The two previously described HLCS knockdown MDA-MB-231 cell lines, KD 868 and KD 1950, and scrambled control cell lines were used in this study (Siritutsoontorn et al., 2022). They were cultured in Minimal Essential Medium (MEM) supplemented with 10% (v/v) fetal bovine serum (FBS) and 1% (v/v) penicillin/streptomycin. Cells were cultured at 37°C with 5% CO₂ and routinely sub-cultured every 3 days.

Five hundred cells of HLCS knockdown and scrambled control cell lines were seeded into a 100-mm culture dish containing MEM supplemented with 10% (v/v) fetal bovine serum and maintained at 37°C with 5% CO₂. Following 10 days, the culture media were removed, and the colonies were washed with phosphate buffer saline (PBS). Cells were fixed with 100% (v/v) methanol for 20 min at room temperature and gently washed with water. The colonies were stained with 0.5% (w/v) crystal violet in 25% (v/v) methanol for 45 min at room temperature in the dark, rinsed with water, air-dried overnight, and counted.

Approximately 1 × 10⁶ cells/mL were suspended in MEM media supplemented with 0.1% (v/v) BSA and incubated with 1 μg/mL of fluorogenic calcein-AM (Invitrogen) for 30 min at room temperature in the dark. Cells were then centrifuged at 1,500 rpm for 5 min, washed three times with MEM supplemented with 0.1% (v/v) BSA, and resuspended at the density of 8 × 10⁵ cells/mL with MEM containing 0.1% (v/v) BSA. Afterwards, 4 × 10⁴ cells were suspended in 49 mL of MEM before placing onto the filter (12-µm pore size) of 96-well ChemoTx^® chemotaxis system (Neuro Probe), which was pre-coated with Geltrex (Life Technologies). Cells were allowed to migrate from the filter to the bottom compartment containing 29 µL of MEM supplemented with 10% (v/v) FBS at 37°C for 6 h. The non-migrated cells on the top of the filter were gently removed with a paper towel before the fluorescence intensity generated from the calcein-labeled cells in the bottom compartment was measured with excitation and emission wavelength at 485 and 520 nm, respectively, in a Triad series multimode detector (Dynex Technologies).

Whole-cell lysates of HLCS knockdown and scramble MDA-MB-231 cells were extracted in 100 μL of 1x radio-immunoprecipitation assay (RIPA) buffer containing 50 mM Tris-HCl pH 7.4, 1% (v/v) NP-40, 0.25% (w/v) sodium deoxycholate, 150 mM NaCl, 1 mM EDTA, and 1x protease inhibitor cocktail (Roche) on ice for 30 min before centrifugation at 12,000 × g for 15 min at 4°C. Protein concentrations were determined using Bradford regent (Bio-Rad). Protein (30 μg) was separated on a 7.5% discontinuous SDS-PAGE mini gel using 1x glycine buffer (25 mM Tris-HCl pH 8.3, 193 mM glycine, 0.1% (w/v) SDS) under reducing conditions. The protein was transferred from polyacrylamide gel to polyvinylidene difluoride membranes by Trans-Blot®Turbo™ Transfer system (Bio-Rad) with 1x transfer buffer (25 mM Tris-HCL pH 8.3, 192 mM glycine, 0.1% (w/v) SDS and 20% (v/v) methanol) at constant 25 V for 20 min. Then, the membranes were blocked in a blocking buffer containing 5% (w/v) skim milk and 1% (v/v) Tween 20 in 1x PBS pH 7.4 at 4°C overnight. The blots were then incubated with appropriate primary antibodies: 5 µg of rabbit anti-HLCS antibody (Bailey et al., 2010), 1:1,000 dilution of rabbit anti-MMP1 polyclonal antibody (E9S9N) (Cell Signaling Technologies), or 1:5,000 dilution of rabbit anti-SerpinB2 polyclonal antibody (ab47742, Abcam) at room temperature for 2 h. Excess antibodies were washed four times with 1x PBS buffer containing 1% Tween 20 before incubating with 1:20,000 dilution of mouse anti-rabbit IgG conjugated with horseradish peroxidase (Dako) at room temperature for 1 h. The immunoreactive bands representing each protein were detected using a chemiluminescence substrate (Merck, Millipore). Following MMP1 detection, the blot was incubated with stripping buffer (20 mM glycine, 0.1% SDS and 1% Tween 20, pH 2.2) and blocked in a blocking buffer at room temperature for 1 h. The membrane was incubated with 1:1,000 dilution of rabbit anti-ASS1 polyclonal antibody (D4O4B) (Cell Signaling Technologies) at room temperature for 2 h. Then, the membrane was incubated with a secondary antibody and developed as above.

