Non‐canonical phosphoglycerate dehydrogenase activity promotes liver cancer growth via mitochondrial translation and respiratory metabolism

Abstract

Phosphoglycerate dehydrogenase (PHGDH) is a key serine biosynthesis enzyme whose aberrant expression promotes various types of tumors. Recently, PHGDH has been found to have some non‐canonical functions beyond serine biosynthesis, but its specific mechanisms in tumorigenesis remain unclear. Here, we show that PHGDH localizes to the inner mitochondrial membrane and promotes the translation of mitochondrial DNA (mtDNA)‐encoded proteins in liver cancer cells. Mechanistically, we demonstrate that mitochondrial PHGDH directly interacts with adenine nucleotide translocase 2 (ANT2) and then recruits mitochondrial elongation factor G2 (mtEFG2) to promote mitochondrial ribosome recycling efficiency, thereby promoting mtDNA‐encoded protein expression and subsequent mitochondrial respiration. Moreover, we show that treatment with a mitochondrial translation inhibitor or depletion of mtEFG2 diminishes PHGDH‐mediated tumor growth. Collectively, our findings uncover a previously unappreciated function of PHGDH in tumorigenesis acting via promotion of mitochondrial translation and bioenergetics.

A novel non‐catalytic function of serine biosynthesis enzyme PHGDH spurs cancer cell protein translation for enhanced bioenergetics.

Introduction

Metabolic reprogramming is a widely recognized hallmark of cancer cells (Ward & Thompson, 2012; Pavlova & Thompson, 2016). Cancer cells alter their metabolic properties and preferences to meet the increased bioenergetic and biosynthetic demands of proliferation (Faubert et al, 2020; Dey et al, 2021). It is known that metabolic enzymes regulated by key signaling pathways in cancer cells can fulfill cellular metabolism and growth requirements by performing classic metabolic functions. However, an increasing number of studies have found that many metabolic enzymes can also meet the needs of rapid cancer cell growth through non‐canonical or nonmetabolic functions (Xu et al, 2021).

Phosphoglycerate dehydrogenase (PHGDH), the first rate‐limiting enzyme of the glucose‐derived de novo serine synthesis pathway (SSP), is a major determinant in various human cancers and drug resistance models (Wei et al, 2019; Muthusamy et al, 2020; Tajan et al, 2021). PHGDH promotes tumor growth by providing serine, glycine, glutathione (GSH), and one‐carbon units (Locasale et al, 2011; Sun et al, 2015), and several specific inhibitors targeting PHGDH have been shown to have antitumor activity in vivo when used alone or in combination with dietary restriction of serine and glycine (Pacold et al, 2016; Wang et al, 2017; Ngo et al, 2020). A previous study showed that exogenous serine supplementation does not restore cell growth inhibition caused by PHGDH knockdown (Possemato et al, 2011), suggesting that PHGDH may play an important role in tumor development through nonmetabolic functions. Recently, Ma et al (2021) found that the nuclear translocation of PHGDH contributes to tumor growth by reducing nuclear NAD⁺ levels and the activity of the transcription factor c‐Jun under nutrient stress conditions. These observations suggest specific cellular and molecular mechanisms by which different subcellular localizations of PHGDH contribute to tumor progression. However, to date, we have been unable to visualize the panoramic distribution of PHGDH in cells.

Mitochondria perform a central function in cellular energy metabolism to meet the bioenergetic demands of cells by producing the majority of cellular ATP through oxidative phosphorylation (OXPHOS) (Fulda et al, 2010; Murphy & Hartley, 2018). Considered ancient bacterial symbionts, mitochondria have their own DNA (mitochondrial DNA, mtDNA) and maintained an internal system for translation and protein synthesis during evolution (Roger et al, 2017). In humans, mtDNA encodes 13 transmembrane proteins, all of which are core subunits of the respiratory chain complex, comprising complexes I, III, IV, and V (Kummer & Ban, 2021). Another 1,000 mitochondrial proteins are nuclear‐encoded; their synthesis is controlled by cytosolic ribosomes, and they are then imported into mitochondria (Dennerlein et al, 2017). Since mtDNA specifically encodes transmembrane proteins, mitochondrial translation tends to proceed in a spatially close to the mitochondrial inner membrane so that newly synthesized proteins can be inserted into the inner membrane and integrated into the OXPHOS complex, thus improving the efficiency of mitochondrial translation and respiration (Englmeier et al, 2017; Lee et al, 2020). Mitochondrial respiratory chain protein synthesis is regulated by systems at two distinct genetic origins, the nuclear and mitochondrial translation systems, among which nuclear‐regulated mitochondrial protein synthesis is well understood, but the mechanism and regulation mode of mitochondrial translation are still unclear. The mitochondrial translation process comprises four phases: initiation, elongation, termination, and recycling. To start a new round of the translation cycle, mitochondrial ribosome recycling factor (mtRRF) dissociates ribosomes into individual subunits with the assistance of mitochondrial translation elongation factor G (mtEFG2) so that the ribosome can reenter the translation cycle (Ott et al, 2016). Therefore, operational mitochondrial translation is required for effective respiration and oxidative ATP production. The factors regulating mitochondrial ribosome recycling have been verified by in vitro assays on purified mitochondria from mammalian liver tissues or yeast (Pfeffer et al, 2015; Kummer et al, 2021). However, considering the complexity of mitochondrial translation, relatively little is known about how in vivo recycling is regulated and whether other proteins are involved in the regulatory patterns of mtRRF and mtEFG2.

Our previous study showed that SSP activated by cMyc is critical for cancer cell survival under nutrient deprivation conditions (Sun et al, 2015). Here, we extended this exploration and found that mitochondrial inner membrane localization of PHGDH is also important for tumorigenesis by enhancing mitochondrial translation efficiency. We revealed that the PHGDH/ANT2/mtEFG2 complex enhances mitochondrial translation by increasing the efficiency of mitochondrial ribosome recycling, which in turn promotes the expression of 13 mtDNA‐encoded peptides, increases mitochondrial respiratory capacity, and enhances cancer cell proliferation. Taken together, our findings show a previously unappreciated role for PHGDH in facilitating cancer progression by boosting mitochondrial translation efficiency.

Results

PHGDH localizes to the mitochondrial inner membrane in liver cancer cells

Metabolic enzymes usually mediate tumorigenesis by performing their primary metabolic functions. However, recent studies have shown that several metabolic enzymes acquire nonmetabolic functions when their subcellular localization is altered (Yang et al, 2011; Li et al, 2016). To determine whether PHGDH has subcellular compartment‐dependent functions other than serine biosynthesis, we first examined the subcellular localization of PHGDH by subcellular fractionation. Intriguingly, our results showed that endogenous PHGDH is obviously enriched in mitochondrial components in human liver cancer cells, including PLC and HepG2 cells, but rarely found in human normal liver THLE3 cells (Fig 1A). Similar results were observed in human lung cancer A549 cells and MCF7 and MDA‐MB‐468 human breast cancer cells (Fig EV1). We then examined the intracellular distribution of PHGDH in clinical hepatocellular carcinoma (HCC) tissues and found that endogenous PHGDH is distributed in mitochondria, and significant accumulation of mitochondrial PHGDH in tumor tissues compared to adjacent non‐tumor tissues was further observed (Fig 1B). In addition, enrichment of exogenous Flag‐tagged PHGDH was also found in the mitochondria of PLC cells overexpression PHGDH (Fig 1C). Moreover, colocalization of Flag‐tagged PHGDH with mitochondria was confirmed by immunofluorescence analysis (Fig 1D), which further supported our Western blot data of the subcellular fractions. Then, we tried to determine which domain of PHGDH is responsible for its mitochondria localization. According to the function of PHGDH, its domains are divided into N‐terminus (substrate binding domain 1, SBD1), nucleotide‐binding domain (NBD), substrate binding domain 2 (SBD2), and the C‐terminus (regulatory domain, RD; Liu et al, 2020). By separating mitochondria components, our Western blot results showed that deletion of C‐terminus of PHGDH (PHGDH^∆RD) obviously prevents its mitochondrial localization (Fig 1E and F).

Figure 1. PHGDH localizes to the mitochondrial inner membrane in liver cancer cells.

