The phospholipid flippase ATP9A enhances macropinocytosis to promote nutrient starvation tolerance in hepatocellular carcinoma

Macropinocytosis is an effective strategy to mitigate nutrient starvation. It can fuel cancer cell growth in nutrient‐limited conditions. However, whether and how macropinocytosis contributes to the rapid proliferation of hepatocellular carcinoma cells, which frequently experience an inadequate nutrient supply, remains unclear. Here, we demonstrated that nutrient starvation strongly induced macropinocytosis in some hepatocellular carcinoma cells. It allowed the cells to acquire extracellular nutrients and supported their energy supply to maintain rapid proliferation. Furthermore, we found that the phospholipid flippase ATP9A was critical for regulating macropinocytosis in hepatocellular carcinoma cells and that high ATP9A levels predicted a poor outcome for patients with hepatocellular carcinoma. ATP9A interacted with ATP6V1A and facilitated its transport to the plasma membrane, which promoted plasma membrane cholesterol accumulation and drove RAC1‐dependent macropinocytosis. Macropinocytosis inhibitors significantly suppressed the energy supply and proliferation of hepatocellular carcinoma cells characterised by high ATP9A expression under nutrient‐limited conditions. These results have revealed a novel mechanism that overcomes nutrient starvation in hepatocellular carcinoma cells and have identified the key regulator of macropinocytosis in hepatocellular carcinoma. © 2023 The Authors. The Journal of Pathology published by John Wiley & Sons Ltd on behalf of The Pathological Society of Great Britain and Ireland.


Introduction
Hepatocellular carcinoma (HCC) is one of the most frequent neoplasms and the third leading cause of cancer-related death [1]. Worse, most patients are diagnosed at an advanced stage that is not amenable to surgical treatment, and current therapeutic strategies have pressure, resulting in structurally and functionally abnormal vessels in HCC tissue. Leaky tumour vasculature compromises tumour perfusion and blood-borne nutrient delivery, leading to nutrient limitation. Therefore, it is important for HCC cells to obtain nutrients from the microenvironment to meet their increased bioenergetic requirements of cell proliferation and survival.
Macropinocytosis is a nutrient scavenging pathway that collects macromolecules from the tumour microenvironment (TME) for reuse after degradation, which in turn promotes tumour cell proliferation and growth [10][11][12]. A recent study revealed that hypoxia induces macropinocytosis in HCC cells [13]. Moreover, macropinocytosis allows HCC cells to overcome sorafenib-induced ferroptosis by augmenting the cysteine supply [14]. However, it is unclear whether macropinocytosis occurs universally in HCC cells under nutrient-deprived conditions and whether it fuels HCC cell proliferation when nutrients are limited.
Macropinocytosis is a regulated and non-selective form of endocytosis. Actin-dependent plasma membrane ruffling is the initiating and essential step of macropinocytosis, which is mainly controlled by the activation of Rac family small GTPase 1 (RAC1) and cell division control protein 42 homologue (CDC42) [15][16][17]. The plasma membrane ruffles form on the cell surface in a wave-like fashion and then fold back onto themselves to form macropinocytic cups that close to create macropinosomes containing abundant extracellular soluble substances [18,19]. Subsequently, early macropinosomes undergo maturation and are trafficked to lysosomes for degradation [20]. The underlying regulatory mechanisms are elaborate and complex. Therefore, revealing the underlying mechanism to provide a basis for targeting macropinocytosis in HCC would be of great value.
In this study, we demonstrated that macropinocytosis occurred universally in some HCC cells under nutrientdeprived conditions, which fuelled their proliferation. Repressing macropinocytosis with a pharmacological inhibitor [5-(N-ethyl-N-isopropyl)-amiloride (EIPA)] inhibited HCC cell proliferation and growth, indicating that targeting macropinocytosis may be a promising HCC therapeutic strategy. Thus, identifying the specific regulators of macropinocytosis in HCC would be of great value. Here, we found that the phospholipid flippase ATP9A was a key factor for regulating macropinocytosis HCC cells. ATP9A promotes ATP6V1A membrane trafficking, causing increased plasma membrane cholesterol levels that promote plasma membrane RAC1 enrichment and initiate macropinocytosis. More importantly, ATP9A was overexpressed in HCC cells, and high ATP9A expression was significantly related to a poor outcome in patients with HCC. Herein, we identified ATP9A as a novel regulator for macropinocytosis and revealed that macropinocytosis is a targetable vulnerability of HCC.