Raw data was processed with Proteome Discoverer v2.4 (Thermo Scientific). Searches were performed against the human FASTA database using the SEQUEST HT search engine, with the precursor and fragment mass tolerances set to 10 ppm and 0.02 Da, respectively. Two missed cleavage sites were allowed, with the minimum peptide length specified at 6 amino acids. Oxidation (+15.995), deamidation (+0.984), N-terminal acetylation (+42.011), N-terminal methionine loss (−131.040) and N-terminal methionine loss and acetylation (−89.030) were included as variable modifications, and cysteine carbamidomethylation was included as a fixed modification. To understand the differentially expressed proteins (DEPs) involvement in various biological processes, molecular functions, and cellular components, a list of all DEPs was subjected to gene annotation analysis using PANTHER database analysis tool version 16.0 (http://www.pantherdb.org/) (Thomas et al., 2003).

All experiment results are expressed as means ± S.D. The statistical analysis was performed using GraphPad Prism software (version 8.4.0; GraphPad Software, Inc., La Jolla, CA). Significant differences between the sample groups were calculated by one-way factorial analysis of variance (ANOVA), followed by Dunnett’s test. A p-value of <0.05 indicated statistically significant.

Two previously generated independent HLCS knockdown MDA-MB-231 cell lines (KD 868 and KD 1950) were used for proteomic analysis. These two HLCS knockdown cell lines possessed approximately 70%–80% reduction of HLCS protein corresponding to 86, 82, and 72 kDa isoforms (Bailey et al., 2010) (Figure 1A) compared to the scrambled control cell line. The decreased HLCS expression in these two knockdown clones was accompanied by 40%–60% reduction of clonogenic growth (Figure 1B). To gain further insight into how suppression of HLCS expression impacted global cellular changes, proteomic analysis was performed on the knockdown cells using LC-MS. The proteomic study was performed on these two HLCS knockdown cell lines to eliminate the clonal bias compared to the scrambled control cell line. All LC-MS data were obtained from three biological replicates, each with two technical replicates. According to the 1% false discovery rate (FDR) criteria with at least one unique peptide (Zhang et al., 2018; Yang et al., 2019), 5,306 proteins were identified across samples. Figure 2A shows the heat map of expression profiles and hierarchical clustering analysis of 5,306 proteins identified in all samples. Principle component analysis of all samples revealed a distinct cluster of scrambled control and HLCS knockdown clones KD 868 and KD 1950, respectively. The greatest component of variation of the data (PC1, 22.4%) showed a clear separation between the scrambled control and the knockdown groups, whereas the distinct cluster-based PC2 (16.4%) separation indicated a variable biological response between HLCS knockdown clones, KD 868 and KD 1950 (Figure 2B). Figure 2C shows the volcano plots of differentially expressed proteins identified between control and each HLCS knockdown cell lines [control vs. KD 868 (left panel) and control vs. KD 1950 (right panel)], separately. With the cut-off threshold of minimal 2-fold change with p-value <0.05, 196 proteins were considered differentially expressed. Of this number, 95 proteins were downregulated (spots in green box), while 101 proteins were upregulated (spots in red box). For the KD 1950 cell line, 151 proteins were differentially expressed, with 59 proteins being downregulated (spots in green box) and 92 proteins being upregulated (spots in red box).

