The mechanisms underlying metabolic adaptation of pancreatic ductal adenocarcinoma (PDA) cells to pharmacologic inhibition of RAS–MAPK signaling are largely unknown. Using transcriptome and chromatin immunoprecipitation profiling of PDA cells treated with the MEK inhibitor (MEKi) trametinib, we identify transcriptional antagonism between c-MYC and the master transcription factors for lysosome gene expression, the MiT/TFE proteins. Under baseline conditions, c-MYC and MiT/TFE factors compete for binding to lysosome gene promoters to fine-tune gene expression. Treatment of PDA cells or patient organoids with MEKi leads to c-MYC downregulation and increased MiT/TFE-dependent lysosome biogenesis. Quantitative proteomics of immunopurified lysosomes uncovered reliance on ferritinophagy, the selective degradation of the iron storage complex ferritin, in MEKi-treated cells. Ferritinophagy promotes mitochondrial iron–sulfur cluster protein synthesis and enhanced mitochondrial respiration. Accordingly, suppressing iron utilization sensitizes PDA cells to MEKi, highlighting a critical and targetable reliance on lysosome-dependent iron supply during adaptation to KRAS–MAPK inhibition.
Reduced c-MYC levels following MAPK pathway suppression facilitate the upregulation of autophagy and lysosome biogenesis. Increased autophagy–lysosome activity is required for increased ferritinophagy-mediated iron supply, which supports mitochondrial respiration under therapy stress. Disruption of ferritinophagy synergizes with KRAS–MAPK inhibition and blocks PDA growth, thus highlighting a key targetable metabolic dependency.
Bypass of oncogene addiction is a major hurdle to effectively treating and eliminating cancer cells (1, 2). Pancreatic ductal adenocarcinoma (PDA) is driven by KRAS mutation and hyperactivation of MAPK signaling, which promotes downstream changes in transcriptional and metabolic pathways that favor tumor progression and growth (3–5). Suppression of the KRAS–MAPK pathway can delay tumor growth in vitro and in vivo but has shown limited efficacy long term due to the activation of compensatory feedback mechanisms that promote rapid emergence of drug resistance and tumor relapse (1, 6–8). One example of a resistance mechanism is the induction of autophagy—a process in which intracellular material is packaged within vesicles for transport to the lysosome for degradation. At baseline, PDA cells and tumors are highly dependent on autophagy to maintain metabolic homeostasis and remodel the cellular proteome (9, 10). Recent studies have also shown that in response to inactivation of KRAS, MEK, or ERK, PDA cells further induce autophagy as an adaptive response that promotes drug resistance and survival (11–15), suggesting that context-dependent autophagy may confer unique advantages to PDA cells. How KRAS–MAPK inhibition or its bypass leads to autophagic enhancement in drug-treated cells has not been established.
We previously showed that the high levels of autophagy and lysosome activity in PDA (hereafter referred to as “baseline” activity) occur via constitutive activation of the microphthalmia/transcription factor E (MiT/TFE) family of master transcription factors—MITF, TFE3, and TFEB (16). These basic helix–loop-helix (bHLH) transcription factors belong to the MYC superfamily and recognize a palindromic 10-base-pair motif (GTCACGTGAC) closely related to canonical E-box elements (17). This motif, termed the coordinated lysosomal expression and regulation (CLEAR) element, is present in the promoter region of numerous lysosomal genes including those that encode for hydrolases, lysosomal membrane permeases, and associated proteins and is necessary for MiT/TFE-mediated activation of gene expression (18–20). Given the high level of homology of the DNA binding domain of MiT/TFE factors, all family members can recognize and bind to the CLEAR element. The MiT/TFE proteins also regulate several autophagy genes and therefore are considered master regulators of catabolism (21). In PDA, TFE3, TFEB, and MITF show increased expression levels, escape negative regulation by the mTORC1 kinase, and are constitutively localized in the nucleus, leading to elevated baseline autophagy and lysosome biogenesis (16). Accordingly, suppression of MiT/TFE factors causes a pronounced decrease in expression of lysosome- and autophagy-associated proteins, leading to defects in lysosome morphology and function and to significant inhibition of tumor growth (16).
Activation of the autophagy–lysosome system confers a growth advantage to PDA cells through degradation, recycling, and refueling of key biosynthetic and bioenergetic pathways critical for PDA growth (10, 22). Autophagy sequesters all classes of cellular macromolecules either in bulk or via a selective process that employs specific autophagy receptors. Whether autophagy–lysosome activity confers unique metabolic advantages via specific sequestration of cargos in different contexts—for example, under baseline conditions versus in response to MAPK suppression—remains unknown. Moreover, the precise mechanism for activation of the pathway in response to KRAS–MAPK inhibition remains unclear.
Using transcriptomic profiling and chromatin immunoprecipitation (ChIP) studies, we uncovered a previously unrecognized antagonism between the transcription factors, c-MYC (hereafter referred to as MYC), and MiT/TFE proteins in regulation of autophagy and lysosomal gene expression in PDA. Under baseline conditions, MYC is capable of binding to the CLEAR element, effectively restricting the full activation of lysosomal genes by MiT/TFE factors. Upon KRAS–MAPK pathway inhibition, MYC levels are decreased, enabling unrestricted access of MiT/TFE factors to CLEAR elements and a corresponding increase in autophagy and lysosomal gene expression and organelle biogenesis. Proteomic profiling of intact lysosomes captured from MEK inhibitor (MEKi)–treated cells further identified ferritinophagy—the selective autophagy pathway for degradation of the iron storage protein, ferritin—as being enhanced in MEKi-treated PDA cells. Increased ferritinophagy in turn was necessary for promoting an iron-dependent increase in mitochondrial iron–sulfur cluster (ISC) proteins required for respiration, thereby enabling PDA cell proliferation and survival. Accordingly, combining MEKi treatment with respiratory complex I inhibition, iron chelation, disruption of ferritinophagy via knockdown of the ferritinophagy receptor NCOA4, or lysosome inhibition leads to increased blockade of PDA growth, which can be rescued with iron supplementation. This signature of autophagy–lysosome-mediated cellular adaptation is retained following prolonged KRAS pathway suppression and is a conserved hallmark in other KRAS-driven cancers treated with MEKi. Finally, an anticorrelation between MiT/TFE- and MYC-dependent activity is common to KRAS-driven PDA, colorectal cancer, and non–small cell lung cancer (NSCLC). Together, these findings highlight a critical interplay between master transcriptional regulators of metabolism that drives context-specific autophagy–lysosome–mitochondria cross-talk to enhance iron utilization and enable adaptation to therapy stress.
MEK Inhibition Triggers MiT/TFE Factor–Dependent Upregulation of Lysosome Gene Expression in PDA
Treatment of human PDA cell lines with a MEKi (trametinib) or ERK inhibitor (ERKi; SCH772984) increases autophagy, as evidenced by increased LC3B puncta, LC3B-II abundance, and autophagic flux in multiple human PDA cell lines (Fig. 1A–C; Supplementary Fig, S1A–S1E). In addition, RNA sequencing (RNA-seq) analysis of MEKi-treated PDA cells indicated an increase in genes associated with lysosomal biogenesis relative to DMSO-treated cells (Fig. 1D; Supplementary Fig. S1F). Gene set enrichment analysis (GSEA) identified “Lysosome” as a significantly enriched pathway in MEKi-treated cells compared with control cells (Fig. 1E and F; Supplementary Fig. S1G and S1H). Western blot analysis in MiaPaca2 cells confirmed increased expression of lysosomal proteins (Supplementary Fig. S1D). Consistent with these findings, expansion of the lysosome compartment was also observed via immunofluorescence staining for LAMP2 in MEKi-treated cells across multiple PDA lines (Fig. 1G and H). ERKi treatment similarly led to an increase in lysosome gene expression, lysosome-associated protein levels, and lysosome biogenesis in multiple PDA cell lines (Supplementary Fig. S1E and S1I–S1K). To test whether suppression of oncogenic KRAS similarly induces an increase in lysosome gene expression, we treated MiaPaca2 cells with AMG 510, a clinically approved KRASG12C inhibitor, and identified an increase in several autophagy and lysosome genes and proteins in drug-treated relative to control-treated cells (Supplementary Fig. S2A and S2B). Analysis of a published proteomics data set (23) similarly showed an enrichment of autophagy- and lysosome-related proteins in MiaPaca2 cells treated with the KRASG12C inhibitor compound 4 (ref. 24; Supplementary Fig. S2C; Supplementary Table S1). These data suggest that upregulation of autophagy and lysosome gene and protein signatures are common hallmarks following the suppression of different nodes of the KRAS–MAPK pathway.