1. Western blot analysis of the localization of endogenous PHGDH in PLC, HepG2, and THLE3 cells. Lamin B1, Tubulin, and Tom20 were used as markers for nuclear (Nuc), cytosolic (Cyto), and mitochondrial (Mito) proteins, respectively.
2. Western blot analysis of the localization of endogenous PHGDH in tumor tissues and adjacent normal tissues. Lamin B1, GAPDH, and Tom20 were used as markers for Nuc, Cyto, and Mito proteins, respectively.
3. Western blot analysis of the localization of endogenous and exogenous Flag‐tagged PHGDH in PLC cells. Lamin B1, Tubulin, and Tom20 were used as markers for Nuc, Cyto, and Mito proteins, respectively.
4. Immunofluorescence images showing colocalization of Flag‐PHGDH (green) and mitochondria (red) acquired with a laser confocal microscope system in PLC cells. Nuclei were stained with DAPI (blue), and mitochondria were stained with MitoTracker Red. Scale bar, 10 μm (left panel), 2 μm (right panel).
5. The PHGDH structure is displayed with the N‐terminus (substrate binding domain 1, SBD1), nucleotide‐binding domain (NBD), substrate binding domain 2 (SBD2), and the C‐terminus (regulatory domain, RD). The PHGDH^∆SBD1 and PHGDH^∆RD mutant variants lack the SBD1 and RD, respectively.
6. Subcellular localization of PHGDH was determined by Western blot in PLC cells overexpressing of Flag‐tagged wild‐type PHGDH, PHGDH^∆SBD1, or PHGDH^∆RD mutants. Mito, mitochondria; WCL, whole‐cell lysates. Tom20 was used as loading control.
7. Proteinase K protection assays were performed on purified mitochondria isolated from PLC and HepG2 cells. The outer mitochondrial membrane (OMM) protein Tom20, the inner mitochondrial membrane (IMM) protein ANT2, and the mitochondrial matrix (MM) protein TFAM were used as markers of the OMM, IMM, and MM, respectively.
8. Purified mitochondria of PLC or HepG2 cells were incubated with Na₂CO₃ (pH 11.5) for 30 min at 4°C. The soluble component (Supernatant/Sup.) and membrane component (Pellet) were separated by centrifugation prior to Western blot analysis. COX2, ATP8, and ANT2 are markers for the membrane component; CYTC and TFAM are markers for the soluble component.

Source data are available online for this figure.

Figure EV1. PHGDH localizes to mitochondria.

Western blot analysis of the localization of endogenous PHGDH in A549, MCF7, and MDA‐MB‐468 cells. Lamin B1, Tubulin, and Tom20 were used as markers for Nuc, Cyto, and Mito proteins, respectively. Source data are available online for this figure.

Next, to determine the submitochondrial localization of PHGDH, we carried out a proteinase K protection assay on mitochondria isolated from PLC cells. Our results showed that PHGDH was protected from degradation by proteinase K when mitochondria remained intact (without the addition of swelling buffer and Triton X‐100), while the outer mitochondrial membrane (OMM) protein Tom20 was degraded by proteinase K. The subsequent degradation trend of PHGDH was the same as that of the inner mitochondrial membrane (IMM) protein adenine nucleotide translocase 2 (ANT2), indicating that PHGDH is localized in the intermembrane space or the inner membrane of mitochondria (Fig 1G). To further clarify its submitochondrial localization, a carbonate extraction assay was performed, and we found that PHGDH was localized in insoluble membrane components (IMM proteins: COX2, ATP8, and ANT2), but not in soluble components (intermembrane space/IMS marker: CYTC; mitochondrial matrix/MM marker: TFAM), suggesting that PHGDH is an IMM protein (Fig 1H). Taken together, these results show that PHGDH is translocated to the inner mitochondrial membrane in liver cancer cells.

PHGDH promotes the translation of mtDNA‐encoded proteins

To determine whether mitochondrial PHGDH affects mitochondrial function, we measured the oxygen consumption rate (OCR) with a Seahorse metabolic analyzer, and our results showed that knockdown of PHGDH reduces basal respiration, ATP‐linked respiration, and maximal respiration in PLC cells (Fig 2A). Further blue native polyacrylamide gel electrophoresis (BN‐PAGE) experiments confirmed that knockdown of PHGDH decreased the enzyme activity of electron transport chain (ETC) complexes I, IV, and V in PLC cells (Fig EV2A).

Figure 2. PHGDH promotes the translation of mtDNA‐encoded proteins.

1. The oxygen consumption rate (OCR) was measured by successive injections of oligomycin (oligo), FCCP, and antimycin A/rotenone (AA/Rot) in PLC cells infected with viruses expressing nontargeting control (NTC) shRNA or shPHGDH. Basal respiration, ATP‐linked respiration, and maximal respiration were analyzed with the OCR curve.
2. Western blot analysis of mtDNA‐encoded proteins in PLC cells stably expressing NTC or shPHGDH and empty vector (EV) or PLVX‐PHGDH. β‐Actin was used as the loading control.
3. PLC cells with endogenous PHGDH knockdown were further infected with viruses expressing EV or PLVX‐PHGDH prior to Western blot analysis of mtDNA‐encoded proteins. β‐Actin was used as the loading control.
4. PLC cells expressing NTC or shPHGDH were treated with 100 μg/ml cycloheximide (CHX) for 20 min prior to puromycin treatment for 15 min. Purified mitochondrial lysates were subjected to a SUnSET assay with an anti‐puromycin antibody to measure the actively translated polypeptide chains. Western blot analysis of PHGDH confirmed the knockdown efficiency of the PHGDH shRNAs. β‐Actin was used as the loading control.
5. SUnSET assays were performed in the indicated cell lines using whole‐cell lysates.
6. PLC cells with or without PLVX‐Flag‐PHGDH overexpression were treated with DMSO or 50 μM chloramphenicol (CAP) in combination with 50 μM thiamphenicol (TAP) for 24 h prior to SUnSET assays. The assays were performed using whole‐cell lysate.
7. PLC cells with or without PLVX‐PHGDH overexpression were treated with DMSO or 50 μM CAP in combination with 50 μM TAP for 24 h prior to Western blot analysis of mtDNA‐encoded proteins. β‐Actin was used as the loading control.
8. The translation efficiency of mtDNA‐encoded proteins was determined by a SUnSET assay in PLC cells with endogenous PHGDH knockdown that were further infected with viruses expressing EV, wild‐type PHGDH, or the catalytically inactive PHGDH mutant PHGDH^R236E as indicated. β‐Actin was used as the loading control.
9. mtDNA‐encoded proteins were analyzed by Western blotting in PLC cells overexpressing EV, wild‐type PHGDH, or PHGDH^R236E. β‐Actin was used as the loading control.

Data information: n = 3 independent experiments and the data are presented as the mean ± SD values in (A). Statistical significance was determined by one‐way ANOVA (A). *P < 0.05 as compared to NTC group.

Source data are available online for this figure.

Figure EV2. PHGDH promotes mitochondrial translation.

1. In‐gel activity assays were performed in purified mitochondria from digitonin‐treated PLC‐NTC and PLC‐shPHGDH cells using blue native polyacrylamide gel electrophoresis (BN‐PAGE). The bands in the red boxes represent the activity of CI (complex I), CIV (complex IV), and CV (complex V).
2. Western blot analysis of mtDNA‐encoded proteins in HepG2 cells stably expressing NTC or shPHGDH. β‐Actin was used as the loading control.
3. Western blot analysis of nuclear‐encoded mitochondrial proteins in PLC cells stably expressing NTC or shPHGDH. β‐Actin was used as the loading control.
4. Quantitative real‐time PCR (qRT–PCR) analysis of the mRNA levels of the indicated mitochondrial genes in PLC cells stably expressing EV or PHGDH.
5. qRT–PCR analysis of the mRNA levels of the indicated mitochondrial genes in PLC cells stably expressing NTC or shPHGDH.
6. Pulse‐labeling translation assay of the 13 mitochondrial polypeptides (seven subunits of complex I [ND], three subunits of complex IV [COX], two subunits of complex V [ATP], and one subunit of complex III [cyt b]) showed a specific translation defect in PLC cells with PHGDH knockdown. Ponceau S staining was performed as a loading control.
7. SUnSET assays were performed using whole‐cell lysates from PLC cells expressing NTC or shPHGDH to measure the translation efficiency of mtDNA‐encoded proteins. β‐Actin was used as the loading control.
8. SUnSET assays were performed using whole‐cell lysates from HepG2 cells expressing NTC or shPHGDH to measure the translation efficiency of mtDNA‐encoded proteins. β‐Actin was used as the loading control.
9. mt‐DNA‐encoded protein expression was determined by Western blot in PLC cells overexpressing of Flag‐tagged wild‐type PHGDH, PHGDH^∆SBD1, or PHGDH^∆RD mutants. β‐Actin was used as loading control.