To gain further insights into the biological response in HLCS knockdown, the functional annotation of the DEPs in each HLCS knockdown cell line was performed using the GO annotation tool. Based on the biological processes classification, the top three biological response changes associated with the downregulation of DEPs in the KD 868 clone were cellular process (GO:0009987, 30%), metabolic process (GO:0008125, 24%), and biological regulation (GO:00650007, 15%), with the other individual processes accounting for less than 10% (Figure 3A, left panel). Likewise, most changes associated with the downregulation of DEPs in the KD 1950 clone were cellular process (32%), metabolic process (24%), and biological regulation (16%), with the other processes listed similar to those of the KD 868 clone (Figure 3A, right panel). Regarding the upregulated proteins, the three most biological changes associated with the KD 868 clone were cellular process (GO: 0009987, 29%), metabolic process (GO: 0008152, 20%), and biological regulation (GO: 0065007, 16%). For the KD 1950 clone, the upregulated proteins were associated with cellular process (27%), metabolic process (18%), and biological regulation (16%).

According to the molecular function classification, the downregulated proteins in both HLCS knockdown clones were mostly involved in catalytic activity (GO:0003824; 65% in KD 868 (Figure 3B, left panel), and 42% in KD 1950 (Figure 3B, right panel), followed by binding activity (GO:0005488, 22% in KD 868, and 36% in KD 1950), and molecular function regulator (GO:0098772, 9% in KD 868, and 19% in KD 1950). However, the molecule transducer activity and structural molecule activity were molecular functions that exclusively related to the downregulated proteins in KD 868 and KD 1950, respectively. In contrast, the upregulated proteins in both knockdown clones were predominantly related to the binding activity (GO:0005488; 41% in KD 868, and 36% in KD 1950), followed by catalytic activity (GO:0003824; 33% in KD 868, and 28% in KD 1950), and molecular function regulator (GO:0098772; 15% in KD 868, and 23% in KD 1950).

To see whether the affected pathway changes in both HLCS knockdown cell lines were mediated through common proteins, the Venn diagram of DEPs of both knockdown clones was generated (Figure 3C). Both HLCS knockdown clones shared 64 common DEPs, with 23 being downregulated and 41 being upregulated. The list of these upregulated and downregulated and common proteins and their associated biological processes are shown in Table 1 and Table 2, respectively. According to the candidate proteins shown in Table 1, suppression of HLCS expression appears to be associated with increased programmed cell death, as shown by the increased expression of proteins involved in apoptosis [Apoptotic protease-activating factor 1 (APAF1)], immunogenic cell death [caspase-1 (CASP1)], autophagy [stimulator of interferon genes protein, (STING1)] and increased expression of proteins that possess tumor suppressor function such as tetraspanin-6 (TSPAN6), mitogen-activated protein kinase 6 (MAPK6), protein sprouty homolog 2 (SPRY2), HAUS augmin-like complex subunit 4 (HAUS4). The other common DEPs include those involved in metabolic processes, transcriptional regulation, cellular transport. Regarding the downregulated DEPs (Table 2), these proteins were associated with cell motility, such as myosin ID (MYO1D), leupaxin (LPXN), Interleukin-1β (ΙL-1β), plastin-2 (LCP1), transglutaminase 2 (TGM2), metallothionein 1X (MT1X), small p97/VCP-interacting protein (SVIP), and peroxiredoxin-like 2A (PRXL2A). The other affected biological processes include metabolic processes, cellular organization and biogenesis, and DNA-binding proteins.

To validate some of the candidate proteins shown in both Tables 1, 2, some candidates from both upregulated proteins and downregulated which showed the most significant fold changes, including argininosuccinate synthase 1 (ASS1) (log₂ + 3.3 and + 2.8-fold in KD 868 and KD 1950 cells, respectively), plasminogen activator inhibitor 2 (SerpinB2) (−3.0 and −2.2-fold), interstitial collagenase/matrix metalloproteinase-1 (MMP1) (−2.2 and −1.5-fold) were validated by Western blot analysis. ASS1, one of the urea cycle enzymes, catalyzes the condensation of citrulline and L-aspartate to form argininosuccinate, and this enzyme is downregulated in many types of cancer (Hajaj et al., 2021). SerpinB2 has been shown to regulate stroma remodeling and metastasis (Harris et al., 2017), and MMP1 is one of the metalloproteases involved in cell invasion (Liu et al., 2012). As shown in Figure 4A, ASS1 expression level was extremely low in the scrambled control cell line, while it was markedly increased in both KD 868 and KD 1950 cell lines (p = 0.005 and p = 0.004, respectively). SerpinB2 expression was reduced by 95% in the KD 868 and was reduced by 75% in KD 1950 (p < 0.001) (Figure 4B). Similar to SerpinB2, MMP1 expression level was barely detectable in KD 868 and decreased by approximately 94% (p = 0.00001) in KD 1950 (Figure 4C). The marked reduction of MMP1 was accompanied by 40% decrease of invasion in both HLCS knockdown clones (Figure 4D).