We previously showed that PDA cells maintain an elevated level of basal autophagy and lysosomal biogenesis via upregulation and constitutive nuclear localization of MiT/TFE factors (TFEB, TFE3, and MITF), the master transcription factors that control autophagy and lysosomal gene expression (16). Therefore, we hypothesized that induction of autophagy and lysosomal gene expression upon MAPK pathway suppression may be due to further alterations in MiT/TFE factor levels or activity. Consistent with our hypothesis, we observed an increase in TFEB and/or TFE3 mRNA and protein levels in PDA cells following inhibition of MAPK signaling as well as induction of MiT/TFE target genes within 24 hours of MEKi treatment (Fig. 1I; Supplementary Fig. S2D–S2G). Moreover, MiT/TFE factors were constitutively localized in the nucleus in PDA cells treated with MEKi (Fig. 1J; Supplementary Fig. S2H) or ERKi (Supplementary Fig. S2I and S2J). Similarly, treatment of MiaPaca2 cells with AMG 510 led to increased expression of TFEB and TFE3 (Supplementary Fig. S2A and S2B). To determine whether MiT/TFE factors are required for mediating the MEKi-induced increase in autophagy–lysosome gene expression, we suppressed TFEB via siRNA knockdown in KP4 cells in the presence or absence of MEKi treatment. Knockdown of TFEB led to downregulation of lysosomal genes under control conditions and was sufficient to block MEKi-induced upregulation of autophagy–lysosome gene and protein expression (Fig. 1K; Supplementary Fig. S2K). Finally, we tested if MEKi treatment leads to increased lysosomal biogenesis in PDA patient-derived ex vivo cultured organoids. qPCR analysis showed that mRNA levels of TFEB and TFE3 and several lysosomal genes were significantly upregulated in MEKi-treated organoids compared with DMSO-treated samples (Fig. 1L). Together, these results indicate that MiT/TFE factors are necessary for mediating the upregulation of the autophagy–lysosome program following MAPK pathway suppression.
MYC Is Downregulated upon MAPK Inhibition to Enable Increased Autophagy–Lysosome Activation
Further analysis of our RNA-seq data indicated that in addition to upregulation of lysosome gene expression, a significant downregulation in MYC-dependent transcriptional signatures occurred in response to MEKi or ERKi treatment (Fig. 2A–C; Supplementary Fig. S3A–S3D). MYC is commonly overexpressed and is a protumorigenic driver of PDA (25–27). Moreover, MAPK signaling is known to regulate MYC mRNA (28) and protein levels (29–31). Accordingly, we noted an overall decrease in MYC mRNA (Fig. 2D; Supplementary Fig. S3E) and protein (Fig. 2E; Supplementary Fig. S3F) and the level of nuclear-localized MYC (Fig. 2F and G; Supplementary Fig. S3G–S3I) in several PDA lines upon MEKi or ERKi treatment. Acute treatment of MiaPaca2 cells with AMG 510 similarly led to decreased expression of MYC (Supplementary Fig. S2B) and its target genes (Supplementary Fig. S3J). Analysis of published whole-cell proteomics data (23) also revealed a significant decrease in several MYC targets following compound 4 treatment [48 proteins with fold change (FC) <0.7, P < 0.05; Supplementary Fig. S2C; Supplementary Table S1]. Consistent with these findings, treatment of PDA patient-derived ex vivo organoid cultures with MEKi led to a decrease in MYC expression and its target genes relative to control-treated samples (Fig. 2H).
MYC and MiT/TFE transcription factors belong to the MYC superfamily and preferentially bind to canonical E-box elements (CANNTG) or modified E-box elements (CLEAR motif; GTCACGTGAC), respectively (19, 20, 32). The overlap between the canonical E-box and the CLEAR element (underlined above) suggests that these transcription factors may also display overlapping promoter occupancy. Therefore, we hypothesized that downregulation of MYC following KRAS–MAPK pathway suppression may enable increased promoter occupancy by MiT/TFE factors in PDA cells. To test this hypothesis, we first analyzed the effect of acute knockdown of MYC in PDA cells in the presence or absence of MEKi treatment. Consistent with an antagonistic role for MYC in the regulation of autophagy and lysosome gene expression (33), siRNA-mediated knockdown of MYC under baseline conditions led to a significant increase in the expression of several autophagy–lysosome genes, including TFEB (Fig. 2I). Moreover, downregulating MYC in conjunction with MEKi treatment led to a further increase in TFEB and lysosome gene expression (Fig. 2I). These results suggest that MEKi treatment has opposite effects on MYC and MiT/TFE factor activity that ultimately favor increased induction of lysosome gene expression.
MiT/TFE and MYC Transcription Factors Bind to CLEAR Motifs in Lysosome Gene Promoters in PDA
To directly test whether downregulation of MYC increases lysosome gene expression by allowing increased occupancy of MiT/TFE proteins at their target gene promoters, we analyzed promoter occupancy via ChIP of endogenous MYC or TFE3 in control- and MEKi-treated KP4 cells followed by next-generation sequencing. We detected MYC and TFE3 binding at 7,465 and 4,511 promoter regions, respectively, under baseline conditions—2,821 of which were common to both transcription factors (Fig. 3A). Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis of bound genes common to both factors identified “Lysosome” and “Regulation of autophagy” as two significantly enriched pathways (Fig. 3B). Accordingly, MYC bound to the proximal promoter of several lysosomal genes (Fig. 3C and D), including the TFE3 and TFEB promoters (Supplementary Fig. S4A), in addition to its canonical targets (e.g., HK2, FASN, and LDHA; Supplementary Fig. S4B) in PDA cells. Importantly, treatment of PDA cells with MEKi led to decreased MYC occupancy at several of its known target promoters and several lysosomal genes (Fig. 3D and E; Supplementary Fig. S4A–S4D). In parallel, we observed an increase in TFE3 occupancy at several lysosomal CLEAR elements following MEKi treatment (Fig. 3F and G; Supplementary Fig. S4E and S4F), which was validated by direct ChIP-qPCR (Fig. 3H). These data suggest that MYC occupancy at CLEAR elements serves to limit the magnitude of lysosomal gene expression by MiT/TFE factors in PDA cells.
Under MEKi treatment conditions, analysis of TFE3 promoter binding coupled to our RNA-seq analysis of upregulated genes identified 1,174 genes as direct targets of TFE3 (Supplementary Table S2). KEGG pathway analysis of these 1,174 TFE3 direct targets that were upregulated in MEKi-treated cells identified “Lysosome” and “Regulation of autophagy” as two of the top significantly enriched terms (Fig. 3I). Of these targets, 38 core genes belonged to the lysosomal gene set (16) and were significantly upregulated following MEKi treatment (Fig. 3J; Supplementary Table S3). To confirm that reciprocal expression changes in MYC and MiT/TFE targets result in changes at the protein level, we additionally conducted quantitative whole-cell proteomics from DMSO- and MEKi-treated cells. In line with increased MiT/TFE promoter occupancy and lysosomal gene expression, several lysosomal proteins were significantly upregulated (48 proteins with FC >1.2; P < 0.05) following MEKi treatment, whereas several MYC targets (57 proteins with FC <0.7, P < 0.05) were downregulated upon MEKi treatment compared with control (Fig. 3K; Supplementary Table S4). Together, our ChIP sequencing (ChIP-seq), transcriptome, and proteome analysis establish that MYC, through binding to MiT/TFE target promoters, limits the expression of autophagy and lysosome genes. Following MEKi treatment, reduction of MYC levels facilitates unimpeded binding of TFE3 to target promoters vacated by MYC, leading to increased lysosomal gene expression.
Ferritin Is an Enriched Lysosomal Cargo in Cells Treated with MEKi
To understand how induced autophagy and increased lysosome biogenesis promotes adaptation to KRAS–MAPK pathway suppression, we isolated intact lysosomes (Lyso-IP) from DMSO- and MEKi-treated cells and performed mass spectrometry–based proteomics to identify enriched lysosomal cargo proteins. KP4 cells stably expressing the TMEM192-mRFP-3xHA tag (referred to as LysoTag) were treated with DMSO or MEKi for 48 hours prior to affinity-based capture on HA-conjugated beads as previously described (34). Eluted protein was quantified, and identical concentrations were analyzed by mass spectrometry–based proteomics (Fig. 4A). This analysis identified approximately 616 proteins that were enriched (>1.2-fold) in MEKi-treated samples, of which 97 were statistically significant (Fig. 4B; Supplementary Table S5). Consistent with increased autophagy flux, several autophagy-related proteins (LC3B and GABARAPL2) and autophagy receptors (p62/SQSTM1) were enriched in lysosomes isolated from MEKi-treated cells relative to DMSO treatment, whereas LAMP1 and NPC1 served as protein normalization controls, and were equally detected in each sample by both proteomics and immunoblotting (Fig. 4C and D).
Of note, ferritin heavy chain (FTH1) and ferritin light chain (FTL) were significantly enriched in lysosomes isolated from MEKi-treated cells (Fig. 4B–D). FTH1 and FTL are subunits of ferritin—the primary cellular iron storage complex (35). Ferritin is degraded by autophagy-dependent capture via the adapter NCOA4, a process referred to as ferritinophagy (36, 37). Lysosomal degradation of ferritin and conversion of ferric iron (Fe3+) to bioactive ferrous iron (Fe2+) via an endolysosomal iron reductase generates a critical source of cellular bioavailable iron—a key micronutrient essential for respiration, energy metabolism, and DNA synthesis (38–40). Direct Lyso-IP followed by immunoblotting confirmed significant enrichment of FTH1 and FTL in lysosome fractions of MEKi-treated cells (Fig. 4D). In addition to FTH1 and FTL, an increase in the level of NCOA4 was detected in MEKi-treated lysosomes (Fig. 4D). Moreover, immunofluorescence-based measurement of the percentage of lysosome localization of FTL in the presence or absence of acute lysosome inhibition with the V-ATPase inhibitor bafilomycin A1 (BafA1) confirmed a higher rate of ferritinophagy flux in MEKi-treated cells (Fig. 4E and F). These results were recapitulated via Lyso-IP from cells treated with MEKi or MEKi combined with BafA1 (Supplementary Fig. S5A), suggesting that an increase in ferritinophagy occurs in response to MEKi treatment. In contrast, no change in mitophagy, as evidenced by lack of mitochondrial proteins in lysosome fractions, was observed in MEKi-treated cells (Supplementary Fig. S5A).
