Abstract

   The RAB interacting factor (RABIF) is a putative guanine nucleotide
   exchange factor that also functions as a RAB-stabilizing holdase
   chaperone. It has been implicated in pathogenesis of several cancers.
   However, the functional role and molecular mechanism of RABIF in
   hepatocellular carcinoma (HCC) are not entirely known. Here, we
   demonstrate an upregulation of RABIF in patients with HCC, correlating
   with a poor prognosis. RABIF inhibition results in decreased HCC cell
   growth both in vitro and in vivo. Our study reveals that depleting
   RABIF attenuates the STOML2-PARL-PGAM5 axis-mediated mitophagy.
   Consequently, this reduction in mitophagy results in diminished
   mitochondrial reactive oxygen species (mitoROS) production, thereby
   alleviating the HIF1α-mediated downregulation of glycolytic genes HK1,
   HKDC1, and LDHB. Additionally, we illustrate that RABIF regulates
   glucose uptake by controlling RAB10 expression. Importantly, the
   knockout of RABIF or blockade of mitophagy sensitizes HCC cells to
   sorafenib. This study uncovers a previously unrecognized role of RABIF
   crucial for HCC growth and identifies it as a potential therapeutic
   target.

   Subject terms: Hepatocellular carcinoma, Oncogenes
     __________________________________________________________________

   RABIF upregulation promotes HCC growth, mitophagy, and glycolysis, and
   enhances sorafenib resistance, suggesting it as a potential therapeutic
   target.

Introduction

   Hepatocellular carcinoma (HCC) is the sixth most common human cancer
   and fourth leading cause of cancer-related mortality, with an annual
   rise worldwide^[38]1. Although several targeted therapies have been
   developed for the treatment of HCC, the prognosis is still poor due to
   therapy resistance and a high rate of recurrence^[39]2. Hence, there is
   a pressing imperative to pinpoint innovative molecular targets and
   unravel the underlying mechanisms, fostering the advancement of novel
   and efficacious therapeutic approaches.

   Hepatocytes exhibit a strong regenerative capacity and demand elevated
   levels of energy. It has been shown that metabolic reprograming is an
   important feature of HCC^[40]3. Mitochondria are the central organelles
   of cells, which play an essential role in cellular metabolism by
   producing ATP via the oxidative phosphorylation (OXPHOS) pathway.
   Numerous reports indicate that a substantial number of hepatocellular
   carcinoma cells possess malfunctioning mitochondria, suggesting their
   reliance on mitochondrial function and a potential association with
   mitochondrial abnormalities^[41]4,[42]5. Mitophagy is a crucial process
   orchestrated by a pathway involving the PINK1 kinase and the PARKIN
   ubiquitin ligase, selectively removes malfunctioning mitochondria to
   preserve a functional mitochondrial network in tumor cells. The
   breakdown products of these dysfunctional mitochondria can be further
   utilized as bioenergetic intermediates to sustain unlimiting
   growth^[43]6. Mitophagy has been implicated in various cancer types,
   including HCC^[44]7–[45]9.

   The RAB GTPase family consists of more than 60 members, is the largest
   branch of the Ras superfamily to date^[46]10, and plays a pivotal role
   in regulating vesicular transport, membrane trafficking, and
   fusion^[47]11–[48]13. It has been well documented that RAB proteins are
   involved in mitophagy pathway^[49]14. For example, one study has
   reported that RAB7 is involved in mitophagosome formation in
   Parkin-mediated mitophagy^[50]15. Some other RAB proteins such as
   RAB8A, RAB8B, and RAB13 were found to be phosphorylated by PINK1 and
   function as downstream effectors of PINK1^[51]16.

   RAB proteins are activated by guanine exchange factors (GEFs) in the
   GTP-bound state. RABIF was one of the first putative RAB GEFs
   identified^[52]17. It binds to a subset of secretory RAB proteins
   (Rab1b, Rab3a, Rab8a, Rab10, Rab12, Rab13 and Rab18) and possesses GEF
   activity towards Rab1, Rab3a, and Rab8a^[53]18. Recent studies indicate
   that RABIF functions as a holdase chaperone rather than a
   GEF^[54]19,[55]20. Several RABIF-interacting RAB proteins have been
   found play important roles in HCC^[56]21,[57]22, suggesting that RABIF
   may be involved in liver cancer development. However, the role and
   underlying mechanisms of RABIF in HCC progression remain to be
   investigated.

   Sorafenib, a multi-target tyrosine kinase inhibitor (TKI), demonstrates
   antiangiogenesis and antiproliferation effects, extending median
   survival in advanced HCC patients from 7.9 to 10.7 months^[58]23. As
   the FDA-approved first-line therapy for advanced HCC, sorafenib
   initially shows promise, but nearly all patients develop resistance
   within a few months. Recent studies indicate that epigenetics,
   transport processes, regulated cell death, and the tumor
   microenvironment play roles in the initiation and development of
   sorafenib resistance in HCC^[59]24. Understanding these molecular
   mechanisms is essential to identifying new therapeutic targets and
   improving outcomes for HCC patients.