Mice and in vivo studies
All animal experimental procedures were performed in accordance with the institutional ethical guidelines and approved by the Sun Yat-sen University Institutional Animal Care and Use Committee (L102042021000J).
As a subcutaneous xenograft model, mice were injected with 1 Â 10 6 MHCC97H cells and then randomly and equally allocated to two groups when the average tumour volume [1/2 Â (width 2 Â length)] was 50-100 mm 3 . Subsequently, the mice were treated with vehicle (DMSO in PBS) or EIPA (7.5 mg/kg, HY-101840, MedChemExpress, Princeton, NJ, USA) by intraperitoneal injection every other day for 1 month. Eight weeks after inoculation, the mice were euthanised, and the tumours were excised and measured. The tumour tissue was divided and used for macropinocytosis detection and ATP level evaluation. The remaining tumour tissue was fixed and paraffin-embedded for immunohistochemical staining.
Orthotopic HCC cell implantation model. A 25 μl cell suspension containing 2 Â 10 5 HCC cells in a PBS-Matrigel mixture (1:1, BD Biosciences, San Diego, CA, USA) was injected into the liver parenchyma. Tumour growth was monitored weekly with the IVIS Imaging System (Calliper Life Sciences, Hopkinton, MA, USA). Three weeks after inoculation, the mice were randomly and equally allocated as indicated (n=6 per group), and the experimental groups were treated with EIPA as described earlier. Eight weeks after inoculation, the mice were euthanised, and the liver was removed intact and weighed. The tumour tissues were treated as described earlier.
patient consent were obtained to use the clinical specimens (B2022-415-01). Details of their clinical and pathological features are listed in supplementary material, Table S1. Tumour clinical stage was determined according to the 8th edition of the tumournode-metastasis (TNM) staging system.

Macropinocytosis assay in vitro and ex vivo
An in vitro macropinocytosis assay was performed as described previously [21]. In brief, HCC cells were seeded on 24-well glass coverslips and treated as indicated for 24 h. Subsequently, TRITC-dextran (T1162, Sigma-Aldrich, St Louis, MO, USA) was added to the medium at a final concentration of 1 mg/ml and incubated at 37 C for 30 min. Then the cells were rinsed with cold PBS five times and fixed in 4% formaldehyde away from light for 30 min. Finally, after 4',6-diamidino-2-phenylindole (DAPI, Sigma-Aldrich) nuclear staining, coverslips were mounted using Dako mounting medium. Fluorescence images were captured with a laser scanning confocal microscope (LMS880, Zeiss, Wetzlar, Germany), analysed, and quantified as described previously [21]. At least five random fields were assessed per sample.
In the ex vivo macropinocytosis assay, the tumours were cut into approximately 125 mm 3 cuboidal shapes, immediately injected with 150 μl TRITC-dextran (10 mg/ml), and immersed in 250 μl TRITC-dextran solution at room temperature protected from light for 30 min. Then the tumours were rinsed with PBS five times and immediately embedded in OCT compound to produce frozen sections, which were air-dried in a fume hood for 10 min at room temperature. Subsequently, the sections were fixed, DAPI-stained, and mounted as described earlier. Fluorescence images were obtained at 40Â magnification and analysed and quantified as described earlier.

Filipin staining
Filipin staining was used to detect the distribution of cholesterol [23]. Cells were cultured in confocal dishes overnight, serum-starved for 24 h, and fixed for study by immunofluorescence. Then the cells were incubated for 2 h with 500 μg/ml filipin complex solution (HY-N6716, MedChemExpress) containing 10% FBS at room temperature in the dark. Images were captured using a Zeiss LMS880 microscope at excitation wavelengths of 340-380 nm.
To add cholesterol to the membranes, cholesterol (C4951, Sigma-Aldrich) was used at 10 μM and added for 90 min prior to fixation or dextran uptake. To decrease membrane cholesterol, 5 mM methylβ-cyclodextrin (MβCD, C4555, Sigma-Aldrich) was incubated with the indicated cells at 37 C for 30 min.

