Abstract

   Rationale: Hepatocellular carcinoma (HCC) is one of the most severe
   cancers worldwide, with few effective targeted therapies for HCC. Lipid
   metabolic reprogramming is emerged as a hallmark of cancer metabolism
   that guides response to antitumoral therapies. Such lipid metabolic
   alteration in cancers is critically regulated by the mammalian target
   of rapamycin mTOR, which is considered as a promising therapeutic
   target. Despite efforts, mTOR inhibitors (mTORi) have produced limited
   response clinically, partly due to incomplete knowledge of mTORC1
   addiction in cancers.

   Methods: CRISPR-Cas9 system was used to establish Hpcal1 null mice. The
   liver cancer model in mice was generated using Hpcal1-deficient mice
   with diethylnitrosamine (DEN) /CCL4 or MYC/Trp53^-/- via hydrodynamic
   tail-vein injection. RNA-sequencing (RNA-seq) was used to identify
   potential signaling pathways. The expression of HPCAL1 and mTOR
   signaling were determined using quantitative polymerase chain reaction
   (qPCR), western blot and immunohistochemistry. The role of Hpcal1 in
   liver tumorigenesis and its response to mTORi was assessed by CCK-8
   measurements, colony formation assay and in mouse model.

   Results: In this study, we identified hippocalcin-like protein 1
   (HPCAL1) as an important negative regulator of de novo lipid
   biosynthesis and mTOR signaling activation, limiting liver
   tumorigenesis and establishing a metabolic vulnerability of HCC in
   mice. Genetic loss of HPCAL1 rendered HCC mTORC1-addicted and sensitive
   to mTORi AZD-8055 in vitro and in vivo. Importantly, HPCAL1 expression
   was inversely correlated with the levels of mTOR phosphorylation and
   several critical lipid biosynthesis enzymes in human specimens.
   Mechanistically, HPCAL1 directly bound to RuvB Like AAA ATPase 1
   (RUVBL1), inhibiting the assembly of TEL2-TTI1-TTI2 (TTT)-RUVBL complex
   and subsequent leading the mTOR signaling suppression.

   Conclusion: We uncover a metabolic vulnerability and mTOR addiction in
   HCC with HPCAL1 loss that provides a selective therapeutic window for
   HCC with mTORC1 hyperactivation using mTORi.

   Keywords: Hippocalcin-Like 1, RUVBL1, mTOR addiction, fatty acid
   biosynthesis, Cholesterol synthesis

Introduction

   Hepatocellular carcinoma (HCC) is the third most frequent cause of
   cancer-related death worldwide and is associated with a significant
   clinical, economic, and psychological burden [45]^1^-[46]^3. Although
   emergence of several new targeted therapies for the treatment of HCC
   has shown promise and therapeutic benefit over past 5 years, the
   long-term patient survival times remains to be improved [47]^1^,
   [48]^4^, [49]^5. Developing additional novel therapeutic drugs or
   understanding the underlying mechanism of tumor progression and
   therapeutic resistance to available targeted drugs will be helpful to
   identify a selective therapeutic window and improve the survival of
   patients with HCC [50]^6.

   Lipid metabolic reprogramming is an established hallmark of cancer
   metabolism that provides lipids and lipid precursors for membrane
   synthesis, signaling molecules and energetic substrates for rapidly
   growing tumor cells that are continuously adapting to unfavorable
   conditions, including therapeutic stresses [51]^7^-[52]^9. Emerging
   evidences have suggested that aberrant lipid metabolism imposes a
   potential metabolic vulnerability that could be exploited to guide
   response and resistance to antitumoral therapies [53]^8.

   The mammalian target of rapamycin (mTOR) is a master regulator of lipid
   biosynthesis and lipolysis via its diverse downstream signaling
   pathways [54]^10^-[55]^12. Constitutive mTORC1 hyperactivation is
   frequently detected in cancers that establish mTOR addiction to drive
   metabolic reprogramming and sustain rapid growth and survival, and has
   been considered an attractive therapeutic target [56]^13^, [57]^14.
   This addiction to mTORC1 signaling in certain cancers has created a
   substantially increased sensitivity to mTOR inhibitors (mTORi), which
   inspired the clinical development of mTORi [58]^15. Indeed, targeting
   mTOR signaling for mTOR-addicted tumors has shown beneficial effects in
   preclinical models [59]^16^-[60]^18. However, only modest or short-term
   clinical response of several tumors to mTORi was observed in some
   clinical trials [61]^19^-[62]^21. This is partly due to the incomplete
   understanding of the mechanisms underlying mTOR activation and mTORC1
   addiction in cancer cells, which compromises the identification of
   effective biomarkers for predicting potentially responsive patients and
   guiding clinical implications.

   mTOR is found in two complexes (mTORC1 and mTORC2), each with distinct
   protein components as well as substrates [63]^22^, [64]^23. Activation
   of mTORC1 occurs at the lysosomal membrane by Rag GTPase heterodimers
   and is known to be mediated by the small GTPase Ras homologue enriched
   in brain (RHEB), which is itself inhibited by the tuberous sclerosis
   heterodimer (TSC1/2) in response to amino acids and glucose [65]^12. In
   addition, recent studies revealed that mTORC1 assembly, lysosomal
   localization and activation are regulated by the TEL2-TTI1-TTI2
   (TTT)-RuvB Like AAA ATPase complex, which is independent of TSC1/2 and
   Rag upon energy stress [66]^24^-[67]^28. Further study demonstrated
   that RUVBL1/2 proteins are pivotal for the vulnerabilities of
   mTORC1-addicted cancer cells. Targeting RUVBL1/2 causes synthetic
   lethality in mTORC1-hyperactive cancer cells, providing a selective
   therapeutic opportunity for tumors with high mTORC1 activation [68]^27.

