Nqo1 ablation inhibits activation of the PI3K/Akt and MAPK/ERK pathways and blocks metabolic adaptation in hepatocellular carcinoma

Abstract

Cancer cells undergo metabolic adaptation to sustain uncontrolled proliferation. Aerobic glycolysis and glutaminolysis are two of the most essential characteristics of cancer metabolic reprogramming. Hyper-activated PI3K/Akt and MAPK/ERK signaling pathways play central roles in cancer cell metabolic adaptation since their downstream effectors, such as Akt and c-Myc, control most of the glycolytic and glutaminolysis genes. Here, we report that the cytosolic flavoprotein Nqo1 is strongly overexpressed in mouse and human hepatocellular carcinoma (HCC). Knockdown of Nqo1 enhanced activity of the serine/threonine phosphatase PP2A, which operates at the intersection of the PI3K/Akt and MAPK/ERK pathways and dephosphorylates and inactivates PDK1, Akt, Raf, MEK and ERK1/2. Nqo1 ablation also induced the expression of PTEN, a dual protein/lipid phosphatase that blocks PI3K/Akt signaling, via the ERK/CREB/c-Jun pathway. Together, Nqo1 ablation triggered simultaneous inhibition of the PI3K/Akt and MAPK/ERK pathways, suppressed the expression of glycolysis and glutaminolysis genes and blocked metabolic adaptation in liver cancer cells. Conversely, Nqo1 overexpression caused hyper-activation of the PI3K/Akt and MAPK/ERK pathways and promoted metabolic adaptation. In conclusion, Nqo1 functions as an upstream activator of both the PI3K/Akt and MAPK/ERK pathways in liver cancer cells, and Nqo1 ablation blocked metabolic adaptation and inhibited liver cancer cell proliferation and HCC growth in mice. Therefore, our results suggest that Nqo1 may function as a therapeutic target to inhibit liver cancer cell proliferation and inhibit HCC.

Introduction

The most fundamental feature of cancer cells is their ability to sustain uncontrolled proliferation via disruption of proliferative signaling pathways and loss of cell cycle regulation (1). Due to nonstop proliferation, cancer cells face a substantial bio-energetic challenge to produce sufficient energy and synthesize various biomolecules concurrently. Inevitably, this forces cancer cells to readapt their energy metabolism and undergo metabolic reprogramming, one of the central hallmarks of cancer (1, 2). A large number of cancer cells such as breast, colon, liver, lung, pancreas, prostate, thyroid, and bladder undergo metabolic adaptation (2, 3). The most well-known example of metabolic adaptation in cancer cells is the alteration of glucose metabolism. Irrespective of oxygen status, many cancer cells depend on high rates of glucose uptake and convert most of the glucose into lactate instead of completely catabolizing glucose by oxidative phosphorylation (OXPHOS), a phenomenon known as the Warburg effect (2, 3). Due to high glycolytic activity, the expression and activity of numerous glycolytic enzymes such as HK1, PFK1, enolase, LDH and PDK1 are highly elevated in cancer cells (2–4). In addition to aerobic glycolysis, glutaminolysis is also an essential characteristic of cancer metabolic reprogramming (5). Glutamine enters cells through SLC38A5 and SLC1A5 transporters and is converted to glutamate by glutaminase (GLS). Glutamate is converted to alpha-ketoglutarate (α-KG) by glutamate dehydrogenase, which enters the TCA cycle. Overall, cancer cells need to efficiently coordinate glycolysis and glutaminolysis to satisfy both bioenergetic and biosynthetic requirements for their proliferation and survival (2–4).

The signal transduction pathways that regulate aerobic glycolysis and glutaminolysis in cancer cells have been under extensive investigation. Accumulating evidence demonstrates that metabolic adaptation of cancer cells is a primary function of dysregulated, hyper-activated proliferative signaling pathways and loss of tumor suppressor functions (2, 4). Growth factors usually stimulate two key signal-transducing kinase pathways: the Ras/Raf/MAP kinase (MAPK/ERK) pathway and the phosphatidylinositol 3-kinase (PI3K) pathway. Most cancers harbor activating mutations of the master regulators (Ras, Raf, the p110a PI3K subunit, and RTKs) or their downstream effectors (Akt, PDK1 and c-Myc) of the PI3K/Akt and MAPK/ERK pathways. The constitutive hyper-activation of the PI3K/Akt and MAPK/ERK signaling pathways can cause cancer-associated metabolic reprogramming (1, 6, 7). For example, activated PI3K/Akt leads to increased glucose transporter expression, activation of hexokinase and PFK1 leading to enhanced glucose uptake and glycolysis (8). On the other hand, c-Myc is one of the most frequently overexpressed genes in human cancers, and it promotes metabolic adaptation in cancer cells (9, 10). c-Myc regulates an array of genes that participate in both glucose and glutamine transport and metabolism, such as HK2, PFK, ENO1, GAPDH, PGK1, LDHA, PDK1, Glut1, glutaminase and the glutamine transporters, thereby exerting tremendous influence on cancer cell metabolic adaptation (9–12). Tumor suppressors, such as p53 and PTEN, negatively regulate activation of the PI3K/Akt and MAPK/ERK pathways and suppress metabolic adaptation (6, 13, 14). However, they are often inactivated or lost in most cancers, making them unable to block metabolic reprogramming. Therefore, identification of factors that operate upstream of the PI3K/Akt and MAPK/ERK pathways could function as potential molecular targets to block cancer cell metabolic adaptation and inhibit tumor growth.

NAD(P)H Quinone Dehydrogenase 1 (Nqo1) is a cytosolic flavoprotein, and its expression is strongly induced in various cancers such as breast, pancreas, liver, bladder, ovary, thyroid, colorectal, cholangiocarcinoma, uterine cervical cancer, melanoma, and lung cancer (15–19). Increased expression of Nqo1 protein was also correlated with tumor size, advanced clinical stage and decreased patient survival rates in breast, lung and liver cancers (17, 19, 20), suggesting that Nqo1 could function as an oncogene. Nevertheless, how Nqo1 exerts its oncogenic effects, and through which pro-tumorigenic signaling pathways Nqo1 controls cancer cell proliferation, is not fully understood. Here, we have unraveled a novel function of Nqo1 in the regulation of the PI3K/Akt and MAPK/ERK pathways and metabolic adaptation in liver cancer cells.

