SIRT2‐dependent IDH1 deacetylation inhibits colorectal cancer and liver metastases

Abstract

Protein lysine acetylation affects colorectal cancer (CRC) distant metastasis through multiple pathways. In a previous proteomics screen, we found that isocitrate dehydrogenase 1 (IDH1) is hyperacetylated in CRC primary tumors and liver metastases. Here, we further investigate the function of IDH1 hyperacetylation at lysine 224 in CRC progression. We find that IDH1 K224 deacetylation promotes its enzymatic activity and the production of α‐KG, and we identify sirtuin‐2 (SIRT2) as a major deacetylase for IDH1. SIRT2 overexpression significantly inhibits CRC cell proliferation, migration, and invasion. IDH1 acetylation is modulated in response to intracellular metabolite concentration and regulates cellular redox hemostasis. Moreover, IDH1 acetylation reversely regulates HIF1α‐dependent SRC transcription which in turn controls CRC progression. Physiologically, our data indicate that IDH1 deacetylation represses CRC cell invasion and migration in vitro and in vivo, while the hyperacetylation of IDH1 on K224 is significantly correlated to distant metastasis and poor survival of colorectal cancer patients. In summary, our study uncovers a novel mechanism through which SIRT2‐dependent IDH1 deacetylation regulates cellular metabolism and inhibits liver metastasis of colorectal cancer.

SIRT2 inhibits colorectal cancer cell invasion and metastasis by regulating IDH1 acetylation on K224.

Introduction

Colorectal cancer (CRC) is one of the most common malignancies and one of the major causes of morbidity and mortality worldwide. Approximately 50% of patients with CRC develop liver metastases during their course of disease 1. Despite remarkable progress in cancer therapy, the prognosis for CRC patients with distant metastases remains unsatisfactory 2, 3. In addition, the molecular mechanisms underlying the progression and liver metastasis of CRC still remain unclear.

Protein post‐translational modification (PTM), especially lysine acetylation, has been reported to be essential in cancer progression. Protein acetylation regulates cancer proliferation through controlling multiple cellular pathways, such as cell cycle, apoptosis, and different metabolism pathways. Metabolism reprogramming is a hallmark for tumors maintaining proliferation rates accompanying the malignant phenotype. The acetylation of key oncogenic enzymes affects the enzymatic and non‐enzymatic activity and subsequently promotes tumor progression.

IDH1 is a NADP⁺‐dependent enzyme located in the cytoplasm that converts isocitrate to α‐ketoglutarate (α‐KG). Its mutant form converts α‐KG to the oncometabolite 2‐hydroxyglutarate (2‐HG) and induces epigenetic and cell signaling alterations, thus causing tumorigenesis 4, 5. Compared to primary gliomas, isocitrate IDH1 mutations have been detected in more than 70% of secondary gliomas; among those, arginine 132 has been found in nearly all cases 6, 7. IDH1 has recently received great attention because of its involvement in glutamine metabolism in cancer cells. Despite that rare IDH1 mutations are found in colon cancer samples, the glutamine addiction of colon cancer cells suggests that IDH1 may also have an important role in CRC progression. Nevertheless, IDH1 acetylation modification and its function in tumorigenesis have not yet been reported.

Sirtuin‐2 (SIRT2) is a cytoplasmic NAD‐dependent protein deacetylase, which can serve as a histone deacetylase with a preference for histone H4 lysine 16 (H4K16Ac), and can affect chromosomal condensation during cell mitosis 8, 9. SIRT2 deacetylates several notable metabolic enzymes or transcription factors, including glucose‐6‐phosphate 1‐dehydrogenase (G6PD), phosphoglycerate mutase (PGAM), ATP‐citrate synthase (ACLY), forkhead box protein O3 (FOXO3), and peroxiredoxin‐1 (Prdx‐1) 10, 11, 12, 13, 14, which affects cell function. In addition, preclinical studies have shown that SIRT2 deficit mice may develop mammary tumors, hepatocellular carcinomas, and gastrointestinal tumors 15, 16, which suggests that SIRT2 could be a target for cancer therapeutic strategy. Yet, other studies have reported on oncogenic characteristic of SIRT2 in leukemia and breast cancer 17, 18. Therefore, the precise role of SIRT2 acting as oncogene or tumor suppressor remains elusive and is probably regulated by its downstream targets.

In our previous study, we used mass spectrometry‐based quantitative proteomics to quantify dynamic changes of non‐histone protein acetylation between matched primary colorectal cancer and liver metastatic tumor specimens. We found aberrant acetylation levels in several cellular metabolism enzymes during colorectal cancer progression and liver metastasis 19. In the present study, we further investigated the function of IDH1 hyperacetylation at lysine 224 in colorectal cancer progression and liver metastases. We discovered that SIRT2 could deacetylate IDH1 at K224 and exhibit tumor suppression function in colon cancer cell model through IDH1 enzymatic activities and HIF1α‐SRC transcription axis. Deacetylation of IDH1 K224 by SIRT2 significantly inhibited the malignant behaviors of CRC cells in vitro and in vivo and predicted poor survival in colorectal cancer samples. These findings further elucidate SIRT2‐dependent IDH1 acetylation treatment of liver metastasis in CRC.

