Comparative proteomic analysis of SLC13A5 knockdown reveals elevated ketogenesis and enhanced cellular toxic response to chemotherapeutic agents in HepG2 cells

Abstract

Solute carrier family 13 member 5 (SLC13A5) is an uptake transporter mainly expressed in the liver and transports citrate from blood circulation into hepatocytes. Accumulating evidence suggests that SLC13A5 is involved in hepatic lipogenesis, cell proliferation, epilepsy, and bone development in mammals. However, the molecular mechanisms behind SLC13A5-mediated physiological/pathophysiological changes are largely unknown. In this regard, we conducted a differential proteome analysis in HepG2 and SLC13A5-knockdown (KD) HepG2 cells. A total of 3826 proteins were quantified and 330 proteins showed significant alterations (fold change ≥ 1.5; p < 0.05) in the knockdown cells. Gene ontology enrichment analysis reveals that 38 biological processes were significantly changed, with ketone body biosynthetic process showing the most significant upregulation following SLC13A5-KD. Catalytic activity and binding activity were the top two molecular functions associated with differentially expressed proteins, while HMG-CoA lyase activity showed the highest fold enrichment. Further ingenuity pathway analysis predicted 40 canonical pathways and 28 upstream regulators (p < 0.01), of which most were associated with metabolism, cell proliferation, and stress response. In line with these findings, functional validation demonstrated increased levels of two key ketone bodies, acetoacetate and β-hydroxybutyrate, in the SLC13A5-KD cells. Additional experiments showed that SLC13A5-KD sensitizes HepG2 cells to cellular stress caused by a number of chemotherapeutic agents. Together, our findings demonstrate that knockdown of SLC13A5 promotes hepatic ketogenesis and enhances cellular stress response in HepG2 cells, suggesting a potential role of this transporter in metabolic disorders and liver cancer.

1. Introduction

Solute carrier family 13 member 5 (SLC13A5), the mammalian sodium-coupled citrate transporter (NaCT) initially cloned from rat brain in 2002, is expressed on the plasma membrane of mammalian cells and mediates the cellular uptake of citrate in the body [1]. As a key intermediate in the tricarboxylic acid (TCA) cycle, citrate plays pivotal roles in the cellular metabolism of carbohydrates and fatty acids, as well as energy production in the mitochondria. Cytosolic citrate is cleaved to oxaloacetate and acetyl coenzyme A (acetyl-CoA) by ATP citrate lyase, which is used for the biosynthesis of triglycerides, fatty acids, cholesterol, and low-density lipoproteins [2]. Relatively high citrate concentrations (100–150 μM) were found in human blood circulation, providing a sufficient source for cytosolic citrate uptake [3]. Given the high transport capacity of human SLC13A5 towards citrate (Km = 2,254 μM, Vmax = 25,117 pmol/min/mg protein) [4], it has been speculated that uptake of citrate by SLC13A5 from the blood stream plays a significant role in regulating the intracellular citrate level and cellular energy metabolism.

In mammals, SLC13A5 mRNA is predominantly expressed in the liver, followed by the brain, testis and ovary with moderate expression levels [4–6]. In line with its abundance in the liver, accumulating evidence reveals that SLC13A5 plays an important role in regulating hepatic citrate level and is associated with metabolic disorders. Slc13a5 knockout in mice by whole-body genetic deletion or liver selective siRNA knockdown, decreased hepatic citrate uptake and effectively protected the animals from hepatic insulin resistance and high-fat diet-induced lipid accumulation [6, 7]. In rat primary hepatocytes, induction of Slc13a5 by benzo[a]pyrene increased the hepatic citrate uptake and resulted in an increased incorporation of citrate into sterol and fatty acid synthesis [8]; while the expression of Slc13a5 was also elevated in a rat model of type 2 diabetes [9]. Moreover, our previous work showed that silencing SLC13A5 in human liver cancer cells, HepG2 and Huh7, led to significant reduction of intracellular levels of citrate and phospholipids [10]. We and others also reported the involvement of SLC13A5 upregulation in lipid accumulation in human primary hepatocytes after exposure to either rifampicin or interleukin-6 [4, 11]. Together, these studies firmly establish SLC13A5 as an important energy sensor regulating cytosolic citrate level in hepatocytes, insulin sensitivity, and hepatic de novo lipogenesis [6, 8, 10].

In addition to its role in energy homeostasis, reduced expression of the SLC13A5 homologue (also known as Indy) extended the life span of D. melanogaster and C. elegans [12, 13]. Silencing SLC13A5 inhibited the proliferation of human hepatoma cells both in vitro and in vivo using a xenograft nude mice model, with induced G1 phase cell cycle arrest and an increased expression of p21, a potent cyclin-dependent kinase inhibitor [10]. Clinical studies further extended the association between SLC13A5 single nucleotide polymorphisms (SNPs) and early-onset epileptic encephalopathy, teeth hypoplasia, and developmental delay in children [14, 15]. The mechanisms by which SLC13A5 SNPs lead to severe seizure and other accompanying clinical signs are unclear. In animal studies, although Slc13a5-null mice showed no deleterious neurological phenotype, defective enamel and bone development have been reported in young knockout mice, including lack of mature mineralized enamel, decreased bone mineral density, and impaired bone mineralization and formation [16]. Collectively, these findings suggest SLC13A5 has a variety of physiological and pathophysiological functions across different species that influence energy metabolism, cell proliferation, bone formation, and central nervous function. Nevertheless, the molecular mechanisms underlying these biological functions of SLC13A5 are not fully elucidated, and the systematic impact of SLC13A5 expression at the whole proteome level remains unknown.

In this study, we conducted a quantitative analysis comparing the proteome profile in HepG2 and SLC13A5-knockdown (KD) HepG2 cells. Leading pathways identified from investigation of proteome-wide changes-mediated by SLC13A5 are experimentally validated. Our results uncovered that silencing of SLC13A5 in HepG2 cells significantly changes the expression of more than 300 proteins involved in 40 canonical pathways and 28 upstream regulators with most of them associated with metabolism, cell proliferation and stress responses. Functional validation indicated that SLC13A5-KD increases hepatic ketogenesis while sensitizes cellular response to chemotherapeutic agents in HepG2 cells.

2. Materials and Methods

2.1. Materials

CCK-8 assay kit (ALX-850–039) was purchased from Enzo Life Sciences (Farmingdale, NY). 5-fluorouracil (5-FU, F6627), doxorubicin (D1515) and ketone body assay kit (MAK134) were obtained from Sigma-Aldrich (St. Louis, MO). Cisplatin (S1166) and sorafenib (S7397) were from Selleck Chemicals (Houston, TX). The antibodies used in this study included anti-SLC13A5 (ab89181) from Abcam, anti-cleaved Caspase-3 (9661S) from Cell Signaling Technology and anti-β-actin (A1978) from Sigma-Aldrich. Oligonucleotide primers were synthesized by Integrated DNA Technologies (Coralville, IA). Unless otherwise specified, all cell culture reagents were purchased from Sigma-Aldrich or Life Technologies.