Experimental procedures

Mice

All animal protocols were reviewed and approved by the Institutional Animal Care and Use Committee of Augusta University, Georgia. Mice were bred and maintained under specific pathogen-free conditions with a 12 h light-dark cycle. Nqo1^+/− mouse frozen sperm were obtained from Dr. Minho Shong, Korea Research Institute of Bioscience and Biotechnology, South Korea. Using the Sperm Cryorecovery services of the Jackson Laboratory (project # 109311), we derived Nqo1^+/− mice. Nqo1^−/− and Nqo1^+/+ litter-mate controls were generated by crossing Nqo1^+/− mice. A single intra-peritoneal injection of DEN to 15-day-old mice at a dose of 25mg/Kg body weight was given to induce liver tumors in male mice.

Human HCC samples

To analyze the expression of Nqo1 in human normal liver and in liver tumor tissues, 25 formalin-fixed, paraffin-embedded cases of liver cancer (American Joint Committee on Cancer stages I–IV) and 15 normal control liver samples were obtained from the pathology archives of Augusta University. The study protocol was reviewed and approved by the Institutional Review Board.

In vivo xenograft assay

Huh7 control and Nqo1-KD cells (6 X 10⁶) were injected subcutaneously in the flank region of athymic nude mice (strain code # 490, Charles River Laboratories), and tumor growth was monitored daily. Mice were sacrificed after one month. Tumor weight and diameter were measured at the time of harvesting. A digital Vernier caliper was used to measure the tumor diameter.

Statistical analysis

All experiments were performed in triplicate. Statistical analyses were performed by 2-tailed Student’s t test, and a value of p<0.05 was considered statistically significant.

Results

Nqo1 is strongly upregulated in HCC

To discover novel genes that are involved in liver tumorigenesis, we performed RNA-Sequencing (RNA-Seq) by comparing non-tumorous normal livers versus diethylnitrosamine (DEN) chemical-induced liver tumors in mice. After applying a fold-change of 4 (FC>4) and FDR<0.001 (false discovery rate) and ranking the genes based on significance (p value), we identified 318 downregulated and 648 upregulated genes in liver tumors compared to non-tumorous tissues (Fig 1A, Suppl. Tables 1 & 2). Out of 648 upregulated genes, NAD(P)H Quinone Dehydrogenase 1 (Nqo1) ranked at 138 with a 14.64 fold upregulation (Fig 1A–C, Suppl. Table 1). We selected Nqo1 for further investigation since it fulfilled all the other criteria we have set forth such as: A) The Human Protein Atlas database showed strong Nqo1 expression in ~40–50% of human liver tumors and no expression in normal livers. This information is either not available for several genes that ranked above Nqo1 or they are not strongly expressed. B) Re-confirmation of RNA-Seq data by Western blotting revealed strong expression of Nqo1 in mouse liver tumors (Fig 1D). C) Liver cancer cell lines HepG2 and Huh-7 showed strong expression of Nqo1 where as its expressed is very low in H2.35 primary hepatocyte cell line (Fig 1E). D) An extensive literature search revealed no previous studies showing the specific function of Nqo1 in HCC. E) Increased expression of Nqo1 protein was correlated with tumor size, advanced clinical stage and decreased patient survival rate in liver cancers (20). Genes that ranked above Nqo1 did not meet one or more of the above criteria, whereas Nqo1 fulfilled all the criteria providing a strong rationale for thoroughly investigating its function in HCC.

Figure 1. Upregulation of Nqo1 in HCC tissues and in cell lines.

(A) Heat map. RNA sequencing analysis of mouse non-tumorous (NT) versus liver tumor (T) tissues (n=3), Green – downregulated and Red – upregulated genes. A 4-fold differential expression with a false discovery rate (FDR) of 0.001 were applied. With that criteria, 318 genes were downregulated and 648 genes were upregulated in liver tumors compared to normal livers. (B) Magnified heat map showing Nqo1 differential expression in 3 independent liver tumors compared to normal livers. (C) Nqo1 expression is 14.64 fold upregulated in liver tumors compared to non-tumorous tissues. (D) Western blot showing the expression level of Nqo1 in normal mouse liver and in DEN-induced liver tumors, (n=5). (E) Expression of Nqo1 protein in primary hepatocyte cell line H2.35 and human liver cancer cell lines HepG2 and Huh-7. (F–G) Representative Nqo1 IHC showing no staining on human non-tumorous tissues (F) but strong intense cytoplasmic staining (G) in HCC samples. Approximately 60% of HCC samples stained strongly positive for Nqo1. (H) Representative human HCC sample showing no Nqo1 staining. Approximately 40% of HCC samples displayed very low to no detectable Nqo1 staining. Scale bar 300µm. (I–J) Representative Nqo1 IHC staining in paired human HCC and surrounding normal liver tissue. Liver tumor tissues (T) showed strong Nqo1 staining whereas non-tumorous (NT) areas are mostly negative. Scale bar 200µm. (K) Nqo1 staining intensity was determined semi-quantitatively and was graded as absent (0), weak (1), strong (2) and very strong (3), n=14, ***p<0.0005. (L–M) Representative Nqo1 IHC showing weak staining in the sinusoidal cells of mouse non-tumorous tissues (L) but strong intense cytoplasmic staining (M) in all DEN-induced HCC samples, n=6. Scale bar 100µm.

To determine if strong expression of Nqo1 in mouse liver tumors translate into the human condition, we analyzed Nqo1 expression in human HCC samples. Non-tumorous liver tissues (15 samples) did not show any significant Nqo1 staining (Fig 1F). Out of 25 HCC samples stained, we detected strong and predominantly cytosolic expression of Nqo1 in 14 (~60%) HCC samples (Fig 1G–J). All the 14 samples that stained positive for Nqo1 predominantly showed strong (IHC score 2) to very strong (IHC score 3) staining intensity (Fig 1K). Additionally, we also performed IHC on DEN-induced mouse HCC samples and again detected predominantly strong cytosolic staining in all 6 mouse HCC samples, whereas non-tumorous tissues showed weak Nqo1 staining in the sinusoidal cells (Fig 1L–M). These observations suggest that Nqo1 might play a role in HCC initiation and/or progression.