GO analysis indicates that HLCS knockdown mainly affects cell death and other biological processes associated with cell motility and metabolic pathways. Regarding cell death, several proteins implicated in cell death pathways were increased—for example, the APAF1, CASP1, and STING1 (Table 1). APAF1 is a core component of the apoptosome, essential for caspase-9 activation (Pop et al., 2006). Increased APAF1 expression is well-described in many types of cancer exposed to chemotherapeutic drugs (Yuan et al., 2019; Xiang et al., 2020; Bakhshoudeh et al., 2021). CASP1 is a pro-inflammatory cytokine that cleaves pro-ΙL-1β, and pro-IL-18 to their active forms. This activity is essential for pyroptosis, a cell death process triggered by inflammation. A recent study shows that the chemotherapeutic drug, cisplatin, induces CASP1-mediated pyroptosis in MDA-MB-231 cells (Yan et al., 2021). CASP1-deficient mice showed enhanced colon epithelial and tumor proliferation, indicating its role as a tumor-suppressor protein (Hu et al., 2010). STING1, an ER-bound protein, is crucial for autophagy and programmed cell death (Zhang et al., 2021a). A recent study shows that Eribulin, an anti-cancer drug used to treat breast cancer and liposarcoma, triggers cancer cell death via the STING1-dependent signaling axis (Fermaintt et al., 2021; Takahashi-Ruiz et al., 2022), indicating its role as a cell-death-promoting factor. Upregulation of these pro-apoptotic proteins may be induced in response to HLCS suppression.

The reduced clonogenic growth of MDA-MB-231 herein may be attributed to the upregulation of some tumor suppressor proteins such as TSPAN6, MAPK6, SPRY2, HAUS4, stomatin (STOM) and conserved oligomeric golgi complex subunit 8 (COG8) (Table 1). TSPAN6, a tetraspanin family member, is a tumor suppressor protein that interferes with TGF-α signaling in colorectal cancer (Andrijes et al., 2021). Similarly, MAPK6 or ERK3, an atypical MAPK member, controls cell growth (Chen et al., 2000). A recent study shows that MAPK6/ERK3 exerts its tumor suppressor function by inhibiting cell cycle progression (Julien et al., 2003) and proliferation and invasion of melanoma (Chen et al., 2019a). SPRY2, a receptor tyrosine kinase inhibitor, has been shown to interfere with oncogenic signaling by antagonizing the ERK signaling pathway (Fong et al., 2006). Overexpression of SPRY2 in several types of cancer inhibits their growth and progression (Kawazoe and Taniguchi, 2019). HAUS4 is part of the HAUS augmin-like complex, which contributes to mitotic spindle assembly, maintenance of chromosome integrity, and completion of cytokinesis. Avelar et al., 2020 showed that elevated HAUS4 expression is associated with senescence and a tumor-suppressor gene. Stomatin, an integral membrane, regulates ion channels and transporter (Rahman et al., 2021). A recent study showed that stomatin suppresses proliferation and induces apoptosis by inhibiting Akt signaling in prostate cancer (Rahman et al., 2021; Sato et al., 2021). The decreased stomatin expression was associated with poor prognosis of prostate, breast, and non-small cell lung cancer patients, indicating its role as a tumor suppressor protein (Chen et al., 2016; An et al., 2019; Rahman et al., 2021). COG8 is the structural component of golgi-apparatus that functions in intracellular transport and glycosylation. COG8 level is negatively correlated with the survival of renal clear carcinoma patients (Zhang et al., 2021b).