2.2. Cell culture and lentiviral infection

HepG2 cells were obtained from the American Tissue Culture Collection (Manassas, VA). The authenticity of HepG2 cells was confirmed by short tandem repeat profiling analysis. The cells were maintained in DMEM, supplemented with 10% fetal bovine serum (FBS), 100 U/mL penicillin and 100 μg/mL streptomycin in a humidified incubator at 37°C with 5% CO₂. For SLC13A5-KD plasmids, DNA oligonucleotides encoding SLC13A5-shRNA (GATCCGAGATCAACGTGCTGATCTGCTTCTCGAGAAGCAGATCAGCACGTTGATCTCTTTTTG, the targeted sequences were underlined) were subcloned into the BamHI and EcoRI sites of the pGreenPuro™ shRNA expression lentiviral vector from System Biosciences (Mountain View, CA). Lentivirus packaging was carried out according to a previous protocol [17], and virus titering was performed using the Lenti-XTM p24 Rapid Titer Kit (Clontech Laboratories) according to the manufacturer’s instructions. For virus infection, HepG2 cells plated in 100 mm culture dishes at the density of 5 × 10⁴ cells/mL (10 mL per dish) were infected with lentivirus at 2.5 × 10⁷ lentiviral particles/mL. Cells infected with lentivirus carrying the empty pGreenPuro vector were served as the control group (shControl). Three dishes/per group were independently infected and total proteins were extracted for further analysis 72 h after infection.

2.3. Liquid chromatography-tandem mass spectrometric data acquisition and analysis

SLC13A5-KD and control HepG2 cells were lysed in 4% sodium deoxycholate after rinse in phosphate-buffered saline. Lysates were washed, reduced, alkylated and trypsinized on filter as described by previously [18, 19]. Tryptic peptides were separated on a nanoACQUITY UPLC analytical column (BEH130 C18, 1.7 μm, 75 μm × 200 mm, Waters) over a 180 min linear acetonitrile gradient (3 – 43%) with 0.1% formic acid on a Waters nano-ACQUITY UPLC system and analyzed on a coupled Waters Synapt G2S HDMS mass spectrometric system. Spectra were acquired using an ion mobility linked parallel mass spectrometry (UDMS^e) and analyzed as described by Distler et al. [20]. Peaks were resolved using Apex3D and Peptide3D algorithms [21]. Tandem mass spectra were searched against a UniProt human reference proteome and its corresponding decoy sequences using an ion accounting algorithm [22]. Resulting hits were validated at a maximum false discovery rate of 0.04. Protein abundance ratios between the SLC13A5 knockdown and control cells were measured by comparing the MS1 peak volumes of peptide ions at the low collision energy cycle, whose identities were confirmed by MS2 sequencing at the elevated collision energy cycle as described above. Label-free quantifications were performed using an aligned AMRT (Accurate Mass and Retention Time) cluster quantification algorithm developed by Qi et al. [23]. The non-targeted global proteome analysis was normalized to total protein.

2.4. Reverse transcription and quantitative real-time PCR

Quantitative real-time PCR was performed to compare the relative mRNA expression levels of genes in HepG2 and SLC13A5-KD HepG2 cells. Total RNA was isolated using Trizol reagent (15596026, Invitrogen, Carlsbad, CA) and same amount of total RNA (1 μg) was used to synthesize cDNA by reverse transcription using High Capacity cDNA Archive Kit (4368813, Applied Biosystems, Foster City, CA). The primers used in this study were listed in Supplementary Table 1. The human GAPDH mRNA was amplified in parallel as the internal control. Real-time PCR was performed on an Applied Biosystems StepOnePlus (Applied Biosystems) at 95°C for 30 seconds, followed by 40 cycles of 95°C for 5 seconds and 60°C for 30 seconds. Data analysis was carried out using the 2^−ΔΔCt method for relative quantification. All samples were normalized to GAPDH.

2.5. Ketone body analysis

HepG2 cells were seeded into 6-well plate at the density of 1 × 10⁵ cells/mL (2 mL per well) and were infected with control or SLC13A5-shRNA lentivirus as detailed above. After infection for 72 hours, cells were subject to ketone body analysis to measure the intracellular concentrations of two ketone bodies acetoacetate and β-hydroxybutyrate. Determination of ketone body levels was carried out using Sigma-Aldrich’s ketone body assay kit (MAK134) according to the manufacturer’s instructions and the absorbance of the samples was measured at 340 nm using a SpectraMax M5 microplate reader (Molecular Devices, San Jose, CA).

2.6. Western blotting

Equal amounts of total proteins (30 μg) were resolved by SDS-polyacrylamide gels and transferred onto PVDF membranes. The membranes were incubated at 4°C overnight with antibodies against SLC13A5 (1:200), cleaved Caspase-3 (1:1000) or β-actin (1:5000) diluted in 5% BSA in washing buffer. Afterwards, the membranes were incubated with HRP-conjugated secondary antibodies at room temperature for 2 hours. Chemiluminescent signals were then developed using SuperSignal™ West Pico PLUS Chemiluminescent Substrate (Thermo Fisher Scientific).

2.7. Cell viability assay

CCK-8 assay was used to measure the cell viability of HepG2 cells. Briefly, the cells were seeded into 96-well plate at the density of 3,000 cells/well and were infected with control or SLC13A5-shRNA lentivirus. After infection for 72 hours, cells were treated with chemotherapeutic agents including 5-FU (0.4–400 μM), doxorubicin (0.05–25 μM), cisplatin (0.2–200 μM) and sorafenib (0.2–100 μM) for another 48 hours. CCK-8 reagent (10 μL) was then added into each well and incubated at 37°C for 2 hours, the absorbance was determined at 450 nm using a SpectraMax M5 microplate reader (Molecular Devices, San Jose, CA). To evaluate the cytotoxicities of these compounds on HepG2 cells, their IC₅₀ values were calculated based on the dose-response curves.

2.8. Bioinformatic and statistical analysis

Protein samples from three independent replicates in the same experiment were used for the proteomic analysis. Statistically significant proteome changes between HepG2 and SLC13A5-KD HepG2 cells were identified by ANOVA test with a significance threshold of p < 0.05. Perturbations caused by SLC13A5-KD to canonical pathways and upstream regulators were analyzed with Qiagen Ingenuity database using Fisher’s exact test, as described by Krämer et al. [24]. Gene ontology analyses were performed with DAVID database using an EASE enrichment p-value, a slightly modified Fisher’s exact test as described by Huang et al. [25], to identify significant changes in biological processes, molecular functions, and cellular components associated with the SLC13A5-KD. Fisher’s exact test examines whether there is a greater than expected proportion of genes within one category than expected by chance. The resulting p-value measures the probability that the association between an experimental gene set showing significant changes and a defined pathway or biological process is due to random coincidence. The smaller the p-value the less likely that the resulting hit is random. Furthermore, to predict directional effects of proteome changes, z-score (standard score) was calculated as described by Krämer et al. [24]. A Z-score is a statistical measure of the match between expected relationship direction from published literature and observed gene expression from an experimental data set. It was used to infer likely activation states of pathways or upstream regulators based on comparison with a model that assigns random regulation directions. |z|>2 is empirically considered significant.

All functional validation data were expressed as Mean ± standard deviation (S.D.) from three independent experiments. Statistical analysis was carried out using Prism 5.0 (GraphPad Software, CA). The significance of the difference between groups was estimated by Student’s t-test or one-way analysis of variance (ANOVA) followed by Dunnett’s post-hoc test, p < 0.05 indicated statistically significant difference. IC₅₀ values were compared using extra sum-of-squares F test.