Statistical analyses
All statistical analyses were carried out using SPSS 25 (IBM, Armonk, NY, USA) and GraphPad Prism 8 (GraphPad Inc., San Diego, CA, USA). The relationship between the ATP9A expression level and each HCC clinicopathological characteristic was tested by the χ 2 test (two-sided). Survival curves were plotted using the Kaplan-Meier method and compared by a log-rank test. Survival data were evaluated using univariate and multivariate Cox regression analyses. Other data were compared with Student's t-test (two-sided). P < 0.05 was considered statistically significant in all cases.
Details for plasmids, lentivirus production and transduction, Flow cytometry analysis, immunofluorescence, CCK-8 assay of cell number, protein IP assays, plasma membrane fractionation, glutathione S-transferase (GST) pulldown assay and GTPase activation assay, ATP detection assay, total cholesterol measurement, proximity ligation assays and RNA isolation, and

ATP9A-enhanced macropinocytosis counters nutrient starvation in HCC 19
Reverse Transcription-quantitative PCR (RT-qPCR) are provided in Supplementary materials and methods.

HCC cell nutrient starvation tolerance involves macropinocytosis
A previous study demonstrated that hypoxia induced macropinocytosis in HCC cells [13]. However, it was unclear whether macropinocytosis occurred universally in HCC cells under nutrient deprivation and whether it supported cell proliferation. Evaluation of the HCC cell potential to take up macromolecules (TRITC-dextran) under serum-starved conditions found that the macropinocytosis level differed greatly among the HCC cell lines ( Figure 1A). The MHCC97H and SNU-398 cells exhibited increased macropinocytosis levels under nutrient stress ( Figure 1B,C). The success of serum starvation was verified by detecting decreased p-mTOR levels in cells cultured with serum-free medium (supplementary material, Figure S1A). To explore whether macropinocytosis supported nutrientdepleted HCC cell survival, we evaluated the proliferation of MHCC97H and SNU-398 cell lines cultured in serum-free medium and demonstrated that BSA, the most abundant extracellular protein which could be taken up via macropinocytosis, supported HCC cell growth under nutrient-deprived conditions and markedly increased the S-phase fraction of serum-starved HCC cells, while the macropinocytosis inhibitor EIPA inhibited the proliferation of HCC cells with BSA treatment ( Figure 1D,E and supplementary material, Figure S1B). Western blotting for p-mTOR revealed that macropinocytosis reversed serum starvationinduced mTOR inactivation (supplementary material, Figure S1C). Macropinocytosis dramatically increased cellular ATP levels under nutrient-deprived conditions, while EIPA caused a large decrease in cellular ATP ( Figure 1F,G). We estimated the therapeutic potential of targeting macropinocytosis in vivo using EIPA ( Figure 1H). EIPA administration significantly decreased the tumour volume ( Figure 1H,I), inhibited the tumour proliferative capacity ( Figure 1J), and strongly decreased the tumour intracellular ATP levels ( Figure 1K), indicating that macropinocytosis was indispensable for tumour growth in vivo.
Taken together, these data indicated that macropinocytosis supported HCC cell proliferation under nutrient deprivation conditions and was required for HCC growth in vivo. Restricting macropinocytosis is a promising HCC treatment strategy.