Results

Identification of functional acetylation sites in IDH1

In our previous study 19, we found eight proteins with eight acetylated sites which showed more than 2.5‐fold changes at acetylation levels, indicating the potential function of these acetylated proteins in colorectal cancer progression and liver metastases (Table 1). In this study, we further evaluated the acetylation levels of these proteins in CRC cell lines. Briefly, we found decreased acetylation levels of K224R on IDH1 or K112R on CSRP1 (Fig EV1A). Furthermore, after performing a cell invasion assay, we found that the overexpression of IDH1 K224R was able to inhibit cell invasive behavior, while no effect was observed for CSRP1 K112R (Fig EV1B). Hence, IDH1 K224 acetylation was selected for further investigation.

Table 1.

Differentially expressed acetylation sites obtained in 3 paired samples according to acetylation |Diff| of metastases vs. tumor ≥ 2.5

ProtDesc	Position	Meta/Tumor	Acetylation |Diff|	Total protein |Diff|
TPM2	152	Down	4.031	1.188
ADH1B	331	Up	3.807	1.077
GLUD1	84	Up	3.711	1.098
ASS1	58	Up	3.067	1.153
CSRP1	112	Down	2.973	1.126
TAGLN	21	Down	2.946	1.219
IDH1	224	Up	2.911	1.055
VTN	275	Up	2.603	1.162

ProtDesc: protein name; Position: acetylation site; Meta/Tumor: liver metastases/colorectal primary tumor; |Diff|: absolute value of the differences.

Figure EV1. IDH1 K224 is confirmed as the object for further investigation.

• A
  Acetylation levels of protein candidates were assessed by Western Blot. K224R on IDH1 or K112R on CSRP1 showed decreased acetylation levels comparing to others (Acetylation level of ADH1B or VTN was failed to be detected, data not shown).
• B
  Transwell assay showed IDH1 K224R could inhibit cell invasion but CSRP1 K112R had no such effect. Results were expressed as mean ± SD of three independent experiments (n ≥ 3 per experimental condition). Differences between groups were evaluated with a two‐tailed unpaired Student's t‐test. **P < 0.01.
• C
  IDH1 K224 is evolutionarily conserved from C. elegans to humans. The sequences of IDH1 were aligned of five species. Acetylation lysine 224 identified by proteomic study was highlighted in red.
• D
  IDH1 protein levels after knockout or treated with empty vector, wild‐type, K224R and K224Q.

To confirm IDH1 acetylation, we treated HEK293T cells with trichostatin A (TSA) and nicotinamide (NAM), the inhibitors of HDAC class I and II or sirtuins, respectively. The immunoprecipitated acetylated IDH1 was increased under treatment of NAM, suggesting that IDH1 acetylation was controlled by sirtuins (Fig 1A). We mutated all the candidate acetylated lysine sites of IDH1 (from UniProt database) to arginine (R) which mimicked the deacetylated states of protein. As shown in Fig 1B, only K224R showed significant lower acetylation levels, indicating that K224 is the major site for IDH1 acetylation.

Figure 1. SIRT2 deacetylates IDH1 on K224.

• A
  IDH1 acetylation levels upon treatment with NAM or TSA. Flag‐tagged IDH1 was ectopically expressed in HEK293T cells treated with NAM (5 mM) and/or TSA (0.5 mM) for the indicated time period.
• B
  Five putative lysine residues were mutated. Acetylation levels of Flag‐bead‐purified IDH1 were determined by Western blot analysis using a pan‐anti‐acetyl lysine antibody. Relative IDH1 acetylation ratios were calculated after normalizing against Flag.
• C
  Mutation of IDH1 K224R and K224Q resulted in altered acetylation levels using Western blot.
• D
  IDH1 catalytic activity of IDH1 K224 mutants in vitro. Wild‐type IDH1 and K224R, K224Q mutants were expressed in HEK293 cells. Proteins were purified by immunoprecipitation (IP), IDH1 levels were normalized for protein, and activity assays were performed.
• E, F
  Steady‐state kinetic analysis of IDH1 and variants in HEK293T cells. Comparison of IDH1 WT and K224R and K224Q showed that the K224 site was important in catalysis.
• G
  IDH1 directly interacted with SIRT2 but not SIRT1. SIRT2 deacetylated IDH1 effectively. Flag‐tagged IDH1 and HA‐tagged SIRT1 or SIRT2 were co‐transfected into cells. The acetylation level of Flag‐bead‐purified IDH1 and the protein association between IDH1 and SIRT1 or SIRT2 were determined by Western blot.
• H
  Co‐immunoprecipitation showed endogenous proteins of IDH1 and SIRT2 interacted with each other.
• I
  SIRT2 interacted with IDH1 in vitro. GST pull‐down assay with GST‐fused SIRT2 and Flag‐tagged IDH1 proteins.
• J
  Acetylation level of endogenous IDH1 K224 when treated with SIRT1 or SIRT2 by Western blot analysis.
• K
  IDH1 K224 deacetylation was dependent on SIRT2 catalytic activity. SIRT2 H187Y was a catalytically inactive mutant.