3. Results

3.1. Quantitative analysis of the proteome of SLC13A5 knockdown HepG2 cells

As shown in Figure 1A, both mRNA and protein levels of SLC13A5 were down-regulated after SLC13A5 shRNA transfection, indicating the successful knockdown of SLC13A5 in HepG2 cells. To evaluate the effect of SLC13A5-KD on a wide range of proteins, a quantitative proteomic analysis of the proteome of HepG2 and SLC13A5-KD HepG2 cells was performed using high resolution LC-MS/MS. In total, we detected 5297 proteins, with which 3826 proteins were quantified with false discovery rate (FDR) less than 0.04. Of the 3826 proteins, 330 proteins showed significant changes in abundance in the SLC13A5-KD HepG2 cells, with fold change ≥ 1.5 and p < 0.05. Of the 330 proteins significantly changed, 210 proteins increased while 120 proteins decreased in abundance in the SLC13A5-KD cells (Figure 1B). The global proteome changes, in particular significantly altered proteins in the SLC13A5-KD HepG2 cells, were visualized in a volcano plot as shown in Figure 1C. Interestingly, with SLC13A5-KD, the alterations of ABC transporters and other SLC transporters were minor in HepG2 cells. Of note, the change of SLC25A1, which transports mitochondrial citrate derived from TCA cycle to cytosol [26], was less than 1.5 fold. This is consistent with our previous report showing no significant change of SLC25A1 mRNA expression after SLC13A5-KD in HepG2 cells [10]. Hence, no significant transporter-associated compensation on cytosol citrate level is expected, as evidenced by the decreased level of citrate in HepG2 cells after SLC13A5-KD [10]. A heat map was presented to illustrate the top changed proteins (fold change ≥ 2; p < 0.05) in SLC13A5-KD HepG2 cells, with average linkage hierarchical clusters indicating the degree of similarity in protein alterations (Figure 1D).

Figure 1.

Comparative analysis of the differential proteome in HepG2 and SLC13A5-KD HepG2 cells. (A) mRNA and protein expression levels of SLC13A5 were down-regulated after SLC13A5 shRNA transfection in HepG2 cells. ***p < 0.001 compared to the shControl group. (B) Results of protein abundance analysis in HepG2 and SLC13A5-KD HepG2 cells. As detailed in “Materials and Methods”, tryptic peptides were separated on a nanoACQUITY UPLC analytical column over a 180 min linear acetonitrile gradient (3 – 43%) with 0.1% formic acid on a Waters nano-ACQUITY UPLC system and analyzed on a coupled Waters Synapt G2S HDMS mass spectrometric system. (C) Volcano plot indicates 330 proteins showing significant changes in abundance (fold change ≥ 1.5; p < 0.05). (D) Heat map demonstrates the top changed proteins (fold change ≥ 2; p < 0.05) in SLC13A5-KD HepG2 compared with HepG2 cells with average linkage hierarchical clustering. Triplicate samples from each group were analyzed.

3.2. Biological processes associated with altered proteins after SLC13A5 knockdown

Gene ontology (GO) annotation and enrichment analyses were carried out to identify the biological processes of differentially expressed proteins in HepG2 and SLC13A5-KD HepG2 cells. Proteins with fold change ≥ 1.5 and p < 0.05 were included in the GO annotation. Of the 330 genes showing significant changes, 391 biological processes defined in gene ontology could be assigned to 247 genes. Based on the number of attributed genes, cellular process (GO: 0009987, 121 hits) and metabolic process (GO: 0008152, 109 hits) were the top two biological processes of altered proteins after SLC13A5-KD, followed by cellular component organization or biogenesis, localization, biological regulation, developmental process, multicellular organismal process, response to stimulus, biological adhesion, immune system process, reproduction and locomotion, as shown in Figure 2A. Moreover, secondary and tertiary level annotation showed that significantly changed cellular processes included cell communication, cell cycle, cell proliferation, signal transduction, etc. While the metabolic processes of nucleobase-containing compound, protein, cellular amino acid, lipid, carbohydrate and tricarboxylic acid are among the top metabolic processes significantly changed in SLC13A5-KD cells (Supplementary Figure 1).

Figure 2.

Biological processes of differentially expressed proteins in HepG2 and SLC13A5-KD HepG2 cells by GO annotation and enrichment analysis. (A) Biological processes associated with differentially expressed proteins (fold change ≥ 1.5, p < 0.05) by GO annotation. (B) Biological processes of proteins significantly increased (fold change ≥ 1.5, p < 0.05) in SLC13A5-KD cells. Red bars indicate upregulation and color intensity indicates the levels of enrichment (1.6–43.3). (C) Biological processes of proteins significantly decreased (fold change ≥ 1.5, p < 0.05) in SLC13A5-KD cells. Green bars indicate down-regulation and color intensity indicates the levels of enrichment (3.6–62.2).

Utilizing the same criteria of proteins with fold change ≥ 1.5 and p < 0.05, GO enrichment analysis suggested 19 biological processes that were significantly upregulated in the SLC13A5-KD HepG2 cells (Figure 2B). Among the 19 biological processes, ketone body biosynthesis, mitochondrial transport, adenine transport, metaphase/anaphase transition of mitotic cell cycle, negative regulation of mRNA splicing via spliceosome, protein K63-linked deubiquitination, and fatty acid catabolic process exhibited enrichment greater than 10-fold (Supplementary Table 2). The most significant difference was found in the ketone body biosynthetic process. In line with the GO biological process annotation, the majority of these significantly upregulated biological processes were associated with primary metabolic process, signal transduction and cell cycle regulation.

On the other hand, 19 biological processes that were significantly down-regulated in the SLC13A5-KD cells were also identified. Biological processes with fold depletion > 10 include wound healing and spreading of cells, response to purine-containing compound, actin filament network formation, actin crosslink formation, purine ribonucleoside monophosphate biosynthetic and microtubule-based process, with wound healing and spreading of cells being the most significant down-regulated two processes (Figure 2C).

3.3. Molecular functions of altered proteins after SLC13A5 knockdown

Regarding the molecular functions of altered proteins, GO annotation showed that the 247 genes have generated a total of 209 molecular function hits (Figure 3A), showing most proteins are associated with catalytic activity (GO: 0003824, 80 hits) and binding activity (GO: 0005488, 77 hits). Since metabolism was the top altered biological process in the SLC13A5-KD HepG2 cells, as expected, differentially expressed proteins have generated the most molecular function hits in the catalytic activity category, in particular the hydrolase activity, transferase activity and oxidoreductase activity, among others. Another important molecular function of differentially expressed proteins was binding activity, especially nucleic acid binding and protein binding as shown in Supplementary Figure 2. Additional molecular functions including structural molecule activity, transporter activity, and receptor activity were also associated with SLC13A5-KD as shown in Figure 3A.

Figure 3.