Lysosomal degradation of ferritin leads to the liberation of Fe3+, which is subsequently converted to Fe2+ and transported out of lysosomes via endolysosomal transporters (41, 42). Importantly, lysosome function is required for the maintenance of cellular iron homeostasis (43, 44), and PDA is known to be “ferro-addicted” (45, 46). Consistent with increased ferritinophagy, MEKi treatment led to higher levels of intracellular labile iron relative to DMSO-treated PDA cells as measured by live-cell imaging and flow cytometry of the fluorescent iron reporters, FeRhoNox-1 and FerroOrange (Fig. 4G and H; Supplementary Fig. S5B–S5D). Importantly, cotreatment with MEKi and BafA1 for 6 hours blocked the MEKi-induced increase in intracellular iron (Fig. 4G and H; Supplementary Fig. S5D). The addition of exogenous iron [ferrous ammonium sulfate (FAS)] led to an increase in iron levels, whereas treatment with the iron chelator deferoxamine (DFO) led to a decrease in iron levels, thus serving as positive and negative controls, respectively (Fig. 4G and H). These results suggest that increased lysosome biogenesis and ferritinophagy are part of an adaptive response to MEKi treatment to regulate intracellular iron homeostasis.
Of note, we found that MEKi and ERKi treatment led to a significant increase in the levels of FTH1, FTL, and NCOA4 protein and FTH1 and FTL mRNA in several PDA cell lines (Fig. 4I and J; Supplementary Fig. S5E–S5G). Treatment of MiaPaca2 cells with AMG 510 similarly led to an increase in FTH1 and FTL protein levels (Supplementary Fig. S5H). Moreover, MEKi treatment of tumor-bearing mice caused a decrease in ERK1/2 signaling and MYC protein levels that coincided with an increase in the lysosome protein NPC1, FTH1, and FTL (Fig. 4K). Regulation of FTH1 and FTL expression is coordinated with iron levels within the cell, whereby an iron-replete state induces increased translation of FTH1 and FTL (46). Moreover, MYC is a reported negative regulator of FTH1 (47), and we show that siRNA-mediated knockdown of MYC leads to a significant increase in FTH1 and FTL transcript levels in PDA cells consistent with a suppressive role for MYC (Supplementary Fig. S5I). An iron-replete state is also associated with decreased uptake of iron-bound transferrin (Tf) by transferrin receptor (TfR1; ref. 38). Importantly, we found that the total mRNA and protein levels of TfR1 were decreased in MEKi-treated cells (Fig. 4L; Supplementary Fig. S5J and S5K), as well as the levels of plasma membrane TfR1 and Tf uptake (Fig. 4M and N; Supplementary Fig. S5L). We also found that expression of the iron exporter ferroportin (FPN) was decreased 48 hours after MEKi treatment (Fig. 4L). These results suggest that MEKi treatment induces an iron-replete state that is mediated by increased ferritinophagy, whereas decreased FPN ensures reduced flux at the plasma membrane. Taken together, our findings suggest a feed-forward circuit in which increased ferritin synthesis and degradation serve to maintain an iron-replete state in KRAS–MAPK pathway–suppressed PDA cells. In parallel, downregulation of MYC contributes to transcriptional induction of FTH1 and FTL.
Ferritinophagy-Mediated Iron Supply Promotes Increased Levels of Mitochondrial ISC Proteins in MEKi-Treated Cells
Redox-active iron within the cytosolic labile iron pool is utilized for the synthesis of heme and ISC (39, 48). ISCs are essential cofactors for several proteins within complex I, II, and III of the mitochondrial electron transport chain (ETC), whereas additional proteins within complex II, III, and IV contain heme (48). Analysis of our whole-cell proteomics data from KP4 cells treated for 48 hours with MEKi indicated an increase in several ISC-containing proteins, particularly those involved in mitochondrial ETC complexes (Fig. 5A; Supplementary Table S6). Direct immunoblotting in MEKi-treated PDA cells further confirmed an increased abundance of aconitase 2 (ACO2) and respiratory chain complex proteins including NDUFS1, NDUFS2, NDUFS3, NDUFS5, NDUFS7, and UQCRFS1 (Fig. 5B; Supplementary Fig. S6A). Of note, the mRNA levels corresponding to mitochondrial ISC proteins and the levels of outer mitochondrial membrane proteins, VDAC and TOM20, were largely unchanged (Supplementary Fig. S6B and S6C), indicating that increased transcription or mitochondrial number does not contribute to the upregulation of mitochondrial ISC proteins following MEKi treatment. Similarly, treatment of MiaPaca2 cells with AGM 510 also led to upregulation of several ISC proteins (Supplementary Fig. S6D). Consistent with our findings, analysis of a published proteomics data set generated from MiaPaca2 cells treated with compound 4 (23) showed a significant increase in FTH1, FTL, NCOA4, and ISC proteins involved in mitochondrial respiration (Supplementary Fig. S6E; Supplementary Table S7). Unlike mitochondrial ISC proteins, those associated with DNA repair and transcription were not increased in response to MEKi treatment (Supplementary Fig. S6F) or oncogenic KRAS inhibition (Supplementary Fig. S6G), suggesting that selective increase of ISC proteins associated with mitochondrial respiration is a conserved response to KRAS–MAPK pathway inhibition.
Mitochondrial Activity Is Increased in MEKi-Treated Cells
To determine whether increased mitochondrial ISC protein abundance translates to increased mitochondrial activity, we first measured mitochondrial respiratory chain complex assembly into supercomplexes, which increases respiratory efficiency (49). Comparative analysis of supercomplex formation via blue native page (BN-PAGE) 48 hours after MEKi treatment showed greater abundance of high-molecular-weight respiratory supercomplexes relative to DMSO-treated cells (Fig. 5C).
To determine whether the MEKi-induced increase in mitochondrial ISC proteins and supercomplexes supports enhanced mitochondrial function, we first performed an in-gel activity assay to measure complex I activity of isolated mitochondria from KP4 cells. In accordance with our results, MEKi treatment led to an increase in mitochondrial complex I enzymatic activity (Fig. 5D). Additionally, we observed a significant increase in mitochondrial ACO2 activity in MEKi-treated cells (Fig. 5E), suggesting that the increased iron level in MEKi-treated cells supports mitochondrial tricarboxylic acid cycle and contributes to ETC activity.
Consistent with increased supercomplex assembly and in-gel activity, the respiratory activities of individual complex I and complex II (Fig. 5F), as well as basal and maximal respiration (Fig. 5G and H), measured via Seahorse, were increased in MEKi-treated cells. Accordingly, pathways associated with increased mitochondrial activity such as “fatty acid metabolism” were also enriched in our whole-cell proteomics data set generated from MEKi-treated cells (Supplementary Fig. S6H; Supplementary Table S8). Consistent with these findings, MEKi treatment sensitized PDA cells to the mitochondrial complex I inhibitor phenformin (Supplementary Fig. S6I and S6J) or iron chelation with DFO (Supplementary Fig. S6K).
Autophagy–Lysosome Inhibition Blocks MEKi-Induced Upregulation of Mitochondrial ISC Proteins and Mitochondrial Activity
Our findings highlight a pivotal role for transcriptional induction of lysosome biogenesis and activation of ferritinophagy in driving metabolic reprogramming in response to MAPK pathway suppression in PDA. To establish a requirement for selective autophagy and the lysosome, we suppressed ferritinophagy or overall lysosomal activity and measured the level of ISC proteins in MEKi-treated cells. We found that blocking ferritinophagy via knockdown of NCOA4 was sufficient to suppress MEKi-induced upregulation of ISC proteins in KP4 cells (Fig. 6A), whereas iron supplementation with ferric ammonium citrate (FAC) was sufficient to rescue the level of ISC proteins under conditions in which NCOA4 knockdown was combined with MEKi treatment (Fig. 6A). Similarly, direct lysosome inhibition with chloroquine (CQ) also blocked MEKi-induced upregulation of ISC protein expression (Fig. 6B; Supplementary Fig. S7A). Of note, MEKi treatment phenocopied FAC treatment, confirming our finding that increased intracellular iron is associated with MAPK pathway inhibition (Fig. 6B; Supplementary Fig. S7A). CQ treatment also blocked MEKi-induced ETC complex activity as measured by in-gel activity assays, which could be rescued following the addition of FAC (Fig. 6C). Importantly, ferritinophagy blockade via short hairpin RNA (shRNA)–mediated knockdown of NCOA4 was sufficient to suppress MEKi-induced upregulation in mitochondrial respiration as measured by Seahorse (Fig. 6D and E). These results underscore the critical requirement for lysosomal-derived iron as a major contributor to the generation of ISC-containing proteins and in stabilizing and enhancing mitochondrial respiratory supercomplexes in response to MEKi treatment.
Suppression of Ferrintinophagy or Lysosomal Activity Sensitizes PDA Cells to MEKi Treatment and Inhibits Growth
To test whether blocking ferritinophagy together with MEKi treatment is sufficient to block PDA growth, we next knocked down NCOA4 in combination with MEKi treatment. We observed a dramatic decrease in PDA growth (Fig. 6F and G) and induction of apoptosis as evidenced by increased levels of cleaved PARP (Fig. 6H). Importantly, iron supplementation was able to reverse the growth inhibition and apoptotic response induced following combined NCOA4 knockdown and MEKi treatment (Fig. 6F–H). These data highlight the importance of ferritinophagy-mediated iron supply for sustaining mitochondrial respiration and PDA viability upon MAPK pathway suppression.