Data information: All results were expressed as mean ± SD of three independent experiments (n ≥ 3 per experimental condition). For multiple comparisons in (D), a one‐way ANOVA with Newman–Keuls post hoc test was used to test for statistical significance. For (E, F), statistical significance was determined using the two‐way ANOVA followed by Tukey's post hoc test. *P < 0.05 and **P < 0.01.

Then, we determined the in vitro enzymatic activities of IDH1 mutants, K224R and K224Q, expressed in HEK293T cells and found that IDH1 K224Q mutants had a greater than 40% reduction in activity as compared with the wild‐type (WT) IDH1 whereas IDH1 K224R mutants had an enhancement (Fig 1C and D). To investigate the mechanism by which K224 acetylation might affect IDH1 activity, we recombinantly expressed and purified human wild‐type IDH1 and the K224R and K224Q mutants from E. coli. Steady‐state kinetic analysis of IDH1 and variants indicated that substitution of K224 with arginine (R) or glutamine (Q) could alter the Km values for isocitrate and NADP⁺ and exhibited a change in Vmax when compared with WT (Fig 1E and F, Table 2).

Table 2.

Steady‐state kinetic analysis of IDH1 WT and mutants

	NADP+	Isocitrate
WT
Km(μM)	4.7 ± 0.9	5.2 ± 0.6
Vmax (μmol/min/mg)	41.1 ± 2.4	56.5 ± 2.0
K224R
Km(μM)	3.9 ± 0.7	3.9 ± 0.5
Vmax(μmol/min/mg)	44.0 ± 2.2	63.3 ± 2.1
K224Q
Km(μM)	8.6 ± 1.1	7.5 ± 1.0
Vmax(μmol/min/mg)	16.0 ± 1.5	16.4 ± 1.6

SIRT2 deacetylates IDH1 on K224

SIRT1, SIRT2, SIRT6, and SIRT7 in sirtuins family might regulate cytoplasmic protein acetylation. As IDH1 is located in cytoplasm, we explored the relationship between IDH1 acetylation and sirtuins mentioned above. After co‐transfection of Flag‐IDH1 and HA‐SIRT1 or HA‐SIRT2 or HA‐SIRT6 or HA‐SIRT7 in HEK293T cells, we found that IDH1 directly interacted with SIRT2, but not with SIRT1 (Fig 1G), SIRT6 and SIRT7 (Fig EV2B); in addition, IDH1 acetylation levels were downregulated by SIRT2 as well. Consistently, endogenous IDH1 interacted with SIRT2 in HCT116 cells other than SIRT1 (Fig 1H). We also performed the GST pull‐down assay and confirmed that SIRT2 can interact with IDH1 in vitro (Fig 1I).

Figure EV2. IDH1 deacetylation is specifically regulated by SIRT2.

• A
  CRC cells were treated with five different KATs. IDH1 K224 was acetylated by CBP or P300.
• B
  IDH1 interacts with SIRT2, but not SIRT6 and SIRT7. Flag‐tagged IDH1 was ectopically expressed in HEK293T cells together with the individual HA‐tagged SIRT as indicated.
• C
  Knockdown of SIRT2 resulted in upregulation of IDH1 K224 acetylation level.
• D
  K224 acetylation‐specific antibody was generated through rabbit immunization. No. 1 or 2 refers to the acetyl‐K224 peptide while No. 3 stands for the unmodified one as negative control.
• E
  The purified antibody reacted strongly with the 47 KD IDH1 protein on extract from mouse liver tissue and HeLa cells.
• F
  Transwell assay evaluating the invasion abilities of HCT116 cells after transfection of SIRT2 or with IDH1 K224Q.
• G
  Cell migration ability was evaluated by wound healing assay in HCT116 cells.
• H
  Colony formation ability after overexpression of SIRT2 or with IDH1 K224Q.

Data information: All results were expressed as mean ± SD of three independent experiments (n ≥ 3 per experimental condition). For (F, G, and H), differences between groups were evaluated with a two‐tailed unpaired Student's t‐test. *P < 0.05 and **P < 0.01.

To further investigate the role of K224 acetylation on IDH1, we constructed a K224‐specific antibody (Figs EV2D and EV3A). As expected, SIRT2, but not SIRT1, greatly decreased the acetylation levels on K224 site (Fig 1J). HEK293T cells were co‐transfected with IDH1 and wild‐type (WT) SIRT2 or enzymatically defective mutant SIRT2 H187Y; compared with SIRT2 WT, SIRT2 H187Y lost IDH1 K224 deacetylation ability (Fig 1K). Taken together, IDH1 K224 acetylation may have an important role in colorectal cancer progression and liver metastases and SIRT2 is the major deacetylase responsible for the regulation of IDH1 K224 acetylation.

Figure EV3. IDH1 K224Ac, HIF1α, or SRC protein levels are detected by immunohistochemistry methods in clinical samples.