Data information: n = 3 independent experiments and the data are presented as the mean ± SD values in (D) and (E).

Source data are available online for this figure.

As the core subunit of the mitochondrial respiratory complexes is mainly composed of proteins encoded by mtDNA, we next investigated whether PHGDH regulates the expression of mtDNA‐encoded proteins. Interestingly, Western blot analysis revealed that knockdown of PHGDH reduced the expression levels of mtDNA‐encoded proteins, including ND2, ND5, ND6, COX2, and ATP8, while overexpression of PHGDH upregulated the expression of these proteins (Fig 2B). Similar results were observed in HepG2 cells expressing shRNAs targeting PHGDH (Fig EV2B). But knockdown of PHGDH did not affect the expression of nuclear‐coded MRPS15, MRPS16, MRPL44, MRPL48 protein levels (Fig EV2C), suggesting the specific effect of PHGDH on the expression of mitochondrial‐encoded proteins. More importantly, overexpression of PHGDH reversed the decreases in protein levels caused by endogenous PHGDH knockdown (Fig 2C). Subsequent quantitative real‐time PCR (qRT–PCR) analysis showed that PHGDH had no effect on the mRNA levels of mtDNA‐encoded proteins, suggesting that PHGDH regulates the expression of mtDNA‐encoded proteins at the posttranscriptional level (Fig EV2D and E).

To determine whether PHGDH promotes the expression of mtDNA‐encoded proteins by increasing the translation efficiency, we performed a ³⁵S pulse labeling experiment, and the results showed that knockdown of PHGDH significantly reduced the translation efficiency of mtDNA‐encoded proteins (Fig EV2F). To further confirm this phenomenon, we introduced a surface sensing of translation (SUnSET) system to examine the protein synthesis rate by measuring the amount of puromycin incorporation into neosynthesized proteins (Schmidt et al, 2009; Liu et al, 2019a, 2019b). PLC cells with PHGDH knockdown were pretreated with cycloheximide (CHX) to block cytoplasmic protein synthesis and labeled with puromycin. Then, whole‐cell lysate (WCL) or mitochondrial lysate was prepared for detection of puromycin incorporation by Western blotting. Our results revealed that knockdown of PHGDH significantly reduced the translation efficiency of mtDNA‐encoded proteins (Figs EV2G and 2D). Similar results were observed in HepG2 cells by the SUnSET assay (Fig EV2H). Notably, unlike PHGDH^WT group, restored expression of PHGDH^∆RD could not rescue shPHGDH‐inhibited mitochondrial translation (Fig 2E). Consistently, overexpression of PHGDH^∆RD mutant also failed to upregulate mt‐DNA‐encoded protein expression (Fig EV2I). Moreover, both the enhancement of mitochondrial translation efficiency and the accumulation of downstream mtDNA‐encoded proteins induced by PHGDH overexpression were obviously suppressed when cells were treated with the mitochondrial translation inhibitors chloramphenicol (CAP) (Minton et al, 2018) and thiamphenicol (TAP) (Fig 2F and G). These results suggest that PHGDH promotes the expression of mtDNA‐encoded proteins by increasing mitochondrial translation efficiency.

To further clarify whether PHGDH promotes mitochondrial translation in a manner dependent on its metabolic catalytic activity, we ectopically expressed either wild‐type PHGDH or the PHGDH^R236E mutant (Mattaini et al, 2015; Vandekeere et al, 2018; Zhang et al, 2021) (a catalytically inactive PHGDH mutant) in PLC cells with endogenous PHGDH knockdown. Our data showed that both wild‐type PHGDH and the PHGDH^R236E mutant increased the translation efficiency and protein levels of mtDNA‐encoded proteins (Fig 2H and I). Taken together, these data demonstrate that PHGDH promotes the translation of mtDNA‐encoded proteins in a manner independent of its metabolic enzyme activity.

mtEFG2 and ANT2 are required for PHGDH‐mediated mitochondrial translation

Multiple mitochondrial translation factors participate in different stages of the mitochondrial translation cycle, namely initiation (mtIF2 and mtIF3), elongation (mtEFTu and mtEFG1), termination, and recycling (mtEFG2 and mtRRF) (Hallberg & Larsson, 2014; Kummer & Ban, 2021). To gain further insight into the mechanism by which PHGDH promotes mitochondrial translation, we performed a mini‐screen to search for potential mitochondrial translation factors that might interact with PHGDH. mtIF2, mtEFG1, and mtEFG2 were found to interact with PHGDH, among which mtEFG2 showed the strongest interaction (Fig 3A). This result was further confirmed by Co‐IP experiments in 293T cells cotransfected with Flag‐mtEFG2 and HA‐PHGDH^WT plasmids (Fig 3B). However, GST pull‐down assays showed that there was no direct interaction between PHGDH and mtEFG2 (Fig EV3A).

Figure 3. mtEFG2 and ANT2 are required for PHGDH‐mediated mitochondrial translation.

1. HEK293T cells were transfected with Flag‐tagged PHGDH in combination with HA‐tagged EV, mtIF2, mtIF3, mtEFTu, mtEFG1, mtEFG2, or mtRRF. Co‐immunoprecipitation (Co‐IP) was performed with an anti‐Flag antibody prior to Western blot analysis.
2. HEK293T cells were transfected with HA‐tagged PHGDH^WT or PHGDH^R236E in combination with Flag‐EV or Flag‐mtEFG2, and Co‐IP was performed with an anti‐Flag antibody prior to Western blot analysis.
3. HEK293T cells were transfected with HA‐tagged PHGDH^WT or PHGDH^R236E in combination with Flag‐EV or Flag‐ANT2, and Co‐IP was performed with an anti‐Flag antibody prior to Western blot analysis.
4. HEK293T cells were transfected with HA‐mtEFG2 in combination with Flag‐EV or Flag‐ANT2, and Co‐IP was performed with an anti‐Flag antibody prior to Western blot analysis.
5. GST pull‐down of His‐PHGDH^WT and His‐PHGDH^R236E by GST‐ANT2 using proteins purified in E. coli, followed by Western blot analysis with anti‐His and anti‐GST antibodies.
6. GST pull‐down of His‐mtEFG2 by GST‐ANT2 using proteins purified in E. coli, followed by Western blot analysis with anti‐His and anti‐GST antibodies.
7. PLC cells stably expressing Flag‐mtEFG2 and HA‐PHGDH were further infected with viruses expressing NTC or shANT2, and then, cell lysates were immunoprecipitated with an anti‐Flag antibody or IgG prior to Western blot analysis.
8. PLC cells overexpressing PHGDH were further infected with viruses expressing NTC or shANT2. The translation efficiency was measured by a SUnSET assay.
9. PLC cells overexpressing PLVX‐PHGDH were further infected with viruses expressing NTC or shANT2. The expression of mtDNA‐encoded proteins was determined by Western blotting. β‐Actin was used as the loading control.
10. PLC cells overexpressing PHGDH were further infected with viruses expressing NTC or shmtEFG2. The translation efficiency was measured by a SUnSET assay.
11. PLC cells overexpressing pSin‐3xFlag‐PHGDH were further infected with viruses expressing NTC or shmtEFG2. The expression of mtDNA‐encoded proteins was determined by Western blotting. β‐Actin was used as the loading control.

Source data are available online for this figure.

Figure EV3. PHGDH binds to mtEFG2 and ANT2.