   The present study is designed to explore the functions and underlying
   mechanisms of RABIF in HCC. We found that RABIF was overexpressed in
   HCC and was inversely correlated with the prognosis of HCC patients. We
   further demonstrated that RABIF promotes HCC cell growth both in vitro
   and in vivo. Mechanistic studies indicated that RABIF resides in the
   mitochondria and interacts with STOML2. Depletion of RABIF suppressed
   STOML2-PARL-PGAM5 axis-mediated mitophagy, which in turn reduced mtROS
   production and HIF1α-dependent glycolytic gene expression. Our results
   also revealed that RABIF regulated glucose uptake by controlling RAB10
   expression. In addition, we found that depletion of RABIF or
   suppression of mitophagy can overcome sorafenib resistance in HCC,
   which may provide a novel strategy to improve the therapeutic efficacy
   of sorafenib in HCC patients.

Results

RABIF is upregulated in HCC and predicts poor prognosis

   To test the RABIF dependency of HCC, we first evaluated the expression
   pattern of RABIF across cancers using data from the Cbioportal
   datasets. Our analysis showed that RABIF is amplified in several types
   of cancers, particularly in 8% of breast cancers and 6% of liver
   cancers (Fig. [60]1A). Similarly, the amplified RABIF gene was observed
   in an HCC patient-derived xenograft (PDX) cohort from PDXliver
   (Fig. [61]1B). Copy number amplification of RABIF was significantly
   correlated with its mRNA expression (Fig. [62]1C). We observed elevated
   RABIF expression in both unpaired and paired HCC tissues from the TCGA
   datasets (Fig. [63]1D and E). RABIF expression was also significantly
   upregulated at each pathological stage of HCC compared to normal
   tissues (Fig. [64]1F). Moreover, in silico analysis of two independent
   datasets from GEO revealed that RABIF expression was upregulated in HCC
   tissues compared with that in normal tissues (Fig. [65]1G). To further
   confirm the expression of RABIF in HCC, we conducted real-time
   quantitative PCR analysis using a tumor cDNA microarray, including 20
   paired HCC tissues. The analysis showed that RABIF was upregulated in
   15/20 liver tumor samples compared to the matched nonmalignant liver,
   with 9 of 20 (45%) showing a 1.5-fold or greater increase
   (Fig. [66]1H). Next, we explored the clinical relevance of RABIF for
   HCC. Analysis of HCC patient survival according to RABIF mRNA
   expression using the KM plotter showed that high expression of RABIF
   was associated with worse overall and disease-free survival
   (Fig. [67]1I and [68]J). Taken together, these data establish that
   RABIF is highly expressed in HCC and correlates with poor prognosis in
   patients with liver cancer.

Fig. 1. RABIF upregulation correlates with poor prognosis in HCC.

   [69]Fig. 1
   [70]Open in a new tab

   A cBioportal examination of TCGA databases revealed RABIF genomic
   changes in human malignancies. B RABIF amplification in HCC [71]PDX. C
   RABIF expression levels are positively linked to gene copy counts. D, E
   RABIF expression was compared in HCC to nonmatched (D) and matched
   (E) normal tissues. F RABIF expression levels are upregulated in each
   HCC pathologic stage. G The expression of RABIF in HCC tissues and
   normal tissues was examined based on [72]GSE14520 and [73]GSE36376
   datasets. H RABIF expression in cDNA array containing 20 pairs of liver
   tissue was determined by RT-qPCR. I, J Correlation between RABIF
   expression and overall survival (I) as well as disease-free
   survival (J) was analyzed by Kaplan−Meier Plotter
   ([74]http://www.kmplot.com/analysis/). *p < 0.05, **p < 0.01,
   ***p < 0.001.

RABIF promotes HCC cell growth in vitro and in vivo

   Considering the negative correlation between RABIF expression and HCC
   patient survival, we examined the functional impact of RABIF in HCC
   cells. We first established RABIF-depleted Huh7 cells, a commonly used
   HCC cell line, and SK-Hep1 cells. Although SK-Hep1 has been
   historically classified as a hepatocellular carcinoma cell line, more
   recent studies suggest that it may more accurately represent liver
   sinusoidal endothelial cells^[75]25. For the purposes of this study,
   SK-Hep1 cells were used as a liver cell model in the context of RABIF
   depletion, acknowledging the potential difference in cellular origin.
   Successful knockout of RABIF expression was validated by western
   blotting and Sanger sequencing (Fig. [76]2A, Supplementary Fig. [77]1).
   We observed that the proliferation of RABIF-depleted cells was slowed
   down compared to control cells, although this effect was not evident
   during the first 24 hours in HuH7 cells or the second 24 hours in
   SK-Hep1 cells. (Fig. [78]2B). To further confirm the oncogenic effects
   of RABIF on HCC cells, we performed a colony formation assay in Huh7
   and SK-Hep1 cells that did not express RABIF. As shown in Fig. [79]2C,
   RABIF knockout led to a significant decrease in the colony number of
   HCC cells. Conversely, overexpression of RABIF in Huh7 and SK-Hep1
   cells markedly promoted HCC cell growth (Fig. [80]2D and [81]E). In
   addition, overexpression of RABIF enhanced the colony formation ability
   of HCC cells compared to that of control cells (Fig. [82]2F). Next, we
   introduced a CRISPR/Cas9–resistant RABIF (with a synonymous mutation
   that disrupts the protospacer adjacent motif) in RABIF-KO HCC cells
   (Supplementary Fig. [83]2A). The expression of the exogenous RABIF
   successfully restored the impaired growth and colony formation observed
   in RABIF-KO HCC cells (Supplementary Fig. [84]2B and [85]C).