Nqo1 knockdown (Nqo1-KD) suppressed liver cancer cell proliferation, colony formation and xenograft growth

Strong expression of Nqo1 in liver tumors and in liver cancer cell lines raises the question if Nqo1 is required for the proliferation, survival and/or migration of liver cancer cells. Hence, we knocked down Nqo1 in Huh-7 cells using shRNA (Fig 2A). Nqo1-KD cells proliferated at a significantly slower rate compared to control cells (Fig 2B), and we did not detect any significant cell death in Nqo1-KD cell cultures. Nqo1-KD cells also displayed a decreased migration rate (Fig 2C–D). Moreover, in vitro colony formation ability of Nqo1-KD cells was also significantly impaired (Fig 2E–F). To further investigate if Nqo1 is required for tumor growth in vivo, we performed subcutaneous xenografts in athymic nude mice, which revealed a significant reduction in xenograft tumor growth of Nqo1-KD cells compared to control cells (Fig 2G–J). These observations suggest that Nqo1 is required for liver cancer cell proliferation and tumor growth in vivo.

Figure 2. Nqo1 is required for liver cancer cell proliferation, colony formation and xenograft growth.

(A) Western blot showing the expression of Nqo1 protein in scramble and Nqo1 knockdown (Nqo1-KD) Huh-7 cells. Out of 5 shRNAs tested, 3 shRNAs efficiently knocked down Nqo1. shRNA-1 was used to generate Nqo1-KD stable cells. (B) Relative proliferation of Nqo1-KD Huh-7 cells was significantly reduced compared to control cells as measured by cell counting method for 6 days, n=3, *p<0.05, **p<0.005, ***p<0.0005. (C–D) Representative photos showing the migration (C) and migration area (D) of scramble and Nqo1-KD Huh-7 cells, n=3, **p<0.005, ***p<0.0005. (E–F) Crystal violet-stained colonies formed by control and Nqo1-KD Huh-7 cells 21 days after plating (E); colony formation ability of Nqo1-KD cells was significantly impaired compared to control cells (F), n=3, **p<0.005. (G) Representative photos showing significantly reduced xenograft tumor growth of Nqo1-KD Huh-7 cells compared to control cells. (H) Western blot showing the expression level of Nqo1 in scramble and Nqo1-KD xenografts. (I–J) Xenograft tumor volume (I) and tumor weight (J) were significantly reduced in Nqo1-KD cells compared to scramble cells, n=5, ***p<0.0005. All the quantitative data are represented as mean ± SD.

DEN-induced HCC is strongly inhibited in Nqo1 knockout (Nqo1^−/−) mice

To study the impact of loss of Nqo1 on HCC growth in a more physiologically relevant system, we utilized Nqo1^+/+ control and Nqo1^−/− mice. Nqo1^−/− mice are normal and fertile, and did not show any significant phenotypes. We performed the well-established DEN-induced HCC in Nqo1^+/+ and Nqo1^−/− male mice. We detected a significantly reduced number of liver tumors in Nqo1^−/− compared to Nqo1^+/+ mice 8 months after DEN treatment (Fig 3A). Tumor incidence as well as tumor size were significantly reduced in Nqo1^−/− compared to Nqo1^+/+ mice (Fig 3B–D). Consistent with reduced liver tumor growth, the liver/body weight ratio was not increased in Nqo1^−/− mice (Fig 3E). These results demonstrate that absence of Nqo1 significantly inhibited HCC growth. Histological analysis of liver tumors revealed large tumor cells with hyperchromatic nuclei in a compact growth pattern in Nqo1^+/+ mice, whereas Nqo1^−/− livers appeared normal (Fig 3F). We also detected reduced lipid accumulation (Fig 3G) and fibrosis (Fig 3H) in Nqo1^−/− compared to Nqo1^+/+ mouse livers. Nqo1^−/− liver tumors also displayed significantly reduced proliferation as revealed by PCNA staining (Fig 3I), indicating that reduced tumor growth in Nqo1^−/− mice was due to decreased proliferation. Overall, these results indicate that absence of Nqo1 inhibits liver tumor growth.

Figure 3. DEN-induced HCC is inhibited in Nqo1^−/− mice.

(A) Representative photographs showing liver tumor growth in Nqo1^+/+ and Nqo1^−/− mice 8 months after DEN treatment. (B) Body weight, (C) average number of tumors per liver, (D) Percentage of tumor incidence and liver tumor size and (E) Liver/body weight ratio in Nqo1^+/+ and Nqo1^−/− mice. Liver tumor size in (D) was categorized into 3 groups; small (<2mm), medium (2–5mm) and large (>5mm), n=9, **p<0.005, ***p<0.0005. (F–I) Representative H&E (F), Oil-Red-O (G), and Sirius-Red (H) stained images of Nqo1^+/+ and Nqo1^−/− mouse livers 8 months after DEN treatment. (I) IHC staining of PCNA on the liver tumors of Nqo1^+/+ and Nqo1^−/− mice. Insert showing the percentage of PCNA-positive cells in the liver tumors of Nqo1^+/+ and Nqo1^−/− mice, (n=9, ***p<0.0005). All the quantitative data are represented as mean ± SD.

Nqo1-KD resulted in strong upregulation of PTEN, impaired activation of Akt and ERK1/2 and downregulation of c-Myc

Our gene expression analysis revealed 648 upregulated and 318 downregulated genes in liver tumors compared to normal livers (Fig 1A, Suppl. Table 1–2). Transcriptome Analysis Console (TAC) pathway mapping on these 966 differentially regulated genes revealed that 77 differentially expressed genes belong to the PI3K/Akt/mTOR and MAPK/ERK signaling pathways (Suppl. Fig 3). These two signaling pathways ranked in the top 5 dysregulated pathways in liver tumors based on the total number of differentially expressed genes in each pathway (Suppl. Fig 3). Since Nqo1 is significantly overexpressed (14.64 fold) in liver tumors and its knockdown strongly inhibited liver cancer cell proliferation and tumor growth (Fig 2–3), we wondered if Nqo1 induces its oncogenic effects via some of the major pro-oncogenic signaling pathways such as the PI3K/Akt/mTOR and MAPK/ERK pathways. To investigate this possibility, we analyzed the expression and/or activity status of some of the critical components of these two pathways in control and Nqo1-KD Huh-7 cells. Very fascinatingly, we detected strong upregulation of PTEN and impaired phosphorylation and activation of PDK1 and Akt, whereas mTOR levels were not significantly altered in response to Nqo1-KD (Fig 4A). These observations suggest that Nqo1 is required for activation of the PI3K/Akt pathway. PTEN is a dual protein/lipid phosphatase that blocks PI3K/Akt signaling by dephosphorylating phosphatidyl-inositol,3,4,5 triphosphate (PIP3) to PIP2 (1, 6, 7, 13). To evaluate if Nqo1-KD-associated PTEN upregulation affected PIP2/PIP3 levels, we measured and detected significantly elevated cellular PIP2 levels in Nqo1-KD cells compared to control cells (Fig 4B), further indicating that Nqo1-KD upregulates PTEN and blocks activation of the PI3K/Akt pathway.