• A
  GST pull‐down of His‐mtEFG2 by GST‐PHGDH using proteins purified in E. coli, followed by Western blot analysis with anti‐His and anti‐GST antibodies.
• B
  HEK293T cells were transfected with HA‐ANT2 in combination with Flag‐EV or Flag‐PHGDH, and Co‐IP was performed with an anti‐Flag antibody prior to Western blot analysis.
• C, D
  Purified mitochondrial fractions of PLC cells stably expressing Flag‐PHGDH (C) or Flag‐ANT2 (D) were immunoprecipitated with an anti‐Flag antibody prior to Western blot analysis with anti‐PHGDH and anti‐ANT2 antibodies. COX4 was used as the marker of mitochondrial proteins.
• E
  SUnSET assays were performed using whole‐cell lysates from PLC‐NTC and PLC‐shmtEFG2 cells to measure the translation efficiency of mtDNA‐encoded proteins. β‐Actin was used as the loading control.
• F
  SUnSET assays were performed using whole‐cell lysates from PLC‐EV and PLC‐mtEFG2 cells to measure the translation efficiency of mtDNA‐encoded proteins. β‐Actin was used as the loading control.
• G
  Western blot analysis of the mtDNA‐encoded proteins in PLC cells stably expressing NTC or shmtEFG2. β‐Actin was used as the loading control.

Source data are available online for this figure.

To further identify the potential adaptor protein that mediates the interaction between PHGDH and mtEFG2, we performed IP experiments followed by mass spectrometry analysis (IP‐MS) on the purified mitochondrial component from PLC cells overexpressing PHGDH. Our results showed that PHGDH interacts with multiple mitochondrial proteins (Table EV1), including ANT2. Aac2p, a homolog of ANT1 in yeast, has been reported to promote mitochondrial translation (Ogunbona et al, 2018). To validate the binding between PHGDH and ANT2, Co‐IP experiments were carried out, and the results showed that ANT2 could interact with PHGDH and mtEFG2 in 293T cells (Figs 3C and D and EV3B). Moreover, GST pull‐down assays showed that ANT2 directly interacted with PHGDH and mtEFG2 (Fig 3E and F). The binding of HA‐PHGDH^R236E to Flag‐mtEFG2 (Fig 3B) and Flag‐ANT2 (Fig 3C), or the direct binding of His‐PHGDH^R236E to GST‐ANT2 (Fig 3E) further proved that these binding events were not dependent on the enzymatic activity of PHGDH. To further explore whether PHGDH binds to ANT2 in mitochondria, purified mitochondria from PLC‐Flag‐PHGDH and PLC‐Flag‐ANT2 cells were subjected to Co‐IP with an anti‐Flag antibody, and the results showed that PHGDH interacted with ANT2 in mitochondria (Fig EV3C and D). Notably, knockdown of ANT2 inhibited the interaction between PHGDH and mtEFG2 (Fig 3G). Taken together, our results suggest that ANT2 acts as an adaptor that recruits mtEFG2 to bind to PHGDH in mitochondria.

mtEFG2 is a mitochondrial translation factor that only functions during mitochondrial ribosome recycling (Tsuboi et al, 2009). Indeed, our data showed that mtEFG2 positively regulates mitochondrial translation efficiency and the expression levels of mtDNA‐encoded proteins (Fig EV3E–G). To further investigate whether ANT2 and mtEFG2 are necessary for PHGDH‐mediated mitochondrial translation, we examined the effects of ANT2 and mtEFG2 on mitochondrial translation in PHGDH‐overexpressing PLC cells. Our results showed that knockdown of ANT2 or mtEFG2 significantly attenuated the PHGDH‐induced enhancement of mitochondrial translation efficiency and the expression of mtDNA‐encoded proteins (Fig 3H–K). Collectively, these results indicate that mtEFG2 and ANT2 are required for PHGDH‐mediated mitochondrial translation.

PHGDH promotes mtEFG2‐mediated mitochondrial ribosome recycling in cancer cells

Recent study has shown that mtEFG2 cooperates with mtRRF to release tRNA and mRNA from the ribosome and dissociate the ribosome into subunits for its participation in a new round of mitochondrial translation (Kummer et al, 2021). In addition, structural evidence suggests that mtEFG2 interacts with mtRRF through electrostatic and spatial optimization (Kummer et al, 2021). As the interaction between mtEFG2 and mtRRF is crucial for the mitochondrial ribosome recycling process, we first verified the interaction between mtRRF and mtEFG2 through Co‐IP experiments in 293T cells (Fig 4A). Most importantly, knockdown of PHGDH inhibited the interaction between mtEFG2 and mtRRF (Fig 4B) and further reduced the mtEFG2‐regulated mitochondrial translation efficiency and protein levels of mtDNA‐encoded proteins (Fig 4C and D), indicating that PHGDH is necessary for mtEFG2‐regulated mitochondrial translation.

Figure 4. PHGDH promotes mtEFG2‐mediated mitochondrial ribosome recycling in cancer cells.

1. HEK293T cells were transfected with HA‐mtEFG2 in combination with Flag‐EV or Flag‐mtRRF, and Co‐IP was performed with an anti‐Flag antibody prior to Western blot analysis.
2. PLC cells stably expressing Flag‐mtEFG2 and HA‐mtRRF were further infected with viruses expressing NTC or shPHGDH, and then, cell lysates were immunoprecipitated with anti‐Flag antibody or IgG prior to Western blot analysis.
3. PLC cells stably overexpressing mtEFG2 were further infected with viruses expressing NTC or shPHGDH. The translation efficiency was measured by a SUnSET assay.
4. PLC cells stably overexpressing mtEFG2 were further infected with viruses expressing NTC or shPHGDH. mtDNA‐encoded proteins were analyzed by Western blot. β‐Actin was used as the loading control.
5. PLC cell lysates were immunoprecipitated with an anti‐MRPL44 antibody or IgG prior to Western blot analysis of MRPL44, MRPL46, MRPL48, MRPS15, MRPS16, and MRPS35.
6. Binding of mtDNA‐encoded mRNA by endogenous MRPL44 (representing mitochondrial ribosomes) was evaluated by RNA immunoprecipitation (RIP) in PLC cells with PHGDH knockdown.
7. The same amount of mitochondria from PLC cells expressing NTC or shPHGDH treated with CAP and TAP were loaded onto the sucrose density gradient. After centrifugation, all fractions were subjected to Western blot analysis of MRPS35, MRPL44, and MRPL48 proteins. RNA concentrations of ND6 were measured.

Data information: n = 3 independent experiments in (F), the data are presented as the mean ± SD values. The average of three independent experiments is shown in (G).

Source data are available online for this figure.

We next investigated whether PHGDH promotes mitochondrial translation by facilitating mtEFG2‐mediated mitochondrial ribosome recycling. Among the 80 mitochondrial ribosomal proteins (MRPs) that compose the mitoribosome, MRPL44, a large ribosomal subunit protein, plays an important role in the expression of mtDNA‐encoded proteins and the subsequent OXPHOS capacity (Yeo et al, 2015; Cheong et al, 2020). We first performed an endogenous IP experiment to detect the interaction between endogenous MRPL44 and other MRPs. The results indicated that MRPL44 was incorporated into complete mitoribosomes by binding to other mitochondrial large or small ribosomal subunit proteins (Fig 4E; Busch et al, 2019). Next, an RNA immunoprecipitation (RIP) assay using an antibody against MRPL44 was performed to detect the binding state between the mitochondrial ribosome (represented by MRPL44) and mtDNA‐encoded mRNA by reference to the method ribosome‐associated mRNA capture from tissues (Sanz et al, 2009; Busch et al, 2019). The RIP assay results showed significant binding of the MRPL44 protein to mtDNA‐encoded mRNA in PLC‐shmtEFG2 cells (Fig EV4A), indicating that the mitochondrial translation process was arrested at the step of ribosome binding to mtDNA‐encoded mRNA and could not proceed to the next cycle. In other words, the efficiency of mitochondrial ribosome recycling was decreased when mtEFG2 was knocked down. Similar results were observed when PHGDH was knocked down in PLC cells (Fig 4F), suggesting that PHGDH is crucial in mitochondrial ribosome recycling.

Figure EV4. Knockdown of mtEFG2 induces the accumulation of intact mitoribosomes on the mtDNA‐encoded transcripts.

1. Binding of mtDNA‐encoded mRNA by endogenous MRPL44 (representing mitochondrial ribosomes) was determined by a RIP assay in PLC cells with mtEFG2 knockdown.
2. The same amount of mitochondria from PLC cells expressing NTC or shmtEFG2 treated with CAP and TAP were loaded onto the sucrose density gradient. After centrifugation, all fractions were subjected to Western blot analysis of MRPS35, MRPL44, and MRPL48 proteins. RNA concentrations of ND6 were measured.