Prior studies showed inhibition of PDA growth following cotreatment with MEKi or ERKi and lysosome inhibitors (11, 12). We confirmed that combined treatment with MEKi (Fig. 6I and J) or ERKi (Supplementary Fig. S7B and S7C) and low-dose BafA1 or treatment with MEKi and low-dose CQ (Supplementary Fig. S7D and S7E) was sufficient to block growth across multiple human PDA cell lines. Importantly, we show that exogenous iron supplementation is sufficient to rescue the growth inhibition observed following combined MAPK and lysosome blockade (Fig. 6I and J; Supplementary Fig. S7B–S7E). Altogether, our results pinpoint a key role for autophagy–lysosome-dependent iron delivery in supporting cell proliferation following MAPK pathway inhibition.
The Lysosome–Mitochondria Signature Is Maintained following Long-term MAPK Inhibition
To test whether coordinated lysosome–mitochondrial activity is maintained during long-term MAPK pathway suppression, we chronically treated KP4 cells with MEKi for 60 days. These cells maintained reduced phospho-ERK1/2, confirming pathway suppression. Importantly, chronic MEKi treatment led to increased TFEB, FTH1, FTL, and mitochondrial ISC proteins and a decrease in MYC (Supplementary Fig. S8A). Similarly, mouse iKRAS cells in which oncogenic Kras can be inducibly suppressed following the withdrawal of doxycycline (Dox) from the culture media (5) also showed an increase in TFE3, LAMP1, LC3B, FTH1, and FTL (Supplementary Fig. S8B) and a corresponding increase in mitochondrial ISC proteins (Supplementary Fig. S8C) following 32 days of Dox withdrawal. As with human PDA cells, iKRAS cells also maintained reduced MYC levels (Supplementary Fig. S8B). To test if increased mitochondrial activity is maintained following prolonged exposure to MEKi treatment, we treated KP4 cells for 8 days with MEKi and measured basal and maximal respiration or individual respiratory complex activity using Seahorse. Similar to acute treatment (Fig. 5F–H), 8-day treatment led to increased mitochondrial respiration (Supplementary Fig. S8D and S8E) and individual complex activity (Supplementary Fig. S8F) compared with DMSO treatment. Finally, analysis of publicly available data sets identified upregulation of the lysosome gene signature and downregulation of MYC gene targets in response to MEKi treatment of colorectal cancer (HCT116; Fig. 7A; ref. 50) and NSCLC (A549, H2030, and H460; Fig. 7B; ref. 51) cells, suggesting that reciprocal regulation of these two pathways may be a broad feature of MAPK pathway inhibition.
MYC and MiT/TFE Activity Signatures Are Anticorrelated across KRAS-Driven Cancers
MYC and MiT/TFE factors display increased activity in PDA (16, 26, 27) due in part to their increased expression. However, our data suggest that upon KRAS–MAPK pathway suppression, a reduction in MYC levels effectively eliminates competition for binding to CLEAR elements, favoring enhanced promoter occupancy by MiT/TFE factors and increased lysosome gene activation. Based on this finding, we examined The Cancer Genome Atlas (TCGA) data sets to determine whether, a priori, MYC and MiT/TFE signatures are inversely correlated. We first generated a MYC score (MYC_score) and a MiT/TFE score (Lyso_score) based on the sum of the z-score of 240 MYC target genes and 135 MiT/TFE target genes, respectively. We observed an anticorrelation between high MYC_score and Lyso_score across three KRAS-driven cancer types [pancreatic adenocarcinoma (PAAD), NSCLC, and colon adenocarcinoma (COAD); Fig. 7C]. Likewise, stratification of PAAD, NSCLC, and COAD into high and low Lyso_score based on the median z-score also showed an inverse correlation with MYC_score (Fig. 7D).
Collectively, our findings uncover an early adaptive response to KRAS–MAPK pathway suppression that involves enhanced MiT/TFE transcriptional activity, which supports increased ferritinophagy flux and lysosome degradation to sustain iron-dependent mitochondrial activity. Our data also highlight a more general anticorrelation between MiT/TFE and MYC transcription factor activity, which may serve to modulate the magnitude of lysosome gene expression. However, in response to KRAS–MAPK suppression, a shift favoring the MiT/TFE program may contribute to establishing a permissive cell state that enables survival.
Mutational activation of KRAS and increased downstream signaling via the MAPK pathway are near universal in PDA and function to increase proliferation in part through rewiring of cellular metabolism. Prior studies have shown that KRAS activation drives increased glycolytic flux to provide key intermediates of glucose metabolism that fuel anabolic pathways including hexosamine biosynthesis and generation of nucleic acids (5, 10, 52). Accordingly, genetic suppression of KRAS or treatment with small-molecule inhibitors targeting downstream signaling nodes including MEK or ERK show a significant decrease in glucose uptake and glycolysis (5, 11) and a corresponding increased reliance on autophagy (11, 12) and mitochondrial activity (15) in PDA. These studies point toward a switch in metabolic dependency following KRAS–MAPK pathway suppression, and our findings provide mechanistic insight into how this switch occurs (Fig. 7E).
First, we show that despite high levels of baseline autophagy and lysosome activity in PDA (16), this pathway can be further elevated via increased engagement of the MiT/TFE transcription factors in response to MAPK pathway inhibition. Our findings indicate that, at baseline, nuclear-localized TFE3 binds to lysosome gene promoters but is restricted from reaching maximal occupancy due to the competitive binding of MYC to the same loci. MiT/TFE transcription factors belong to the same superfamily as MYC, but they do not heterodimerize (53), and so their common presence at lysosome gene promoters is unlikely due to physical interaction with one another. Instead, both MYC and MiT/TFE factors are highly expressed in PDA (16, 54), which may allow for broader occupancy across canonical and noncanonical E-box elements. Accordingly, downregulation of MYC—via genetic suppression of MYC or KRAS or inhibitor-based suppression of KRAS, MEK, or ERK—allows MiT/TFE proteins unrestricted occupancy at autophagy and lysosome gene promoters, leading to further upregulation of autophagy and lysosome gene expression in PDA. Although signaling-mediated mechanisms that impinge on ULK1 activity have also been shown to increase autophagy flux following MAPK pathway inhibition (11, 12), our findings suggest that transcriptional induction ensures sustained and coordinated autophagy–lysosome pathway induction in response to stress.
Recent studies have highlighted a stringent dichotomy between MiT/TFE proteins and MYC in the regulation of lysosome gene expression in acute myeloid leukemia (AML; ref. 55), Hela cells, and induced pluripotent stem cells (33) and in the maintenance of stem cell renewal (56). These studies show mutual exclusivity of MiT/TFE and MYC expression patterns, corresponding gene signatures, and cellular programs. Our analysis of TCGA data sets indicates that this dichotomous state exists at baseline in several KRAS-driven tumors. MAPK pathway suppression and the subsequent decrease in MYC transcript and protein levels tilt the balance in favor of MiT/TFE factors by effectively eliminating competition for promoter binding and allowing a persistent engagement of MiT/TFE proteins with their target gene promoters. Interestingly, Yun and colleagues show that overexpression of TFEBS211A—a constitutively active nuclear form of TFEB—was able to bypass MYC repression and induce lysosome gene expression in AML cells (55). This finding is in line with our prior study showing that MiT/TFE proteins are upregulated and constitutively nuclear in PDA cells, which may enable partial bypass of MYC-driven suppression to maintain high baseline levels of autophagy and lysosome activation (16). Unlike the study by Yun and colleagues (55), in which TFEB was proposed as a tumor suppressor in AML due to its ability to regulate programs favoring differentiation, MiT/TFE proteins are protumorigenic and sustain tumor cell metabolism and high rates of autophagic and macropinocytic flux in PDA (10, 17). Hence, the role of MiT/TFE factors in tumorigenesis may depend on cell type and cell state and the specific gene programs that are activated.
A second key insight is our observation that NCOA4-mediated ferritinophagy is upregulated in the context of unrestricted MiT/TFE activation following MEKi treatment. Immunoisolation of lysosomes followed by proteomics-based profiling identified increased degradation of ferritin in MEKi-treated cells relative to control cells, whereas other forms of selective autophagy, such as mitophagy, were not induced. Increased ferritinophagy led to a subsequent increase in intracellular labile iron levels. Paradoxically, despite increased degradation of ferritin, we also noted that FTH1 and FTL transcript and protein levels are elevated in MEKi-treated cells. This result is consistent with a prior study showing that MYC negatively regulates ferritin expression (47). We confirmed that knockdown of MYC or treatment with MEKi, which leads to downregulation of MYC, causes an increase in FTH1 and FTL transcripts in PDA cells. Taken together, we propose that ferritin is dually regulated at the transcriptional and protein levels in response to MAPK pathway suppression. The purpose of a dual increase in ferritin production and degradation may be to ensure that all iron is captured by ferritin and delivered to lysosomes prior to distribution elsewhere in the cell. Recent work supports a role for the lysosome as an obligate conduit for mitochondrial iron uptake by ensuring that iron transfer to mitochondria is facilitated by direct contact with endolysosomes (57).