Molecular functions of differentially expressed proteins in HepG2 and SLC13A5-KD HepG2 cells by GO annotation and enrichment analysis. (A) Molecular functions associated with differentially expressed proteins (fold change ≥ 1.5, p < 0.05) by GO annotation. (B) Molecular functions of proteins significantly increased (fold change ≥ 1.5, p < 0.05) in SLC13A5-KD cells. Red bars indicate upregulation and color intensity indicates the levels of enrichment (1.2–85.7). (C) Molecular functions of proteins significantly decreased (fold change ≥ 1.5, p < 0.05) in SLC13A5-KD cells. Green bars indicate down-regulation and color intensity indicates the levels of enrichment (1.3–45.1).

In line with the GO annotation, the enrichment analysis indicated that proteins significantly upregulated after knockdown of SLC13A5 exhibited functions mainly in catalytic activity and binding activity, including nucleotide binding, serine-type carboxypeptidase activity, protein binding, carboxypeptidase activity, hydroxymethylglutaryl-CoA lyase activity (HMGCL), phosphatidylinositol-4,5-bisphosphate binding, protein complex binding, natriuretic peptide receptor activity, motor activity, p53 binding, fatty-acyl-CoA binding and guanyl-nucleotide exchange factor activity (Figure 3B). Among these functions, HMGCL showed the highest enrichment of 85.7-fold (Supplementary Table 3). Of note, HMGCL converts 3-hydroxy-3-methylglutaryl-CoA into acetoacetate and acetyl-CoA in mitochondria and is a key enzyme in ketogenesis [27]. It is therefore speculated that such high fold enrichment of HMGCL activity underlies the most significantly upregulated ketone body biosynthetic process in the SLC13A5-KD cells.

On the other hand, the molecular functions of proteins significantly down-regulated were poly(A) RNA binding, protein binding, ATPase binding, identical protein binding, cadherin binding involved in cell-cell adhesion, large ribosomal subunit rRNA binding, receptor signaling protein serine/threonine kinase activity, ATP binding and structural constituent of ribosome, with large ribosomal subunit rRNA binding showing the highest enrichment of 45.1-fold (Figure 3C & Supplementary Table 3).

3.4. Cellular distributions of altered proteins after SLC13A5 knockdown

GO annotation and enrichment analysis were also carried out to locate the cellular distributions of differentially expressed proteins in HepG2 and SLC13A5-KD HepG2 cells. As shown in Figure 4A, 179 cellular component hits were generated from the 247 genes. In this regard, the majority of altered proteins were localized in cell part (GO: 0044464, 78 hits), followed by organelle (GO: 0043226, 48 hits), macromolecular complex (GO: 0032991, 25 hits), membrane (GO: 0016020, 17 hits), extracellular region (GO: 0005576, 7 hits), cell junction (GO: 0030054, 2 hits) and extracellular matrix (GO: 0031012, 2 hits).

Figure 4.

Cellular components of differentially expressed proteins in HepG2 and SLC13A5-KD HepG2 cells by GO annotation and enrichment analysis. (A) Cellular components associated with differentially expressed proteins (fold change ≥ 1.5, p < 0.05) by GO annotation. (B) Cellular components of proteins significantly increased (fold change ≥ 1.5, p < 0.05) in SLC13A5-KD cells. Red bars indicate upregulation and color intensity indicates the levels of enrichment (1.4–18.7). (C) Cellular components of proteins significantly decreased (fold change ≥ 1.5, p < 0.05) in SLC13A5-KD cells. Green bars indicate down-regulation and color intensity indicates levels of enrichment (1.3–5.8).

Enrichment analysis further demonstrated that proteins significantly increased after SLC13A5-KD were primarily localized in centrosome and extracellular exosome, followed by exocytic vesicle, microtubule cytoskeleton, membrane, extracellular space, cytosol, melanosome, nuclear envelope, mitochondrion, mitochondrial inner membrane and primary cilium (Figure 4B). The presence of mitochondrion and mitochondrial inner membrane on this list is expected, since it is the organelle where fatty acid β-oxidation, TCA cycle and ketogenesis occur [28, 29]. In contrast, cellular distributions of proteins significantly decreased in the SLC13A5-KD cells were found in cytoplasm, extracellular exosome, cytosol, focal adhesion, ribosome, nucleoplasm, cytoskeleton, intracellular ribonucleoprotein complex, actin cytoskeleton, nucleus and cell-cell adherens junction (Figure 4C & Supplementary Table 4).

3.5. Canonical pathways and upstream regulators altered in SLC13A5 knockdown cells

To understand the key initiating molecular events after knockdown of SLC13A5 in HepG2 cells, canonical pathways and upstream regulators were predicted by ingenuity pathway analysis (IPA) based on the proteins that significantly changed (fold change ≥ 1.5 and p < 0.05). A total of 40 canonical pathways have been projected to be significantly altered in the SLC13A5-KD HepG2 cells (p < 0.01) (Figure 5). Based on the activation z-score, activated pathways mainly included LXR/RXR activation, corticotropin releasing hormone signaling, relaxin signaling, ERK/MAPK signaling, cardiac hypertrophy signaling, NGF signaling, synaptic long-term potentiation, protein kinase A signaling, GNRH signaling and FGF signaling. While pathway inhibition was observed in EIF2 signaling, cAMP-mediated signaling, actin cytoskeleton signaling, Gαi signaling, and acute phase response signaling. Notably, ketogenesis was also predicted by IPA as one of the top perturbed pathways in the SLC13A5-KD cells, with p < 0.01 (Table 1). From a molecular signaling perspective, these predicted canonical pathways provide explanations for the functional changes observed in SLC13A5-KD HepG2 cells. In particular, the alterations in energy metabolism, as most of the perturbed pathways are associated with regulation of the metabolism of cholesterol, lipid, glucose and protein.

Figure 5.

Ingenuity pathway analysis showing the canonical pathways perturbed in SLC13A5-KD HepG2 cells. Forty statistically significant canonical pathways identified in SLC13A5-KD HepG2 cells are ranked according to their p values (−log P). Red bars indicate positive z-score with corticotropin releasing hormone signaling showing the highest value of 1.63; blue bars indicate negative z-score with actin cytoskeleton signaling showing the lowest value of −0.71; white bars indicate no activity pattern available.

Table 1.

Pathways perturbed in the SLC13A5-KD HepG2 cells.