• A
  Acetyl‐IDH1 K224 antibody was used in clinical samples. After treatment with corresponding peptides, the signal of IDH1 K224Ac was significantly diminished, indicating that IDH1 K224 peptides successfully blocked the acetylation antibody.
• B
  HIF1α or SRC protein expression in CRC and the adjacent normal tissues.

SIRT2 is a tumor suppressor in CRC

To investigate the function of SIRT2 in CRC tumorigenesis, we performed a number of cellular tumor biological assays. We found that the overexpression of SIRT2 could dramatically decrease CRC cell proliferation, migration, and cell invasive behavior (Fig 2A–C). Moreover, overexpression of SIRT2 induced cell cycle S phase arrest phenotype (Fig 2D). However, after treated with AGK2 (exclusive SIRT2 inhibitor) or SIRT2 siRNA in HCT116 cells, significantly increased cell invasion, migration, and colony formation were detected. All these experiments demonstrated that SIRT2 inhibits colorectal cancer cell proliferation and invasion.

Figure 2. SIRT2 is a tumor suppressor in CRC .

• A
  Colony formation ability after transfection of SIRT2 plasmids, AGK2, and SIRT2 siRNA.
• B
  Transwell assay was performed to evaluate the invasion abilities of HCT116 cells after overexpression or downregulation of SIRT2 or treated with AGK2. Stained cells in the lower chambers were quantified.
• C
  Cell migration ability was assessed by wound healing assay in HCT116 cells.
• D
  Cell cycle distribution was evaluated flow cytometrically after treatment with SIRT2 plasmids, SIRT2 siRNA, or AGK2.
• E
  Immunohistochemical analysis of SIRT2 protein in representative colorectal tissue specimens is shown (magnification ×100). Weak cytoplasmic staining for SIRT2 was observed in CRC tissues and metastatic tissues, whereas strong cytoplasmic staining was observed in non‐cancerous tissues.

Data information: All results were expressed as mean ± SD of three independent experiments (n ≥ 3 per experimental condition). For (A, B, and D), statistical significance was assessed with the one‐way ANOVA with Newman–Keuls post hoc test. For (C), statistical significance was assessed with the two‐way ANOVA followed by Tukey's post hoc test. *P < 0.05 and **P < 0.01.

Next, SIRT2 antibodies were used to detect the expression levels of SIRT2 in CRC primary tissues and their corresponding adjacent normal tissues, as well as liver metastatic tumors. As shown in Fig 2E, SIRT2 expression was significantly decreased in CRC tissues or liver metastases than in corresponding colorectal normal tissues. These data suggest that the downregulation of SIRT2 could lead to CRC progression and liver metastases.

IDH1 acetylation senses metabolite concentration and regulate cellular redox hemostasis

The benefit of IDH1 for rapidly growing tumorigenic cells is believed to result from a decreased IDH1 activity. To further define the physiologic functions of the SIRT2‐IDH1 regulation, we first examined the effect of IDH1 K224 acetylation and enzymatic activities under different metabolite concentrations. Given that glucose is the major carbon and energy source for cancer cell proliferation, we treated HCT116 cells with high glucose levels. High glucose concentration resulted in increased K224 acetylation levels of IDH1 with reduced IDH1 activity in a dose‐dependent manner (Figs 3A and B, and EV4A and B). Recent studies have underlined that cancer cells survive for their glutamine addiction 20, 21. Consistent with these data, IDH1 K224 acetylation was increased with high glutamine in a dose‐dependent manner with decreased IDH1 activity (Figs 3C and D, and EV4C). Thus, IDH1 acetylation sensed the metabolite concentration and tightly controlled IDH1 enzymatic activities. We further investigated the function of SIRT2 in this regulation process. Knockdown of SIRT2 in HCT116 cells abrogated the effect of changing the K224 acetylation levels under high glucose or glutamine treatment (Fig 3E and F), suggesting the pivotal regulating role of SIRT2 in IDH1 enzymatic activities.

Figure 3. IDH1 acetylation senses metabolite concentration and regulate cellular redox hemostasis.

• A, B
  IDH1 K224 acetylation level was enhanced with increasing glucose concentration. Flag‐IDH1 was overexpressed in cells treated with increased concentrations of glucose for 6 h. IDH1 proteins were purified by Flag beads, and then, IDH1 K224 acetylation level was determined by Western blot and IDH1 catalytic activity was assessed.
• C, D
  IDH1 K224 acetylation level was upregulated, and IDH1 activity was weakened with increasing glutamine concentration. Flag‐IDH1 was overexpressed in cells treated with increased glutamine for 6 h. IDH1 proteins were purified by Flag beads, and the K224 acetylation level of IDH1 was determined by Western blot analysis, and then, IDH1 catalytic activity was evaluated.
• E, F
  Downregulation of SIRT2 diminished the effect of glucose or glutamine on changing IDH1 K224 acetylation. Flag‐tagged IDH1 was overexpressed in HCT116 cells with or without transient SIRT2 knockdown. The cells were treated with different concentrations of glucose (E) and glutamine (F).
• G
  IDH1 K224 deacetylation regulated cellular NADPH/NADP⁺ redox in cells. In HCT116 stable cells with IDH1 knockout and re‐expressing the indicated proteins, the ratio of NADPH/NADP⁺ in cells was measured as followed by manufacturer's instruction.
• H
  IDH1 K224 deacetylation promoted GSH production in HCT116 cells. In HCT116 stable cells with IDH1 knockout and re‐expressing the indicated proteins, the ratio of GSH/GSSG was assessed as followed by manufacturer's instruction.
• I
  IDH1 K224 deacetylation suppressed cellular ROS levels in HCT116 cells. ROS was determined in cells under non‐stressed condition or exposed to menadione. *P < 0.05 and **P < 0.01.