Extracellular matrix detachment is a fundamental process for cancer cell migration. This process involves metalloproteinase secretion and other proteases such as urokinase plasminogen activator. MMP1 is a member of matrix metalloproteinases, which degrades extracellular matrix during an invasion, while SerpinB2 regulates stromal remodeling, local invasion and metastasis (Valiente et al., 2014; Harris et al., 2017). Overexpression of MMP1 is associated with tumor invasion and metastasis (Chambers and Matrisian, 1997) and is highly expressed in triple-negative breast cancer (Wang et al., 2019). The decreased abundances of MMP1 and SerpinB2 in HLCS knockdown cells are consistent with impaired cell invasion and migration (Siritutsoontorn et al., 2022). The reduced cell invasion of HLCS deficient cells may also be associated with the decreased expression of cytoskeletal proteins, including SerpinB2, LPXN, MYO1D, LCP1, or their regulators, such as ΙL1B and TGM2 (Table 2). LPXN, a focal adhesion adaptor protein, is essential for cell adhesion and migration (Alpha et al., 2020). Recent studies show that LPXN is overexpressed in many cancers (Kaulfuss et al., 2008; Kaulfuss et al., 2015; Hou et al., 2018). Depletion of LPXN expression reduces prostate cell migration and invasion, while its overexpression promotes prostate cancer progression (Kaulfuss et al., 2009). MYO1D, a member of the myosin motor protein, controls cell movement (Diaz-Valencia et al., 2022) and promotes the growth and invasion of cancer cells (Amcheslavsky et al., 2018). MYO1D interacts with EGFR and contributes to oncogenesis (Ko et al., 2019; Mu et al., 2022). ΙL-1Β plays various roles in physiological processes, including inflammation, cell adhesion, migration, and angiogenesis (Rebe and Ghiringhelli, 2020). Although ΙL-1B is primarily produced from immune cells, it can be synthesized and secreted from several types of solid tumors (Jin et al., 1997; Wu et al., 2016), where it stimulates the expression of IL-6 and COX-2, which are promoting factor for breast cancer aggressiveness (Reed et al., 2009; Oh et al., 2016). Plastin-2 (LCP1), an actin-binding protein, regulates cytoskeleton movement. Plastin-2 is overexpressed in most types of cancer (Park et al., 1994; Tiedemann et al., 2019), supporting invasion and metastasis (Foran et al., 2006). TGM2 catalyzes the posttranslational modification of glutamine residue-bound proteins. TGM2 also plays a non-enzymatic function by interacting with fibronectin and integrin to stabilize ECM structure (Condello et al., 2022). TGM2 is overexpressed in many types of cancer (Agnihotri and Mehta, 2017), promoting proliferation, migration, and invasion (Wang et al., 2016).

In addition, HLCS suppression also downregulates some proteins involved in intracellular transport and stress response, i.e., copper metabolism domain containing 1 (COMMD1), SVIP, and PRXL2A. COMMD1 is a copper-binding protein with additional non-copper-related function (Weiskirchen and Penning, 2021). COMMD1 was elevated in lymphoma, and its level was correlated with poor prognosis (Taskinen et al., 2014). SVIP involves the ER-associated protein degradation of misfolded proteins. In cancer cells, SVIP was overexpressed in prostate cancer tissue, while its inhibition reduces migration and malignant transformation (Erzurumlu and Ballar, 2017). PRXL2A is a member of the peroxiredoxin antioxidant protein and functions in removing reactive oxygen species. PRXL2A is an essential protein in cancer stem cells that modulates redox status and maintains stemness properties. PRXL2A was overexpressed in oral squamous cell carcinoma and renal clear cell carcinoma, where high PRXL2A was associated with poor prognosis (Chen et al., 2019b; Ren et al., 2021). MT1X, a member of the metallothionein family, functions as a copper/zinc binding protein to maintain metal homeostasis and control oxidative stress and DNA damage (Krizkova et al., 2018). MT1X supports proliferation while inhibiting apoptosis and p53 expression in several types of cancer (Liu et al., 2018). MT1X was overexpressed in many cancers, including breast cancer (Si and Lang, 2018), and was a prognostic marker for invasive ductal carcinoma (Zhang et al., 2000) and renal carcinoma (Ding et al., 2022).