Fig. 2. RABIF promotes HCC proliferation.

   [86]Fig. 2
   [87]Open in a new tab

   A CRISPR/Cas9-mediated knockout of RABIF was confirmed by western blot
   in Huh7 and SK-Hep1 cells. B Cell proliferation was determined by CCK-8
   assay. C Representative images (left) and quantification of colonies
   (right) from control and RABIF KO cells (n = 3 per cell type). D
   Overexpression of RABIF in Huh7 and SK-Hep1 cells was validated by
   immunoblot assay. E Cell proliferation of control and RABIF
   overexpression cells was compared by CCK-8 assay. F Representative
   images (left) and quantification of colonies (right) from control and
   RABIF overexpression cells (n = 3 per cell type). G–L Tumor growth
   curve (G, J), xenograft tumors (H, K) and tumor weight (I, L) in the
   subcutaneous implantation mouse model (n = 7 for each group). B, E, G,
   J Two-way ANOVA; (C, F, I, L) Unpaired t-test. *p < 0.05, **p < 0.01,
   ***p < 0.001.

   Next, we determined the role of RABIF in the development of HCC. Nude
   mice were inoculated with either control or RABIF depletion HCC cell
   lines. Although RABIF depletion appears to inhibit Huh7 xenograft tumor
   growth, a marked reduction in both tumor growth and tumor weight was
   observed in Sk-Hep1 xenograft models (Fig. [88]2G–[89]L). Collectively,
   these results validated the oncogenic function of RABIF in HCC.

RABIF regulates mitophagy through interaction with STOML2

   To gain insight into the potential mechanism that mediates the
   oncogenic function of RABIF in liver cancer cells, we performed
   immunoprecipitation mass spectrometry (IP/MS) analysis to identify
   proteins associated with RABIF. The previously known RABIF interacting
   partner RAB10 as well as several other RAB proteins such as RAB1B and
   RAB8B were immunoprecipitated by RABIF (Fig. [90]3A and Supplementary
   Table [91]2). Unexpectedly, this analysis also identified many
   previously unknown mitochondrial proteins such as STOML2, ATP5B, and
   ATP5D, suggesting that RABIF may regulate mitochondrial function in HCC
   (Fig. [92]3A). To address the involvement of RABIF in regulating
   mitochondrial function, we first examined its subcellular localization.
   Using confocal microscopy, we observed that endogenous RABIF was
   primarily located in the cytoplasm and, to a lesser extent, in the
   mitochondria. This was confirmed by co-staining with MitoTracker
   (Fig. [93]3B). To confirm the presence of RABIF in mitochondria, we
   fractionated the cytoplasm and mitochondria and then conducted
   immunoblotting with antibodies against RABIF and three other
   mitochondrial proteins, TIM23, TOM20, and ATP5B (Fig. [94]3C). We
   detected RABIF not only in the cytoplasm but also in the mitochondria,
   as confirmed by knockout of RABIF using the Crispr/Cas9 system
   (Fig. [95]3C).

Fig. 3. RABIF regulates mitophagy in HCC.

   [96]Fig. 3
   [97]Open in a new tab

   A RABIF-interacting proteins were identified using immunoprecipitation
   mass spectrometry. Non-tagged RABIF was expressed in Huh7 cells, and
   co-purified proteins were identified by MS. Representative top hits are
   displayed. B Immunofluorescence study of RABIF protein subcellular
   localization (green). Mitotracker marks mitochondria. Scale bars,10 μm.
   Co-localization index of RABIF and mitochondria was calculated by Image
   J (right). C RABIF and the specified components of mitochondrial
   complex proteins were detected in the cytosolic and mitochondrial
   fractions using immunoblotting. D Immunoprecipitation assay as applied
   to confirm the interaction between RABIF and STOML2. E Cell lysates
   from control and RABIF KO cells were immunoprecipitated with
   anti-STOML2 antibody. Bound proteins and total cell lysates were
   analyzed by immunoblotting with indicated antibodies. F Huh7 cells were
   treated with CCCP for 4 hours, then cell lysates were analyzed by
   Western blotting with indicated antibodies. The lower band of PGAM5 is
   the cleavage form. Relative intensities of cleaved PGAM5 were shown
   below the band. G Huh7 control and RABIF KO cells were treated with
   CCCP for 4 hours. To assess mitophagy, the cells were fixed and
   immunostained for LC3 (red) and TOM20 (green). The graph shows the
   LC3/TOM20 colocalization puncta per cell calculated by ImageJ. Scale
   bar: 10 μm. H, I Seahorse measurements of the oxygen consumption rate
   (OCR) (H) and their corresponding metabolic rates in (I) in control and
   RABIF KO Huh7 cells. G Two-way ANOVA. I Unpaired t-test. **p < 0.01,
   ***p < 0.001.