   HPCAL1 is a neuronal calcium sensor with three EF-hand domains that
   belongs to a member of the visinin-like proteins (VILIP) superfamily,
   which include visinin-like protein 1 (VILIP-1) (VSNL1), VILIP-2
   (HPCAL4), VILIP-3 (HPCAL1), hippocalcin (HPCA), and neurocalcin δ
   (NCALD). VILIP-1 is identified a tumor suppressor that inhibits the
   epidermal-mesenchymal transition in esophageal squamous cell carcinoma
   and non-small cell lung carcinoma [69]^29^, [70]^30. Other members in
   this family have rarely been reported in cancers. We have previously
   identified that HPCAL1 was frequently lost in HCC, resulting cell cycle
   progression and hepatocarcinogenesis via modulating Cyclin-dependent
   kinase inhibitor 1 p21^WAF1/CIP1 stabilization [71]^31. By contrast,
   HPCAL1 was reported to facilitate the differentiation and proliferation
   of glioblastoma via distinct mechanisms [72]^32^, [73]^33. In this
   study, we have uncovered that HPCAL1 serves as an important negative
   regulator of de novo lipid and cholesterol biosynthesis and tumor
   progression in mice. Through binding to RUVBL1/2, HPCAL1 diminishes the
   assembly of TTT-RUVBL complex and mTOR signaling activation and renders
   HCC therapeutic resistant to mTORi.

Materials and Methods

Sample collection

   90 cases of pathologically examined liver cancerous tissues were
   collected from Renji Hospital between 2014 and 2018. Tissue microarrays
   (TMA) containing these cases were prepared by Shanghai ZUOCHENG Ltd.
   All experiments and analyses were conducted with the understanding and
   written consent of each participant. All manipulations were performed
   under the approval of the Research Ethics Committee of Renji Hospital,
   School of Medicine, Shanghai Jiao Tong University.

Generation of Hpcal1^-/- mice

   C57BL/6J mice were constructed by CRISPR-Cas9 system. Briefly, four
   sgRNAs were designed to target the Hpcal1 gene. The oligonucleotide
   sequences are as follows: sgRNA1, 5′- AAGGGTCATTCAATCCCCACAGG -3′;
   sgRNA2, 5′- GGGTCATTCAATCCCCACAGGGG -3′; sgRNA3, 5′-
   TGCAATCTGAACCCTTCCAATGG -3′; sgRNAs were then cloned into T7-cas9
   vector and linearized with for in vitro transcription using the
   mMACHINE T7 Ultra kit (Ambion), followed by purification using RNeasy
   Mini kit (Qiagen). A mixture of Cas9 mRNA and four sgRNAs was injected
   into the cytoplasm and male pronucleus of the zygote, which were
   transferred into pseudopregnant C57BL/6J female mice to yields F0
   generation mice. Genomic DNA was isolated from tails and analyzed by
   PCR amplification using the following primers:
   5′-TGTCCCCTTATACCACGATTT-3′ and 5′-CCAACCCTCACTGCTTTTG-3′. All mice
   were maintained in specific pathogen-free facilities and housed in
   single-sex cages at 20 ± 2 °C with 40-60% humidity and a 12-h
   light/12-h dark photoperiod. All animal experiments were performed
   according to the National Institute of Health Guide for the Care and
   Use of Laboratory Animals under the approval of the Research Ethics
   Committee of Renji Hospital, School of Medicine, Shanghai Jiao Tong
   University.

Diethylnitrosamine (DEN) /CCL4 liver cancer model

   For the HCC model induced by DEN and CCl4, 2-week-old male mice were
   intraperitoneally injected with 25 mg/kg DEN. At age of 8 weeks, mice
   were injected intraperitoneally with 25% CCl4 (2 ml/kg) in corn oil
   once a week for additional 12 weeks. Mice were then killed, and livers
   and sera were collected for subsequent analysis.

MYC/Trp53^-/- liver cancer model via hydrodynamic tail-vein injection

   8-week-old was injected with 2 mL sterile 0.9% NaCl solution/plasmid
   mix containing 12 μg of MYC-Luc, 12 μg of sgp53, 12 μg of sgCON or
   sgHpcal1 or sgRuvbl1 and 3 μg of SB13 transposase-encoding plasmid via
   tail vein within 5 to 7 s [74]^34. 30 days post injection, mice were
   then killed, and livers were collected for subsequent analysis. For
   survival analysis, mice were observed regularly every day after
   injection for 30 days until all mice were dead. Survival curves were
   constructed according to the survival times. For sgHpcal1, four tandem
   sgRNA (same as above) driven by separate U6 promoters were prepared in
   plenti-CRISPR V2 (Addgene Plasmid 49535). For sgRuvbl1, three tandem
   sgRNA by separate U6 promoters was used. The sgRNA sequence was as
   follow: sgRuvbl1-1, 5'-CACATCTAGATAAACTTCGA-3'; sgRuvbl1-2,
   5'-GCCGCGCATTAGCCACATCC-3'; sgRuvbl1-3, 5'-GCTCGGATCTTAATGATCTT-3'.

Analysis of liver enzymes

   Serum analyses of ALT and AST from WT and Hpcal1 mice were determined
   using an Alanine Aminotransferase Assay kit and Aspartate
   Aminotransferase Assay kit (Nanjing Jiancheng Ltd., China).

Hematoxylin and eosin (HE), immunohistochemistry (IHC) and Oil red O staining

   Formalin-fixed and paraffin-embedded mouse livers or human liver
   cancers were analyzed by HE as previously reported [75]^35^, [76]^36.
   For IHC staining, liver sections were preincubated in normal goat serum
   for 15 min, and were incubated with anti-phos-mTOR (Cell signaling,
   #2976,1: 50), anti-phos-4EBP1 (Cell signaling, #2855, 1:100),
   anti-HPCAL1 (Sigma, SAB1307075; 1:200), anti-SCD1 (Abclonal, A16429;
   1:400), anti-ACSS2 (Proteintech, 16087-1-AP, 1:400), RUVBL1
   (Proteintech,10210-2-AP,1:400) and Ki-67 (Proteintech, 27309-1-AP,
   1:5000) at room temperature. Two hours later, secondary antibody
   (JACKSON, 111-035-003, 1:500) was used, followed by incubation with DAB
   Chromogen dilution solution. For Oil red O staining, Oil Red O (0.5%
   (v/v) in isopropanol) was diluted with water (3:2), filtered through a
   0.45 mm filter, wished 3 times with PBS and fixed for 30 min with 4%
   (v/v) paraformaldehyde (PFA), then the fixed cells were incubated with
   filtered Oil Red O for 1 h at RT. Quantification of Oil red O staining
   was performed by Image J software as previously described [77]^37.