Figure 4. Enhanced PP2A activity and inactivation of PI3K/Akt and MAPK/ERK pathways in response to Nqo1-KD.

(A) Western blots showing expression and activity levels of the indicated proteins in scramble and Nqo1-KD Huh-7 cells. (B) PIP2 levels in scramble and Nqo1-KD Huh-7 cells, n=3, **p<0.005. (C) Western blots showing expression and activity status of the indicated proteins in scramble and Nqo1-KD Huh-7 cells. (D) Western blots showing the expression and activity levels of the indicated proteins in Nqo1^+/+ and Nqo1^−/− mouse liver tumors. (E) Expression levels of p53, AMPK and the sub-units of PP2A in scramble and Nqo1-KD cells. (F) PP2A enzymatic activity in scramble and Nqo1-KD Huh-7 cells, n=4, ***p<0.0005. (G) PP2A enzymatic activity in scramble and Nqo1-KD Huh-7 cells after treatment with 10µM of PP2A inhibitor LB-100 for 48 h, n=3, * p<0.05, **p<0.005. (H) Western blots showing the expression and activity levels of the indicated proteins in scramble and Nqo1-KD Huh-7 cells after treatment with 10µM of LB-100 for 48 h. Red * in Western blots indicates strong detectable difference. All the quantitative data are represented as mean ± SD.

Next, to investigate if Nqo1 also regulates the MAPK/ERK pathway, we analyzed the activity status of some of the main components of this pathway. Very interestingly, Nqo1-KD resulted in a significant reduction in the phosphorylation and activation of MEK1 and ERK1/2 and reduced inhibitory phosphorylation at Ser9 of GSK3β (active form) (Fig 4C), suggesting that Nqo1 is required for stimulation of the MAPK/ERK pathway, and in the absence of Nqo1, MAPK/ERK activation is blunted. Both the PI3K/Akt and MAPK/ERK pathways regulate c-Myc expression and stability (21). Since Nqo1-KD blunted the activation of both of these pathways, consequently, we detected significant downregulation of c-Myc (Fig 4A). To determine if these in vitro observations translate into in vivo, we analyzed the expression and/or activity status of the components of the PI3K/Akt and MAPK/ERK pathways in Nqo1^+/+ and Nqo1^−/− mouse liver tumors. Again, we detected upregulation of PTEN, and impaired phosphorylation and activation of PDK1, Akt, MEK1 and ERK1/2, and downregulation of c-Myc in Nqo1^−/− compared to Nqo1^+/+ liver tumors (Fig 4D). Overall, our observations suggest that Nqo1 is required for the activation of both the PI3K/Akt and MAPK/ERK signaling pathways, and in the absence of Nqo1, activation of these pathways are blunted.

Nqo1-KD enhanced the activity of PP2A, which is a negative regulator of the PI3K/Akt and MAPK/ERK pathways

There is a substantial amount of cross-regulation between the PI3K/Akt and MAPK/ERK signaling pathways, and at several critical regulatory points these pathways also intertwine with other signaling pathways (22, 23). Some of the factors that are involved in the cross-regulation of the PI3K/Akt and MAPK/ERK pathways are PTEN, P53, AMPK, GSK3β and PP2A (22, 23). To evaluate if Nqo1 regulates any of these factors, we analyzed their expression which revealed enhanced expression of PTEN in Nqo1-KD cells (Fig 4A and D). PTEN is not only a negative regulator of the PI3K/Akt pathway, but also inhibits the MAPK/ERK pathway in certain cell types (22, 24). We did not detect any significant difference in the expression of p53, AMPK and pAMPK (Fig 4E). We then examined if Nqo1 also influences the expression and/or activity of the serine/threonine phosphatase PP2A, one of the strongest molecular links between the PI3K/Akt and MAPK/ERK pathways (25–28).

PP2A comprises a scaffolding subunit (A), a catalytic subunit (C) and a regulatory subunit (B) that provides functional specificity to the PP2A hetero-trimer. PP2A operates at the intersection of the PI3K/Akt and MAPK/ERK pathways, and dephosphorylates and inactivates PDK1, Akt, Raf, MEK and ERK1/2, thereby shutting down these pathways simultaneously (28, 29). Analysis of PP2A subunit levels revealed no difference between control and Nqo1-KD cells (Fig 4E). Hence, we deliberated whether Nqo1-KD affected PP2A activity directly or indirectly, which in turn caused reduced activation of MEK, ERK and Akt. Analysis of PP2A enzymatic activity revealed significantly enhanced PP2A phosphatase activity in Nqo1-KD cells (Fig 4F), suggesting that Nqo1 interferes with PP2A activity, and absence of Nqo1 causes enhanced PP2A activity, leading to reduced phosphorylation of MEK, ERK and Akt. To further validate that enhanced PP2A activity in Nqo1-KD cells was responsible for reduced phosphorylation of the kinases in the PI3K/Akt and MAPK/ERK pathways, we treated Nqo1-KD cells with different concentrations of the PP2A inhibitor LB-100 for 24 h (Suppl. Fig 5A). The phospho PDK1, Akt, MEK1 and ERK1/2 levels were increased in Nqo1-KD cells in response to LB-100 treatment in a concentration-dependent manner (Suppl. Fig 5B). Since we detected higher PP2A activity in Nqo1-KD cells compared to control cells (Fig 4F), we asked if we reduce PP2A activity in Nqo1-KD cells to control cell levels will it rescue phosphorylation of kinases in the PI3K/Akt and MAPK/ERK pathways to control cell levels. Thus, we treated Nqo1-KD cells with LB-100 (10µM) for 48 h, which reduced PP2A activity in Nqo1-KD cells to approximately equal to control cell levels (Fig 4G). Phospho PDK1, Akt, MEK1/2 and ERK1/2 in LB-100 treated Nqo1-KD cells was significantly rescued and almost comparable to control cells (Fig 4H). These findings suggest that increased activity of PP2A in Nqo1-KD cells contributed to decreased phosphorylation and activation of kinases in the PI3K/Akt and MAPK/ERK pathways.