Canonical Pathways	p-value	Activation z-score	Genes
LXR/RXR Activation	0.000	1.13	SCD, ALB, LYZ, APOH, APOA2, ARG2, FGA, HMGCR, APOC3
EIF2 Signaling	0.003	−0.38	RPS15, RPLP1, RPL36A, RPL34, EIF4A3, SOS1, RPL10, MAP2K1, RPLP0
FXR/RXR Activation	0.003		ALB, APOH, APOA2, CREBBP, FGA, SULT2A1, APOC3
Atherosclerosis Signaling	0.003		ALB, LYZ, APOA2, COL18A1, PLA2G12B, APOC3, ITGA4
Caveolar-mediate d Endocytosis Signaling	0.005		ALB, FLNA, HLA-A, ITGA4, MAP3K2
Pyrimidine Deoxyribonucleot ides De Novo Biosynthesis I	0.005		DUT, RRM2, CMPK1
Corticotropin Releasing Hormone Signaling	0.007	1.63	NPR1, CREBBP, PRKAR2A, NPR2, RAP1A, MAP2K1
Leucine Degradation I	0.008		HMGCLL1, HMGCL
Relaxin Signaling	0.008	1.34	PDE6A, PDE4C, NPR1, PRKAR2A, NPR2, RAP1A, MAP2K1
HGF Signaling	0.009	0.00	MET, SOS1, RAP1A, MAP2K1, ITGA4, MAP3K2
Ketogenesis	0.010		ACAT1, HADHA, HADHB, HMGCL, HMGCLL1
Phospholipase C Signaling	0.011	0.00	RHOQ, RALA, ARHGEF16, SOS1, CREBBP, PLA2G12B, RAP1A, MAP2K1, ITGA4
ERK/MAPK Signaling	0.011	0.71	SOS1, CREBBP, PRKAR2A, MAPKAPK5, PLA2G12B, RAP1A, MAP2K1, ITGA4
Purine Nucleotides De Novo Biosynthesis II	0.012		ADSS, GART
Dolichyl-diphosp hooligosaccharid e Biosynthesis	0.012		DPM1, ALG1
Gap Junction Signaling	0.015		TUBA1B, NPR1, SOS1, PRKAR2A, NPR2, MAP2K1, MAP3K2
Histamine Biosynthesis	0.015		HDC
FAK Signaling	0.018		CSK, SOS1, MAP2K1, GIT2, ITGA4
cAMP-mediated signaling	0.020	−0.38	GABBR2, PDE6A, PDE4C, AKAP9, CREBBP, PRKAR2A, RAP1A, MAP2K1
phagosome maturation	0.023		CTSD, TUBA1B, CTSA, HLA-A, ATP6V0A2, ATP6V0D1
Actin Cytoskeleton Signaling	0.024	−0.71	MYH2, FLNA, CSK, SOS1, ARHGAP35, GSN, MAP2K1, ITGA4
RAR Activation	0.027		BRD7,CSK,SMARCB1,CREBBP,PRKAR2A,MAP2K1,RDH13
Cardiac Hypertrophy Signaling	0.028	0.38	ADSS,RHOQ,SOS1,CREBBP,PRKAR2A,CACNA1C,MAP2K1,MAP3K2
BMP signaling pathway	0.029		SOS1,CREBBP,PRKAR2A,MAP2K1
Glycine Biosynthesis III	0.030		AGXT
Spermidine Biosynthesis I	0.030		SRM
Clathrin-mediate d Endocytosis Signaling	0.032		MET,ALB,LYZ,STON2,APOA2,CLTA,APOC3
NGF Signaling	0.034	0.45	SOS1,CREBBP,RAP1A,MAP2K1,MAP3K2
Granzyme A Signaling	0.037		H1FX, CREBBP
Synaptic Long Term Potentiation	0.037	1.00	CREBBP, PRKAR2A, CACNA1C, RAP1A, MAP2K1
Gai Signaling	0.037	−0.45	GABBR2, RALA, SOS1, PRKAR2A, RAP1A
Protein Kinase A Signaling	0.038	0.33	PDE6A, MYH2, PDE4C, FLNA, H1FX, AKAP9, CREBBP, PRKAR2A, RAP1A, MAP2K1, CDC27
5-aminoimidazol e Ribonucleotide Biosynthesis I	0.045		GART
D-glucuronate Degradation I	0.045		DCXR
Methionine Salvage II (Mammalian)	0.045		BHMT
Thyroid Hormone Biosynthesis	0.045		CTSD
Acute Phase Response Signaling	0.046	−0.45	ALB, APOH, APOA2, SOS1, FGA, MAP2K1
Neuregulin Signaling	0.046		SOS1, ERBB3, MAP2K1, ITGA4
GNRH Signaling	0.048	0.45	SOS1, CREBBP, PRKAR2A, MAP2K1, MAP3K2
FGF Signaling	0.049	1.00	MET, SOS1, CREBBP, MAP2K1

Pathway analysis also suggested that 28 upstream regulators were changed after knockdown of SLC13A5 in HepG2 cells. Based on the activation z-score, activated upstream regulators included TP53, HNF4A, ESR1, Cg, RORA, PPARα, LEP, SOX4, GATA1, MITF, SP1, PPARδ, SMAD4, TCF7L2, RB1, IL3, and POMC. While MAP4K4, HRAS, MYC, MYCN, CST5, PML, SREBF1, SREBF2, ESR2 and THRB were repressed (Table 2). To further understand the molecular functions of these 28 upstream regulators, they were categorized based on their participation in specific biological processes, including metabolism, cell proliferation, and cellular stress response, as shown in Figure 6.

Table 2.

Altered upstream regulators and their target genes in SLC13A5-KD HepG2 cells.