IHC staining and scoring

   IHC staining was independently assessed by two experienced
   pathologists. The staining intensity was graded from 0 to 2 (0, no
   staining; 1, weak; 2, strong). The staining extent was graded from 0 to
   4 based on the percentage of immunoreactive tumor cells (0%, 1%-5%,
   6%-25%, 26%-75%, 76%-100%). A score ranging from 0 to 8 was calculated
   by multiplying the staining extent score with the staining intensity
   score, resulting in a low (0-4) level or a high (6-8) level for each
   sample.

Reagents and Antibodies

   AZD-8055 was purchased from MedChemExpress (Shanghai, China).
   Antibodies for western blotting against HPCAL1, ACSS2, RUVBL1, GAPDH,
   his tag and HA tag were obtained from Proteintech (Wuhan, China).
   Antibody against phos-mTOR, and phos-4EBP1 and phos-4EBP1 were
   purchased from Cell Signaling Technology (USA). SCD1 antibody was
   obtained from ABclonal (Wuhan, China). HPCAL1 antibody for IHC staining
   and anti-FLAG M2 gel was purchased form Sigma-Aldrich (USA).

Cell culture, RNA interference, Cell proliferation and Colony Formation Assay

   Human Embryonic Kidney (HEK) 293T and Huh7 were purchased from Cell
   Bank of Shanghai Institutes of Biological Sciences. HEK-293T and Huh7
   were maintained in DMEM high glucose medium (Shanghai BasalMedia
   Technologies) in 5% CO2 at 37 °C incubator. All cell lines underwent
   detection for mycoplasma contaminants using Mycoplasma Detection Kit
   (40611ES25, YEASEN, Shanghai, China) and confirmed mycoplasma-negative.
   STR authentication were performed to confirm the identity of Huh7.
   RUVBL1, RUVBL2, RPTOR, RICTOR and HPCAL1 knockdown was achieved by RNA
   interference using short interfering RNA (siRNA) and lentiviral
   vector-based shRNA, respectively. The siRNA sequence used was as
   follows: siRUVBL1, 5'-AAGGAACCAAACAGTTGAAACTG-3'; siRUVBL2,
   5'-TAACAAGGATTGAGCGAAT-3'; siRPTOR: 5'-GGACAACGGCCACAAGTACTT-3';
   siRICTOR 5'- AGACAAGGCCAATCTTCATGC-3'; shHP#1,
   5'-GGTGACATGCAGGGTTCAAGT-3' and shHP#2, 5'-GCCGCTTGCACGTATAGATAC-3'.
   shRNA lentivirus particles were prepared as previously described
   [78]^38. Cell proliferation was performed by CCK8 assay in Huh7 cells
   with indicated doses of AZD-8055, and all the experiments were
   performed in triplicate with each group in quintuplicate as previously
   reported [79]^38. For colony formation, cells infected with indicated
   lentivirus was prepared in suspension medium with cell density of 1 ×
   10^3 cells/mL. 1-3 mL cell suspension medium were seeded into 6-well
   plates in triplicate overnight. Cells were then treated with for
   indicated doses of AZD-8055 for additional 2-3 weeks. Plates were
   washed gently twice with PBS, and fixed with 4% methanol for 30 min.
   The number of colonies was counted and statistically analyzed.

Purification of HPCAL1 Interacting Proteins and Protein Identification

   Huh7 cells infected with vector or FLAG-HPCAL1 were lysed to a
   homogenous mixture. The supernatant was incubated with 20 μL FLAG-M2
   gel overnight on a gentle rotating machine. The resulting pellet was
   washed five times with RIPA buffer (50 mM Tris (pH 7.4), 250 mM NaCl,
   1% TritonX-100, 1% sodium deoxycholate, 1 mM EDTA and 20% glycerol),
   and was then eluted with 500μL FLAG peptide (5 mg/mL) in TBS (10 mM
   Tris HCL, 150 mM NaCl, pH7.4). The eluates were further recovered using
   ultrafiltration spin columns (3 kDa), and were subjected to sodium
   dodecyl sulfate-polyacrylamide gel electrophoresis and Coomassie blue
   staining. Protein band cutting, in-gel trypsin digestion, peptide
   extraction, LC-MS/MS analysis, and protein identification were
   performed as previously described [80]^31^, [81]^38.

Statistical analysis

   All statistical analyses were carried out using GraphPad Prism 7 or
   SPSS for Windows 25 software (SPSS Inc., Chicago, IL, USA). The
   comparisons of measurement data between two groups were performed using
   unpaired t test. The comparisons among three or more groups were
   firstly performed by One-Way ANOVA test if the variation between groups
   were comparable. Overall survival was evaluated by the Kaplan-Meier
   survival curve and the Log-rank test using GraphPad Prism 7. The
   correlation of the two proteins were examined by Pearson's correlation
   test. T statistical tests and p-values were two-sided. Differences were
   considered significant with a value of P < 0.05.