Data information: n = 3 independent experiments and the data are presented as the mean ± SD values in (A). the average of three independent experiments is shown in (B).

Source data are available online for this figure.

To further validate this phenomenon, we performed mitochondrial polysome profiling assay to decipher the effect of PHGDH or mtEFG2 depletion on association of mtDNA‐encoded transcripts with ribosomes (Ruzzenente et al, 2012; Antonicka et al, 2013; Tu & Barrientos, 2015). We isolated 14 fractions of mitochondrial ribosomes by sucrose density gradient centrifugation. Based on the expression of small subunit protein MRPS35 and large subunit proteins MRPL44 and MRPL48, we found that fraction 5 represents a small subunit (28S), fraction 7 represents a large subunit (39S), and fraction 10 represents an intact mitochondrial ribosome (55S). Our results showed that protein levels of fraction 10 are obviously accumulated when we knocked down PHGDH or mtEFG2 compared to NTC group (Figs 4G and EV4B, left), suggesting that the intact mitoribosome is retained on the mRNAs. More importantly, when we isolated RNA from the sucrose gradient fractions and determined the distribution of mitochondrial transcripts, we found that mtDNA‐encoded transcripts ND6 were significantly enriched in fraction 10 and its subsequent series of fractions, which were considered as the polysomes for mitochondrial translation (Figs 4G and EV4B, right). However, the binding of ND6 mRNA to MRPL44 was increased (Figs 4F and EV4A), but the protein expression of ND6 was significantly decreased when we knocked down PHGDH or mtEFG2 (Figs 2B and EV3G). Taken together, these data strongly indicate that depletion of PHGDH or mtEFG2 prevents the dissociation of intact mitochondrial ribosomes, that is, the small subunit and large subunit of ribosomes cannot effectively separate and proceed to the next translation cycle. Our results showed for the first time that in addition to mtEFG2 and mtRRF, the metabolic enzyme PHGDH is involved in the process of mitochondrial ribosome recycling to promote efficient mitochondrial translation.

PHGDH promotes the proliferation of liver cancer cells by facilitating mitochondrial activity

To explore whether PHGDH increases mitochondrial respiratory capacity and promotes cell growth through mtEFG2, we measured the oxygen consumption rate in PLC cells. Our results showed that knockdown of mtEFG2 diminished the enhancing effect of wild‐type PHGDH or PHGDH^R236E on the capacity for mitochondrial respiration, including basal respiration, ATP‐linked respiration, and maximal respiration (Fig 5A). Cell growth assays showed that mtEFG2 significantly promoted the proliferation of cancer cells (Fig EV5A). Moreover, when mtEFG2 was knocked down, the promoting effect of wild‐type PHGDH or PHGDH^R236E on cell proliferation was obviously inhibited (Fig 5B), indicating that the promoting effect of PHGDH on cell growth is dependent on mtEFG2. Importantly, by preventing mitochondrial localization, overexpression of PHGDH^∆RD mutant could not compensate for the proliferation disadvantage caused by endogenous PHGDH depletion as efficiently as wild‐type PHGDH (Fig EV5B).

Figure 5. PHGDH promotes the proliferation of liver cancer cells by facilitating mitochondrial activity.

• A
  PLC cells overexpressing EV, PHGDH^WT, or PHGDH^R236E were further infected with viruses expressing NTC or shmtEFG2. The OCR was measured after successive injections of oligomycin (oligo), FCCP, and antimycin A/rotenone (AA/Rot). Basal respiration, ATP‐linked respiration, and maximal respiration were analyzed with the OCR curve.
• B
  PLC cells overexpressing EV, PHGDH^WT, or PHGDH^R236E were further infected with viruses expressing NTC or shmtEFG2 prior to Western blot analysis. β‐Actin was used as the loading control (left). Cell growth curves were constructed by cell counting with trypan blue exclusion (right).
• C
  PLC cells overexpressing EV, PHGDH^WT, or PHGDH^R236E were further treated with H₂O or tigecycline prior to Western blot analysis. β‐Actin was used as the loading control (left). Cell growth curves were determined by cell counting with trypan blue exclusion (right).
• D–F
  PLC cells overexpressing EV or PHGDH were further infected with viruses expressing NTC or shmtEFG2. Then, these cells were subcutaneously injected into male nude mice (n = 5 in each group). Tumor sizes were measured beginning 14 days after inoculation (D). At the end of the experiment, the tumors were excised (E), and PHGDH and mtEFG2 expression in tumor lysates was analyzed by Western blotting (F).
• G–I
  PLC cells overexpressing EV or PHGDH were subcutaneously injected into male nude mice (n = 5 in each group). Mice were intraperitoneally injected with 100 mg/kg tigecycline daily beginning 14 days after inoculation. Tumor sizes were measured beginning 14 days after inoculation (G). At the end of the experiment, the tumors were excised (H), and PHGDH expression in tumor lysates was analyzed by Western blotting (I).

Data information: n = 3 independent experiments in (A–C), and data are presented as the mean ± SD values in (A–C, D, G). Statistical significance was determined by one‐way ANOVA. *P < 0.05 between the indicated groups. NS, not significant.

Source data are available online for this figure.

Figure EV5. PHGDH promotes tumor progression through mtEFG2 and mitochondrial translation.

1. Growth curves of PLC cells stably expressing NTC or shmtEFG2 were determined by cell counting with trypan blue exclusion. Western blot showed protein levels of mtEFG2 in PLC cells stably expressing NTC or shmtEFG2. β‐Actin was used as the loading control.
2. The growth curves of the indicated PLC cells were determined by trypan blue staining.
3. The tumors shown in Fig 5E were weighed at the end of the experiment (n = 5 in each group).
4. The tumors shown in Fig 5H were weighed at the end of the experiment (n = 5 in each group).

Data information: n = 3 independent experiments and data are presented as the mean ± SD values in (A) and (B). Data in (C) and (D) are presented as the mean ± SEM values. Statistical significance was determined by one‐way ANOVA. *P < 0.05 between the indicated groups. NS, not significant.

Source data are available online for this figure.

Next, to further investigate whether mitochondrial translation is essential for PHGDH‐mediated liver cancer proliferation, we treated wild‐type PHGDH or PHGDH^R236E‐overexpressing PLC cells with tigecycline, an FDA‐approved broad‐spectrum antibiotic that inhibits the translation of mtDNA‐encoded proteins, and the results showed that a reduction in mitochondrial translation efficiency significantly decreased the PHGDH‐mediated cell growth advantage (Fig 5C). Collectively, these data suggest that mitochondrial translation is essential for PHGDH‐mediated liver cancer proliferation.

Xenograft experiments in nude mice further revealed that knockdown of mtEFG2 diminished the PHGDH‐mediated promoting effects on tumor growth and tumor mass (Figs 5D and E and EV5C). Using tumor tissue lysates, we further confirmed the knockdown of endogenous mtEFG2 and overexpression of PHGDH (Fig 5F). Moreover, we further examined the effect of tigecycline on PHGDH‐mediated tumor growth. Mice‐bearing xenografts formed from EV and PHGDH liver cancer cells were intraperitoneally injected daily with 100 mg/kg tigecycline. The results showed that tigecycline significantly retarded PHGDH‐promoted tumor growth and decreased the tumor mass (Figs 5G and H and EV5D). Overexpression of PHGDH was confirmed by Western blot analysis of tumor lysates (Fig 5I). Taken together, these data demonstrate that PHGDH‐mediated mitochondrial translation is mtEFG2‐dependent and critical for tumor growth in vivo.