As HLCS knockdown cells showed marked reduction of invasion accompanied by reduced MMP1 expression, we checked whether these defective phenotypes were attributed to impaired epithelial-mesenchymal transition program and stemness of cancer cells. Western blot analysis of vimentin and Oct4 showed that the levels of these two proteins were unchanged, suggesting the impaired clonogenic growth and invasion were not due to the loss of EMT and stemness (Supplementary Figure S1).

In addition to the changes in cellular processes, HLCS knockdown also affects metabolic pathways related to both glycolysis and mitochondrial metabolism. One of the most apparent changes is the increased expression of ASS1, suggesting that the urea cycle may be compromised in response to HLCS knockdown. The urea cycle is not only a cycle that disposes ammonium from amino acid catabolism, but it is also linked to the TCA cycle through aspartate. Transamination of oxaloacetate produces L-aspartate, which enters the urea cycle, where it condenses with L-citrulline to form argininosuccinate mediated by ASS1. Because L-aspartate is an essential nitrogen donor for purine biosynthesis, modulation of ASS1 expression would determine the cellular aspartate pool, which supports nucleotide synthesis during proliferation. ASS1 also plays a non-metabolic role by suppressing Akt signaling and attenuating cancer growth (Miyamoto et al., 2017). Growing evidence indicates that many cancers inactivate urea cycle enzymes, including ASS1 (Rabinovich et al., 2015; Lee et al., 2018b), so the L-aspartate is available for nucleotide synthesis. Ectopic expression of the ASS1 gene in cancer cells inhibits their growth, indicating its role as a tumor suppressor enzyme (Khare et al., 2021; Kim et al., 2021). The marked increase of ASS1 expression in HLCS knockdown cells suggests that urea cycle activity may be increased as previously reported in hepatocytes of liver-specific PC knockout mice (Cappel et al., 2019). The increased ASS1 expression can also affect immediate downstream reactions in the urea cycle, i.e., the conversion of argininosuccinate to arginine catalyzed by arginosuccinate lyase. Because arginine is a precursor of nitric oxide, a chief mediator of inflammatory cytotoxicity, elevated arginine levels due to increased ASS1 would enhance NO production (Kim et al., 2021). Interestingly, PC KO pancreatic cells showed elevated urea cycle activity accompanied by increased arginine and NO production (Fu et al., 2020). The increased NO, in turn, promotes oxidative stress in PC KO pancreatic beta cells, contributing to cell death (Fu et al., 2020). Because HLCS regulates PC activity through biotinylation, it is not unexpected to see a similar perturbation of urea cycle activity in both HLCS and PC-deficient cells. However, it is still unclear how suppression of HLCS expression induces ASS1 expression. Although the expression of ASS1 was markedly increased in MDA-MB-231 cells, we did not see the same response in HLCS knockdown MCF-7 cells (Supplementary Figure S2). However, this outcome was anticipated as the basal level of ASS1 expression in MCF-7 cell line was already high. The lack of ASS1 change in HLCS knockdown MCF-7 cells may recapitulate the early stage of cancer, where ASS1 may not be essential during early stage of tumor progression. This is consistent with previous studies demonstrating that not all breast cancer subtypes inactivate ASS1 expression to enhance their oncogenic property (Qiu et al., 2014; Zou et al., 2021). Qiu et al. (2014) showed that highly invasive MDA-MB-231 cells regulate their oncogenic property by down-regulating ASS1 expression, but this is not the case for low invasive cell lines, i.e., MCF-7 and T47D, which possess much higher levels of endogenous ASS1. This indicates that the low and high invasive breast cancer cell lines regulate their oncogenic potentials via ASS1 differently (Qui et al., 2014). Qui et al. also showed that breast cancer patients with stage 0-I possess higher ASS1 expression, while those with advanced stage (III and IV) possess considerably lower ASS1 levels, suggesting the inactivation of ASS1 expression may occur during the transition from low to advanced stages of breast cancer.