Nqo1-KD enhanced Pten transcription

PTEN is upregulated in Nqo1-KD Huh-7 cells and in Nqo1^−/− mouse liver tumors (Fig 4A and 4D), suggesting that Nqo1 regulates activation of the PI3K/Akt pathway by controlling PTEN expression. Next, we asked at what level Nqo1 regulates PTEN. Using qPCR, we detected a significant increase in Pten mRNA levels in Nqo1-KD cells (Fig 5A), suggesting that Nqo1 regulates PTEN at the transcriptional level. Since Nqo1 is not a transcription factor, we postulated that Nqo1 may regulate Pten expression via other factors that either induce or suppress Pten transcription. EGFR, PPARγ and p53 induce Pten expression, whereas TGFβ, NFκB, Hes1 and c-Jun bind to the Pten promoter and suppress its expression (30, 31). Analysis of expression levels of some of these factors revealed a significant reduction in c-Jun expression in Nqo1-KD cells (Fig 5B).

Figure 5. Nqo1 regulates Pten expression via ERK/CREB/c-Jun pathway.

(A) Relative expression level of Pten in control and Nqo1-KD Huh-7 cells, n=3, **p<0.005. (B) Western blots showing expression levels of the indicated proteins in scramble and Nqo1-KD Huh-7 cells. (C) According to our proposed model, Nqo1 inhibits or interferes with PP2A activity, leading to constitutive activation of ERK1/2, which phosphorylates and activates CREB. Activated CREB induces c-Jun transcription. c-Jun directly binds to the Pten promoter and suppresses its transcription. (D) Expression levels of indicated proteins in scramble and Nqo1-KD Huh-7 cells. (E) Expression levels of the indicated proteins in Nqo1-KD Huh-7 cells after treatment with 10µM LB-100 for 48h. (F) ChIP-qPCR analysis of c-Jun binding to the Pten promoter in scramble and Nqo1-KD Huh-7 cells after normalizing to 18S binding, n=3, **p<0.005. (G) Western blots showing expression levels of the indicated proteins after expressing vector control or constitutively active form of Creb (VP16-CREB) in Nqo1-KD Huh-7 cells. (H) Relative mRNA expression level of Pten in Nqo1^+/+ and Nqo1^−/− mouse liver tumors. n=6, **p<0.005. (I) Expression levels of the indicated proteins in Nqo1^+/+ and Nqo1^−/− liver tumors. (J) IHC showing strong intense PTEN staining in Nqo1^−/− mouse liver tumor tissues, whereas PTEN expression is very low to undetectable in Nqo1^+/+ liver tumors. (K) Relative mRNA expression level of Pten in Nqo1^+/+ and Nqo1^−/− normal mouse livers, n=4, **p<0.005. (L) Expression level of PTEN in Nqo1^+/+ and Nqo1^−/− normal mouse livers. Red * in Western blots indicates strong detectable difference. All the quantitative data are represented as mean ± SD.

How does Nqo1 regulate Pten transcription via c-Jun? The ERK/CREB/c-Jun pathway is known to inhibit Pten transcription. ERK1/2 phosphorylates and activates CREB, which can induce c-Jun expression. c-Jun directly binds to the Pten promoter and inhibits its transcription (30–32). Thus, we postulated that Nqo1 maintains ERK1/2 in its active phospho form by inhibiting PP2A, which phosphorylates and activates CREB. Activated CREB induces c-Jun expression, which binds to the Pten promoter and inhibits its expression (Fig 5C). In the absence of Nqo1, active PP2A dephosphorylates and inactivates ERK1/2, leading to blunted CREB activation, diminished expression and reduced binding of c-Jun to the Pten promoter, causing elevated expression of Pten. Consistent with this idea, we detected reduced levels of pCREB and c-Jun and elevated expression of PTEN in Nqo1-KD cells (Fig 5D). To evaluate if PP2A-mediated inactivation of CREB is responsible for downregulation of c-Jun and elevated expression of PTEN, we inhibited PP2A activity in Nqo1-KD cells by LB-100, which resulted in increased pCREB and c-Jun levels and reduced expression of PTEN (Fig 5E). In accordance with the reduced levels of c-Jun in Nqo1-KD cells, by CHIP-qPCR, we detected reduced c-Jun binding to the Pten promoter in Nqo1-KD cells (Fig 5F). To further evaluate if the absence of Nqo1 terminated the ERK/CREB/c-Jun pathway, leading to enhanced expression of Pten, we overexpressed the constitutively active form of Creb (VP16-CREB) in Nqo1-KD cells, which caused c-Jun induction and PTEN suppression (Fig 5G), suggesting that Nqo1 controls Pten expression via the ERK/CREB/c-Jun pathway. Similar to our in vitro observations, we also detected increased expression of Pten mRNA (Fig 5H), and reduced expression of pCREB and c-Jun and increased expression of PTEN in Nqo1^−/− compared to Nqo1^+/+ liver tumors (Fig 5I). Moreover, IHC also revealed more intense PTEN staining in Nqo1^−/− compared to Nqo1^+/+ liver tumors (Fig 5J). We detected increased Pten mRNA and protein levels not only in the liver tumors but also in normal livers of Nqo1^−/− compared to Nqo1^+/+ mice (Fig 5K–L). Overall, these observations suggest that PTEN is one of the main downstream targets of Nqo1.

Nqo1-KD suppressed aerobic glycolysis

The downstream effectors of the PI3K/Akt and MAPK/ERK pathways play a central role in cancer cell metabolic reprogramming (1, 6, 7). For example, hyperactivation of Akt promotes aerobic glycolysis and glutaminolysis (4, 33), and cMyc activates the transcription of various glycolytic and glutaminolysis genes (9–11). Since Nqo1 ablation caused significant impairment in the activation of Akt and expression of c-Myc (Fig 4A and D), we wondered whether Nqo1 controls metabolic adaptation in cancer cells, which would explain impaired proliferation of liver cancer cells and HCC growth in response to Nqo1 ablation (Fig 2–3). To investigate this possibility, we first measured glucose uptake and lactate production rates between control and Nqo1-KD Huh-7 cells, which revealed reduced glucose uptake in Nqo1-KD cells (Fig 6A). Consequently, lactate production was significantly reduced in Nqo1-KD cells compared to control cells (Fig 6B). These observations encouraged us to speculate whether Nqo1-KD caused the cells to switch from aerobic glycolysis to OXPHOS. To this end, we measured oxygen consumption rates (OCRs) and detected significantly increased basal OCR and maximal respiratory capacity in Nqo1-KD cells compared to control cells (Fig 6C). Nqo1-KD cells also displayed reduced extracellular acidification rate (ECAR) (Fig 6D), an index of lactate production. These observations suggest that Nqo1-KD suppresses aerobic glycolysis and promotes OXPHOS in liver cancer cells.

