominant role of MCT1-positive cancer cells in tumor environment 12,49 . Here, we found that MCT1 is overexpressed in CRC tissue and cell lines. Importantly, MCT1 is upregulated to activate the LKB1/AMPK signaling pathway and subsequently initiates protective autophagy in OSI-treated CRC cells, conferring resistance to OSI. Our findings together with previous studies indicate that intervening MCT1 under OSI treatment could improve treatment efficacy and outcome in CRC tumors 50 . However, our further data suggests that MCT1 knockdown, rather than pharmacological inhibition of its lactate transporting function by AZD3965, suppresses autophagy induction and aggravates the antitumor effect of OSI in CRC cells. This could be attributed to MCT4 heterogeneous expression as AZD3965 has been reported to kill preferentially tumor cells with high MCT1 and low MCT4 expression 51 . In addition, the role of MCT1 in modulating autophagy might be in a non-canonical role of MCT1, independent of its lactate transporter activity. Indeed, MCT1 has been reported to activate the transcription factor NF-κB to promote tumor metastasis, beyond its role as a lactate transporter 14 . Further studies should be conducted to investigate the efficacy of combinatorial use of OSI with MCT1 inhibitors in vivo.
In conclusion, we have demonstrated that OSI has a pronounced anti-CRC effect both in vitro and in vivo by stimulating apoptosis. Meanwhile, OSI upregulates the expression of MCT1 in CRC cells, which in turn promotes the induction of protective autophagy ( Supplementary  Fig. 10). Our study indicates that MCT1 might be an attractive therapeutic target for CRC treatment, and reveals the potency of OSI as a promising therapeutic agent for CRC treatment.