Results

Hpcal1 deletion promotes Hepatocarcinogenesis

   To investigate the role of HPCAL1 in hepatocarcinogenesis of
   genetically engineered mouse models (GEMM), we prepared Hpcal1 knockout
   (Hpcal1^-/-) mice using CRISPR/Cas9-mediated gene editing to delete
   exon 2 and 3 of Hpcal1 ([82]Figure S1A). The Hpcal1-KO status was
   validated PCR using specific primers, gene sequencing and western
   blotting ([83]Figure S1B-C). HCC mouse model was then established using
   diethylnitrosamine (DEN)/CCl4 consisting of the injection of 25 mg/kg
   of DEN, followed by intraperitoneal injection of CCL4 for 12 weeks as a
   tumor promoter, and the tumor burden was analyzed after 12 weeks
   (Figure [84]1A). Compared with the wide type counterparts (WT),
   Hpcal1^-/- mice had higher liver weight and liver-body ratio, and more
   tumor numbers (Figure [85]1B-C, [86]Figure S1D-E). Consistently, we
   observed bigger histological lesions and enhanced proliferative
   activity (Ki67 reactivity) in the Hpcal1-deficient livers (Figure
   [87]1D and [88]Figure S1F). Moreover, biochemical parameters such as
   alanine aminotransferase (ALT) and aspartate aminotransferase (AST)
   were significantly increased in serum of Hpcal1-deficient mice than
   those of WT (Figure [89]1E). These data indicate that Hpcal1 loss
   promotes hepatocarcinogenesis in mice. To validate this hypothesis in
   liver-specific deletion of Hpcal1 and established the clinical
   relevance between HPCAL1 and key driver genes of human HCC, we take
   advantage of hydrodynamic tail-vein injections to generate a
   MYC/Trp53^-/- liver cancer model in which oncogenic MYC can be
   genomically integrated, and that TP53 is deficient to recapitulate the
   features of HCC, as previously described [90]^34. Hpcal1 was
   genetically deleted using 4 sgRNA (small guide RNA), which was the same
   as those in Hpcal1^-/- mice (Figure [91]1F-G). The knockout efficiency
   was confirmed by western blotting ([92]Figure S1G). As expected,
   liver-specific loss of Hpcal1 significantly facilitated tumorigenesis
   with evidence of increased liver weight, liver-body ratio, tumor number
   and liver tumor cell proliferation activity (Figure [93]1H-[94]1K and
   [95]Figure S1H-1J). More importantly, Hpcal1 loss in liver greatly
   shortened the overall survival times of mice compared to control group
   (Figure [96]1L). These data indicate that HPCAL1 deficiency in mice
   promotes hepatocarcinogenesis.

Figure 1.

   [97]Figure 1
   [98]Open in a new tab

   Genetic loss of Hpcal1 results in enhanced liver tumorigenesis. (A)
   Scheme of DEN/CCL4 induced murine liver cancer model. Two-week old mice
   were intraperitoneally (i.p.) injected with DEN at dose of 25 mg/kg in
   saline. At age of eight weeks, mice were administrated with CCl4 in a
   volume of 2 ml/kg body weight (CCL4 was mixed with corn oil at ration
   1:3 once a week for additional 12 weeks. W, week. This figure was drawn
   by using BioRender (JX24G8PWZC). (B) Representative images of gross
   morphology from livers of WT (n=9) and Hpcal1-/- (n=7) mice. The tumor
   lesions are indicated by black arrows. (C) Analyses of the ratio of
   liver to body weight (left) and the number of tumors (right). (D)
   Representative images of hematoxylin and eosin (HE), HPCAL1 and Ki67
   staining of livers from WT and Hpcal1-/- mice. scale bar, 50 µm. (E)
   ALT and AST levels in the sera of WT and Hpcal1-/- mice. AST, aspartate
   aminotransferase. ALT, alanine aminotransferase. (F) Schematic
   representation of constructs for MYC-Trp53 liver cancer model. (G)
   Scheme of MYC-Trp53 liver cancer model murine liver cancer model. hd.i,
   hydrodynamic injection. Mice were injected with 12 µg of MYC-Luc, 12 μg
   of sgp53, 12 µg of sgCON or sgHpcal1 and 3 µg of SB13
   transposase-encoding plasmid mixed in 2 mL of 0.9% NaCl solution within
   5 to 7 seconds. This figure was drawn by using BioRender (JX24G8PWZC)
   (H) Representative images of gross morphology from livers of sgCON
   (n=5) and sgHpcal (n=5) mice. (I) Analyses of the ratio of liver to
   body weight (left) and the number of tumors (right). (J) Representative
   images of HE, HPCAL1 and Ki67 staining of livers from sgCON and sgHpcal
   mice. scale bar, 50 µm. (K) Statistical analysis of the percentage of
   Ki67+ cells from indicated mice. (L) Kaplan-Meier survival curves of
   sgCON (n=8) and sgHpcal (n=8) mice. Log-rank Mantel-Cox test was used
   to calculate significance. Scale bar, 50 µm. Unpaired t test was used
   to determine statistical significance. *, **, *** means p < 0.05, p <
   0.01, and p < 0.001.

Hpcal1 loss increase lipid biosynthesis

   To explore the underlying mechanisms by which Hpcal1 mediates
   tumorigenesis, we performed RNA-sequencing of liver tumor derived from
   WT and Hpcal1^-/- mice to identify signaling pathways potentially
   involved. By pathway enrichment analysis, it was found that lipid
   metabolic pathways, including cholesterol, fatty acids and steroids,
   were remarkably enriched upon Hpcal1 loss in mice (Figure [99]2A),
   which was also supported by the result of gene set enrichment analysis
   (GSEA) (Figure [100]2B). The heat map of top upregulated gene
   implicated in cholesterol and fatty acids metabolism was shown (Figure
   [101]2C). The mRNA levels of selected 23 key genes implicated in lipid
   metabolism were further verified by real-time PCR. In line with the
   observation in RNA-seq datasets, these genes were remarkably
   upregulated in liver tumors of Hpcal1^-/- mice relative to WT tumors
   (Figure [102]2D). Furthermore, some key genes including Fasn, Scd1 and
   Acss2, were examined by western blotting, which also showed higher
   protein abundance in Hpcal1-depleted tumors (Figure [103]2E). These
   evidences are supportive of overall upregulated genes in lipid
   biosynthesis. To confirm whether observed gene upregulation ultimately
   results in lipid biosynthesis in tumors upon Hpcal1 deletion, we
   determined the hepatic levels of triglyceride and cholesterol. Indeed,
   higher levels of triglyceride and cholesterol were observed in
   Hpcal1-depleted tumors relative to WT (Figure [104]2F). This
   observation was supported by bigger and greater lipid droplets,
   evidenced by Oil Red O staining and visual estimation by phase contrast
   microscopy (Figure [105]2G). In addition, we found that
   Hpcal1-deficient liver tumors had much higher level of ACSS2 and SCD1
   than WT tumors (Figure [106]2H and [107]Figure S1K). Taken together, we
   demonstrate that Hpcal1 loss in liver tumors enhances lipid
   biosynthesis.