   We selected STOML2 for further study due to its reported role in the
   mitophagy process and its essential function in Hepatocellular
   Carcinoma (HCC)^[98]8,[99]26,[100]27. The interaction between RABIF and
   STOML2 was confirmed by immunoprecipitation in Huh7 and Sk-Hep1 cells
   (Fig. [101]3D). It has been reported that STOML2 regulates the activity
   of rhomboid protease presenilin-associated rhomboid-like protein (PARL)
   via forming a protease complex with it^[102]28. To test whether RABIF
   mediates this interaction, we immunoprecipitated STOML2 in control and
   RABIF KO cells and detected PARL by immunoblotting. Intriguingly, our
   results showed that the knockout of RABIF markedly mitigated the
   physical interaction between endogenous STOML2 and PARL (Fig. [103]3E).

   The mitochondrial Ser/Thr protein phosphatase phosphoglycerate mutase 5
   (PGAM5) is a substrate of PARL^[104]29. Processed PGAM5 by PARL has
   been proposed to be involved in the regulation of mitophagy^[105]30.
   Therefore, we next attempted to determine whether knockout of RABIF
   impairs mitophagy in HCC cells through the PARL-PGAM5 axis. Indeed, we
   observed less cleaved PGAM5 in RABIF-deficient cells than in wild-type
   cells treated with CCCP (Fig. [106]3F, Supplementary Fig. [107]3A).
   Expression of the CCCP-induced autophagy marker LC3B II was also
   decreased when RABIF was knocked out (Fig. [108]3F, Supplementary
   Fig. [109]3A). In addition, colocalization of endogenous LC3 with TOM20
   was markedly decreased in RABIF KO cells treated with CCCP
   (Fig. [110]3G, Supplementary Fig. [111]3B). These results indicate that
   depletion of RABIF inhibits mitophagy in HCC.

   Mitophagy is a type of selective autophagy that modulates mitochondrial
   content to maintain energy metabolism^[112]31. It has been reported
   that mitophagy is involved in HCC progression by regulating metabolism
   reprogramming^[113]32. To explore whether RABIF regulates HCC
   metabolism, we conducted seahorse analysis to measure the oxygen
   consumption (OCR) in HCC cells. The results showed that depletion of
   RABIF led to a decrease in OCR, including basal respiration and ATP
   production (Figs. [114]3H, [115]I, Supplementary Fig. [116]3C, [117]D),
   revealing that knockout of RABIF also impaired mitochondrial
   metabolism.

RABIF regulates glycolysis in liver cancer cells

   The metabolic shift from mitochondrial to glycolytic metabolism for
   adapting to the surrounding environment is a distinctive feature of
   most cancer cells, including HCC^[118]33,[119]34. To study whether
   depletion of RABIF is involved in a metabolic shift in HCC cells, we
   performed genome-wide transcriptomic analysis by RNA sequencing in
   RABIF-knockout Huh7 cells. We identified 342 differentially expressed
   genes between RABIF knockout cells and control cells, comprising 287
   upregulated and 55 downregulated genes (Fig. [120]4A). KEGG pathway
   analyses revealed that the metabolic pathway was the most affected
   pathway upon RABIF knockout (Fig. [121]4B), consisting of 42
   differentially expressed genes (Fig. [122]4C). Among these, we found
   three of the most downregulated genes, HK1, LDHB, and HKDC1, were
   involved in glycolysis (Fig. [123]4C). As dysregulation of glycolysis
   has been shown to play an important role in HCC, we therefore examined
   the expression of HK1, LDHB and HKDC1 in control and RABIF KO cells by
   qPCR. The results clearly showed that RABIF depletion significantly
   reduced HK1, LDHB and HKDC1 expression (Fig. [124]4D, [125]E).

Fig. 4. RABIF regulates glycolysis in HCC.

   [126]Fig. 4
   [127]Open in a new tab

   (A) The volcano plot of statistical significance (p < 0.05) versus fold
   change (ratio of KO/Ctr group) shows the most substantially
   differentially expressed genes by genome-wide transcriptome analysis
   between Ctr and RABIF KO Huh7 cells (n = 3). B KEGG pathway enrichment
   analysis of the entire list of differentially expressed genes. C
   Color-coded heat map of differentially expressed genes from the glucose
   metabolism pathway. D, E qRT-PCR assay to validate the expression
   change of glycolytic genes in Huh7 (D) and SK-Hep1 (E) cells. F
   Microscopic analysis of mitoROS production in HCC cells with or without
   RABIF knockout. HCC cells were stained with 1 μM MitoSOX, fixed and
   subjected to imaging. Scale bar: 50μm. G HIF1α and glycolytic enzymes
   were determined by western blotting in control or RABIF depletion HCC
   cells treated with CoCl[2] for 48 h. (H-K) Extracellular acidification
   rate (ECAR) (H, J) and their corresponding metabolic rates (I, K) were
   measured by seahorse in Huh7 (H, I) and SK-Hep1 cells (J, K). D, E, I,
   K Unpaired t-test. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