Discussion

A striking feature of tumors is the great versatility of metabolic enzymes in cancer cells during tumor progression and evolution, including their performance of metabolic and nonmetabolic or non‐canonical functions (Faubert et al, 2020; Dey et al, 2021; Xu et al, 2021). Here, we report that PHGDH, a committed serine biosynthesis enzyme that has been extensively studied, is unexpectedly localized to the mitochondrial inner membrane and then recruits mtEFG2 via ANT2 to improve mitochondrial translation efficiency in cancer cells. During conventional mitoribosome recycling, mtEFG2 specifically interacts with mtRRF to dissociate the ribosomal subunits in mammalian mitochondria (Tsuboi et al, 2009). Interestingly, PHGDH was found to promote the dissociation of mitoribosome into its subunits by enhancing the binding state between mtEFG2 and mtRRF, thereby facilitating OXPHOS, cell proliferation, and tumor progression. This binding mode of PHGDH/ANT2/mtEFG2/mtRRF explains why PHGDH localizes to the mitochondrial inner membrane and acts as a recruiter to recruit other translation‐related factors to the vicinity of the inner membrane, thus promoting the assembly of mitochondrially translated proteins into the OXPHOS complex, which can maximize the smooth progression of mitochondrial respiration while improving the efficiency of mitochondrial translation. Our findings shed new light on the role of PHGDH as a moonlighting enzyme and demonstrate that PHGDH functions as an activator of mitochondrial translation in cancer cells.

As the first catalytic enzyme in the SSP pathway, PHGDH participates in tumor progression through various mechanisms in the tumor microenvironment. For example, PHGDH is phosphorylated by PKC‐ζ to inhibit its enzymatic activity under nutrient stress conditions (Ma et al, 2013). Our previous study found that cMyc promotes cancer progression by activating the serine synthesis pathway under nutrient deprivation conditions (Sun et al, 2015). In addition to being responsible for serine synthesis, PHGDH also promotes tumor proliferation in a nonmetabolic manner. Glucose deprivation induces PHGDH to undergo nuclear translocation and promotes tumorigenesis by repressing c‐Jun PARylation in the nucleus (Ma et al, 2021). PHGDH is considered to interact with the translation initiation factors eIF4A1 and eIF4E to promote translation initiation in the cytoplasm, thereby accelerating the development of pancreatic cancer (Ma et al, 2019). However, it is still unclear whether PHGDH localizes to mitochondria and regulates mitochondria‐related processes. In this study, we show that mitochondria‐translocated PHGDH recruits the mitochondrial ribosome recycling factor mtEFG2 via ANT2 to facilitate subsequent cycles of mitochondrial translation, thereby promoting cancer cell progression. Our study not only provides a link between a metabolic enzyme and mitochondrial translation but also reveals that metabolic enzymes can regulate mitochondrial translation through nonmetabolic functions.

Adenine nucleotide translocases (ANTs) exchange ADP for ATP through the mitochondrial inner membrane (Kokoszka et al, 2004). In addition, ANTs are involved in the regulation of a variety of mitochondrial functions, such as apoptosis, oxidative phosphorylation, and mitophagy (Prabhu et al, 2015; Hoshino et al, 2019; Seo et al, 2019). There are four ANT isoforms in humans, namely, ANT1, ANT2, ANT3, and ANT4, among which ANT2 is abundantly expressed in cancer cells, in contrast to the other isoforms (Chevrollier et al, 2011). Previous studies have suggested that ANT2 functions as an oncogenic protein, but the mechanism is not clear (Chevrollier et al, 2011; Jang et al, 2012; Li et al, 2020). In this study, we found that ANT2 mediates the promoting effect of PHGDH on mitochondrial translation. Aac2p, the major mitochondrial ADP/ATP carrier in yeast, regulates cytochrome c oxidase activity in a translation‐dependent manner and is the only mitochondrial ADP/ATP carrier required for OXPHOS and growth (Lawson et al, 1990; Ogunbona et al, 2018). Although the role of Aac2p in mitochondrial translation is known in yeast, we revealed that ANT2 promotes mitochondrial translation in human cancer cells. However, the direct role of ANT2 in mitochondrial translation and its underlying mechanism remain to be further studied.

Cancer cells are usually metabolically heterogeneous. Small subsets of cancer cells with tumorigenic potential or quiescent/slow‐cycling cancer cells are more dependent on oxidative respiration than on glycolysis (Viale et al, 2015). Emerging studies have shown that targeting mitochondrial energy metabolism is a promising therapeutic strategy for cancer. The mitochondrial ribosome inhibitor tigecycline selectively kills human acute myelocytic leukemia stem cells compared to their normal counterparts and thus has clinical therapeutic potential in human leukemia (Skrtic et al, 2011). Metformin, a commonly used drug in the clinical treatment of diabetes, has been shown to improve the survival rate of patients who have already developed cancer by targeting mitochondrial ATP production (Saraei et al, 2019). Several preclinical drugs also have the potential to treat cancers, and VLX600 and gamitrinib exert anticancer effects by targeting the ETC complex (Chae et al, 2012; Ekstrom et al, 2021). Therefore, elucidating the specific mechanism by which OXPHOS is upregulated in cancer cells and targeting the related components accordingly may achieve promising effects. In our study, PHGDH elevated the expression of mtDNA‐encoded proteins by promoting mitochondrial mRNA translation and enhancing mitochondrial ribosome recycling, thereby accelerating mitochondrial respiration and tumor proliferation. Moreover, tigecycline treatment significantly inhibited the promoting effect of PHGDH on tumor progression. Therefore, these results suggest that it is possible to maximize tumor growth inhibition if we target both the cytoplasm‐dependent metabolic function and mitochondria‐dependent translation function of PHGDH.

Given that PHGDH is upregulated and amplified in different cancer cells and is associated with tumor metastasis and drug resistance, our findings demonstrate a new mechanism by which PHGDH promotes cancer cell proliferation, extending the understanding of the new function of the metabolic enzyme PHGDH to the mitochondrial level and proposing a potential strategy to treat cancer by inhibiting PHGDH‐induced mitochondrial translation.

Materials and Methods

Cell culture and reagents

All cells were purchased from ATCC. Human PLC, HepG2, MDA‐MB‐468, A549, MCF7, and HEK293T cells were cultured in Dulbecco's modified essential medium (DMEM) (Gibco, 12800‐017) supplemented with 10% Certified Fetal Bovine Serum (FBS) (VivaCell, Shanghai, China, C04001‐500) and 1% Pen/Strep (Sangon Biotech, E607011). THLE3 cells were cultured in Bronchial epithelial cell growth medium (BEGM) (Clonetics Corporation, Walkersville, CC‐3170) containing 10% FBS. All cell lines were authenticated using short tandem repeat profiling (by GENEWIZ Co. Ltd.) and tested for and found to be free of mycoplasma contamination using PCR. Cells were cultured at 37°C in a humidified atmosphere with 5% CO₂ in air. Cycloheximide (CHX, Santa Cruz, sc‐3508), chloramphenicol (CAP; Sangon Biotech, A600118), thiamphenicol (TAP; Sangon Biotech, A506242), hygromycin B (Sangon Biotech, A600230), digitonin (Sangon Biotech, A601152), a Seahorse XF Cell Mito Stress Test Kit (Agilent, 103010‐100), and tigecycline (Selleck, S1403) were obtained from the indicated vendors.

RNA isolation and quantitative real‐time PCR

Cellular total RNA was isolated with TRIzol (Invitrogen, Thermo Fisher Scientific) and treated with DNase (Ambion), and cDNA synthesis was then performed using a HiScript II 1^st Strand cDNA Synthesis Kit (Vazyme, R211‐01). cDNA samples were used for quantitative real‐time PCR (qRT–PCR) analysis using SYBR Green Master Mix (Vazyme, Q111‐02) on a Bio‐Rad iCycler. The sequences of the primers used are shown in Table EV2. Expression in all samples was normalized to 18S rRNA expression, and the fold change in the target mRNA expression was calculated based on the threshold cycle (C _t) value, where $\mathrm{\Delta }{C}_{\mathrm{t}}=C_{{\mathrm{t}}_{\text{target}}}-C_{{\mathrm{t}}_{18\mathrm{S}}}$ and $\mathrm{\Delta }\left(\mathrm{\Delta }{C}_{\mathrm{t}}\right)=\mathrm{\Delta }{C}_{{\mathrm{t}}_{\text{Control}}}-\mathrm{\Delta }{C}_{{\mathrm{t}}_{\text{Indicated condition}}}.$ The data are presented as the mean ± standard deviation (SD) of three biological replicates.