Figure 6. Nqo1-KD caused downregulation of glycolysis and glutaminolysis genes and suppressed aerobic glycolysis.

(A) Glucose uptake in control and Nqo1-KD Huh-7 cells, n=3, ***p<0.0005. (B) Cellular lactate levels in control and Nqo1-KD Huh-7 cells, n=3, ***p<0.0005. (C–D) Oxygen consumption rate (OCR) and extracellular acidification rate (ECAR) of control and Nqo1-KD cells measured by Seahorse Bioscience XF96 analyzer, n=3, **p<0.005, ***p<0.0005. (E) Representative flow cytometry analysis showing no significant difference in mitochondrial content between control and Nqo1-KD cells. n=3. (F-G) Relative mRNA expression levels of glycolytic genes (F), and glutamine transporters (G) in control and Nqo1-KD cells, n=3, *p<0.05, **p<0.005. (H) Western blots showing expression and activity levels of the indicated proteins in scramble and Nqo1-KD Huh-7 cells. (I) Hexokinase activity levels in control and Nqo1-KD cells, n=3, **p<0.005. (J) Expression levels of the indicated proteins in Nqo1^+/+ and Nqo1^−/− liver tumors. (K–L) Glucose and lactate levels in Nqo1^+/+ and Nqo1^−/− liver tumors, n=6, **p<0.005. Red * in Western blots indicates strong detectable difference. All the quantitative data are represented as mean ± SD.

To investigate if increased O₂ consumption in Nqo1-KD cells is due to an increase in mitochondrial content, we measured and observed no clear differences in mitochondrial mass and expression levels of the mitochondrial marker protein Cyt C between control and Nqo1-KD cells (Fig 6E, H), suggesting that Nqo1-KD could increase OXPHOS without affecting total mitochondrial content. To determine if glycolytic activity was indeed reduced in response to Nqo1-KD, we analyzed the relative expression levels of glycolytic genes by qPCR and detected significant downregulation of HK2, Glut1, PKM2, aldolase-A, enolase-1, enolase-2, LDHA and PDHK1 mRNA in Nqo1-KD cells (Fig 6F). We also detected significantly reduced expression of glutaminase and glutamine transporters SN1, SN2 and SLC1A5 at the mRNA level in Nqo1-KD cells compared to control cells (Fig 6F–G). Further analysis at the protein level also revealed downregulation of glycolytic enzymes such as HK2, aldolase A, enolase-1, enolase-2 and glutaminase in Nqo1-KD cells (Fig 6H). Consistent with the reduced expression, HK2 enzymatic activity was significantly reduced in Nqo1-KD cells compared to control cells (Fig 6I). Not only Nqo1-KD Huh-7 cells but also Nqo1^−/− liver tumors displayed reduced expression of glycolytic enzymes and GLS (Fig 6J), and glucose and lactate levels were significantly less in Nqo1^−/− compared to Nqo1^+/+ liver tumors (Fig 6K–L). Overall, these results suggest that Nqo1 ablation caused reduced expression of glycolytic and glutaminolysis genes, leading to suppression of aerobic glycolysis and glutaminolysis.

Nqo1 overexpression (Nqo1-OE) downregulated PTEN, activated Akt and ERK1/2, and induced c-Myc expression, leading to enhanced aerobic glycolysis

To further validate if Nqo1 is, in fact, promoting metabolic adaptation in liver cancer cells by activating the PI3K/Akt and MAPK/ERK pathways, we transiently overexpressed human Nqo1 in Huh-7 cells. Nqo1-OE cell proliferation is increased (Fig 7A), and glucose uptake and lactate production are also increased in Nqo1-OE cells compared to control cells (Fig 7B–C). Nqo1-OE cells also displayed reduced OCR and increased ECAR compared to control cells (Fig 7D–E). These observations suggest that Nqo1-OE enhances aerobic glycolysis in liver cancer cells. To determine if increased aerobic glycolysis in response to Nqo1-OE is due to activation of the PI3K/Akt and MAPK/ERK pathways, we analyzed and detected significantly diminished expression of PTEN and increased expression of c-Jun (Fig 7F), which suppresses Pten expression. We also observed increased activation of ERK and Akt and induced expression of c-Myc and glycolytic genes in Nqo1-OE cells compared to control cells (Fig 7F). These results further confirm that Nqo1 promotes metabolic adaptation in liver cancer cells by simultaneously activating the PI3K/Akt and MAPK/ERK pathways. Based on these observations, we questioned if enhanced expression of Nqo1 is sufficient to stimulate the PI3K/Akt and MAPK/ERK pathways and drive metabolic adaptation even in the primary hepatocytes. We overexpressed Nqo1 in H2.35 primary hepatocytes and observed induction of pCREB and c-Jun and downregulation of PTEN, and increased expression of pERK1/2, pAkt and c-Myc (Fig 7G). Nqo1-OE H2.35 cells also displayed elevated expression of glycolytic enzymes (Fig 7H) and increased glucose uptake and lactate production (Fig 7I–J). Also, Nqo1-OE cells displayed reduced OCR and increased ECAR compared to control cells (Fig 7K–L). These observations suggest that Nqo1-OE is sufficient to enhance aerobic glycolysis and drive metabolic adaptation in primary hepatocytes.

Figure 7. Nqo1-OE caused upregulation of glycolysis and glutaminolysis genes and enhanced aerobic glycolysis.