   HIF-1α is a shared element in pathways responsible for regulating
   cellular metabolism. Many glycolytic genes have been identified as
   targets of HIF-1α in HCC^[128]33,[129]35. It has been reported that
   mitoROS are responsible for the stabilization of
   HIF-1α^[130]36,[131]37. As mitophagy is a widely acknowledged process
   to remove compromised mitochondria that generate elevated levels of
   mitoROS^[132]38, we therefore hypothesized that RABIF regulated
   glycolytic gene expression through mitoROS-mediated HIF-1α activation.
   To validate this hypothesis, we used MitoSOX to measure mitoROS
   production in HCC cells. We found that depleting RABIF significantly
   decreased mitoROS production (Fig. [133]4F). In line with this
   discovery, we noted a decrease in the protein expression of HIF-1α and
   its glycolytic target genes, including HK1, LDHB, and HKDC1
   (Fig. [134]4G).

   To further test whether RABIF regulates glycolytic flux, we
   investigated the effect of RABIF depletion on glycolysis by measuring
   the extracellular acidification rate (ECAR). We observed a marked
   decrease in the basal glycolysis and glycolytic capacity in RABIF
   knockout cells compared to control cells (Fig. [135]4H-[136]K),
   suggesting that RABIF is essential for glycolysis in liver cancer cell.
   Taken together, these data indicated that RABIF depletion affects
   glycolysis in HCC cells.

RABIF enhances RAB10-mediated glucose uptake in HCC

   RABIF has been reported to act as a RAB-stabilizing chaperone of RAB10
   to mediate glucose uptake in adipocytes^[137]19. To test whether RABIF
   was involved in glucose uptake in HCC cells, we first validated the
   interaction between RABIF and RAB10 in HCC cells. Immunoprecipitation
   experiments confirmed their interaction in Huh7 and SK-Hep1 cells
   (Fig. [138]5A). We also found that RABIF depletion led to a significant
   decrease in RAB10 expression (Fig. [139]5B). The knockout of RABIF
   markedly inhibited glucose uptake, whereas overexpression of RAB10 in
   RABIF depletion cells could partially rescue glucose uptake
   (Fig. [140]5C, D). Moreover, enforced expression of RAB10 could also
   restore the cell growth and colony formation defects mediated by RABIF
   depletion (Fig. [141]5E, F). Collectively, our results highlight the
   important role of the RABIF-RAB10 axis in mediating glucose metabolism
   and tumor cell growth in HCC.

Fig. 5. RABIF regulates glucose uptake via RAB10.

   [142]Fig. 5
   [143]Open in a new tab

   A The interaction between RABIF and RAB10 was confirmed by
   immunoprecipitation assay. B RAB10 expression was detected in control
   and RABIF KO cells by western blotting. C Rescue expression of RAB10 in
   RABIF KO cells was validated by western blot. D Glucose uptake was
   measured in control and RABIF KO cells. E, F Overexpression of RAB10
   restored cell proliferation (E) and colony formation (F) of RABIF KO
   cell. D, E, F Two-way ANOVA. *p < 0.05, **p < 0.01, ***p < 0.001.

RABIF confers sorafenib resistance to HCC by regulating mitophagy

   Previous studies have demonstrated that mitophagy promotes resistance
   to sorafenib in HCC^[144]39,[145]40. Considering the role of RABIF in
   mitophagy, we investigated whether RABIF was involved in sorafenib
   resistance. We first analyzed the correlation between RABIF expression
   and the prognosis of sorafenib-treated HCC patients. The results showed
   that high expression of RABIF was correlated with worse patient
   survival (Supplementary Fig. [146]4A). To further confirm the role of
   RABIF in sorafenib resistance, we compared the IC50 between wild-type
   cells and RABIF-knockout cells. We found that RABIF depletion
   significantly decreased the IC50 in HCC cells (Fig. [147]6A, B). In
   addition, sorafenib treatment induced much more apoptosis in
   RABIF-knockout cells than in control cells (Fig. [148]6C-[149]F).
   Noteworthy, knockout RABIF induced a slighter more apoptosis of HCC
   cells without sorafenib treatment, suggesting that the comprised
   mitochondria may be undergo apoptosis when loss of RABIF blocking
   mitophagy. Moreover, we used xenograft tumor models to evaluate whether
   the knockout of RABIF could enhance sorafenib-induced tumor inhibition
   in vivo. Consistent with the in vitro data, the depletion of RABIF
   markedly promoted the tumor-inhibitory effects of sorafenib
   (Fig. [150]6G-[151]J). To determine whether knockout of RABIF
   sensitizes HCC cells to sorafenib by regulating mitophagy, we treated
   HCC cells with sorafenib and examined mitophagy by western blotting.
   Our results indicated that sorafenib treatment induced more PGAM5
   cleavage and LC3B II in control cells than in RABIF KO cells
   (Fig. [152]6K). The immunofluorescence assay also showed that knockout
   of RABIF significantly reduced sorafenib-induced co-localization of
   LC3B and Tom20 (Fig. [153]6L, [154]M). To further confirm that
   RABIF-mediated mitophagy is involved in the anticancer effects of
   sorafenib, we treated HCC cells with Mdivi-1, a mitophagy inhibitor,
   and found that Mdivi-1 treatment inhibited HCC cell proliferation in a
   dose-dependent manner (Supplementary Fig. [155]4B). Co-treatment of HCC
   cells with Mdivi-1 and sorafenib synergistically inhibits HCC growth.
   More importantly, RABIF knockout further enhanced the synergistic
   effects of Mdivi-1 and sorafenib (Supplementary Fig. [156]4C). Taken
   together, these data suggest that RABIF-mediated mitophagy is involved
   in sorafenib resistance in HCC cells.