Macropinocytosis in HCC cells requires ATP9A
The phospholipid flippases belonging to the P4 subfamily of P-type ATPases (P4-ATPases) are a protein family involved in membrane ruffle generation during cell motility [24]. To investigate whether P4-ATPases were involved in HCC progression by regulating macropinocytosis, we screened the key macropinocytosis-involved P4-ATPases by silencing the 14 P4-ATPases in turn and determined that knockdown of ATP8A1 and ATP9A obviously inhibited macropinocytosis in MHCC97H cells (Figure 2A, supplementary material, Figure S2A). Both ATP8A1 and ATP9A mRNA were significantly upregulated in HCC tissues in The Cancer Genome Atlas (TCGA) Liver HCC (TCGA-LIHC) dataset, but ATP8A1 expression was exceedingly low in the HCC tissues (supplementary material, Figure S2B). ATP9A mRNA levels were significantly upregulated in the tumour samples in the public TCGA cohort (supplementary material, Figure S2C) and Gene Expression Omnibus datasets (supplementary material, Figure S2D). Further experiments demonstrated that ATP9A expression was notably higher in HCC cells than in normal liver cells (supplementary material, Figure S2E). The expression of ATP9A in HCC cells positively correlated with their macropinocytic abilities (supplementary material, Figure S2F), indicating a potential biological regulation of macropinocytosis by ATP9A. Intriguingly, ATP9A ectopic expression significantly increased the macropinocytosis capacity of HCC cells, while EIPA significantly inhibited its promoting effect ( Figure 2B,C, supplementary material, Figure S2G). ATP9A depletion impaired macropinocytosis as much as EIPA treatment ( Figure 2D,E, supplementary material, Figure S2G).
We then evaluated the role of macropinocytosis in ATP9A-depleted or ATP9A-overexpressed HCC cells and found that BSA-rescued cell proliferative capability, increased S-phase fraction, and intracellular ATP levels were weakened in ATP9A-silencing HCC cells ( Figure 2F-K), whereas the opposite was true with ATP9A overexpression (supplementary material, Figure S3A-F), and the suppressive effect of EIPA was more prominent in ATP9A-high HCC cells. These results indicated that ATP9A was critical in maintaining HCC cell proliferation under nutrient depletion by enhancing macropinocytosis to support nutrient acquisition and increase the energy supply.

ATP9A promotes HCC growth by inducing macropinocytosis in vivo
We then assessed the critical role of ATP9A in HCC tumour growth in vivo. ATP9A ectopic expression dramatically promoted HCC tumour growth in vivo ( Figure 3A,B), decreased survival times ( Figure 3C), and supported the energy supply ( Figure 3D). More importantly, EIPA treatment robustly inhibited the ATP9Amediated tumour growth and extended the survival time of the tumour-bearing mice and decreased the energy supply in mice tumours ( Figure 3A-D). Knockdown of ATP9A markedly inhibited HCC tumour growth, prolonged mouse survival ( Figure 3E-G), and decreased tumour ATP levels ( Figure 3H). Moreover, ATP9A overexpression enhanced tumour macropinocytosis in vivo, while ATP9A silencing weakened it ( Figure 3I-K). The Ki67-positive cells was markedly 20 X Wang, Y Li, Y Xiao, X Huang et al ATP9A-enhanced macropinocytosis counters nutrient starvation in HCC 21

ATP9A-enhanced macropinocytosis counters nutrient starvation in HCC 23
increased in the ATP9A-overexpressing HCC tumours and was decreased in the ATP9A-silenced HCC tumours ( Figure 3L-O). Similarly, EIPA dramatically suppressed the growth of ATP9A-overexpressing HCC tumours ( Figure 3L,M). Altogether, these results demonstrated that ATP9A promoted HCC tumour growth by enhancing macropinocytosis.