Figure 2.

   [108]Figure 2
   [109]Open in a new tab

   Hpcal1 deletion promotes lipid metabolism in livers. (A) Pathway
   enrichment analysis of the differential expression genes from the
   RNA-sequencing datasets of liver tumors of WT (n=3) and Hpcal1-/- (n=3)
   mice. (B) Gene set enrichment analysis (GSEA) showed enrichment of
   cholesterol and fatty acid biosynthesis in liver tumors of Hpcal1-/-
   (n=3) mice compared with WT mice. (C) Heatmap demonstration of the gene
   expression related to cholesterol and fatty acid biosynthesis. (D)
   Realtime PCR analysis of the mRNA levels of genes related to
   cholesterol and fatty acid biosynthesis. (E) Western blotting analysis
   of the cell lysates from liver tumors of WT and Hpcal1-/- mice with
   indicated antibodies. (F) Determination of hepatic total cholesterol
   (TG) and total triglyceride (TG) from liver tumors of WT (n=5) and
   Hpcal1-/- (n=5). (G) Representative images and statistical analysis of
   Oil Red O staining of liver sections from WT and Hpcal1-/-. Staining
   intensity of 2 fields per liver section from 5 mice of indicated group
   were calculated using Image J software. Scale bar, 25 µm. (H)
   Representative IHC images of HPCAL1, ACSS2 and SCD1 of liver sections
   from WT and Hpcal1-deficient mice. Unpaired t test was used to
   determine statistical significance. Scale bar, 50 µm. *, **, *** means
   p < 0.05, p < 0.01, and p < 0.001.

HPCAL1 directly interacts with RUVBL1/2 in an Ca^2+ independent manner

   To dissect the molecular basis of HPCAL1 in mediating lipid metabolism,
   we performed co-immunoprecipitation (Co-IP) and mass spectrometry (MS)
   to identify potential signaling molecules related to lipid biosynthesis
   (Figure [110]3A). Notably, RUVBL1/2, the AAA+ ATPases (ATPases
   associated with diverse cellular activities) that are in complex with
   TTT controlling mTOR assembly, activation and lipid metabolism
   [111]^25^, [112]^39, was identified as the potential binding partners
   (Figure [113]3B, [114]Figure S2A-B and [115]Table S1). Co-IP assays
   showed that HPCAL1 indeed interacted with RUVBL1/2 in the HEK293T
   cells, and it appears that HPCAL1 bound to RUVBL1 with higher intensity
   relative to RUVBL2 (Figure [116]3C). Moreover, this interaction was
   further confirmed in the mouse liver tumors with endogenous proteins
   (Figure [117]3D). To better understand the nature of the RUVBL /HPCAL1
   interaction, we defined the protein regions responsible for their
   reciprocal binding. To this end, serially truncated HPCAL1 or RUVBL1
   motif sequence were prepared and co-transfected into HEK293T cells. It
   was shown that multiple domains of HPCAL1 likely mediated its
   interaction with RUVBL1 (Figure [118]3E and [119]3F). Also, mutation of
   several phosphorylation sites identified previously did not clearly
   affected HPCAL1/RUVBL1 interaction (Figure [120]3G), indicating that
   multiple domains of HPCAL1 likely bind to RUVBL1 independently.
   Likewise, it seems that all 3 domains of RUVBL1 were required for
   HPCAL1 binding as deletion of either one will affect the association
   (Figure [121]3H). In an effort to investigate whether this binding is
   direct, we prepared bacterial-expressed and purified HPCAL1 and RUVBL1
   with GST and his tag, respectively. GST pulldown analysis revealed a
   direct binary association between HPCAL1 and RUVBL1 (Figure [122]3I).
   Furthermore, considering that HPCAL1 undergoes post-translational
   N-terminal myristylation, conformational change and membrane
   localization in response to cellular calcium (Ca^2+) levels [123]^40,
   we next test whether the interaction between HPCAL1 and RUVBL1 is
   Ca2^+-dependent. Addition of either calcium chloride (CaCl[2]) or EGTA,
   which increases or decreases Ca^2+ levels, respectively, did not
   clearly affect HPCAL1/RUVBL1 binding (Figure [124]3J). Thus, this
   interaction is independent of the Ca^2+ concentration.

Figure 3.

   [125]Figure 3
   [126]Open in a new tab

   HPCAL1 directly binds to RUVBL1. (A) HEK293T transfected with
   FLAG-HPCAL1 or control plasmid were subjected to Co-IP/MS. Co-IP,
   co-immunoprecipitation. MS, mass spectrometry. (B) Protein information
   and representative mass spectrum peak of RUVBL1. (C) HEK293T cells were
   co-transfected with EGFP-HPCAL1 and FLAG-RUVBL1 or RUVBL2 plasmids.
   Immunoblot analysis of the cell lysates and immunoprecipitates with the
   indicated antibodies. (D) Endogenous HPCAL1 derived from MYC-induced
   liver tumors was immunoprecipitated using anti-HPCAL1 antibody or
   isotype IgG control and subjected to immunoblot analysis using
   indicated antibody. (E) Scheme of HPCAL1 domain organization. EF,
   EF-hand. (F) HEK293T cells were co-transfected with EGFP-HPCAL1
   fragment and FLAG-RUVBL1 plasmids. Immunoblot analysis of the cell
   lysates and immunoprecipitates with the indicated antibodies. (G)
   HEK293T cells were co-transfected with various EGFP-HPCAL1 mutants and
   FLAG-RUVBL1 plasmids. Immunoblot analysis of the cell lysates and
   immunoprecipitates with the indicated antibodies. (H) Scheme of RUVBL1
   domain structures and Co-IP assay. HEK293T cells were co-transfected
   with EGFP-HPCAL1 and various FLAG-RUVBL1 fragment plasmids. Immunoblot
   analysis of the cell lysates and immunoprecipitates with the indicated
   antibodies. DI, domain I; DII, domain II; DIII, domain III. (I)
   Purified recombinant GST-HPCAL1 was incubated with purified his-RUVBL1.
   Coomassie blue staining and immunoblot analysis of the cell lysates or
   immunoprecipitates with indicated antibody. (J) Co-IP and immunoblot
   analysis of RUVBL1/HPCAL1 interaction in the presence of NaCl (10 mM),
   CaCl2 (10 mM), and ETGA (10 mM).