Upstream Regulator	Activation z-score	p-value	Genes
MAP4K4	−1.46	0.000	ACADS, ATP6V0A2, CYC1, HMGCL, LDHB, PDHX, PRKAR2A, SCD
PPP3R1	0.00	0.000	CACNA1C, CLIC5, LYZ, SCD
TP53	1.34	0.000	ALB, APBB2, ATAD2, BICD2, COL14A1, COL18A1, CSK, CTSD, DCTN2, DUT, ECH1, EPHX1, GART, GSN, HDC, HMGCR, IDH1, KIAA0368, KPNA2, LYZ, MAP2K1, MET, NAP1L4, PALLD, PDE6A, PDHX, PECAM1, PPIC, PRKAR2A, RACGAP1, RRM2, SMARCB1, SYN1, TDO2, ZEB2
HNF4A	0.27	0.001	ADSS, ANKZF1, APOA2, APOC3, APOH, ARG2, ASAH1, BHMT, CLCN6, CLTA, COPS8, CSK, CTSA, CUL2, DPF3, DPM1, EPHX1, EVPL, FGA, FLNA, GRWD1, GSN, GSTK1, MBD4, NBAS, NUAK1, OTUD7B, PELO, PLA2G12B, PTK7, R3HDM1, RBKS, RBM6, RPL10, RPLP1, SCD, SERPINB3, SLC7A6OS, STK24, SULT2A1, TDO2, THOC6, UCP2, USP15, ZKSCAN5, ZNHIT6
HRAS	−0.13	0.001	COL18A1, EPHX1, FAM213A, FLNA, GRN, GSN, HDC, ITGA4, KRT16, LYZ, MET, PCDHGA3, PIN1, PPIC, RPLP1, RRM2, SYN1, TACC3
MYC	−1.20	0.001	ALB, APOC3, CLUH, COL14A1, CTSD, DNTT, EVPL, EXOSC8, FCGRT, FLNA, GART, GREB1, HLA-A, HNRNPAB, IDH1, KAT2A, LDHB, LYZ, MAPKAPK5, NAP1L1, PECAM1, RPL10, RPLP1, RRM2, SNRPD1, SRM, USP54
MYCN	−1.93	0.001	CHPF2, COL18A1, DPYSL3, HLA-A, NACA, RPL10, RPLP0, RPLP1, RPS15, TUBA1B, ZEB2
CST5	−1.51	0.002	COPS8, CREBBP, EDC4, EVPL, NIFK, PDCL3, PPIC, RRP1, RRP15, RTN3, SRSF4
PML	−1.85	0.002	ACADS, CPE, HMGCR, LYZ, MAP2K1, SCD, UCP2
SREBF1	−0.05	0.002	AACS, ACADS, APOA2, APOC3, HMGCR, IDH1, RPLP0, SCD, UCP2
SREBF2	−0.73	0.003	AACS, APOA2, HMGCR, IDH1, SCD
ESR1	1.31	0.004	ATP6V0D1, CDC27, COTL1, CPE, CREBBP, CTSD, DST, ERBB3, FLNA, GABBR2, GIT2, GREB1, GSN, HMGCR, LGALS3BP, MYO9B, NUPL2, PALLD, POLR3A, RAB35, RALA, RHOQ, SLC25A4, SLC9A3R1, SOS1, SP100, SRM, TDO2, TRPM2, USP8
Cg	1.92	0.004	BHMT, CNNM1, HLA-A, HMGCR, ITGA4, KCNT2, LGALS3BP, NPR1, PECAM1, PLXND1, RNF130, SULT2A1
RORA	1.97	0.005	APOC3, BHMT, HMGCR, LPIN2, SCD, SLC25A24, SULT2A1, UCP2
PPARA	0.60	0.006	ACADS, AGXT, APOA2, APOC3, CD63, ECH1, FGA, GSTK1, HMGCR, SCD, SRM, SULT2A1, UCP2
ESR2	−1.03	0.006	CLIC5, CNNM1, CTSD, EPHA5, GABBR2, GREB1, KCNT2, KRT16, RALA, RHOQ, SLC25A4, SLC9A3R1, USP8
LEP	0.54	0.007	APOA2, APOH, ASAH1, CD63, ECH1, HMGCR, IDH1, NPR1, NPR2, PECAM1, SCD, SYN1, UCP2
SOX4	2.65	0.013	FAM213A, H1FX, LMO7, LYZ, NUAK1, PECAM1, PPIC
GATA1	2.24	0.013	DNTT, DUT, EXOC1, GART, HNRNPAB, LYZ, PECAM1, SRM
MITF	1.50	0.017	ASAH1, GREB1, ITGA4, MET, RAD51D, RHOQ, TACC3, ZFYVE16
SP1	1.66	0.019	APOC3, CRYBB2, CTSD, CYC1, FLNA, HDC, HMGCR, KRT16, MET, MFN2, NPR1, PDHX, SLC25A4, ZEB2
PPARD	1.21	0.022	APOA2, ECH1, LDHB, LPIN2, PDE4C, SCD, UCP2
THRB	−1.73	0.022	APOA2, APOC3, CTSD, ERBB3, GSN, LYZ, MET
SMAD4	1.18	0.030	APOA2, APOC3, CTSD, MET, NPR2, SCD, SLC25A4
TCF7L2	1.00	0.035	CTNNAL1, ERBB3, G2E3, GSN, IDH1, OTUD7B, RAP1A, SYTL2, ZNF536
RB1	1.01	0.036	ATAD2, COA3, MCMBP, MET, MFN2, MRPL9, MYH2, PGRMC1, RRM2, ZEB2
IL3	1.00	0.042	CD63, ECH1, GART, HLA-A, NACA, NIFK, RALA, RPL10, UCP2
POMC	1.07	0.042	KRT16, SLC9A3R1, SULT2A1, USP15

Figure 6.

Ingenuity pathway analysis showing the upstream regulators of differentially expressed proteins in SLC13A5-KD HepG2 cells. A total of 28 upstream regulators are significantly changed in SLC13A5-KD HepG2 cells (p < 0.05). Red and blue bars indicate positive and negative activation z-score, respectively. (A) 26 out of 28 upstream regulators are related to metabolism. (B) 16 out of 28 upstream regulators are related to cell proliferation. (C) 16 out of 28 upstream regulators are related to cellular stress response.

In line with the importance of SLC13A5 in regulating cellular energy metabolism, 26 out of the 28 upstream regulators were metabolism-related, except Cg and CST5 (Figure 6A). The most significant difference was observed in the suppression of MAP4K4 (Table 2). This gene was reported to promote insulin resistance in obesity [30]. Hence, the prediction of MAP4K4 suppression was consistent with the findings showing the protective effect of SLC13A5-KD against insulin resistance [6]. Moreover, 16 upstream regulators were related to the regulation of cell proliferation (Figure 6B), including 12 activated regulators (TP53, HNF4A, ESR1, RORA, LEP, SOX4, GATA1, PPARδ, SMAD4, TCF7L2, RB1, IL3) and 4 repressed regulators (HRAS, MYC, MYCN, PML). Among the 16 regulators, tumor suppressor TP53 was activated and showed the highest statistical difference (Table 2). Activation of TP53 is known to induce apoptosis and cell cycle arrest, decreasing cell proliferation rate [31]. Oncogenes HRAS, MYC and MYCN known to promote cell proliferation were predicted to be repressed [32]. This was consistent with our previous findings, in which knockdown of SLC13A5 in HepG2 cells reduced cell proliferation rate via G0/G1 phase cell cycle arrest [10]. There were also 16 upstream regulators involved in regulating cellular response to stress (Figure 6C), including 10 activated regulators (TP53, HNF4A, ESR1, RORA, PPARα, LEP, SOX4, GATA1, PPARδ, SMAD4) and 6 repressed regulators (MAP4K4, HRAS, MYC, PML, SREBF1, SREBF2). MAP4K4 was found to be overexpressed in hepatocellular carcinoma samples and its down-regulation has been reported to decrease cell proliferation, inhibit cell cycle progression and induce apoptosis in HepG2 cells [33]. This beneficial effect was associated with the inhibition of JNK phosphorylation [33], which is known to sensitize hepatocellular carcinoma cells to apoptosis [34]. Furthermore, dysfunction of TP53 is also an important mechanism conferring resistance of cancer cells to cellular stress, in particular the stress caused by chemotherapeutic agents [35]. Given the strong prediction of MAP4K4 repression and TP53 activation, it is therefore speculated that knockdown of SLC13A5 is capable of sensitizing the cancer cells to cellular stress. Together, the predicted upstream regulators not only agreed with previous findings showing the significant role of SLC13A5 in regulating cellular energy metabolism, but also indicated its involvement in regulating cell proliferation and cellular stress response.

3.6. Increased ketone body levels in SLC13A5 knockdown HepG2 cells

Based on the proteomic analysis, ketone body biosynthetic process was the most significantly upregulated biological process in the SLC13A5-KD HepG2 cells. Hence, we measured the expression levels of genes related to ketogenesis and the levels of ketone bodies in the HepG2 cells and SLC13A5-KD HepG2 cells, as a functional validation of the proteomics data.

In the SLC13A5-KD cells, as shown in Supplementary Table 2, ketone body biosynthetic process showed the greatest significance (p < 0.0001) and the highest enrichment of 32.5-fold, based on the upregulation of five key proteins involved in ketone body biosynthesis: ACAT1, HADHA, HADHB, HMGCL, HMGCLL1. Markedly, the fold changes of these proteins were relatively minor (1.3 – 1.9 fold, p < 0.001), except for HMGCLL1 which showed a change of > 3.5-fold. In our quantitative PCR analysis, mRNA expression of genes encoding these five proteins showed similarly moderate changes after SLC13A5-KD (Figure 7A). Comparatively, the intracellular levels of two key ketone bodies acetoacetate and β-hydroxybutyrate were significantly increased by 3- and 2.5-fold, respectively (Figure 7B), in line with the proteomics data.

Figure 7.