Accordingly, our studies highlight an essential role for lysosome-derived iron in the maintenance of ISC-containing proteins critical for mitochondrial ETC activity and respiration capacity. ISCs serve as critical cofactors for several proteins involved in DNA replication and repair, the tricarboxylic acid cycle, and mitochondrial respiration (38, 39). Mitochondrial ISC proteins are components of complexes I, II, and III of the ETC and are required for the assembly, stabilization, and function of these complexes (48). We find that, in response to MEKi treatment, several ISC proteins of the ETC are upregulated and assemble into stable supercomplex structures that promote enhanced respiratory output. Further analysis of our proteomics data set also showed enrichment for fatty acid metabolism signatures in MEKi-treated cells, which may generate metabolites that fuel increased oxidative phosphorylation. Increased reliance on mitochondrial output was previously shown to be a hallmark of PDA cells that no longer depend on oncogenic KRAS for growth (15). Our studies provide important mechanistic insight into how this reliance is established and the role of enhanced autophagy–lysosome pathway activity in this switch. Likewise, our studies suggest that the synergistic antitumor activity resulting from combined MAPK pathway and lysosome inhibition reported previously (11, 12) is due, in large part, to the inability of MEKi-treated cells to mobilize lysosome-derived iron following cotreatment with CQ. Consistent with this idea, iron supplementation was able to rescue the growth and viability of MEKi-treated cells in which ferritinophagy or lysosome activity was simultaneously suppressed. Although the lysosome plays a central role in the regulation of proliferative signaling (58), cholesterol homeostasis and lipid transfer (59), metabolic regulation (60, 61), and maintenance of ion homeostasis including calcium (62) and iron (43, 44, 62), our studies suggest that lysosome dependency following MEKi treatment centers on the regulation of iron. Together, our findings highlight a critical role for multiorganelle cross-talk, incorporating NCOA4-mediated ferritinophagy, lysosome-dependent iron supply, and mitochondrial iron-dependent respiration, in promoting adaptation and survival following MAPK pathway inhibition.
Upregulation of ferritinophagy was also observed following treatment of PDA cells with ERKi, genetic suppression of oncogenic KRAS, or treatment with the KRASG12C inhibitor AMG 510. Thus, reliance on ferritinophagy and lysosomal iron supply to fuel mitochondrial activity appears to be a broadly applicable dependency following KRAS–MAPK pathway suppression and may therefore have important therapeutic implications. Moreover, we show that upregulation of MiT/TFE proteins and ferritinophagy is retained after 60 days of MEKi treatment, suggesting that cells may utilize this stress response pathway to sustain viability long enough to acquire genetic alterations that enable chronic resistance to KRAS–MAPK inhibition. Results from ongoing clinical trials combining MEKi/ERKi treatment with the lysosomal inhibitor hydroxychloroquine will help to evaluate the long-term potential of this combination in suppressing PDA progression. Moreover, directly targeting ferritinophagy in the context of KRAS–MAPK inhibition may lead to more potent or sustained antitumor activity.
Cell Culture and Reagents
MiaPaCa2, KP4, PaTu8902, PaTu8988T, Panc-1, PSN1, YAPC, and HEK293T cell lines were obtained from ATCC. MiaPaca2, KP4, PaTu8902, PaTu8988T, Panc-1, YAPC, and HEK293T were cultured in DMEM (Gibco) supplemented with 10% FBS (Atlanta Biologicals), 1% penicillin/streptomycin (Gibco), and 15 mmol/L HEPES (Gibco). PSN1 cells were grown in RPMI media (Gibco) supplemented with 10% FBS (Atlanta Biologicals) and 1% penicillin/streptomycin (Gibco). Mouse-inducible KRAS (iKRAS) cell lines were a gift from Dr. Ronald DePinho (MD Anderson Cancer Center) and were cultured as previously described (63). All cell lines were grown in a humidified chamber at 37°C in 5% CO2. Cells were trypsinized using TrypLE (Gibco). Cell lines are tested for Mycoplasma using the MycoAlert Detection Kit (Lonza, LT07-418) at least once a month, and the cell lines were authenticated by short tandem repeat fingerprinting. Cell lines were passaged for a maximum of 15 passages upon thawing prior to replacement.
Trametinib and SCH779284 were purchased from Selleckchem. AMG 510 (sotorasib) was purchased from MedChem Express. Hydroxychloroquine was purchased from Sigma. BafA1 was purchased from CST. FAC, FAS, NADH, deferoxamine, and ferrostatin were purchased from Sigma. Iron binding dye FeRhoNox-1 was purchased from Goryo Chemical, and FerroOrange was purchased from Dojindo. Mitochondrial complex I inhibitor phenformin was purchased from Cayman Chemicals. Puromycin was purchased from Gibco. All chemicals were used at the indicated concentrations. Subcellular fractionation was performed using the NE-PER nuclear and cytoplasmic extraction kit (Thermo Fisher, 78833).
pLJM1-TMEM192-mRFP-3xHA was generated as previously described (34, 64). pMRX-IP-GFP-LC3-RFP-LC3ΔG (Addgene, plasmid #84572) was a gift from Noboru Mizushima, University of Tokyo, Japan.
The shRNA vector pLKO.1-TRC-puromycin was obtained from the Sigma MISSION TRC shRNA library. pLKO.1-TRC vector containing the human shRNA targeting NCOA4 (TRCN0000236186; 5′-TCAGCAGCTCTACTCGTTATT-3′) was cloned according to the Addgene shRNA cloning protocol.
Lentivirus was produced by transfecting HEK293T cells with lentiviral vector pLKO.1 containing shRNA sequence and packaging plasmids psPAX2 (Addgene, plasmid #12260) and pMD2.G (Addgene, plasmid #12259) at a ratio of 4:3:1 using X-tremeGENE transfection (6365787001, Sigma-Aldrich) reagent following the manufacturer's instructions. The supernatant containing the virus was collected after 48 hours by passing through a 0.45-μm filter, aliquoted, and stored at −80°C until further use. Cells were infected with media containing virus using polybrene reagent (TR-1003-G, EMD Millipore) following the manufacturer's protocol and selected for at least 48 hours in 2 μg/mL of puromycin.
Predesigned Silencer select siRNAs were purchased from Invitrogen. Cells were transfected with Silencer select siRNA using Lipofectamine RNAiMAX transfection reagent (Invitrogen, 13778100) following the manufacturer's instruction. Briefly, 250,000 cells were plated in 6-well plates and allowed to attach overnight. siRNA (0.5 μmol/L) in 100 μL of optiMEM media (Gibco) was mixed with 5 μL of RNAiMax solution in OptiMEM and incubated at room temperature (RT) for 15 minutes. The mixture was added dropwise to the cells in media without antibiotics and incubated for at least 8 hours. Cells were transfected on 2 consecutive days, and downstream assays were conducted for 48 hours. The IDs for the siRNAs used in this study are as follows: siCTRL (4390846, Invitrogen), siTFEB (s15495, Invitrogen), and siMYC (s9129, Invitrogen).
Autophagic Flux Assay
Stable cell lines expressing the GFP-LC3-RFP-LC3ΔG construct were subjected to DMSO (control) or trametinib treatment for 48 hours. Following treatment, cells were lifted using TrypLE and resuspended in FACS buffer (1× PBS containing 2% FBS), and the fluorescence was measured by flow cytometry. The ratio of RFP over GFP fluorescence was reported as autophagic flux.
Cells were washed twice with ice-cold PBS and lysed in either RIPA lysis buffer (50 mmol/L Tris-HCl pH 8.0, 150 mmol/L sodium chloride, 1% NP-40, 0.5% sodium deoxycholate, 0.1% SDS supplemented with 10 mmol/L β-glycerophosphate and protease inhibitor; Fisher Scientific; A32965) or regular lysis buffer (50 mmol/L HEPES pH 7.4, 40 mmol/L sodium chloride, 10 mmol/L sodium-pyrophosphate, 1 mmol/L β-glycerophosphate, 50 mmol/L sodium fluoride, 2 mmol/L EDTA, 1% Triton X-100 supplemented with protease inhibitor). The protein concentration was quantified using the Pierce BCA Protein Assay Kit (Life Technologies, 23227) following the manufacturer's instructions. Protein (15–20 μg) was resolved on SDS-PAGE (8%–15%) and transferred onto PVDF membrane (EMD Millipore; IPVH00010). The membrane was blocked with 5% skim milk and incubated overnight with primary antibody (see Supplementary Table S9 for antibody details) diluted in 5% skim milk or 5% BSA at 4°C. Membranes were washed in 1× TBS-T and incubated for a 1 hour at RT with species-specific horseradish peroxidase–conjugated secondary antibody. Membranes were subsequently washed and developed using supersignal West Pico Chemiluminescent substrate (Fisher Scientific, 34080), and the images were captured using the ChemiDoc XRS+ System (Bio-Rad).
RNA Extraction and Quantitative PCR
Total RNA was extracted using the PureLink RNA Mini Kit (12183025, Thermo Fisher). cDNA synthesis was performed using iScript Reverse Transcription Supermix (1708841, Bio-Rad) according to the manufacturer's instruction. RT-qPCR was performed using iTaq Universal SYBR Green Supermix (1725122, Bio-Rad) on a CFX384 Touch Real-Time PCR Detection System (Bio-Rad) following standard protocols. The relative gene expression level was analyzed by the comparative Ct method and was normalized to 18s rRNA. Sequences of primers used are listed in Supplementary Table S10.