Plasmids and established stable cells

Lentiviral shRNAs targeting PHGDH, ANT2, and mtEFG2 in the PLKO.1 vector were commercially purchased (Sigma‐Aldrich). shRNAs targeting the 3′UTR of PHGDH and ANT2 were constructed in the PLKO.1 vector. The target sequences of all shRNAs we used are listed in Table EV3. PHGDH, PHGDH^R236E, ANT2, and mtEFG2 were subcloned into the pSin‐EF2 vector with a 3× Flag tag. PHGDH, ANT2, mtIF2, mtIF3, mtEFTu, mtEFG1, mtEG2, and mtRRF were subcloned into the pSin‐EF2 vector with an HA tag. The PLVX‐Flag‐PHGDH, PLVX‐Flag‐PHGDH^R236E, and PLVX‐Flag‐mtEFG2 vectors were used to construct G418 resistance plasmids. All plasmids targeting specific genes were cotransfected into HEK293T cells with plasmids encoding Δ8.9 and VSVG using PEI (Polysciences, Inc, 23966‐2) for 48 h. Then, the collected viruses were used to infect target cells in the presence of 8 μg/ml polybrene (Sigma‐Aldrich). The transduced cells were selected with 0.5 mg/ml hygromycin B, 0.8 mg/ml G418, or 1 mg/ml puromycin to establish the corresponding stable cell lines.

Western blotting

Proteins were isolated from cells using RIPA buffer (50 mM Tris–HCl (pH 8.0), 150 mM NaCl, 5 mM EDTA, 0.1% SDS, and 1% NP‐40) supplemented with protease inhibitor cocktail (Roche, #43203100, Mannheim, Germany) and 1 mM PMSF, 1 mM Na₃VO₄, 2 mM DTT, and 10 mM NaF. The lysate supernatant was collected after centrifugation at 13,000 rpm for 10 min at 4°C, and the protein concentration was quantified with a Bradford assay kit (Sangon Biotech, C000164‐0200). Then, the sample was placed in a 100°C metal bath for 5 min until the proteins were completely denatured. Equal amounts of protein were subjected to 6–12% SDS–PAGE. Detailed information for all antibodies used is provided in Table EV4.

Immunoprecipitation

Immunoprecipitation was carried out as described previously (Wu et al, 2017). In brief, 293T cells transfected with the indicated plasmids were collected and lysed in IP buffer (20 mM HEPES [pH 7.5], 150 mM NaCl, 2 mM EDTA, and 1% NP‐40) for 1–2 h, and the supernatant was then transferred to a 10 μl volume of solution containing protein A/G beads and precleared for 1 h. After protein quantification, the corresponding primary antibody was added and incubated with the lysate overnight (for endogenous IP) or for 4–6 h (for exogenous IP). Next, protein A/G beads were added, incubated for another 1 h, and washed four or five times at 4°C with IP buffer. Finally, 5× SDS loading buffer was added to the samples, and the beads were placed in a 100°C metal bath for 5 min to fully denature proteins. Western blot analysis was used to detect binding between the indicated proteins.

Mitochondria isolation

PLC cells were suspended in ice‐cold buffer containing 100 mM sucrose, 50 mM Tris–HCl (pH 7.4), 100 mM KCl, 1.5 mM MgCl₂, 1 mM EGTA, 100 μg/ml chloramphenicol, and 100 μg/ml thiamphenicol and homogenized by 35–40 passes in a prechilled Dounce tissue grinder tube (Sigma, T0566). The postnuclear supernatant was obtained by centrifugation of the samples twice for 20 min at 2,000 g. Mitochondria were pelleted by centrifugation for 30 min at 4,000 g and washed twice in the same buffer. The protein concentration was determined with the Bradford assay kit.

Proteinase K protection assay and alkaline sodium carbonate extraction

For the proteinase K protection assay, mitochondria were resuspended with a final concentration of 50 μg/ml proteinase K in mitochondrial isolation buffer or swelling buffer (10 mM HEPES‐KOH [pH 7.4] and 1 mM EDTA) or 50 μg/ml proteinase K in swelling buffer with 1% (v/v) Triton X‐100, and the mixture was then incubated on ice for 15 min. PMSF (4 mM) was then added to inhibit the activity of proteinase K and incubated for another 20 min. Then, the sample was placed in a 100°C metal bath for 5 min to completely denature the proteins, and the proteins were subjected to SDS–PAGE.

For alkaline sodium carbonate extraction, mitochondria were divided into two equal parts, one of which was resuspended in ice‐cold 100 mM Na₂CO₃ (pH 11.5), incubated on ice for 30 min, and centrifuged for 30 min at 100,000 g to pellet the membranes. Then, the supernatant was labeled Sup., and the precipitate was labeled Pellet and resuspended with the same volume of lysate as Sup. The other portion of mitochondria was suspended with the same volume of lysate and labeled T (Total). Finally, the samples were placed in a 100°C metal bath for 5 min to completely denature proteins, and the proteins were then separated by SDS–PAGE.

Mitochondrial translation assays

For SUnSET assays (Schmidt et al, 2009), PLC and HepG2 cells were seeded at 4–6 × 10⁵ cells/ml in a 6 cm dish for 24 h for adherence and were then cultured for 6–8 h under serum starvation conditions. For serum stimulation, cells were maintained in regular medium containing 10% FBS for 1 h and then exposed to 100 μg/ml CHX for 20 min to inhibit cytoplasmic translation. Puromycin pulses were performed by incubating the cells with 10 mg/ml puromycin for 15 min at 37°C. Cells or isolated mitochondria were then washed with cold PBS and lysed in RIPA buffer. Equal amounts of the lysates were analyzed by Western blot analysis using the anti‐puromycin antibody.

For the pulse labeling assay of mitochondrial translation products, the translation efficiency of mitochondrial DNA‐encoded protein was evaluated by pulse labeling of 80% confluent PLC cells in 6 cm dishes. Cells were cultured in medium lacking methionine and cysteine in the presence of 100 μg/ml emetine to inhibit cytoplasmic protein synthesis, and 200 μCi/ml EasyTag™ EXPRESS ³⁵S protein labeling mix (NEG772; PerkinElmer) was then added for ³⁵S labeling for 1 h. Finally, the cells were washed with PBS and lysed in RIPA buffer. Equal amounts of the lysates were analyzed by Western blot analysis and transferred to a nitrocellulose membrane prior to autoradiography (Sasarman & Shoubridge, 2012; Antonicka et al, 2013; van Voss et al, 2018).

Mitochondrial RNA immunoprecipitation

Mitochondria from PLC cells treated with puromycin (to synchronize mitochondrial translation at the ribosomal recycling stage) were lysed in 1 ml of RIP lysis buffer (0.5% NP‐40, 20 mM Tris–HCl (pH 7.5), 200 mM NaCl, 2.5 mM MgCl₂, and 10% glycerol) supplemented with protease inhibitor cocktail (Roche, #43203100, Mannheim, Germany), and 400 U/ml RNase inhibitor (Invitrogen, AM2684) on ice for 30 min with occasional vortexing. The extract was centrifuged at 16,000 g for 10 min at 4°C, and the supernatant was transferred to a new tube containing 20 μl of protein A/G beads and precleared for 1 h. After protein quantification, the prepared cell lysate was mixed with protein A/G beads coated with an antibody (2 μg) and incubated overnight at 4°C with rotation. To isolate RNA following the immunoprecipitation reaction, the beads were washed six times with RIP buffer, and RNA bound to the antibody‐coated protein A/G beads was then extracted with TRIzol (Invitrogen). Finally, the amount of RNA bound to RNA‐binding proteins was determined by qRT–PCR.

Polysome profiling assay

PLC cells were treated with a final concentration of 100 μg/ml CAP, 100 μg/ml TAP for 20 min and then resuspended in mitochondrial extraction buffer to obtain approximately 1 mg of mitochondria. Mitochondrial extracts was lysed in lysis buffer (10 mM Tris‐Cl [pH 7.5], 100 mM KCl, 20 mM MgCl₂, 1% Triton X‐100, 5 mM β‐mercaptoethanol, 100 μg/ml CAP, 100 μg/ml TAP, protease inhibitor cocktail (EDTA‐free) (Roche), and RNase inhibitor RNasin (Invitrogen)) on ice for 30 min. Lysates were precleared by centrifugation at 13,000 rpm for 15 min at 4°C, then 1 mg lysates were loaded on a 13 ml 10–30% continuous sucrose gradient (10 mM Tris‐Cl [pH 7.5], 100 mM KCl, 20 mM MgCl₂, 5 mM β‐mercaptoethanol, 100 μg/ml CAP, 100 μg/ml TAP) (Liu et al, 2019b). The gradients were centrifuged in a SW41 Ti Rotor (Beckman) at 22,000 g for 4 h at 4°C. Subsequently, the gradients were fractionated using a Biocomp gradient fractionator. Aliquots of each of 14 fractions were used for Western blot analysis of mitoribosomal proteins. The rest was used for RNA extraction, and the mt‐mRNAs were analyzed by cDNA synthesis followed by qRT–PCR.