PPARα-mediated increase of ketone body levels in SLC13A5-KD HepG2 cells. HepG2 cells were seeded into 6-well plate at the density of 1 ×10⁵ cells/mL and were infected with control or SLC13A5-shRNA lentivirus. Cells were subject to ketone body analysis 72 hours after infection. (A) Real-time PCR showing the mRNA expression levels of genes related to ketogenesis. *p < 0.05 compared to the shControl group. (B) Increased levels of ketone bodies acetoacetate (AcAc) and β-hydroxybutyrate (BOH) in SLC13A5-KD HepG2 cells. *p < 0.05; ***p < 0.001 compared to the shControl group. (C) Increased levels of AcAc and BOH in SLC13A5-KD HepG2 cells were abolished by GW6471, a PPARα antagonist. *p < 0.05; **p < 0.01 compared to the shSLC13A5 group.

Intriguingly, peroxisome proliferator activated receptor α (PPARα), a known key transcription factor regulating ketogenesis [36], was one of the predicted upstream regulators (Figure 6), suggesting the activation of PPARα in SLC13A5-KD cells. Indeed, the elevation of acetoacetate and β-hydroxybutyrate mediated by SLC13A5-KD in HepG2 cells was abolished in the presence of GW6471, a potent PPARα antagonist (Figure 7C). These results provided additional experimental evidence to support the proteomics results that ketogenesis activity was significantly increased in the SLC13A5-KD HepG2 cells.

3.7. Knockdown of SLC13A5 sensitized HepG2 cells to chemotherapy

Given that 16 of the 28 predicted upstream regulators were associated with the regulation of cellular stress response (Figure 6C), we subsequently compared the responses of HepG2 and the SLC13A5-KD HepG2 cells to chemotherapy-induced cell damages. Four chemotherapeutic agents commonly used in liver cancer treatment, including 5-FU, doxorubicin, cisplatin, and sorafenib, were selected to test the sensitivity of HepG2 cells to chemotherapy following knockdown of SLC13A5. After 48 hours of drug exposure at the indicated concentration ranges, parallel IC₅₀ values in HepG2 vs. SLC13A5-KD HepG2 cells were generated for 5-FU (134.9 vs. 8.57 μM, p < 0.0001), doxorubicin (1.06 vs. 0.16 μM, p < 0.0001), cisplatin (19.95 vs. 8.45 μM, p = 0.0001), and sorafenib (22.39 vs. 10 μM, p < 0.0001) (Figure 8A). Hence, the cytotoxicities of 5-FU, doxorubicin, cisplatin and sorafenib against HepG2 cells were potentiated by SLC13A5-KD by 15.74, 6.63, 2.36 and 2.24 fold, respectively. Since induction of apoptosis is an important mechanism of action underlying these chemotherapeutic agents, the levels of cleaved Caspase-3 in HepG2 cells and SLC13A5-KD HepG2 cells were also compared after treatments. As shown in Figure 8B, 5-FU (25 μM), doxorubicin (0.5 μM), cisplatin (10 μM) and sorafenib (10 μM) treatments led to more cleavage of Caspase-3 in SLC13A5-KD cells, when compared with HepG2 cells. This finding suggests that significantly higher level of apoptosis can be achieved by these chemotherapeutic agents following SLC13A5-KD HepG2 cells, which is consistent with the cytotoxicity results.

Figure 8.

Enhanced cellular response of SLC13A5-KD HepG2 cells to stress caused by chemotherapeutic agents. (A) HepG2 cells were seeded into 96-well plate at the density of 3,000 cells/well and were infected with control or SLC13A5-shRNA lentivirus. After infection for 72 hours, cells were treated with 5-fluorouracil (5-FU), doxorubicin, cisplatin or sorafenib at indicated concentrations for another 48 hours. CCK-8 assay was used to measure the cell viability. (B) Protein expression levels of cleaved Caspase-3 in HepG2 cells and SLC13A5-KD HepG2 cells after treatment with 5-FU (25 μM), doxorubicin (DOX, 0.5 μM), cisplatin (CIS, 10 μM) and sorafenib (SOR, 10 μM) for 48 hours.

4. Discussion

To the best of our knowledge, this is the first report showing the proteome-wide changes that occur after knockdown of SLC13A5 in human cells. HepG2 cells, as mentioned previously, represent an excellent model system for studies on the endogenous di- and tri-carboxylate transporters [37]. Thus, proteomic analysis based on this model cell line is expected to provide valuable information regarding the biological functions of human SLC13A5.

Citrate, the most favorable SLC13A5 substrate, stands at the crossroad of a number of cellular metabolic processes, such as fatty acid synthesis, glucose metabolism, and β-oxidation [38]. In line with the crucial role of citrate in energy metabolism regulation in vivo, our proteomics data showed that the majority of alterations caused by SLC13A5-KD in the HepG2 cells were related to metabolic processes, in particular the upregulation of ketogenesis, fatty acid β-oxidation, and catabolic process. Hepatic ketogenesis was reported to prevent diet-induced fatty liver injury and hyperglycemia [39]. The proteomics data hence suggest that down-regulation of SLC13A5 could exhibit beneficial effects against hepatic lipid accumulation and metabolic disorders, and this is in agreement with the previous findings in SLC13A5 knockout mice and knockdown human primary hepatocytes [6, 7, 11]. Based on pathway analysis, molecular signaling pathways associated with lipid metabolism including LXR/RXR activation, corticotropin-releasing hormone signaling, and protein kinase A signaling were among the most significantly activated pathways in the SLC13A5-KD cells. LXR/RXR is known to play central roles in regulating whole-body cholesterol homeostasis [40]. LXRα knockout mice were found to exhibit marked cholesteryl ester accumulation in the liver when challenged with a cholesterol-rich diet [41]. Activation of LXR/RXR was shown beneficial against inflammation, atherosclerosis, and type II diabetes [42]. Intriguingly, the role of LXR in ketogenesis appears to be controversial. LXRα ablation has been reported to increase the fasting plasma levels of ketone body 3-hydroxybutyrate in mice [43]. In contrast, Archer et al. reported that LXRα knockout mice showed lower serum ketone bodies level when compared with their wild-type littermates [44]. Additionally, protein kinase A signaling and corticotropin releasing hormone signaling have been reported to inhibit lipogenesis and promote lipid metabolism [45, 46]. The activation of these signaling pathways is speculated to contribute to the decreased lipid accumulation in the SLC13A5-KD cells. Importantly, 26 of the 28 predicted upstream regulators were related to metabolism, with MAP4K4 showing the highest statistically significant difference. Such results provided valuable information regarding the molecular mechanisms underlying the functional roles of SLC13A5 in the regulation of metabolic processes. Since HepG2 cells are known to secrete a wide range of proteins involved in various physiological functions [47], quantitative analysis of secreted proteins from HepG2 cells after SLC13A5-KD is expected to provide further insights into the functional roles of this transporter in regulating cellular processes.