ATP9A induces macropinocytosis by regulating cholesterol trafficking and plasma membrane RAC1 localisation
ATP9A is mainly localised to the intracellular compartment and has a critical role in cargo transfer from endosomes to the plasma membrane [25]. We speculated that ATP9A promoted macropinocytosis by mediating plasma membrane RAC1 trafficking. Verification of plasma membrane RAC1 localisation revealed that ATP9A upregulation increased plasma membraneassociated RAC1 (PM-RAC1) levels and that ATP9A downregulation decreased PM-RAC1 levels ( Figure 4A). CDC42, another key factor in regulating membrane ruffling, exhibited no change in the plasma membrane fraction or the WCL regardless of ATP9A expression levels (supplementary materials, Figure S4A). To investigate the functional relationship between ATP9A and RAC1, we inhibited RAC1 expression using siRNA (siRAC1) in ATP9A-overexpressing SNU-449 and HCCLM3 cell lines (supplementary materials, Figure S4B). RAC1 depletion strongly impaired ATP9A-mediated macropinocytosis ( Figure 4B,C). Plasma membrane localisation of RAC1 depends on its activation or the presence of abundant plasma membrane cholesterol [26,27]. However, the GST pull-down assay revealed that neither ATP9A upregulation nor downregulation altered the RAC1-GTP levels (supplementary materials, Figure S4C), suggesting that ATP9A did not affect RAC1 activity. Detection of the cholesterol distribution revealed that ATP9A overexpression resulted in plasma membrane cholesterol accumulation ( Figure 4D) but did not affect the intracellular total cholesterol levels (supplementary materials, Figure S5A). ATP9A silencing specifically led to the loss of plasma membrane cholesterol ( Figure 4E, supplementary materials, Figure S5B).
To clarify whether the contribution of ATP9A to plasma membrane RAC1 and macropinocytosis was associated with plasma membrane cholesterol accumulation, we depleted plasma membrane cholesterol using methylβ-cyclodextrin (mβCD) (Figure 4F), which strongly abolished ATP9A-mediated plasma membrane trafficking of RAC1 ( Figure 4H) and inhibited ATP9A-enhanced macropinocytosis ( Figure 4I). Furthermore, exogenous addition of cholesterol to HCC cells [28] (Figure 4G) restored the plasma membrane RAC1 levels reduced by ATP9A depletion ( Figure 4H) and rescued ATP9A-silencing inhibited macropinocytosis ( Figure 4J). These results indicated that plasma membrane cholesteroldependent RAC1 signalling was involved in ATP9Aregulated macropinocytosis in HCC.

ATP9A promotes plasma membrane cholesterol trafficking by interacting with ATP6V1A
Plasma membrane cholesterol enrichment is mainly mediated by endosome-to-plasma membrane cholesterol trafficking, a process reliant on an acidic endosomal pH [29]. Vesicular trafficking requires vacuolar-type H + -ATPase (V-ATPase)-driven organellar acidification [30]. V-ATPases are involved in membrane trafficking of cholesterol from endocytic organelles [31,32]. Mass spectrometry (MS) analysis of the ATP9A interactome indicated that several V-ATPase-associated proteins were significantly enriched (supplementary material, Data S1), among which ATP6V1A exhibited a predominant difference (supplementary material, Figure S6A). We further confirmed the interaction between ATP9A and ATP6V1A using endogenous and exogenous reciprocal immunoprecipitation (IP) assays ( Figure 5A,B). In addition, the proximity ligation assay (PLA) showed that ATP9A and ATP6V1A colocalised in the cytoplasm of HCC cells ( Figure 5C). Moreover, in vitro binding assays using purified proteins revealed that ATP9A bound with ATP6V1A directly ( Figure 5D, supplementary material, Figure S6B). It has been demonstrated that plasma membrane ATP6V1A is critical for macropinocytosis in RAS-transformed cells [23]. Here, we found that plasma membrane ATP6V1A levels were significantly increased in ATP9A-overexpressing HCC cells and decreased in ATP9A-silencing HCC cells ( Figure 5E,F). To investigate the role of ATP6V1A in ATP9A-enhanced plasma membrane cholesterol and macropinocytosis, we depleted ATP6V1A in ATP9A-overexpressing HCC cells (supplementary material, Figure S6C), which dramatically suppressed

24
X Wang, Y Li, Y Xiao, X Huang et al ATP9A-stimulated macropinocytosis ( Figure 5G,H). Meanwhile, knockdown of ATP6V1A in ATP9Aoverexpressing HCC cells clearly decreased the plasma membrane RAC1 levels ( Figure 5I) and abolished the plasma membrane cholesterol distribution ( Figure 5J). More importantly, ATP6V1A depletion dramatically Figure 4 Legend on next page.