HPCAL1 negatively regulates mTOR signaling in mice and human HCC specimens

   Given that HPCAL1 directly binds to RUVBL1, which promotes mTOR
   signaling and lipid metabolism [127]^41^, [128]^42, we reason that
   HPCAL1 may also regulate mTOR signaling via RUVBL1. Not surprisingly,
   genes related to mTOR signaling is significantly enriched in
   transcriptome datasets of liver tumors following Hpcal1 deficiency in
   mice, as revealed by GSEA (Figure [129]4A). This result was in
   agreement with observation that Hpcal1-depleted liver tumor showed
   higher phosphorylation levels of mTOR and its substrates 4EBP1 as well
   (Figure [130]4B). Moreover, HPCAL1 overexpression in Huh7 cells reduced
   mTOR activation, whereas depletion of HPCAL1 resulted in the opposite
   effect (Figure [131]4C). Consistently, IHC examination showed that the
   phosphor-mTOR levels were markedly higher in sections of Hpcal1-deleted
   liver tumors than their counterparts either initiated by DEN/CCL4 or
   MYC/sgp53 (Figure [132]4D). To provide the clinical relevance between
   HPCAL1 and mTOR signaling, we investigated the expression correlation
   among HPCAL1, phosphor-mTOR, phosphor-4EBP1, SCD1, ACSS2 and RUVBL1 in
   the tissue microarray of 90 cases of HCC specimens ([133]Table S2).
   Through IHC staining and grading, we found that HPCAL1 was negatively
   associated with the components of mTOR signaling and lipid metabolic
   enzymes (Figure [134]4E-F and [135]Figure S3A-3B). Although HPCAL1 did
   not affect the expression of RUVBL1, the expression of these two
   proteins was also inversely correlated in HCC tissues. Furthermore,
   while low expression of HPCAL1 predicted unfavorable clinical outcome
   in patients with HCC, high expression of phosphor-mTOR, phosphor-4EBP1,
   SCD1, ACSS2 and RUVBL1 was correlated with worse survival times and
   (Figure [136]4G and [137]Figure S3C-3E). Together, these results
   demonstrate that HPCAL1 represses mTOR signaling and lipid metabolism
   related to HCC progression and patient survival.

Figure 4.

   [138]Figure 4
   [139]Open in a new tab

   HPCAL1 was inversely associated with mTOR signaling in mice and human.
   (A) GSEA showed enrichment of mTOR signaling from the RNA-sequencing
   datasets of liver tumors of WT (n=3) and Hpcal1-/- (n=3) mice. (B)
   Western blotting analysis of the cell lysates from liver tumors of WT
   and Hpcal1-/- mice with indicated antibodies. (C) Immunoblot analysis
   of the cell lysates from Huh7 cells with HPCAL1 overexpression (left)
   or HPCAL1 depletion (right) using indicated antibodies. (D)
   Representative IHC images of HPCAL1 and p-mTOR of liver sections from
   WT and Hpcal1-deficient mice. (E) Heatmap demonstration of IHC score
   and correlation between PROX1, p-mTOR, p-4EBP1, SCD1, ACSS2 and RUVBL1
   in HCC tissues (n=89). The criteria of IHC scoring were described in
   Materials and Methods. Pearson's correlation analysis was used to
   measure the relationship between indicated proteins in human HCC
   tissues. (F) Representative IHC images of indicated proteins derived
   from human HCC tissue microarray (TMA). (G) Kaplan-Meier curves
   analyses of patients with low versus high expression of indicated
   proteins HCC TMA. Scale bar, 50 µm.

HPCAL1 suppresses mTOR signaling and HCC progression via TTT-RUVBL1 assembly

   To clarify whether HPCAL1 regulates mTOR signaling via its binding to
   RUVBL1, we investigate mTOR activity in Huh7 cells with simultaneous
   depletion of HPCAL1 and RUVBL1/2. As showed, HPCAL1 ablation
   significantly increased levels of mTOR and 4EBP1 phosphorylation, which
   was largely abrogated in Huh7 cells following RUBVBL1/2 knockdown
   (Figure [140]5A). Given that RUVBL1/2 in complex with TTT is required
   for mTOR activation, we next examine whether the levels of HPCAL1 would
   affect TTT-RUVBL1 complex formation. Expectedly, the association
   between RUVBL1 and TEL2 or TTI1, two essential components of TTT
   complex, was greatly increased in Huh7 cells upon HPCAL1 depletion
   relative to control cells (Figure [141]5B). Using samples derived from
   DEN-induced liver tumors, we further showed a similar effect in liver
   tumors (Figure [142]5C), indicating that HPCAL1 indeed affects
   TTT-RUVBL1 complex, thus mTOR activation. We next ask whether the
   observed signaling event would ultimately alter HCC progression in
   vivo. To this end, we used a MYC/Trp53^-/- liver cancer model and
   CRISPR/Cas9 to delete individual Hpcal1 and Ruvbl1 or both in mouse
   livers. Ruvbl1 loss alone led to significant repression of liver
   tumorigenesis, as previously reported [143]^41. Consistent with the
   observation that Hpcal1 deficiency resulted in dramatical increase of
   tumor growth above, this effect was partially reversed in tumors with
   simultaneous deficiency of Hpcal1 and Ruvbl1 (Figure [144]5D-E),
   considering that the tumor number was higher compared with that of
   Ruvbl1-deleted tumors. Histological analyses confirmed the knockout
   efficiency and the consistent effect with evidence of Ki-67 reactivity
   (Figure [145]5F and [146]Figure S4). We further performed western
   blotting demonstrating that mTOR activation was significantly
   compromised in Ruvbl1-deleted and or co-deleted tumors (Figure
   [147]5G). These results suggest that HPCAL1 blunts liver tumor growth
   partly through RUVBL1/2-mediated mTOR activation.