Fig. 6. RABIF confers sorafenib resistance in HCC.

   [157]Fig. 6
   [158]Open in a new tab

   A, B The half inhibitory concentration (IC50) of sorafenib in the
   respective groups was determined. C–F Flowcytometry analysis of
   sorafenib-induced apoptosis in control and RABIF KO Huh7 (C, D) and
   SK-Hep1 (E, F) cells. The cells were treated with sorafenib as
   indicated dose for 72 h. The graph shows the percentage of apoptotic
   cells. G-J Tumor growth curve (G), representative xenograft tumors (H),
   tumor volumes and tumor weight (J) in the indicated groups (n = 5).
   Sorafenib was intraperitoneally injected at 30 mg/kg for two weeks. K
   Immunoblot showing expression levels of indicated proteins in control
   or RABIF KO cells treated with sorafenib for 48 h. L Control and KO
   cells were treated with sorafenib for 12 h. Then the cells were fixed
   and stained with LC3 (red) and TOM20 (green). The graph shows the
   LC3/TOM20 colocalization puncta per cell calculated by ImageJ (right).
   M Similar analysis as in K was performed in SK-Hep1 cells. D, F, G, M
   Two-way ANOVA; (I, J) One-way ANOVA. *p < 0.05, **p < 0.01,
   ***p < 0.001, ****p < 0.0001.

Discussion

   In this study, we provide novel evidence for an essential role of RABIF
   in HCC. Our data reveal that RABIF is upregulated in HCC and promotes
   HCC cell growth. We found that RABIF was mainly distributed in the
   cytoplasm and partly in the mitochondria. Cytosolic RABIF regulates
   glucose uptake in a RAB10-dependent manner. Upon localization to the
   mitochondria, RABIF exhibited a discernible interaction with STOML2.
   The ablation of RABIF led to a consequential reduction in PARL
   activity, primarily attributed to the diminished interaction between
   STOML2 and PARL. Consequently, the processing of PGAM5 and the
   subsequent mitophagy orchestrated by PGAM5 were impeded following RABIF
   knockout. This disruption in cellular processes resulted in the
   suppression of mitoROS production and downregulation of HIF1α
   expression. Ultimately, these molecular perturbations culminated in a
   discernible reduction in the expression of glycolytic genes.

   Accumulating evidence indicates that RABIF possesses oncogenic
   potential in cancers, including HCC^[159]41–[160]43. However, the
   underlying mechanism of RABIF in HCC is unexplored. Our IP/MS analysis
   shows that RABIF associates with multiple mitochondrial proteins,
   suggesting it may be involved in mitochondrial regulation. Indeed, we
   confirmed its mitochondrial localization through immunofluorescence and
   subcellular fractionation assays. The interaction between RABIF and
   STOML2 was further validated.

   STOML2 is a mitochondrial protein and acts as a scaffold to promote
   STOML2-PARL-YME1L complex formation. The formation of this complex
   influences PARL-mediated processing of both PINK1 and PGAM5^[161]28.
   Our results indicate that the knockout of RABIF disrupts the
   interaction between STOML2 and PARL and impairs PGAM5 processing.
   Recent research has underscored the significance of PGAM5 in various
   cellular processes associated with cancer, including mitophagy^[162]44.
   Dephosphorylation of FUNDC1 and stabilization of PINK1 have been found
   crucial for PGAM5-mediated mitophagy^[163]45,[164]46. Further study is
   needed to identify the downstream effectors involved in the
   RABIF-STOML2-PGAM5 axis-mediated mitophagy in HCC.