(A) Proliferation rate of Nqo1-OE Huh-7 cells was significantly increased compared to control cells as measured by cell counting method for 6 days, n=3, *p<0.05, **p<0.005. (B) Glucose uptake in control and Nqo1-OE Huh-7 cells, n=3, **p<0.005. (C) Cellular lactate levels in control and Nqo1-OE cells, n=3, **p<0.005. (D-E) Oxygen consumption rate (OCR) and extracellular acidification rate (ECAR) of control and Nqo1-OE Huh-7 cells measured by Seahorse Bioscience XF96 analyzer. n=3, **p<0.005. (F) Western blots showing expression and activity levels of indicated proteins in control and Nqo1-OE Huh-7 cells. (G-H) Western blots showing expression and activity levels of the indicated proteins in control and Nqo1-OE H2.35 cells. (I) Glucose uptake in control and Nqo1-OE H2.35 cells, n=3, *p<0.05. (J) Cellular lactate levels in control and Nqo1-OE cells, n=3, *p<0.05. (K–L) Oxygen consumption rate (OCR) and extracellular acidification rate (ECAR) of control and Nqo1-OE H2.35 cells measured by Seahorse Bioscience XF96 analyzer, n=3, *p<0.05, **p<0.005. Red * in Western blots indicates strong detectable difference. All the quantitative data are represented as mean ± SD.

Akt and c-Myc overexpression partially rescued Nqo1-KD-associated phenotypes

Nqo1-KD caused impaired proliferation of liver cancer cells and downregulation of glycolysis and glutaminolysis genes (Fig 2 and 6). This is due to impaired activation of Akt and downregulation of c-Myc (Fig 4A, D), the downstream effectors of the PI3K/Akt and MAPK/ERK pathways. To determine if Akt and c-Myc are the direct and final downstream targets of Nqo1, we overexpressed constitutively active forms of Akt (Akt-T308D/S473D), and c-Myc (c-Myc T58A mutant) in Nqo1-KD Huh-7 cells, which revealed partial but significant rescue of cell proliferation (Fig 8A), colony formation (Fig 8B–C), and glycolytic gene expression (Fig 8D–E) in Nqo1-KD cells. These data indicate that Akt and c-Myc are two of the main downstream targets through which Nqo1 controls metabolic reprogramming in liver cancer cells.

Figure 8. Forced expression of Akt and c-Myc partially reverse Nqo1-KD-associated phenotypes.

Proliferation rate of indicated cell lines as measured by cell counting method for 6 days, n=3, *p<0.05, **p<0.005. (B–C) Crystal violet-stained colonies formed by indicated cell lines 21 days after plating (B); colony formation ability of Nqo1-KD cells was partially but significantly rescued when constitutively active forms of Akt or c-Myc were re-expressed (C), n=3, **p<0.005, ***p<0.0005. (D–E) Western blots showing the expression levels of glycolytic proteins after expressing vector control or constitutively active forms of Akt or c-Myc in Nqo1-KD Huh-7 cells. Red * in Western blots indicates strong detectable difference. All the quantitative data are represented as mean ± SD. (F) Proposed model: PP2A functions as one of the central players in the regulation of both the PI3K/Akt and MAPK/ERK pathways. PP2A can dephosphorylate PDK1, Akt, Raf and ERK1/2, thereby shutting down both the PI3K/Akt and MAPK/ERK pathways. PP2A also dephophorylates and activates GSK3β leading to GSK3β-mediated phosphorylation and degradation of c-Myc. In the presence of Nqo1, Nqo1 interferes with PP2A, thereby preventing it from dephosphorylating and inactivating protein kinases in the PI3K/Akt and MAPK/ERK pathways, leading to their constitutive activation. Activated ERK1/2 also suppresses Pten expression via the CREB/c-Jun pathway, resulting in further activation of the PI3K/Akt pathway. Activated Akt and ERK1/2 phosphorylate and inactivate GSK3β, thereby preventing GSK3β-mediated phosphorylation and degradation of c-Myc. ERK1/2 can also directly phosphorylate and stabilize c-Myc. Akt and c-Myc together regulate the expression of vast majority of glycolysis and glutaminolysis genes, promoting metabolic adaptation, proliferation and tumor growth. In the absence of Nqo1, activated PP2A dephosphorylates kinases in both the PI3K/Akt and MAPK/ERK pathways, leading to simultaneously shutting down of both pathways. Due to ERK1/2 inactivation and impaired activation of the CREB/c-Jun pathway, Pten is upregulated. PTEN, by targeting PIP3 further inhibits PI3K/Akt signaling. Overall, due to impaired activation of Akt and downregulation of c-Myc, glycolysis and glutaminolysis genes are not induced, leading to blockade of metabolic adaptation, proliferation and tumor growth. Green and Red arrows represent up- and down-regulation of indicated proteins in response to the presence or absence of Nqo1.

Discussion

Nqo1 is a cytosolic flavoprotein that utilizes NADH or NADPH to catalyze the reduction of quinones, thereby preventing the generation of free radicals. Hence, Nqo1 may protect cells from oxidative stress (34, 35). Accordingly, Nqo1 deficiency in mice resulted in increased susceptibility to chemical-induced skin carcinogenesis and radiation-induced myeloproliferative disease (36, 37). However, a large number of studies demonstrated strong upregulation of Nqo1 in various cancers such as breast, pancreas, liver, bladder, ovary, thyroid, colorectal, cholangiocarcinoma, uterine cervical cancer, melanoma, and lung cancer (15–19). Moreover, Nqo1-KD inhibited proliferation of cholangiocarcinoma and adenocarcinoma cells (38, 39). Conversely, Nqo1-OE increased the proliferation of melanoma cells (40), suggesting that Nqo1 may function as an oncogene and promote cell proliferation and/or survival of certain cancer cells. Additionally, increased expression of Nqo1 protein was correlated with tumor size, advanced clinical stage, and decreased patient survival rates in breast, liver and lung cancers (17, 19, 20), signifying that Nqo1 may also be involved in the progression of certain cancers. In the current study, we demonstrate that Nqo1 is strongly overexpressed in mouse and human liver tumors and in liver cancer cell lines. Nqo1-KD significantly inhibited proliferation of liver cancer cells and xenograft tumor growth, and absence of Nqo1 strongly inhibited DEN-induced HCC in mice. Our observations suggest that Nqo1 functions as an oncogene in HCC.