In addition to the urea cycle, HLCS knockdown cells showed increased pyruvate kinase M1 (PKM1) isoform expression. While PKM1 is a highly active isoform, most cancers expressed PKM2 rather than PKM1 to promote Warburg’s effect. The increased PKM1 level in HLCS knockdown could attenuate cancer growth, as reported in prostate adenocarcinoma, where inhibiting receptor tyrosine kinase signaling induces the expression of PKM1 and attenuates the growth of glioblastoma (Yang et al., 2022). The tumor suppressor role of PKM1 was also reported in breast cancer (Choksi et al., 2021) and prostate adenocarcinoma (Davidson et al., 2022).

HLCS knockdown also negatively affects lipid metabolism by altering expression of adipocyte triglyceride lipase (ATGL) (Table 2) and apolipoprotein B-100 (APOB) (Table 1). ATGL catalyzes the hydrolysis of triglycerides to free fatty acid and diacylglycerol. ATGL was overexpressed in several types of cancer, suggesting that fatty acid catabolism is essential to support the growth of tumors (Iftikhar et al., 2021; Yin et al., 2021; Zhang et al., 2022). APOB, a component of low-density lipoprotein, functions in triglyceride transport. APOB has recently been reported to play a role as tumor suppressor protein. The loss of APOB expression in several types of cancers is also associated with the upregulation of oncogenes and poor prognosis (Lee et al., 2018a; He et al., 2022). Mechanistically, overexpression of APOB in MDA-MB-231 cells inhibits their growth and invasion by depleting the lipid supply to the breast cancer cells (Ben Hassen et al., 2020).

The reduced growth of HLCS knockdown cells may result from the perturbation of mitochondrial oxidative phosphorylation, causing mitochondrial stress and apoptosis via downregulation of SURF1-like protein (SURF1), persulfide dioxygenase (ETHE1), and NADH: ubiquinone oxidoreductase subunit B9 (NDUFB9) (Tables 1, 2). SURF1 encodes an assembly factor of cytochrome oxidase of respiratory complex IV. The reduced expression of SURF1 can potentially impair cytochrome C oxidase activity, accelerating mitochondrial apoptosis in response to oxidative stress (Pequignot et al., 2001; Schull et al., 2015). ETHE1, a mitochondrial enzyme, detoxifies hydrogen persulfide, a harmful agent for cytochrome C oxidase. A recent study showed that ETHE1 is overexpressed in colorectal cancer, promoting aerobic glycolysis and mitochondria biogenesis (Witherspoon et al., 2019). Depleting ETHE1 expression induces oxidative stress, which aligns with the increased apoptosis of HLCS knockdown cells. NDUFB9 is a component of NADH dehydrogenase in the respiratory chain complex I. NDUFB9 was downregulated in highly metastatic breast cancer and silencing its expression in MDA-MB-231 cells promotes proliferation, migration, and invasion (Li et al., 2015). This inhibitory effect of NDUFB9 on tumor growth may be attributed to enhanced aerobic glycolysis in cancer cells (Li et al., 2015). The upregulation of NDUFB9 expression may be partly responsible for restricting MDA-MB-231 cell growth in response to HLCS suppression. The ganglioside GM2 activator (GM2A) activator protein is a specific glycolipid transport that binds to ganglioside GM2 and facilitates degradation. GM2A is associated with a programmed-cell death via interaction with tumor-necrosis receptor 1 receptor (Mahata et al., 2015). The increased abundance of GM2 activator may be attributed to the increased apoptosis in HLCS knockdown cells.

Kynureninase (KYNU) is an enzyme that catalyzes the reversible conversion of kynurenine to anthranilate and L-alanine. Kynurenine serves as the immunomodulant that protects cells against strong immune reactions. A recent study indicates that tumor cells hijack the kynurenine pathway to promote their growth and protect them from immune surveillance (Venkateswaran and Conacci-Sorrell, 2020). Kynurenine supports tumor growth by acting as the ligand for the aryl hydrocarbon receptor-transcription factor that activates oncogenic growth. Regarding immunomodulation, kynurenine suppresses T-lymphocyte proliferation by modulating their metabolism (Siska et al., 2021). Kynurenin was overexpressed in ductal and renal carcinoma, which confers ferroptosis-cell death (Liu et al., 2023).

Figure 5 summarizes the biological changes in response to HLCS knockdown. As previously reported, these combined effects likely contribute to growth retardation, migration, invasion, cell cycle arrest, and apoptotic induction.