   Recently, many studies have demonstrated that RAB GTPases function as
   important regulators of hepatocellular metabolism. For example, RAB32
   binds to lysosomes and regulates Hep3B proliferation and cell growth by
   controlling lysosomal mTOR trafficking^[165]47. RAB24 is increased in
   individuals with non-alcoholic fatty liver disease (NAFLD) and
   regulates blood glucose homeostasis by increasing mitochondrial
   plasticity^[166]48. As an RAB-interacting factor, whether RABIF was
   involved in RAB-mediated hepatocellular metabolism is largely unknown.
   Our IP-MS study revealed that RABIF can interact with several RAB
   proteins, including RAB10. We further validated that RABIF regulates
   glucose uptake in a RAB10-dependent manner. Additionally, we
   demonstrated that depletion of RABIF inhibited glycolysis by reducing
   glycolytic gene expression. Mechanistic studies indicated that RABIF
   deficiency led to a decrease of mitoROS production and subsequent
   destabilization of HIF1α, which is critical for activating glycolytic
   gene expression^[167]49,[168]50. These findings uncover a novel role of
   RABIF in glucose metabolism in HCC.

   It has been shown that RAB10 is involved in hepatocellular metabolism
   by regulating low-density lipoproteins (LDLs) uptake and
   lipophagy^[169]51–[170]53. Indeed, in our RNA-seq analysis, we found
   that RABIF knockout also affected fatty acid and cholesterol
   metabolism. This finding implies that RABIF may potentially engage in
   lipid metabolism in hepatocytes and that RABIF may also promotes HCC
   progression through this mechanism.

   Since it was the first treatment to be approved for HCC, sorafenib is
   still one of the most common first-line therapies for this condition.
   However, the impact of sorafenib on tumor response in pivotal clinical
   trials was attenuated, attributed to the presence of cell resistance to
   sorafenib^[171]54,[172]55. Recent studies indicated that mitophagy was
   involved in sorafenib resistance in HCC^[173]39,[174]56,[175]57.
   Consistent with these findings, our results indicated that RABIF
   confers resistance to sorafenib in HCC. Suppression of RABIF expression
   sensitizes HCC cells and xenograft tumors to sorafenib. The combination
   of the mitophagy inhibitor Mdivi-1 and sorafenib has a synergistic
   effect on growth arrest in HCC.

   In conclusion, our study reveals that elevated RABIF plays an essential
   role in HCC progression and sorafenib resistance. Our findings suggest
   that targeting RABIF may provide novel therapeutic strategy for HCC.

Methods

Cell culture

   Huh7 and Sk-Hep1 human liver cancer cell lines were obtained from
   Shanghai Cell Line Bank of the Chinese Academy of Sciences (Shanghai,
   China). All cells were cultured in DMEM supplemented with 10% fetal
   bovine serum and 1% streptomycin/penicillin, as previously
   described^[176]58. To knock out RABIF in HCC cells, a high-scoring
   guide RNA (target sequence:5′- AGCGAGTTAGTGTCAGCCGA-3′) was designed by
   CHOPCHOP^[177]59 and cloned into the lentiCRISPRv2 vector. The CRISPR
   plasmids were transiently transfected into HCC cells for 48 h, after
   which the cells were passaged and selected using puromycin for 1 week.
   After which, HCC cells were harvested to determine the KO by using
   sanger sequence. To prevent the impact of single-clone selection,
   polyclonal RABIF KO cells were employed.

Cell viability and colony formation analysis

   The cells were seeded into 96-well plates at a density of 2 × 10^3
   cells/well. The CCK-8 solution was added to each 96-well at the
   indicated time points and incubated for 2 h. The absorbance was
   measured using a microplate reader at 450 nm. To determine the IC50 in
   HCC cell lines, HCC cells were seeded into 96-well plates and incubated
   overnight, after which the cells were treated with serial
   concentrations of sorafenib for 72 h. Cell viability was measured using
   the CCK-8 assay.

   Cells were sown and cultivated for 14 days in the medium for the colony
   formation experiment. Colonies were stained with 0.5% crystal violet
   after treatment with a 4% paraformaldehyde solution. ImageJ software
   was used to count the colonies.

RT-qPCR

   TRIZOL (Invitrogen) was used to extract total RNA from HCC cells. The
   PrimeScript RT Reagent Kit (TAKARA) was used to synthesize cDNA. qPCR
   tests with SYBR Green probes (Takara) were performed on an Applied
   Biosystems 7500 Fast Real-Time PCR apparatus. Primer sequences used are
   listed in Supplementary Table [178]S1.

RNA sequencing

   For RNA sequencing, an Agilent 2100 Bioanalyzer (Agilent Technologies,
   Palo Alto, CA, USA), NanoDrop (Thermo Fisher Scientific Inc.), and 1%
   agarose gel were used to measure and quantify the total RNA in each
   sample. A total of 1 μg of RNA with a RIN value of at least 6.5 was
   utilized to prepare the following library. Next-generation sequencing
   libraries were prepared following the manufacturer’s instructions.
   Then, in accordance with the manufacturer’s recommendations, libraries
   with various indices were multiplexed and placed on an Illumina HiSeq
   instrument (Illumina, San Diego, CA, USA). A 2 × 150 bp paired-end (PE)
   configuration was used for sequencing, and HiSeq Control Software
   (HCS) + OLB + GAPipeline-1.6 (Illumina) was used for image processing
   and base calling. The sequences were processed and analyzed using
   GENEWIZ.