Figure 5.

   [148]Figure 5
   [149]Open in a new tab

   HPCAL1 suppresses liver tumorigenesis via RUVBL1-dependent mTOR
   activation. (A) Immunoblot analysis of the HPCAL1-depleted Huh7 cell
   lysates with or without RUVBL1 knockdown. HP, HPCAL1. (B) HEK293T cells
   co-transfected with FLAG-RUVBL1 and HA-TEL2 or his-TTI1 in the presence
   or absence of HPCAL1 were subjected to Co-IP and immunoblot analysis
   with indicated antibodies. (C) Co-IP and immunoblot analysis of the
   lysates of liver tumors from WT and Hpcal1-/-. (D) Statistical analyses
   of the ratio of liver to body weight (upper), liver weight (middle) and
   the tumor number (lower) from livers of indicated mice (n-=8). (E)
   Representative images of gross morphology, HE and IHC staining from
   liver section of indicated mice. (F) Statistical analysis of the
   percentage of Ki67+ cells indicated mice. (G) Immunoblot analysis of
   the lysates of liver tumors from indicated mice using indicated
   antibodies. Scale bar, 50 µm.

HPCAL1 renders liver tumors resistant to mTORi

   In an effort to exploit the therapeutic opportunity based on the
   observed mechanism, we test whether HPCAL1 would affect the sensitivity
   of mTORi such as AZD8055, a potent, selective, and orally bioavailable
   ATP-competitive mTOR kinase inhibitor [150]^43. As expected, while
   HPCAL1 overexpression blunts the response of Huh7 cells to AZD8055,
   HPCAL1 depletion increased the sensitivity of Huh7 cells to AZD8055, as
   evidenced by CCK8 assay (Figure [151]6A, 6B and [152]Figure S5A). This
   role of HPCAL1 was demonstrated to be dependent on RUVBL1/2 (Figure
   [153]6C). Similar results were confirmed in Huh7 cells with colony
   formation assay ([154]Figure S5B-D). To extend this observation in
   mouse model, control and Hpcal1-depleted tumors in mice were treated
   with or without AZD8055. Strikingly, compared with control tumors,
   although Hpcal1-depleted mice had larger tumors, they showed higher
   sensitivity to AZD8055, with lower liver weight, liver-body ratio,
   tumor numbers and diminished Ki-67 staining (Figure [155]6D-G and
   [156]Figure S5E-H). While the overall survival times was remarkably
   decreased in sgHpcal1 mice relative to sgCON mice fed with chow, it was
   greatly prolonged in mice deficient for Hpcal1 following AZD-8055
   treatment, although comparable reduction of mTOR and 4EBP1
   phosphorylation was observed in tumors following AZD-8055 treatment
   (Figure [157]6H and [158]Figure S5I). Together, we conclude that HPCAL
   is a novel negative regulator for liver lipid metabolism and Hpcal1
   deficiency renders liver tumors addicted to mTOR signaling and boosts
   the therapeutic response to mTOR inhibitor via RUVBL1/2-dependent mTOR
   assembly.

Figure 6.

   [159]Figure 6
   [160]Open in a new tab

   HPCAL1 suppresses liver tumorigenesis via RUVBL1-dependent mTOR
   activation. Short-term growth inhibition assay of HPCAL1-overexpressed
   (A) or HPCAL1-depleted (B) Huh7 cells treated with indicated
   concentration of AZD-8055. OE, overexpression. HP, HPCAL1. (C)
   Short-term growth inhibition assay of SCR or HPCAL1-depleted Huh7 cells
   transfected with SCR or siRUVBL1 and treated with indicated
   concentration of AZD-8055. (D) Schematic representation of MYC-Trp53
   liver cancer model and mTORi treatment. This figure was drawn by using
   BioRender (JX24G8PWZC) (E) Statistical analyses of the ratio of liver
   to body weight (upper) and the number of tumors (lower) from livers of
   indicated mice fed with regular chow or chow with additional AZD-8055
   at the dose of 67 mg/kg (n=7). (F) Representative images of gross
   morphology, HE and IHC staining from liver section of indicated mice.
   (H) Statistical analysis of the percentage of Ki67+ cells from
   indicated mice. (G) Kaplan-Meier curves analyses of indicated mice with
   or with dietary addition of AZD-8055 (n=9). Scale bar, 50 µm. *, **,
   *** means p < 0.05, p < 0.01, and p < 0.001.