In vitro GST pull‐down assay

The cDNAs encoding PHGDH and mtEFG2 were inserted into the pET‐22b (+) vector, and E. coli (Rosetta (DE3) pLySs strain) transformed with His‐PHGDH or His‐mtEFG2 vectors was then induced to express His‐PHGDH or His‐mtEFG2 proteins by incubation with 0.1 mM IPTG for 4 h at 37°C. The cDNAs encoding PHGDH and ANT2 were inserted into the pGEX‐4T‐1 (GE) vector, and E. coli (Rosetta (DE3) pLySs strain) transformed with GST‐PHGDH or GST‐ANT2 vectors was then induced to express GST‐PHGDH or GST‐ANT2 proteins by incubation with 0.1 mM IPTG for 4 h at 37°C.

Purified GST fusion proteins and His‐tagged proteins were used for pull‐down assays in PBST buffer (137 mM NaCl, 2.7 mM KCl, 10 mM Na₂HPO₄, 2 mM KH₂PO₄, 0.1% Tween 20, and 2 mM EDTA). After incubation, the beads were pelleted and washed with PBST buffer prior to elution of proteins and Western blot analysis using an anti‐His or anti‐GST antibody.

Proliferation assay

Direct cell counting was carried out by plating cells in triplicate in 12‐well plates at 10,000 cells per well in 2 ml of medium on day 0. On Days 2, 4, and 6, dead cells were excluded with trypan blue staining (Invitrogen, 25‐900‐CI), and the number of live cells was determined using a hemocytometer under a light microscope.

Animal studies

All animal studies were conducted with approval from the Animal Research Ethics Committee of the University of Science and Technology of China (USTCACUC1701025). Male BALB/c nude mice (GemPharmatech Animal Company, China) were kept at room temperature under a constant 12 h light/dark cycle. Mice were held five per cage and randomly assigned to experimental groups. For xenograft experiments, 8 × 10⁶ PLC cells stably expressing EV or PHGDH with or without mtEFG2 knockdown were injected subcutaneously into 5‐week‐old male nude mice. For another xenograft experiment with tigecycline treatment, 8 × 10⁶ PLC cells stably expressing EV or PHGDH were subcutaneously injected into 5‐week‐old male nude mice, while 100 mg/kg tigecycline was injected intraperitoneally every day beginning on Day 14. Tigecycline was dissolved in saline solution (0.8% NaCl), and solutions were freshly prepared from dry powder just before each injection. Tumors were measured using digital calipers every 3 days, and tumor volumes were calculated using the following formula: length (mm) × width (mm) × depth (mm) × 0.52.

Clinical specimens

Fresh HCC tissues and corresponding noncancerous tissues that were at least 2 cm distant from the edge of tumors were collected from HCC patients in the First Affiliated Hospital of University of Science and Technology of China. Total protein and different cellular components including nucleus, cytoplasm, and mitochondria were extracted from paired HCC and noncancerous tissues and then detected by Western blot. For using the clinical materials for research purposes, prior written informed consent from the patients and study approval from the Institutional Research Ethics Committee of The First Affiliated Hospital of University of Science and Technology of China were obtained.

Immunofluorescence analysis

PLC cells were cultured on coverslips, incubated for 15 min with 300 nM MitoTracker™ Red FM dye (Molecular Probes, M22426), washed three times with phosphate‐buffered saline (PBS), fixed with 4% paraformaldehyde, and stained with an antibody against PHGDH (Proteintech, 14719‐1‐AP). After another three washes in PBS, staining was performed with an anti‐Alexa Fluor 488 antibody (Invitrogen, A‐11094) for 1 h at room temperature. Nuclei were labeled with DAPI. Finally, the cells were imaged with ZEISS LSM 980 laser scanning confocal microscopy.

BN‐PAGE

Isolated mitochondria were solubilized by incubation of the mitochondrial suspension in ice‐cold BN lysis buffer (20 mM imidazole, 500 mM 6‐amino‐hexanoic acid, 20 mM NaCl, 2 mM EDTA [pH 8.0], 10% glycerol, 5% digitonin, 5% dodecyl maltoside, and 1 mM PMSF) for 30 min, centrifuged, and loaded onto a 6–16% gradient blue native electrophoresis gel. For the in‐gel activity (IGA) assay, Complex I IGA (CI‐IGA) was visualized by incubating the gel with 2.5 mg/ml nitrotetrazolium blue (NBT) and 0.1 mg/ml NADH in 20 mmol/l Tris–HCl (pH 7.4) at room temperature for 60 min. Complex IV IGA (CIV‐IGA) was visualized by incubating the gel with 50 mmol/l 3,3′‐diaminobenzidine (DAB), 0.1% (w/v) cytochrome c, and 24 U/ml catalase in 50 mmol/l Tris–HCl (pH 7.4) at 37°C for 3–6 h. Complex V IGA (CV‐IGA) was visualized by incubating the gel with the following steps: two washes in water; incubation in 35 mmol/l Tris, 270 mmol/l glycine, 14 mmol/l MgSO₄, 8 mmol/l ATP; adjustment of the pH to 7.8; addition of 0.2% Pb(NO₃)₂; and incubation in reaction mix at 37°C for 3–6 h (Zhu et al, 2018).

Oxygen consumption measurements

The oxygen consumption rate (OCR) of PLC cells was measured using an XFe96 Extracellular Flux Analyzer (Seahorse Bioscience, North Billerica, MA, USA) according to the manufacturer's instructions. Briefly, 10,000 cells per well were plated in a 96‐well XF cell culture microplate the night before the experiment. The OCR was measured with an XF96 analyzer in XF base medium containing 2 mM glutamine, 10 mM glucose, and 1 mM pyruvate (pH 7.4) following sequential addition of oligomycin (10 μM), FCCP (10 μM), and rotenone/antimycin A (5 μM). Data were analyzed with a Seahorse XF Cell Mito Stress Test Report Generator.

Statistical analysis

Data are presented as the mean (± SD) of three independent experiments unless otherwise noted. Group differences (P < 0.05) were analyzed by two‐tailed Student's t‐test or one‐way ANOVA.

Author contributions

Ying Shu: Data curation; formal analysis; investigation; methodology; writing—original draft. Yijie Hao: Formal analysis; methodology. Junru Feng: Data curation; formal analysis; visualization. Haiying Liu: Investigation. Shi‐ting Li: Investigation. Jiaqian Feng: Investigation. Zetan Jiang: Investigation. Ling Ye: Investigation. Yingli Zhou: Investigation. Yuchen Sun: Investigation. Zilong Zhou: Investigation. Haoran Wei: Investigation. Ping Gao: Conceptualization; supervision; funding acquisition; project administration; writing—review and editing. Huafeng Zhang: Supervision; funding acquisition; project administration; writing—review and editing. Linchong Sun: Conceptualization; data curation; formal analysis; supervision; funding acquisition; writing—review and editing.

Disclosure and competing interests statement

The authors declare that they have no conflict of interest.

Supporting information

Expanded View Figures PDF

Table EV1

Table EV2

Table EV3

Table EV4

Source Data for Expanded View

Source Data for Figure 1

Source Data for Figure 2

Source Data for Figure 3

Source Data for Figure 4

Source Data for Figure 5

PDF+

The EMBO Journal (2022) 41: e111550

Data availability

The immunoprecipitation mass spectrometry data of PHGDH‐binding proteins relating to Table EV1 have been deposited at PRIDE (Perez‐Riverol et al, 2022) hosted at the EBI (https://www.ebi.ac.uk/pride/archive/projects/PXD036321/). Data are available via ProteomeXchange with identifier PXD036321.