ATP9A-enhanced macropinocytosis counters nutrient starvation in HCC 25
inhibited the effect of ATP9A in promoting nutrientstarved HCC cell proliferation and energy supply ( Figure 5K,L). Collectively, we revealed that ATP9A increased plasma membrane ATP6V1A levels to regulate plasma membrane cholesterol trafficking, thereby promoting macropinocytosis in HCC.
ATP9A is highly expressed in human HCC and predicts a poor outcome To investigate the clinical relevance of ATP9A in HCC, we examined the ATP9A expression by IHC analysis in 264 HCC specimens and analysed the correlation between the expression level of ATP9A and the various HCC clinical characteristics. The clinicopathological characteristics of patients included in this study are summarised in supplementary material, Table S1. ATP9A expression levels were higher in tumour tissues, especially advanced-stage tumours ( Figure 6A). More importantly, ATP9A was highly expressed in the tumour tissues surrounding necrotic lesions, indicating that it may support HCC survival via macropinocytosis, as necrotic cell debris is a macropinocytic fuel source. Moreover, high ATP9A expression was associated with advanced stages (T and TNM stage), tumour recurrence, and poor survival ( Figure 6B and supplementary material, Table S3). Kaplan-Meier survival analysis demonstrated that patients with high ATP9A expression had shorter recurrence-free survival (RFS) and overall survival (OS) ( Figure 6C,D), and high ATP9A expression was an independent predictive factor of RFS and OS (supplementary material, Figure S7A,B, Tables S4 and  S5). Furthermore, we analysed the clinical correlation of the ATP9A-ATP6V1A-RAC1 axis in HCC and demonstrated that ATP9A expression positively correlated with plasma membrane ATP6V1A and RAC1 and Ki67 levels ( Figure 6E,F and supplementary material, Table S3). Altogether, these results suggested that ATP9A overexpression may predict a poor outcome for patients with HCC. In summary, our study found that ATP9A upregulation promoted ATP6V1A-mediated cholesterol trafficking and the accumulation of PM-RAC1, leading to an increased energy supply and starvation tolerance in HCC. Importantly, EIPA has the potential to be applied in precision treatment for patients with ATP9Aoverexpressing HCC.