RNA-seq and Analysis
For transcriptome profiling of PDA cell lines, Illumina-compatible RNA-seq libraries were created from purified RNA using the Tecan Universal mRNA Plus kit (9156-A01). Libraries were sequenced using an Illumina HiSeq 4000 using single-end, 50-bp reads at the UCSF Center for Advanced Technology (http://cat.ucsf.edu/). Sequencing reads were aligned to the human reference genome (GRCh38), and reads per gene matrix were counted using the latest Ensembl annotation build using STAR (65). Read counts per gene were used as input to DESeq2 (66) to test for differential gene expression between conditions using a Wald test while correcting for possible covariates. Genes passing a multiple testing correct P value of 0.1 [false discovery rate (FDR) method] were considered significant.
For public RNA-seq data analysis, .txt files containing normalized counts for the colon cancer cell line (HCT116; GSE118490) and NSCLC cell lines (A549, HCT116, and H460; GSE110397) were downloaded from the Gene Expression Omnibus (GEO) database. GSEA comparing DMSO and MEKi conditions was conducted with Broad Institute GSEA software (version 4.2.0; ref. 67) using the KEGG and Hallmark gene set in the Molecular Signatures Database (MSigDB). Further visualization was performed using dotplot in R (version 1.3.1073).
Cells were cross-linked with 1% formaldehyde for 10 minutes, and the reaction was stopped with 125 mmol/L glycine at RT. Cells were lysed in lysis buffer (150 mmol/L NaCl, 50 mmol/L Tris-HCl pH 7.5, 5 mmol/L EDTA pH 8, 0.5% NP-40, 1.0% Triton X-100) for 10 minutes, and cell slurry was homogenized further with a 20 G needle. The lysed cells were pelleted and resuspended in shearing buffer (50 mmol/L Tris-HCl pH 8, 10 mmol/L EDTA pH 8, 1% SDS) supplemented with the protease inhibitor and sonicated for 25 minutes using a Covaris Focused Sonicator at 4°C. Shearing of the chromatin was verified via agarose gel electrophoresis after reverse cross-linking. Chromatin (100 μg) was diluted 10 times in ChIP dilution buffer (50 mmol/L Tris-HCl pH 8, 0.167 mol/L NaCl, 1.1% Triton X-100, and 0.11% sodium deoxycholate) and 3 μg of antibody (MYC: 9402, CST, TFE3: HPA023881, Sigma, IgG: 2729, CST) was added per immunoprecipitation (total volume 1.9 mL) and incubated overnight at 4°C. Immunoprecipitated chromatin was captured using 60 μL of Protein A Dynabeads (Invitrogen) and incubated with rotation for 2 hours, followed by sequential washing with ice-cold RIPA buffer containing 150 mmol/L salt (50 mmol/L Tris-HCl pH 8, 0.15 mol/L NaCl, 1 mmol/L EDTA pH 8, 0.1% SDS, 1% Triton X-100, 0.1% sodium deoxycholate) or 500 mmol/L salt (50 mmol/L Tris-HCl pH 8, 0.5 mol/L NaCl, 1 mmol/L EDTA pH 8, 0.1% SDS, 1% Triton X-100, 0.1% sodium deoxycholate) and twice with lithium chloride buffer (50 mmol/L Tris-HCl pH 8, 1 mmol/L EDTA pH 8, 1% Nonidet P-40, 0.7% sodium deoxycholate, 0.5 mol/L LiCl2). The final wash was performed with 1× TE buffer (10 mmol/L Tris-HCl pH 8, 1 mmol/L EDTA pH 8) and eluted for 4 hours at 55°C with elution buffer (10 mmol/L Tris-HCl pH 8, 0.3 mol/L NaCl, 5 mmol/L EDTA pH 8, 0.5% SDS) containing RNaseA. Eluted chromatin was separated from the magnetic beads and reverse cross-linked overnight with Proteinase K at 65°C prior to phenol/chloroform extraction using phase lock tubes (Qiagen) and ethanol precipitation. Eluted DNA was used for library preparation and qPCR analysis.
ChIP-seq Library Preparation and Analysis
Illumina-compatible libraries were created from post-ChIP DNA with the Illumina DNA Flex library preparation method (Illumina, 20018705) following the manufacturer's protocol. DNA libraries were then pooled to run on a MiniSeq for quality assurance prior to sequencing on an Illumina HiSeq 4000 with single-end, 50-bp reads. All libraries were sequenced to a depth of at least 28 million reads.
ChIP-seq FASTQs were mapped and peak-called using ENCODE's chip-seq-pipeline2 (https://github.com/ENCODE-DCC/chip-seq-pipeline2) from 2 biological replicates and their replicates. IDR-Optimal peak set at a threshold of 5% was used for ChIP experiment–specific analysis. Using GencodeV38's “All” Comprehensive Gene Annotation (68) for GrCh38.p13 and the lysosomal gene set and MYC target gene set, promoter regions were designated as ±1 kb from each transcription start site. Using deepTools bamCompare (69), ChIP input–normalized bigwig files were generated (70) with options: –binSize 50 –scaleFactorsMethod None –normalizeUsing CPM –smoothLength 1,000 –extendReads 275 –centerReads. Profile plots were generated using deepTools computeMatrix (69) with options: –referencePoint center -a 2000 -b 2000 –skipZeros. The original MACS2 narrowpeak peaksets were also sorted by signalValue score for unique gene promoters from highest to lowest for further analysis of the top-scoring genes.
Lysosome Purification and Proteomics
Intact lysosomes were purified as previously described (34). In brief, cell lines stably expressing TMEM192-mRFP-3xHA were treated with DMSO or 100 nmol/L MEKi for 48 hours alone or in combination with BafA1 (400 nmol/L, 4 hours). The cells were scraped in ice-cold KPBS buffer (136 mmol/L KCl, 10 mmol/L KH2PO4 pH 7.25) supplemented with Pierce protease inhibitor and were collected by centrifugation. The pellets were resuspended in KPBS buffer supplemented with 50 mmol/L sucrose and 0.5 μmol/L TCEP and mechanically lysed, followed by centrifugation at 2,700 rpm for 10 minutes. The supernatant containing the organelles was incubated with 50 μL of anti-HA–conjugated Dynabeads (Thermo Scientific, 88837) for 30 minutes with rotation. Lysosome-bound beads were washed 3 times and eluted overnight at 4°C using KPBS buffer containing 0.1% NP-40 detergent. Protein concentration was measured using the Pierce BCA Protein Assay Kit. Equal amounts of protein from each condition were used for mass spectrometry–based proteomics or immunoblotting.
Mass spectrometry analysis was performed as follows. A trichloroacetic acid precipitation was performed to remove detergents by adding 1 volume of 6.1N trichloroacetic acid to 4 volume of sample and incubated on ice for 10 minutes. The samples were centrifuged at 14,000 rpm at 4°C for 5 minutes, and the supernatant was removed. The pellet was then washed twice with 200 μL of cold acetone at 14,000 rpm at 4°C for 5 minutes, and the residual acetone was allowed to evaporate.
The precipitated protein was resuspended in 100 μL of 6 mol/L urea in 100 mmol/L Tris pH 7.8 and treated with 5 μL of 200 mmol/L DTT in 100 mmol/L Tris pH 7.8 for 60 minutes to reduce the disulfide bonds. The resulting free cysteine residues were subjected to an alkylation reaction. The samples were subjected to tryptic digest at 37°C overnight with gentle shaking. Urea, Tris, and other nonvolatile reagents in the sample were removed by solid-phase extraction using Sep-Pak Plus C18 Cartridges (Waters Corp; WAT020515) according to the manufacturer's specifications.
The peptide solutions were manually injected on a Shimadzu microflow high-performance liquid chromatography (HPLC) system consisting of two LC20AD pumps, a CBM20A controller, and a FRC10A fraction collector. Fractions were analyzed by reversed-phase HPLC using Waters NanoAcquity pumps and autosampler and a Thermo Fisher Orbitrap Elite mass spectrometer using a nano flow configuration. Peptides were identified from the MS data using SEQUEST algorithms. A species-specific database was generated from NCBI's nonredundant (nr.fasta) database and concatenated to a database of common contaminants (keratin, trypsin, etc.). The resulting data were then loaded into Scaffold (Proteome Software), and a minimum of two peptides and a peptide threshold of 95% and protein threshold of 99% were used for identification of peptides and protein-positive identifications.
For comparative analysis between DMSO- and MEKi-treated lysosomal elutes, minimum peptide abundance was set to 1 for all replicates. FC was calculated between the DMSO (3 replicates)- and MEKi (3 replicates)-treated samples, and statistical significance was calculated using a two-tailed unpaired t test. The data were represented as a volcano plot displayed as the log2 of the FC and the −log10 of the P value.
Mass spectrometry–based proteomics was performed as previously described (23). Briefly, cell pellets from KP4 cells were lysed using 8 mol/L urea and 200 mmol/L 4-(2-hydroxyethyl)-1-piperazinepropanesulfonic acid (EPPS) at pH 8.5 with protease inhibitors. Protein (50–100 μg) was aliquoted for each condition and TMT channel for further downstream processing. Proteins were digested overnight with Lys-C (1:100, enzyme: protein ratio) at RT. The next day, trypsin (1:100 ratio) was added and incubated at 37°C for an additional 6 hours. To each digested sample, 30% anhydrous acetonitrile was added and 50 μg of peptides were labeled using 100 μg of TMT reagent. To equalize protein loading, a ratio check was performed by pooling 2 μg of each TMT-labeled sample and analyzed by LC-MS/MS. Normalization factors were calculated from this label check, and samples were mixed 1:1 across all TMT channels and desalted using a 100-mg Sep-Pak solid-phase extraction cartridge. Pooled TMT-labeled peptide samples were fractionated with basic-pH reverse-phase (bRP) HPLC, and fractions were subsequently desalted using StageTips prior to analyses using LC-MS/MS.