Discussion

   The EF hand domain, which sequesters Ca^2 professionally, is one of the
   most common domains encoded by the human genome, in contrast with a
   poor knowledge of their physiological functions [161]^31^, [162]^44^,
   [163]^45. We previously have identified that HPCAL1, a neural calcium
   sensor protein with four EF hand domains, was often lost in patients
   with HCC and correlated with deregulated cell cycle progression and an
   unfavorable prognosis [164]^31. In this study, we have used DEN/CCl4
   and MYC/Trp53-induced liver cancer model and demonstrated that genetic
   loss of HPCAL1 resulted in significantly enhanced liver tumorigenesis
   in mice, reinforcing the notion that HPCAL1 functions as a potential
   tumor suppressor in HCC. Although our prior work has indicated a role
   of HPCAL1 in cell cycle arrest via a direct binding to p21^WAF1/Cip1
   (ref), a potent cyclin-dependent kinase (CDK) inhibitor [165]^31, its
   functions in HCC has been largely unknown. Intriguingly, we found here
   that Hpcal1-deficient tumors displayed molecular signature of lipid
   metabolic reprogramming and triggered numerous genes implicated in de
   novo fatty acid and cholesterol biosynthesis, an emerging hallmark of
   cancer metabolism occurs in aggressive tumors and correlates with
   therapeutic resistance [166]^8^, [167]^12. Such deregulation of lipid
   metabolism was further linked to mTOR hyperactivation, which critically
   regulates many essential biological functions, including lipid
   metabolism. Paradoxically, tumors deficient for Hpcal1 elicits
   hypersensitivity to mTORi AZD-8055, indicating that genetic
   inactivation of Hpcal1 in hepatocytes results in lipid metabolic
   changes that are critical for HCC progression and also a metabolic
   vulnerability that offer therapeutic opportunity to mTORi. This
   observation provides the rationale for HPCAL1 as a surrogate biomarker
   for HCC treatment with mTOR hyperactivation using mTORi, considering
   the frequent loss of HPCAL1 in HCC [168]^31. Given that numerous mTORi,
   including rapamycin analogs and ATP-competitive mTOR inhibitors, have
   been approved and more mTORi are being evaluated in the clinical trials
   [169]^46, it is not yet defined whether Hpcal1-deficient tumors respond
   similarly to these mTORi, and which is more effective. It should be
   noted that while many oncogene-addicted cancers, including mTOR, show
   striking initial responses to acute targeted therapies, most patients
   ultimately develop resistance being targeted [170]^47^-[171]^49. As
   such, this common pattern of resistance to mTOR-addicted therapy
   requires additional target(s) for combined therapy [172]^50^, [173]^51,
   which might be interesting works in the future. Nevertheless, our data
   here establishes a functional link between HPCAL1 and therapeutic
   response via mTOR signaling.

   Mechanistically, we uncovered that HPCAL1 directly bound to RUVBL1,
   inhibiting the assembly of TTT complex and mTOR activation. Such
   interaction occurs not only in cells, but also are present in
   established tumors of mice, underlying the physical importance of
   RUVBL1-HPCAL1 interaction in liver cancers. In agreement with the
   previous observations by us and others [174]^31^, [175]^52, despite
   being as a Ca^2+ sensor, HPCAL1 interacted with RUVBL1 that was clearly
   not affected by Ca^2+ concentration in cancers.

   Of note, RUVBL2, the additional essential component that belongs to
   AAA+ ATPase superfamily, appears to interact with HPCAL1 with lower
   intensity. However, depleting RUVBL1 alone was sufficient to reverse
   the tumorigenesis in Hpcal1-deficient mice, suggesting HPCAL1 may
   function that is largely dependent on the intact hetero-hexamer of
   RUVBL proteins [176]^28^, [177]^39^, [178]^53. As the TTT- RUVBL1/2
   complex is required for the functional assembly of mTOR and activation
   [179]^27, the observation that HPCAL1 deficiency in HCC cells and
   tumors as well facilitates the binding of RUVBL1 to TEL2 and TTI1, two
   essential components of TTT complex, solidifies that HPCAL1 is indeed
   implicated in the TTT-RUVBL1/2 complex formation. Although RUVBL1
   depletion largely overcomes HPCAL1-mediated mTOR sensitivity and tumor
   progression, it is currently unknown whether there is other
   mechanism(s) potential involved, such as TSC1/2, AMPK and Rag upon
   energy stress [180]^23. Since the mTOR activation respond to energetic
   stresses [181]^13, such as glucose and amino acids, we did not observe
   a change in HPCAL1 expression in HCC cells and mice in response to
   glucose deprivation or glutamine (data not shown). Also, in liver
   tumors, which is exposed to glucose deficiency, Hpcal1 deletion boosted
   TTT- RUVBL1/2 complex formation. These lines of evidence suggested that
   HPCAL1 senses energetic signals to mTOR signaling that is irrelevant of
   its protein abundance, implying a potential post-translational
   modification(s) potential involved, which warrants future
   investigation. Clinically, we uncovered a negative association between
   HPCAL1 and mTOR signaling and lipid metabolic enzymes in human liver
   cancers. Notably, HPCAL1 expression is not significantly altered in
   livers of mice with HFD or type 2 diabetes (T2D) (data not shown), and
   we have not observed the clinical relevance of HPCAL1 in non-alcoholic
   fatty liver disease (NAFLD) and NAFLD-associated HCC in patients, due
   to limited clinical information available for relevant analysis. Given
   that NAFLD is increasingly ranking the top risk factor for human liver
   cancer [182]^54, dissecting the role and clinical relevance of HPCAL1
   with NAFLD-associated HCC is of great significance for potential cancer
   subtyping and therapy.

   While the role of HPCAL1 in HCC growth and therapeutic response has
   been investigated here, it is current unknown and will be an
   interesting work in future whether HPCAL1 is implicated in the
   aggressive phenotypes of HCC. In addition, although we used two
   separate HCC mouse model, the observations and conclusions obtained
   from the whole-body knockout mice did not indicate a direct role of
   HPCAL1 in livers. We can't rule out the possibility of potential side
   effect derived from other organs upon HPCAL1 deficiency, particularly
   considering that HPCAL1 is a neural calcium sensor that is expressed at
   higher levels in brain than liver (Human Protein Atlas). Nevertheless,
   this is an unresolved and interesting work in the future using
   conditional-knockout mouse model. Collectively, this study identified
   HPCAL1 as an important regulator of lipid biosynthesis and mTOR
   addiction, and proposed a metabolic vulnerability in Hpcal1-deficient
   HCC that could be exploited for antitumoral therapy using mTORi.

Supplementary Material

   Supplementary figures and tables.
   [183]Click here for additional data file.^ (3.3MB, pdf)

Acknowledgments