Data information: All results were expressed as mean ± SD of three independent experiments (n ≥ 3 per experimental condition). For (B, D, G, and H), statistical significance was assessed with the one‐way ANOVA with Newman–Keuls post hoc test. For (I), statistical significance was assessed with the two‐way ANOVA followed by Tukey's post hoc test. *P < 0.05 and **P < 0.01. n.s. = not significant.

Figure EV4. IDH1 acetylation level regulates α‐KG and metabolite concentration.

• A
  Cellular α‐KG levels decreased with increasing IDH1 K224Q expression. The α‐KG level in cells transfected with empty vector was set as 100%.
• B, C
  The endogenous IDH1 K224 acetylation level was enhanced with increasing glucose or glutamine.
• D
  Knocking out of IDH1 inhibited NADPH production in HCT116 cells.
• E
  Knocking out of IDH1 repressed GSH production in HCT116 cells. **P < 0.01.

Data information: All results were expressed as mean ± SD of three independent experiments (n ≥ 3 per experimental condition). For (A), statistical significance was assessed with a one‐way ANOVA with Newman–Keuls post hoc test. For (D and E), differences between groups were evaluated with a two‐tailed unpaired Student's t‐test. **P < 0.01.

IDH1 and its enzymatic reaction product α‐KG are indispensable for transition between NADP⁺ and NADPH, whose ratio displays cellular redox balance 22. To investigate the role of IDH1 K224 acetylation in this process, we knocked down IDH1 in HCT116 cells by CRISPR/Cas9 and generated four stable re‐expression cell lines: empty vector, Flag‐tagged IDH1 wild type (WT), IDH1 K224R, and IDH1 K224Q (mimicking the acetylated modification). IDH1 WT, K224R, and K224Q reconstituted cells had similar IDH1 expression levels compared to empty vector HCT116 cells (Fig 1C). We found that re‐expression of IDH1 K224R in HCT116 cells led to significantly increased ratios of NADPH/NADP⁺ (by ~25%) (Fig 3G) and GSH/GSSG (by ~20%) (Fig 3H) or to a decrease of ROS levels (Fig 3I). In contrast, re‐expression IDH1 K224Q (or knockdown of IDH1) reversely reduced the ratios of NADPH/NADP⁺ (by ~14%) (Figs 3G and EV4D) and GSH/GSSG (by ~13%) (Figs 3H and EV4E) or an increase of ROS levels (Fig 3I). As a result, IDH1 acetylation enhanced ROS levels through cellular redox imbalance.

IDH1 acetylation controls HIF1α‐SRC axis in CRC progression

Metabolism remodeling in tumor cells always shows converse regulation in transcription level through aberrant metabolite concentration. To explore how IDH1 K224 acetylation affects CRC metastasis, we investigated its downstream signaling pathways with the Human Tumor Metastasis PCR Array by comparing IDH1 K224R HCT116 cells with negative control cells. The expression levels of all these genes are summarized in Dataset EV1. Among all genes, SRC, a proto‐oncogene, showed the highest decrease in expression (Fig 4A). Thus, we postulated that SRC was the critical downstream target gene regulated by IDH1 K224 acetylation.

Figure 4. IDH1 acetylation control HIF1α‐SRC axis in CRC progression.

• A
  Gene expression profile array analysis in HCT116 cells. Representative genes with higher fold difference (up or down) compared IDH1 K224R cells with IDH1 WT cells.
• B
  ChIP assays for HIF1α and its binding motif. Antibodies anti‐IgG and anti‐ HIF1α were used in the ChIP assays. qRT–PCR was performed to quantify the binding activity.
• C, D
  The transcriptional activity of the SRC promoter was greatly increased after co‐expression of HIF1α and SRC reporter in wild type. SRC promoter constructs containing a potential HIF1α binding motif (−1,636 to −1,629 bp and −1,466 to −1,459 bp).
• E
  IDH1 K224 deacetylation led to degradation of HIF1α but upregulation of hydroxy‐HIF1α. In HCT116 stable cells with IDH1 knockout and re‐expressing the indicated proteins, HIF1α and hydroxy‐HIF1α were measured by Western blot assay.
• F
  HCT116 cells of IDH1 K224Q were cotreated with Octyl‐α‐KG and NAC, and the expression of HIF1α and SRC was further evaluated.
• G, H
  HCT116 cells of IDH1 K224Q with or without treatment with Octyl‐α‐KG and/or NAC were analyzed for glucose and lactate levels.