Most significantly, we discovered that Nqo1 is required for the activation of two major pro-oncogenic signaling pathways in liver cancer cells: PI3K/Akt and MAPK/ERK. These signaling pathways are often dysregulated and hyper-activated in HCC and promote cancer cell metabolic adaptation, proliferation and tumor growth (1, 6, 7, 33, 41–43). Nqo1-KD in liver cancer cells blunted activation of both the PI3K/Akt and MAPK/ERK signaling pathways, indicating that Nqo1 functions upstream and is required for the activation of these pathways. How does Nqo1 regulate these two pathways simultaneously? There is a substantial amount of cross-talk and critical regulatory points between the PI3K/Akt and MAPK/ERK pathways (22, 23, 25–28). Therefore, it is entirely possible that Nqo1 may target factors that connect these two pathways. Some of the factors that operate in the cross-regulation of the PI3K/Akt and MAPK/ERK pathways are PTEN, p53, AMPK, GSK3β and PP2A (22, 23, 25–28). Very fascinatingly, we found that Nqo1 interferes with PP2A enzymatic activity which is one of the strongest molecular links between the PI3K/Akt and MAPK/ERK pathways (25–28). PP2A operates at the inter-junction of these two pathways and dephosphorylates and inactivates PDK1, Akt, Raf, MEK and ERK1/2, thereby shutting down the PI3K/Akt and MAPK/ERK pathways simultaneously. Therefore, factors that target PP2A apparently cause constitutive activation of both the PI3K/Akt and MAPK/ERK pathways, and many viral proteins target PP2A to constitutively activate both these pathways and promote tumorigenesis (44). Nqo1-KD increased PP2A activity, suggesting that PP2A is one of the major downstream targets of Nqo1. Therefore, it appears that Nqo1 exerts its influence on metabolic adaptation, cell proliferation and tumor growth via PP2A-linked signaling pathways such as the PI3K/Akt and MAPK/ERK pathways.

Our other substantial observation is that Nqo1-KD enhanced PTEN expression by promoting its transcription. One of the signaling pathways that regulate Pten transcription runs through ERK/CREB/c-Jun. ERK1/2 is known to phosphorylate and activate CREB, which can induce c-Jun expression. c-Jun directly binds to the Pten promoter and inhibits its transcription (30–32). Here, we provide evidence that Nqo1-KD enhanced PP2A activity, inhibited ERK and CREB activation, and suppressed c-Jun expression and its subsequent inhibitory binding to the Pten promoter, leading to enhanced expression of Pten. Inhibition of PP2A activity or re-expression of Creb in Nqo1-KD cells induced c-Jun expression and suppressed PTEN. Based on these observations, we propose that Nqo1 regulates Pten expression via the PP2A/ERK/CREB/c-Jun pathway. Overall, we propose Nqo1 as a novel player in the regulation of both the PI3K/Akt and MAPK pathways, and functions upstream to activate both of these pathways simultaneously (Fig 8F).

The down-stream effectors of the PI3K/Akt and MAPK/ERK pathways, Akt and c-Myc, play central roles in metabolic adaptation. Together Akt and c-Myc regulate the majority of the glycolytic and glutaminolysis genes whose expression and/or activation is essential for reprogramming of glucose and glutamine metabolism and uncontrolled proliferation of cancer cells (9–11, 33, 43). Our work here establishes Nqo1 as a key mediator of metabolic adaptation due to its tremendous influence on the activation of the PI3K/Akt and MAPK/ERK activity in cancer cells. Nqo1 appears to control glycolysis and glutaminolysis via two of the most powerful regulators of metabolic adaptation, PP2A and PTEN. For example, Pten elevation alone is sufficient to induce a tumor-suppressive, anti-Warburg state (13). We demonstrate that Nqo1 ablation not only induced PTEN expression and impaired activation of Akt, but also blunted activation of the MAPK/ERK pathway and downregulated c-Myc, leading to suppression of both glycolytic and glutaminolysis genes and reversal of the Warburg effect. Conversely, Nqo1-OE downregulated PTEN, activated Akt and induced expression of c-Myc, leading to enhanced expression of glycolytic and glutaminolysis genes and promotion of metabolic adaptation. Likewise, a recent study by metabolomics profiling showed that Nqo1 depletion downregulates HK2 and suppresses glycometabolism in NSCLC cells, highlighting the involvement of Nqo1 in glucose metabolism (45). In conclusion, our findings that Nqo1 functions as an upstream activator of the PI3K/Akt and MAPK/ERK signaling pathways and regulates metabolic adaptation, HCC cell proliferation and tumor growth suggest that Nqo1 could function as a potential therapeutic target to inhibit HCC growth.

Abbreviations

Nqo1

NAD(P)H Quinone Dehydrogenase 1

Nqo1-KD

Nqo1 knockdown

Nqo1-OE

Nqo1 overexpression

PI3K

Phosphoinositide 3-kinase

Akt

Akt serine/threonine kinase

MAPK

Mitogen activated protein kinase

ERK

Extracellular signal-regulated kinase

HCC

Hepatocellular carcinoma

PTEN

Phosphatase and tensin homolog

CREB

CAMP responsive element binding protein

c-Jun

Jun proto-oncogene

OXPHOS

Oxidative phosphorylation

HK1

Hexokinase 1

PFK1

Phosphofructokinase 1

LDH

Lactate dehydrogenase

PDK1

Pyruvate dehydrogenase kinase 1

SLC38A5

Solute carrier family 38 member 5

SLC1A5

Solute carrier family 1 member 5

GLS

Glutaminase

TCA cycle

Tricarboxylic acid cycle

α-KG

alpha-ketoglutarate

HK2

Hexokinase 2

PFK

Phosphofructokinase

ENO1

Enolase 1

GAPDH

Glyceraldehyde 3-phosphate dehydrogenase

PGK1

Phosphoglycerate kinase 1

LDHA

Lactate dehydrogenase A

Glut 1

Glucose transporter 1

PP2A

Protein phosphatase 2A

DEN

diethylnitrosamine

NADPH

Nicotinamide adenine dinucleotide phosphate

ROS

Reactive oxygen species

RNS

Reactive nitrogen species

8-OHdG

8-hydroxy-2-deoxyguanosine

mTOR

The mechanistic target of rapamycin

GSK3β

Glycogen synthase kinase-3 beta

EGFR

Epidermal growth factor receptor

PPARγ

Peroxisome proliferator activated receptor gamma

TGFβ

Transforming growth factor beta

NFκB

Nuclear factor kappa B subunit 1

Hes1

Hes family BHLH transcription factor 1

OCR

Oxygen consumption rate

ECAR

Extracellular acidification rate

Cyt C

Cytochrome C

FDR

False discovery rate

IHC

Immunohistochemistry

H&E

Hematoxylin and eosin

ChIP

Chromatin immunoprecipitation

PCNA

Proliferating cell nuclear antigen

CoxIV

cytochrome c oxidase subunit IV