All mass spectrometry data were acquired using an Orbitrap Lumos mass spectrometer in line with a Proxeon NanoLC-1200 UHPLC system. Peptides were separated using an in-house 100 μm capillary column packed with 40 cm of Accucore 150 resin (2.6 μm, 150 Å; Thermo Fisher Scientific) using a 120-minute LC gradient from 4% to 24% acetonitrile in 0.125% formic acid per fraction. Eluted peptides were acquired using a synchronous precursor selection (SPS-MS3) method for TMT quantification as previously described (71). Intelligent data acquisition using real-time searching was performed using Orbiter (72). All acquired data were processed using Comet (73) and a previously described informatics pipeline (74). Briefly, peptide spectral libraries were first filtered to a peptide FDR of less than 1% using linear discriminant analysis using a target decoy strategy. Spectral searches were done using a custom fasta-formatted database that included common contaminants and reversed sequences (Uniprot Human, 2020). Resulting peptides were further filtered to obtain a 1% protein FDR, and proteins were collapsed into groups. For quantitation, a total sum signal-to-noise of all reported ions of 100 was required for analysis. Lastly, protein quantitative values were normalized so that the sum of the signal for all proteins in each channel was equal to account for sample loading. For comparative analysis, the FC was calculated between the DMSO (3 replicates)- and MEKi (3 replicates)-treated samples, and statistical significance was calculated using a two-tailed unpaired t test. The data were represented as a volcano plot displayed as the log2 of the FC and the −log10 of the P value.
Cells were cultured on coverslips coated with fibronectin and treated with DMSO or drug for the indicated time. Following treatment, cells were washed twice with PBS and fixed in ice-cold methanol at −20°C for 5 minutes or 4% PFA for 10 minutes at RT. PFA-fixed cells were permeabilized in 0.1% saponin for 10 minutes at RT. Cells were blocked with 5% normal goat serum for 15 minutes at RT. After blocking, cells were incubated overnight with primary antibody (for antibody details, refer to Supplementary Table S9) at 4°C. Subsequently, cells were washed 3 times with PBS and incubated in species-specific fluorophore-conjugated secondary antibody for 45 minutes at RT. Coverslips were mounted using Fluoromount-G containing DAPI (100–20, SouthernBiotech) and allowed to dry overnight. Cells were imaged using a Zeiss Laser Scanning Microscope (LSM) 710 using a 63× objective. Image processing and quantification were performed using ImageJ software. Measurement of colocalization was performed using thresholded images using the image calculator function in ImageJ. Percent overlap of pixels in the red and green channel was calculated from a minimal of 10 fields per condition. Total fluorescence was measured for at least 60 to 100 cells per condition using the mean fluorescence intensity function in ImageJ following background fluorescence subtraction.
Iron Quantification Using Live-Cell Imaging and Flow Cytometry
Quantification of cellular iron by live-cell imaging was performed using FerroOrange dye (Dojindo, F374) following the manufacturer's instructions. Briefly, cells were seeded on Poly-D-Lysine–coated 35 mmol/L glass-bottom MatTek dishes (MatTek P35GC-1.5-14-C) and treated with DMSO or drugs as indicated. Following treatment, cells were washed twice with 1× PBS and incubated with 1 μmol/L of FerroOrange for 30 minutes at 37°C. After incubation, cells were imaged live using a Leica SP5 microscope equipped with 63× objective at 37°C and 5% CO2. At least 10 representative fields were captured per condition, and the images were processed using ImageJ software. Mean fluorescence intensity was calculated on a per-cell basis for at least 60 to 100 cells after subtracting the background fluorescence of each image.
For iron quantification using flow cytometry, cells were stained with 5 μmol/L FeRhoNox-1 (Goryo Chemical, SCT030) for an hour or 1 μmol/L FerroOrange dye (Dojindo, F374) for 30 minutes at 37°C. Following incubation, cells were washed, trypsinized, and resuspended in FACS buffer. One million cells were counted and used for analysis.
Transferrin Uptake Assay
Two thousand cells were seeded in a 96-well plate and treated with DMSO or drug as indicated. On the day of the experiment, cells were starved for 30 minutes in media without FBS at 37°C. Starved cells were prechilled on ice for 5 minutes and pulsed with prechilled Alexa fluorophore 488–conjugated transferrin (Thermo Fisher, T13342, 50 μg/mL) resuspended in serum-free media for 30 minutes on ice. Following incubation, the media were replaced with prewarmed media containing FBS, and cells were immediately transferred to 37°C for 10 minutes to initiate transferrin uptake. Cells were then washed twice with 1× PBS, fixed using 4% PFA for 10 minutes at RT, and stained with DAPI. Fluorescence was recorded using a plate reader (Cytation5, BioTek) and transferrin fluorescence intensity was normalized to DAPI.
Transferrin Receptor Quantification by Flow Cytometry
To quantify transferrin on the plasma membrane, cells treated with DMSO or drug were washed, detached with TrypLE, and resuspended in ice-cold FACS buffer. Cells were stained with FITC-labeled transferrin receptor antibody (BioLegend, 334104) for 1 hour on ice, and 1 million cells were counted per condition in triplicate.
Crude Mitochondrial Extraction and BN-PAGE
Crude mitochondrial fraction was isolated as described (75) with slight modifications. Cells were pelleted, washed, and resuspended in 500 μl ice-cold PBS supplemented with protease inhibitor prior to mechanical lysis. Ice-cold PBS (500 μL) was added to the suspension prior to centrifugation at 1,000 × g for 10 minutes at 4°C. The supernatant containing mitochondria was transferred to a new tube and centrifuged at maximum speed for 10 minutes at 4°C, and pelleted mitochondria were resuspended in ice-cold PBS. Protein concentration was measured using Pierce BCA Protein Assay Kit.
BN-PAGE was performed using the Invitrogen NativePAGE system. Crude mitochondria (100–200 μg) were pelleted and resuspended in buffer containing 50 mmol/L imidazole and 1 mol/L 6-aminohexanoic acid pH 7. Digitonin (GoldBio D-180–2.5) was added at a ratio of 4 g per gram of protein. The mixture was incubated on ice for 15 to 20 minutes and spun down at maximum speed for 30 minutes at 4°C. The supernatant containing mitochondrial supercomplexes was mixed with 1/3 volume of sample buffer (5% Coomassie brilliant blue G250 in 1 mol/L 6-aminohexanoic acid). Sample (15–20 μg) was loaded on 3% to 12% Bis-Tris Native PAGE gel (Invitrogen, BN1002BOX) with Native PAGE anode (Invitrogen, BN2001) and dark blue cathode buffer (Invitrogen, BN2002) at 90 V on ice for 30 minutes. Dark blue cathode buffer was exchanged with light blue cathode buffer, and the gel was run for an additional 90 minutes at 300 V. Gels were transferred onto PVDF membrane and blotted with mitochondrial complex antibodies (for antibody details refer to Supplementary Table S9). As a loading control, one set of samples was run in parallel and stained with EZ-Blue gel staining solution (Sigma, G1041).
In-gel Activity Assay
To measure the activity of mitochondrial respiratory complexes, the crude mitochondrial extract was isolated and run on a precast Invitrogen Bis-Tris (3%–12%) Native PAGE gel as described above with slight modifications. For in-gel activity, clear PAGE was performed where the gel was run with light blue cathode buffer for 30 minutes and exchanged with the cathode buffer without any blue dye additive. The gel was incubated in ice-cold water for 20 minutes. Complex I activity was performed in dark at RT by adding freshly prepared activity buffer containing the substrate [0.1 mol/L Tris-HCl pH 7.5, 1 mg/mL nitro blue tetrazolium (NBT), 0.14 mmol/L NADH]. The reaction was carried out until the color was fully developed (15–20 minutes) and stopped with 10% acetic acid solution. The gel was washed in distilled water and scanned.
Aconitase Activity Assay
Mitochondrial aconitase activity assay was performed using an aconitase activity kit (Abcam, ab183459) following the manufacturer's protocol. Briefly, cells were harvested and homogenized in a cold assay buffer using a Dounce homogenizer. For mitochondrial aconitase activity, the homogenized cells were spun down at maximum speed for 15 minutes at 4°C and the pellet was resuspended in ice-cold assay buffer and sonicated for 20 seconds. The activity measurement was normalized to the total protein content.
Oxygen Consumption Rate Using Seahorse
The oxygen consumption rate was measured using a Seahorse XFe96 analyzer (Agilent). KP4 cells were treated with DMSO or drug or infected with shCTRL or shNCOA4 as indicated. On the day of the experiment, cells were trypsinized and 50,000 cells resuspended in Seahorse media (DMEM with 8 mmol/L glucose, 2 mmol/L pyruvate, and 2 mmol/L glutamine) were seeded on Seahorse 96-well plate precoated with 25 μg/mL Cell-Tak (Corning, 354240). Respiratory rates were measured with sequential injections of 1 μmol/L oligomycin, 750 nmol/L FCCP twice, and 1 μmol/L antimycin and 1 μmol/L rotenone.