Glucose uptake

   The absorption of glucose in Sk-Hep1 or Huh-7 cells (2 × 10^4) was
   quantified using the Glucose Uptake Assay kit (#ab136955, Abcam, UK),
   as per the guidelines provided by the manufacturer. Briefly, cells were
   incubated with the glucose analog 2-deoxyglucose. The oxidation of the
   accumulated 2-DG6P produced NADPH, which in turn oxidized a substrate.

Western blotting

   Cells were lysed with RIPA buffer containing a protease inhibitor
   cocktail. A mitochondrial protein extraction kit (KGB5401, KeyGEN,
   China) was used to extract and isolate mitochondrial and cytoplasmic
   fractions. The proteins were separated on a 4–12% SDS-PAGE gel,
   transferred to a PVDF membrane (Bio-Rad), and detected with appropriate
   antibodies. The RABIF (sc-390759) and PGAM5 (sc-515880) antibodies were
   obtained from Santa Cruz. The β-actin (AC026), GAPDH (AC002) and
   β-tubulin (AC008) antibodies were purchased from Abclonal. RAB10
   (11808-1-AP), TIM23 (67535-1-Ig), TOM20 (11802-1-AP), ATP5B
   (17247-1-AP), LC3B (14600-1-AP), STOML2 (60052-1-Ig) antibodies were
   obtained from Proteintech Group.

Immunofluorescence

   The cells were plated on glass slides overnight and then treated
   progressively in accordance with the applicable experimental conditions
   for the immunofluorescence test. The cells were then fixed with 3.7%
   paraformaldehyde and permeabilized with a permeabilizing solution (0.2%
   [w/v] Triton X-100 in Phosphate buffered saline [PBS]). After blocking
   with 5% bovine serum albumin C for 30 min at room temperature, the
   cells were treated with specific antibodies and fluorescence was
   measured using confocal microscopy (Zeiss).

Immunoprecipitation assay

   The cells were lysed with NP40 buffer (50 mmol/L Tris-HCl, pH7.5,
   150 mmol/L NaCl, 0.5% NP-40, and 50 mmol/L NaF) containing a cocktail
   of protease inhibitors. Lysates were treated with primary antibody for
   12 h at 4 °C before incubation with protein A/G-Sepharose beads
   (sc-2003, Santa Cruz) for 4 h at 4 °C. The beads were extracted by
   centrifugation. After three washes with NP40 buffer, the beads were
   boiled in 40 μL of loading buffer and analyzed by western blotting.

Seahorse assays

   Seahorse assays were performed as previously described^[179]48. A
   Seahorse XF 24 Analyzer was used to measure mitochondrial respiration
   and glycolysis (Agilent Technologies). As a result, 30,000 cells per
   well were plated on Seahorse cell plates as a monolayer culture.
   Mitochondrial respiration was achieved by injecting 2 μM oligomycin
   (Sigma-Aldrich), 1 μM FCCP (Sigma-Aldrich), and 1 μM antimycin A
   (Sigma-Aldrich) + 1 μM rotenone into the cells (Sigma-Aldrich).
   Glycolysis was monitored by injecting 100 mM of 2-deoxy-d-glucose
   (Sigma-Aldrich). The basal OCR and ECAR values indicated the
   differences observed before and after the addition of oligomycin or
   glucose. ATP production was calculated by subtracting the OCR after
   oligomycin injection from the basal respiration level. The glycolysis
   capacity was determined by subtracting the ECAR before glucose addition
   from the ECAR after oligomycin treatment.

Xenograft tumor assays

   The Xuzhou Medical University Animal Care Committee approved the use of
   animals in this study. Male nude mice (n = 7) were subcutaneously
   injected with 1 × 10^6 hepatocellular carcinoma (HCC) cells. The tumor
   size was measured at the indicated time points using calipers. For
   sorafenib treatment assays, when tumor growth reached approximately 100
   mm^3, the mice were randomly divided into four groups (n = 5) and
   administered either vehicle or sorafenib. The mice were euthanized by
   CO[2] after 21 days of therapy, and the tumors were collected and
   weighed.

Statistics and reproducibility

   All results are reported as the mean ± SD of at least three replicate
   studies. GraphPad Prism version 9.3.0 was used for statistical
   analysis. Student’s t-test and one- or two-way ANOVA were used for
   comparisons between groups. *P < 0.05; **P < 0.01; ***P < 0.001;
   ****P < 0.0001.

Reporting summary

   Further information on research design is available in the [180]Nature
   Portfolio Reporting Summary linked to this article.

Supplementary information

   [181]Supplementary Information^ (1.3MB, pdf)
   [182]42003_2024_7028_MOESM2_ESM.pdf^ (74.7KB, pdf)

   Description of Additional Supplementary Files
   [183]Supplementary Data 1^ (92.8KB, xlsx)
   [184]Reporting Summary^ (72.1KB, pdf)

Acknowledgements