Data information: All results were expressed as mean ± SD of three independent experiments (n ≥ 3 per experimental condition). For (B and D), statistical significance was assessed with a two‐tailed unpaired Student's t‐test. For (G and H), statistical significance was assessed with a one‐way ANOVA with Newman–Keuls post hoc test. *P < 0.05 and **P < 0.01. n.s. = not significant.

IDH1 K224 deacetylation promotes the enzymatic activity of IDH1 that leads to increasing production of α‐KG, which is important for hydroxylation and degradation of HIF1α. To elucidate the mechanism of how IDH1 K224 acetylation affects SRC expression, we utilized JASPAR online software to predetermine the transcription factor of SRC. Interestingly, prediction revealed HIF1α as one of the potential transcription factors with two binding sites localized on SRC promoter region at the positions of −1,636 to −1,629 and −1,466 to −1,459. The recruitment of HIF1α to SRC at its binding motif was observed in SW620 cells in CHIP‐PCR assay (Fig 4B). Luciferase reporter assay was also performed to confirm this result. We constructed the reporter plasmids with both wild type and mutant of these two potential binding sites (Fig 4C) for the assay. The results showed that HIF1α could indeed bind to the SRC promoter position on these two sites (Fig 4D).

α‐KG is an electron donor to PHDs, which enables hydroxylation of HIF1α and VHL‐mediated HIF1α proteasomal degradation 23. In the present study, we examined the effect of IDH1 acetylation on HIF1α expression. Overexpression of IDH1 K224 hypoacetylation‐mimic K224R led to downregulation of HIF1α but upregulation of hydroxy‐HIF1α, whereas the overexpression of IDH1 K224 hyperacetylation‐mimic K224Q resulted in the inverse expression of HIF1α (Fig 4E), suggesting that the IDH1 acetylation status could control HIF1α transcription activity dependent on its enzymatic activity. Treating IDH1 K224Q rescued cells with Octyl‐α‐KG (cell‐permeable α‐KG analog), or N‐acetyl‐L‐cysteine (NAC, ROS scavenger) 24, could significantly reduce the increased expression of HIF1α and SRC (Fig 4F), suggesting that downregulation of α‐KG together with the increase in ROS was cooperatively responsible for IDH1 acetylation‐dependent HIF1α upregulation. To further clarify that IDH1 K224 acetylation regulated α‐KG/HIF1α/SRC axis in CRC cells, we performed the rescue study. Re‐expression of HIF1α in IDH1 K224R cells resulted in upregulation of SRC (Fig EV5C). Moreover, after re‐expression of SRC in IDH1 K224R stable cells, we observed the increased ability of colony formation and invasion in CRC cells (Fig EV5A and B), which indicated that IDH1 K224 deacetylation is achieved by HIF1α‐SRC axis. Consistently, glucose levels were examined to further prove the influence of IDH1 acetylation. IDH1 K224Q cells showed increased glucose or lactate concentration, which could be reversed by Octyl‐α‐KG or NAC treatment (Fig 4G and H). These results further explained the converse regulating function by IDH1 dynamic acetylation levels.

Figure EV5. Effects of IDH1 acetylation can be reversed by overexpression of HIF1α or SRC .

• A
  Cell invasive ability was assessed by transwell method after deacetylation of IDH1 K224 and recovering SRC.
• B
  Colony formation ability after deacetylation of IDH1 K224 and rescue of SRC.
• C
  HIF1α and SRC expression levels upon deacetylation of IDH1 K224 combined with α‐KG treatment or overexpression of HIF1α or SRC. When treated with α‐KG, both HIF1α expression and SRC expression were obviously decreased. However, after hypoacetylation of IDH1, overexpression of HIF1α successfully reversed the expression of SRC in HCT116 cells. **P < 0.01.

Data information: All results were expressed as mean ± SD of three independent experiments (n ≥ 3 per experimental condition). For (A and B), differences between groups were evaluated with a two‐tailed unpaired Student's t‐test. **P < 0.01.

IDH1 K224 deacetylation suppresses CRC functions in vitro

We further investigated the effect of IDH1 K224R mutant on tumor cell growth. IDH1 K224R mutant significantly inhibited the invasion ability of HCT116 cells (Fig 5A). Consistently, IDH1 K224R mutant reduced tumor cell migration (Figs 5B and C) and inhibited its colony formation (Fig 5D). In addition, this phenotype was further confirmed in cell cycle analysis by flow cytometry. The arrested S phase in IDH1 K224R cells indicated the slower cell proliferation ability (Fig 5E). However, when knocking out of IDH1, we observed enhanced cell invasion, movement rate, migration, and colony formation compared with IDH1 K224R overexpression. Taken together, IDH1 deacetylation on K224 impairs cancer cell proliferation and invasion.

Figure 5. IDH1 K224 deacetylation suppresses CRC functions in vitro .