Despite the relatively minor expression change of each individual gene associated with ketone body biosynthesis in the quantitative PCR and proteomic analysis, the number of genes involved is relatively large, which collectively led to the elevation of the intracellular levels of two ketone bodies acetoacetate and β-hydroxybutyrate in the SLC13A5-KD cells. This measurement serves as a direct validation of the proteomics data and confirms that ketone body biosynthetic process is one of the most significantly upregulated biological processes after knockdown of SLC13A5 in the HepG2 cells. We speculate that such an increase in ketone body levels was caused by the synergistic effect of relatively mild changes in multiple key ketogenesis-related proteins, including ACAT1, HADHA, HADHB, HMGCL, HMGCLL1, and potentially others. Ketone bodies are predominantly produced in the liver under certain circumstances such as fasting, starvation, low carbohydrate diet, and prolonged exercise. Once produced, they are released from hepatocytes into the circulation to provide an energy source for extrahepatic tissues including brain, heart, and skeletal muscle [48]. Of note, in whole-body Slc13a5 knockout mouse fed with high fat diet, plasma level of β-hydroxybutyrate, the most abundant ketone body in the circulation, was increased by 62% when compared with the wild-type littermates [6]. This is consistent with our in vitro findings in human cells and provides the in vivo experimental evidence to support our proteomics data. The mechanisms by which SLC13A5-KD leads to increased ketone body levels are largely unknown. Since knockdown of SLC13A5 decreases cellular citrate uptake and mimics caloric restriction, upregulation of ketogenesis may represent a metabolic adaptation of the body to such nutrient deprivation. PPARα is a key transcription factor responsible for the induction of many genes associated with ketone body biosynthesis [36]. Based on our proteomics data and ketone body assays, activation of PPARα has most likely played an important role in the increased ketone body levels in HepG2 cells following SLC13A5-KD.

Recently, the role of SLC13A5 in the development and progression of liver cancer has captured heightened attention from scientists. We previously showed that SLC13A5-KD led to G1 phase cell cycle arrest and decreased proliferation in HepG2 cells, associated with reduced cellular uptake of citrate [10]. These findings were consistent with our current proteome profiling results, where DNA replication was significantly down-regulated after knockdown of SLC13A5 in the HepG2 cells and 16 of the 28 predicted upstream regulators (p < 0.01) were involved in regulating cell proliferation. Of the 16 upstream regulators, oncogenes known to promote cell proliferation including HRAS, MYC and MYCN were projected to be repressed, while tumor suppressor genes that inhibit cell proliferation were suggested to be activated, including TP53, SMAD4, and RB1. The changes of such upstream regulators were in agreement with the decreased cell proliferation reported previously. In addition, 16 of the 28 predicted upstream regulators were associated with cellular stress response, suggesting enhanced cellular response to stress in the SLC13A5-KD cells. Indeed, upon knockdown of SLC13A5, HepG2 cells became more sensitive to cellular stress caused by several chemotherapeutic agents including 5-FU, doxorubicin, cisplatin, and sorafenib. Since most of the 16 upstream regulators, in particular the repressed MAP4K4 and upregulated TP53, are involved in the activation of apoptotic signaling pathways, it seems that evasion of apoptosis may represent an important mechanism conferring resistance of HepG2 cells to apoptosis inducers. On the other hand, although alterations in the function of drug-metabolizing enzymes are also important mechanisms of cancer drug resistance, the contribution of such alterations to the development of drug resistance in HepG2 cells appears to be minimal, presumably due to the extremely low levels of expression and activity of drug-metabolizing enzymes in this cell line [49]. Nevertheless, it is speculated that liver cancer patients with dysfunctional SLC13A5 may have a better therapeutic outcome to chemotherapy. Likewise, wound healing, spreading, and migration of cells were among the top down-regulated biological processes after knockdown of SLC13A5 in the HepG2 cells. In this regard, it would be interesting to further evaluate the role of SLC13A5 in regulating invasion and metastasis of liver cancer cells. Collectively, our previous and current experimental and bioinformatic data have demonstrated SLC13A5 as a potential novel therapeutic target in the liver cancer therapy.

Of particular note, analysis of clinical liver samples has shown that expression of SLC13A5 was positively correlated to body mass index, waist circumference, body fat, hepatic insulin resistance index, as well as hepatic steatosis [4]. In line with our previous report [11], our proteomics data also suggests SLC13A5 as a promising therapeutic target in treating metabolic disorders, such as obesity and type 2 diabetes. The decrease of hepatic citrate uptake through inhibition of SLC13A5 is speculated to benefit patients with metabolic diseases and presumably exhibit metabolic benefits [2]. In fact, reduced hepatic lipid accumulation and improved glycemic control caused by SLC13A5 inhibition have been observed in mice fed with high fat diet [50]. Thus, further investigation of potential therapeutic effect of SLC13A5 inhibition and development of its inhibitors are highly desirable. So far, two hydroxysuccinic acids, F-06649298 and its derivative PF-06761281, have been identified as selective inhibitors of SLC13A5 [50, 51]. Both inhibitors were found to block hepatic citrate uptake and reduce plasma glucose level in mice [50, 51]. F-06649298 was also able to reverse glucose intolerance in mice fed with high fat diet [50]. The protective effects of SLC13A5 inhibitors against metabolic disorders depict the beneficial effects of SLC13A5 suppression. Given the diverse biological functions of SLC13A5 in vivo, it would be of great interest to investigate other pharmacological activities of SLC13A5 inhibitors in the future.

In summary, the current study provides a global landscape of proteome-wide changes in human HepG2 cells after knockdown of SLC13A5 and revealed the functional pathways of this transporter in regulating lipid metabolism, ketogenesis, cell proliferation and cellular stress response. Functional analysis further validated the proteomics results and demonstrated the elevated ketone body levels and the enhanced response to chemotherapy in the SLC13A5 knockdown HepG2 cells. In the meantime, we do realize the limitations of this study including the use of a single hepatoma cell line and the lack of more broad and extensive evaluation of the proteomics findings, which warrant future in-depth investigations. Collectively, our results thus far uncover key functional pathways associated with silencing SLC13A5 in HepG2 cells and suggest that SLC13A5 may represent a potential therapeutic target in the management of metabolic disorders and liver cancer.

Highlights:

• SLC13A5 silencing results in globally altered proteomic profile in HepG2 cells

• Metabolism, cell proliferation and stress response are major pathways predicted

• Ketogenesis is the most significantly upregulated process by SLC13A5 knockdown

• SLC13A5 silencing sensitizes toxic response to chemotherapeutics in HepG2 cells

Abbreviations:

AcAc

acetoacetate

ACAT1

Acetyl-CoA acetyltransferase 1

BOH

β-hydroxybutyrate

ESR1

estrogen receptor 1

ESR2

estrogen receptor 2

FGF

fibroblast growth factor

GO

gene ontology

HADH

hydroxyacyl-CoA dehydrogenase trifunctional multienzyme complex

HMG-CoA

hydroxymethylglutaryl-CoA

HMGCL

HMG-CoA lyase

HNF4A

hepatocyte nuclear factor 4 alpha

IPA

ingenuity pathway analysis

KD

knockdown

LXR

liver X receptor

MAP4K4

mitogen-activated protein kinase kinase kinase kinase 4

MITF

melanocyte inducing transcription factor

NGF

nerve growth factor

PPARα

peroxisome proliferator activated receptor alpha

SLC13A5

solute carrier family 13 member 5

SREBF

sterol regulatory element binding transcription factor

TCF7L2

transcription factor 7 like 2

THRB

thyroid hormone receptor beta