Individual Mitochondrial Complex Activity Using Seahorse
Assessments of individual respiratory chain complexes were performed with a Seahorse XFe96 analyzer (Agilent) according to a previously established protocol (76). Briefly, 40,000 cells were plated in quadruplicate wells of a Cell-Tak–treated XFe96 microplate and allowed to seed for 4 hours. Immediately prior to assay, cells were washed twice with 100 μL of mitochondrial assay solution (220 mmol/L mannitol, 70 mmol/L sucrose, 10 mmol/L KH2PO4, 5 mmol/L MgCl2, 2 mmol/L HEPES, and 1 mmol/L EGTA) and then overlaid with 175 μL of mitochondrial assay solution supplemented with the Seahorse Plasma Membrane Permeabilizer (Agilent), 4 mmol/L ADP, 10 mmol/L sodium pyruvate, and 1 mmol/L malate. Cells were then sequentially subjected to 1 μmol/L rotenone, 10 mmol/L succinate, 1 μmol/L antimycin A, and 10 mmol/L ascorbate with 100 μmol/L N,N,N’,N’-tetramethyl-ρ-phenylene diamine.
To assess cell growth following treatment with MEKi in combination with BafA1, cells were seeded on 12-well plates at the following densities: KP4, 35,000; MiaPaca2, PaTu8988T, 30,000; PSN1, 45,000; PaTu8902, 50,000; YAPC, 80,000 per well) and allowed to attach for 24 or 48 hours for YAPC. Cells were treated with DMSO or MEKi (100 nmol/L for KP4, YAPC, PaTu8988T; 10 nmol/L for PaTu8902, MiaPaca2, and PSN1) with or without low-dose (1–2.5 nmol/L) BafA1. Rescue of growth was performed by the addition of FAC (50–300 μmol/L) in the presence of ferrostatin (1–2 μmol/L). Growth was assessed after 6 to 7 days without media change.
To assess growth with ERKi, cells were seeded at the densities described above and treated with DMSO or ERKi (1 μmol/L for KP4, PaTu8988T, YAPC; 100 nmol/L for PaTu8902, MiaPaca2, and PSN1) in the presence or absence of low-dose BafA1 (1 nmol/L for KP4, PaTu8902, PSN1; 0.75 nmol/L for YAPC; 2 nmol/L for PaTu8988T). Cell growth was rescued with FAC (50–300 μmol/L) and ferrostatin (1–2 μmol/L) for 6 to 7 days without media change. For treatment with CQ, cells (MiaPaca2, 30,000; YAPC, 80,000 cells per well) were treated with DMSO or MEKi (100 nmol/L) in the presence or absence of 12.5 μmol/L CQ and rescued with 50 μmol/L FAC and 1 μmol/L ferrostatin for 7 days. At the conclusion of each assay, cells were fixed in ice-cold methanol and stained with a 0.1% crystal violet stain.
For proliferation assays with NCOA4 knockdown, cells (shCTRL and shNCOA4) were seeded in a 12-well plate (KP4, 35,000; PaTu8902, 50,000 cells per well) and treated with MEKi (100 nmol/L for KP4, 10 nmol/L for PaTu8902) for 6 days. FAC (50 μmol/L for PaTu8902 and 150 μmol/L for KP4) and ferrostatin (1 μmol/L for PaTu8902 and 2 μmol/L for KP4) were added on day 1 without media change. Cells were stained as described above.
Organoids were established based on previously described protocols (77). Organoids were dissociated and 300,000 single cells per well were seeded in 10% growth factor–reduced Matrigel (Corning, 356231) and 90% human organoid feeding medium into 6-well plates. Cells were treated with either DMSO or 100 nmol/L trametinib for 48 hours before harvest. At harvest, organoid cells were washed twice with ice-cold PBS, resuspended in Cell Recovery Solution (Corning, 354253) for 30 to 60 minutes and washed 2 to 3 times with ice-cold PBS to get rid of Matrigel prior to pelleting. Cell pellets were snap-frozen on dry ice until use.
Mouse Xenograft Studies
All experiments were performed in accordance with the University of California, San Francisco Institutional Animal Care and Use Committee–approved protocol (AN181382). Nude mice (10 weeks old) were purchased from The Jackson Laboratory and injected subcutaneously on both flanks with 2.5 × 106 KP4 cells. Once tumors reached ∼150 mm3, mice were randomized and treated with vehicle [5% (hydroxypropyl)methyl cellulose] or MEKi (trametinib 2 mg/kg) for 11 days via oral gavage. The length and width of the tumors were measured 3 times a week using a caliper, and the volumes were calculated according to the following formula: (length × width2)/2. At endpoint, tumors were dissected, proteins were extracted with RIPA lysis buffer, and lysates were generated for analysis of target protein expression.
Analysis of TCGA Patient Data Set
RNA-seq data from TCGA database for PAAD, NSCLC, and COAD tumor tissues carrying KRAS mutations were downloaded from the UCSC Xena portal (http://xena.ucsc.edu). To compare the relative expression between lysosomal gene signature (16) and MYC target genes (Hallmark MYC targets), a z-score was calculated for each gene within the respective gene sets. Next, the z-scores for all genes within a gene set were summed to generate a “Lyso_Score” and “MYC_score” for each tumor. A Pearson correlation test (GraphPad Prism) was then performed between the Lyso_score and MYC_score. Lyso_score-high and -low tumors were stratified based on whether a given score for an individual tumor was above or below the median Lyso_score. Statistical analysis was performed using the Student t test.
Results are represented as mean ± standard deviation unless otherwise specified, and each experiment was repeated at least 3 times. Statistical analysis was performed using GraphPad Prism software (version 9.2.0). The statistical significance between two groups was calculated using an unpaired two-tailed Student t test. For multiple groups, one-way ANOVA or two-way ANOVA was used. A P value less than 0.05 was considered statistically significant.
RNA-seq and ChIP-seq data are available via the accession code GSE206269 in the GEO database. Mass spectrometry data have been deposited in MassIVE with accession codes MSV000089676 and MSV000089741. All other data supporting the findings of this study are available from the corresponding author upon request.
A.J. Aguirre reports grants and personal fees from Mirati Therapeutics, Revolution Medicines, and Syros Pharmaceuticals, grants from Novo Ventures, Deerfield Management, Bristol Myers Squibb, and Novartis, and personal fees from Boehringer Ingelheim, Servier Pharmaceuticals, Reactive Biosciences, Arrakis Therapeutics, T-Knife Therapeutics, AstraZeneca, and Merck outside the submitted work. J.D. Mancias reports grants from the National Institute of Diabetes and Digestive and Kidney Diseases (R01-DK124384), the Burroughs Wellcome Fund (1014767.01), The Sidney Kimmel Foundation, and the Hale Family Center for Pancreatic Cancer Research during the conduct of the study; grants from the Damon Runyon Cancer Research Foundation and Novartis outside the submitted work; and a patent for the modulation of NCOA4-mediated autophagic targeting of ferritin (PCT/US2015/023142) issued. G.M. DeNicola reports grants from the NIH/NCI during the conduct of the study. R.M. Perera reports grants from the NIH during the conduct of the study. No disclosures were reported by the other authors.
M. Ravichandran: Conceptualization, data curation, formal analysis, validation, investigation, methodology, writing–original draft, writing–review and editing. J. Hu: Conceptualization, data curation, formal analysis, validation, investigation, methodology, writing–original draft, writing–review and editing. C. Cai: Formal analysis, methodology. N.P. Ward: Data curation, formal analysis. A. Venida: Data curation. C. Foakes: Data curation. M. Kuljanin: Data curation. A. Yang: Data curation. C.J. Hennessey: Data curation. Y. Yang: Data curation, investigation. B.R. Desousa: Data curation. G. Rademaker: Data curation, investigation. A.A.L. Staes: Data curation. Z. Cakir: Data curation. I.H. Jain: Supervision. A.J. Aguirre: Resources, supervision. J.D. Mancias: Data curation, formal analysis, supervision, methodology. Y. Shen: Data curation, formal analysis, supervision, methodology. G.M. DeNicola: Data curation, formal analysis, supervision, validation, methodology. R.M. Perera: Conceptualization, resources, formal analysis, supervision, funding acquisition, writing–original draft, project administration, writing–review and editing.
This work was supported by NCI grants R01CA240603 and R01CA260249, a Damon Runyon-Rachleff Innovation Award, NIH Director's New Innovator Award (DP2CA216364), the Shorenstein Fund, the Helen Diller Family Comprehensive Cancer Center, and the Ed Marra Passion to Win Fund (to R.M. Perera); NCI grant R37CA230042 (to G.M. DeNicola); and the Lustgarten Foundation, Dana-Farber Cancer Institute Hale Family Center for Pancreatic Cancer Research, the Doris Duke Charitable Foundation, the Pancreatic Cancer Action Network, and NIH/NCI K08 CA21842002 and P50CA127003 (to A.J. Aguirre). We thank Hani Goodarzi and Sohit Miglani for advice on ChIP-seq analysis as well as Suprit Gupta and Grace Hernandez for technical assistance. We thank Lenka Maliskova and Walter L. Eckalbar from the UCSF Genomics CoLab for assistance with RNA-seq and ChIP-seq library preparations.
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.
Note: Supplementary data for this article are available at Cancer Discovery Online (http://cancerdiscovery.aacrjournals.org/).