• A
  Transwell assay was performed to evaluate the invasion abilities of HCT116 cells after transfection of IDH1 K224R or knockout of IDH1. Quantification was performed by counting the stained cells that invaded to the lower chamber under a light microscopy.
• B
  Three‐dimensional imaging analysis was used to assess the velocity of cell movement.
• C
  Cell migration ability was evaluated by wound healing assay in HCT116 cells. Quantification was performed by measuring the smallest clearance distance of the wound.
• D
  Colony formation ability after deacetylation of IDH1 K224 or knockout of IDH1.
• E
  Cell cycle distribution was assessed flow cytometrically after treatment with IDH1 K224R or knockout of IDH1.

Data information: All results were expressed as mean ± SD of three independent experiments (n ≥ 3 per experimental condition). For (A, B, D, and E), statistical significance was assessed with a one‐way ANOVA with Newman–Keuls post hoc test. For (C), statistical significance was assessed with a two‐way ANOVA followed by Tukey's post hoc test. *P < 0.05 and **P < 0.01.

IDH1 K224 deacetylation impairs CRC growth and metastasis in vivo

To gain further insight into the physiologic significance of IDH1 hypoacetylation on colorectal cancer progression, we established a xenograft tumor model and liver metastasis model in nude mice injected with IDH1 WT or IDH1 K224R SW620 cells. Briefly, slower xenograft tumor growth was observed in IDH1 K224R group (Fig 6A–C) compared to the IDH1 WT group. Moreover, as shown in Fig 6D, tumor metastases were found in IDH1 WT group (metastases 4/5) compared to the IDH1 K224R cells (metastases 0/5). These findings support the implication that IDH1 K224 deacetylation exerts suppressive effect in CRC growth and metastasis in vivo.

Figure 6. IDH1 K224 deacetylation impairs CRC growth and metastasis in vivo .

• A
  Pictures of the nude mice and their tumors after inoculation were shown in two subgroups.
• B
  Tumor volumes of each group during the tumor growth process.
• C
  Tumor weight measured from the xenograft model.
• D
  Pictures of livers in nude mice and their respective representative images of the tissues by hematoxylin–eosin (HE) staining.

Data information: All results were expressed as mean ± SD of three independent experiments (n ≥ 3 per experimental condition). Statistical significance was assessed with a two‐tailed unpaired Student's t‐test. **P < 0.01.

IDH1 K224 acetylation correlates with poor prognosis in CRC

To evaluate whether IDH1 K224Ac levels were associated with the clinicopathological characteristics and prognosis in CRC patients, we investigated IDH1 K224 acetylation levels in 90 sets of CRC primary tissues, their liver metastases, and their corresponding adjacent normal tissues by immunohistochemistry. The results showed that increased IDH1 K224Ac levels in liver metastasis tumors and CRC primary tumors (Fig 7A) were closely related to late TNM stage (P = 0.004), distant metastasis (P = 0.036), and liver metastasis (P = 0.029), whereas no significant differences were observed with regard to age, gender, tumor size, tumor differentiation, or lymph node metastasis (Table 3). Meanwhile, CRC patients with higher IDH1 K224 acetylation levels had a poorer prognosis in CRC patients who displayed a shorter median survival P = 0.009 for OS and P = 0.012 for PFS (Fig 7B). Furthermore, Cox's multivariate analysis showed that IDH1 K224Ac levels and distant metastasis were significantly associated with overall survival in CRC patients as independent prognostic factors (Table 4). The results above indicated that enhanced IDH1 K224 acetylation predicted poor prognosis in patients with CRC.

Figure 7. IDH1 K224 acetylation correlates with poor prognosis in CRC .

• A
  Expression levels of IDH1 K224Ac in human CRC (n = 90) and CRN (n = 90) tissue specimens, assessed by immunohistochemistry. CRC, colorectal carcinoma samples; CRN, matched adjacent non‐cancerous colorectal tissue specimens.
• B
  Kaplan–Meier survival curves of CRC cases with elevated (n = 49) and reduced (n = 41) IDH1 K224Ac. CRC, colorectal carcinoma samples. The overall survival was analyzed by the log‐rank test using the Kaplan–Meier method.
• C
  A working model illustrating regulation of the HIF1α‐SRC axis by SIRT2‐dependent IDH1 deacetylation.

Table 3.

The relationship between IDH1 K224 acetylation expression and clinicopathologic characteristics in CRC patients

Parameters	IDH1 K224 acetylation
	High (n = 49)	Low (n = 41)	P‐value
Age (year)
≤ 60	12	9	0.777
> 60	37	32
Gender
Female	29	24	0.951
Male	20	17
Tumor size (cm)
≤ 5	32	21	0.176
> 5	17	20
Tumor differentiation
Well	10	8	0.943
Moderate	27	24
Poor	12	9
TNM stage
I+II	16	26	0.004a
III+IV	33	15
Lymph node metastasis
N0	22	20	0.713
N1–N2	27	21
Distant metastasis
M0	41	40	0.036a
M1	8	1
Liver metastasis
Negative	43	41	0.029a
Positive	6	0

^aStatistically significant (P < 0.05).

Table 4.