Discussion
Accounting for approximately 50% of HCC cases, the proliferative subtype is the major HCC molecular subtype and is characterised by a poor prognosis [33]. HCC cells require adequate nutrients to support their rapid proliferation. However, HCC tumours are often poorly perfused and nutrient-limited because tumour overgrowth outstrips the vasculature. The imbalance between the strong energy demand and an inadequate nutrient supply induces HCC cells to acquire nutrients by scavenging from the TME [10]. Macropinocytosis is an important scavenging process that fuels cancer cell proliferation by scavenging extracellular proteins or necrotic cell debris [34]. Recently, it was clarified that macropinocytosis was specifically induced by hypoxia and supported HCC cell survival. That study also reported that macropinocytosis occurred in tumour regions with fewer hypoxic markers, which indicated that other mechanisms independent of hypoxia were involved in regulating macropinocytosis in HCC cells [13]. Herein, we demonstrated that macropinocytosis occurred universally in HCC cells under nutrient limitation and supported the cell energy supply and proliferation both in vitro and in vivo. Targeting macropinocytosis strongly suppressed HCC cell proliferation. These results revealed a novel pro-survival route in HCC and suggested a novel HCC therapeutic strategy.
Macropinocytosis might confer resistance to current cancer therapies [14,[35][36][37]. A recent study found that HCC cells harnessed macropinocytosis to acquire cysteine and thereby resist sorafenib-induced ferroptosis. Combined sorafenib with the macropinocytosis inhibitor amiloride strongly suppressed HCC growth, but amiloride alone did not inhibit tumour growth [14]. However, we demonstrated that EIPA obviously restricted HCC growth. The interpretation of these contradictory results is that macropinocytosis-targeting therapy might be effective for a subgroup of patients, suggesting the importance of selecting patients who potentially benefit from such therapies. Our results demonstrated that HCC cells with high ATP9A expression benefited remarkably from EIPA treatment, while ATP9A-silencing obviously reduced the effect of EIPA.
A Huh7 subcutaneous tumour model demonstrated that amiloride was ineffective for HCC [14], and we revealed    [13,34,35,[38][39][40]. However, current studies have only uncovered one dimension of the underlying regulatory mechanisms. Plasma membrane ruffling is the first step of macropinocytosis. P4-ATPases are a protein family involved in membrane ruffle generation during cell motility [24]. However, their role in macropinocytosis remains elusive. Here, we demonstrated that ATP9A was a key regulator of macropinocytosis and supported HCC cell nutrient starvation tolerance. These findings suggested a novel therapeutic target and mechanism of macropinocytosis in HCC.
Biological ontology analysis indicated that ATP9A alteration dramatically correlated with HCC cell growth and proliferation and the cell cycle [41]. However, its role in HCC and the underlying mechanism requires further examination. Here, we found that ATP9A promoted HCC nutrient starvation tolerance by inducing macropinocytosis. Surprisingly, ATP9A was overexpressed in necrotic lesions in the peripheral tumour tissues in this study. Necrotic cell debris is a high-quality macropinocytic fuel source because it contains abundant nutrients, including proteins and lipids [35]. Necrosis is a common feature of liver cancer that always predicts a poor outcome [42,43]. This phenomenon suggested that high-ATP9A HCC cells may fuel their own proliferation by ingesting necrotic cell debris via macropinocytosis. However, we did not clarify whether and how necrosis stimulates ATP9A upregulation, which requires further exploration.
Cell surface regulators are involved in mediating macropinocytosis [23,44]. A previous study demonstrated that plasma membrane V-ATPase in RAS-transformed cancer cells strongly promoted macropinocytosis under nutrient starvation [23]. Generally, V-ATPase is mainly distributed on intracellular tubulovesicular membranes [45]. Membrane trafficking regulation is a major mechanism controlling V-ATPase activity [46]. ATP9A is required for endosome-to-plasma membrane protein trafficking [25]. Here, we demonstrated that ATP9A promoted plasma membrane ATP6V1A accumulation, followed by increased plasma membrane cholesterol and RAC1, thereby inducing macropinocytosis. ATP9A also directly interacted with ATP6V1A. We speculated that ATP9A promoted ATP6V1A feeding into endosomes and trafficking to the plasma membrane since ATP9A is mainly distributed in endosomes and acts in endosome-to-membrane recycling. However, the precise mechanism by which ATP9A regulates ATP6V1A transport requires further exploration.
In conclusion, our results demonstrated that macropinocytosis was critical for sustaining HCC proliferation and growth under nutrient limitation. EIPA targeting of macropinocytosis strongly inhibited HCC cell growth. The interaction of ATP9A with ATP6V1A facilitated ATP6V1A membrane trafficking, thereby activating plasma membrane cholesterol-dependent RAC1 signalling and initiating macropinocytosis to increase the energy supply and support HCC cell survival and proliferation. These findings reveal a novel role for macropinocytosis in HCC progression and suggest that targeting macropinocytosis may be a promising HCC therapeutic strategy, especially in patients with high ATP9A levels.

SUPPLEMENTARY MATERIAL ONLINE
Supplementary materials and methods Figure S1. Macropinocytosis supported HCC cells proliferation under nutrient deprivation conditions Figure S2. ATP9A is highly expressed in HCC tissues and cells Figure S3. ATP9A support HCC cell proliferation via macropinocytosis Figure S4. Western blotting for CDC42 and RAC1 Figure S5. Total cholesterol in ATP9A-overexpressing and ATP9A-silencing HCC cells Figure S6. Representative MS plot of ATP6V1A peptides and western blotting for ATP6V1A and ATP9A Figure S7. Multivariate Cox regression analysis of RFS and OS in HCC Data S1. Peptides and counts for ATP9A-interacting proteins analysed by IP/MS assays Table S1. Clinicopathological characteristics of hepatocellular carcinoma specimens Table S2. Primers for RT-qPCR (referred to in Supplementary materials and methods) Table S3. Correlation between ATP9A expression and clinicopathological characteristics in hepatocellular carcinoma patients Table S4. Univariate and multivariate analysis of factors associated with recurrence free survival in hepatocellular carcinoma patients Table S5. Univariate and multivariate analysis of factors associated with overall survival in hepatocellular carcinoma patients ATP9A-enhanced macropinocytosis counters nutrient starvation in HCC 31