Univariate and multivariate analyses of factors related to overall survival in CRC patients

Characteristics	Univariate analysis		Multivariate analysis
	HR (95% CI)	P‐value	HR (95% CI)	P‐value
Age	1.006 (0.977–1.037)	0.676
Gender	1.067 (0.576–1.976)	0.838
Tumor size	0.841 (0.454–1.558)	0.582
Tumor differentiation	1.305 (0.808–2.108)	0.276
TNM stage	1.493 (0.800–2.786)	0.208
Lymphatic metastasis	1.528 (0.819–2.850)	0.183
Distant metastasis	2.705 (1.248–5.865)	0.012a	2.518 (1.014–6.253)	0.047a
IDH1 K224Ac	2.197 (1.191–4.051)	0.012a	2.124 (1.093–4.126)	0.026a

CI, confidence interval; HR, hazard ratio.

^aStatistically significant (P < 0.05).

Discussion

The prognosis for CRC patients with distant metastases, especially liver metastasis, remains unsatisfied [2,3]. In addition, the exact molecular mechanisms underlying the progression and liver metastasis of CRC are not completely elucidated. Protein lysine acetylation affects CRC distant metastasis through multiple pathways. In previous study, we identified a complete atlas of acetylome in CRC and paired liver metastases. Herein, we presented evidence suggesting that SIRT2 suppresses CRC cell progression and, in turn, liver metastases by regulating IDH1 acetylation and its enzymatic activity. Furthermore, we propose that the function of IDH1 acetylation on K224 underpins our observation that IDH1 hypoacetylation antagonizes CRC malignant progression through metabolite α‐KG‐induced HIF1α‐SRC transcription axis. Our results revealed that IDH1 K224 hyperacetylation is correlated with poor prognosis in CRC, which is a novel insight into metabolism remodeling and oncogenesis mechanism based on SIRT2‐dependent IDH1 acetylation (Fig 7C).

IDH1 has been previously reported and mainly studied in glioma 25 and acute myelogenous leukemia 26. IDH1 R132 mutation is frequently observed in these two types of malignant tumors. However, in colorectal cancer, IDH1 R132 mutation has been rarely found 27, 28. Thus, understanding how IDH1 functions in colorectal cancer may provide a new insight in CRC tumorigenesis. Nevertheless, thus far no prior study has assessed IDH1 post‐translational modification and its functions in cancers. In our previous work, we used mass spectrometry to quantify dynamic changes of non‐histone protein acetylation between matched colon cancer primary and liver metastatic tumor specimens, finding that IDH1 K224 acetylation levels rise in metastatic sites 19. In the present study, we acquired both clinical and experimental evidence supporting the critical role of IDH1 K224 acetylation in the promotion of α‐KG/HIF1α/SRC signaling pathway in the metastatic process of CRC and elucidate the mechanisms of IDH1 acetylation regulation in primary CRC tissues and liver metastatic tissues.

The clinicopathological importance of IDH1 K224 acetylation in CRC was assessed in the present study. IDH1 K224 acetylation levels in patients with CRC were correlated with the pathological stage, distant metastasis, and liver metastasis (Fig 7A). In addition, high IDH1 K224 acetylation levels were associated with poor overall survival in CRC (Fig 7B). Furthermore, multivariate analysis showed that the high IDH1 K224 acetylation level and distant metastasis were two independent prognostic factors for a poor survival in CRC patients (Table 4). Additionally, in vivo and in vitro colon cancer behavior analysis supported the fact that IDH1 K224 deacetylation may significantly repress liver metastasis of CRC cells (Figs 5 and 6). Thus, these clinical and experimental findings clearly and strongly support the conclusion that IDH1 K224 acetylation is a promoter of CRC metastasis.

Previous studies have identified the SIRT2 as a tumor suppressor, primarily due to its ability to maintain genomic fidelity during mitosis by deacetylating and stabilizing BUBR1 and APC/C activity 15, 29, 30. Chalkiadaki et al 30 discovered that SIRT2 knockout mice may develop cancers; in addition, reduced SIRT2 levels were found in variety of cancer types, including breast, liver, renal, and prostate cancers. However, the exact role of SIRT2 in colorectal cancer remains unclear. In our study, we found lower levels of SIRT2 in CRC samples and liver metastases (Fig 2). We also demonstrated that SIRT2 had a tumor suppressor role in colorectal cancer by mediating IDH1 deacetylation.

Reprogramming of metabolic pathways, including abnormal glycolysis and amino acid metabolism, is a hallmark of cancer 31. As a key metabolic enzyme, IDH1 has an important role in cell metabolism. Recent studies have shown a high metabolic dependency of CRC cells on increased usage of both glucose and glutamine to maintain cell growth 32, 33. Our results showed that both glucose and glutamine increase IDH1 K224 acetylation and decrease IDH1 enzymatic activity (Fig 3A–D). In addition, we discovered that SIRT2‐mediated IDH1 K224 deacetylation stimulates the production of NADPH and GSH, which protects cells against ROS produced during rapid cell proliferation (Fig 3E–I).

As an isocitrate dehydrogenase, IDH1 catalyzes isocitrate to produce α‐KG 34. IDH1 enzymatic activation by SIRT2‐mediated IDH1 K224 deacetylation could increase α‐KG production and decrease ROS levels thus resulting in the degradation of HIF1α. HIF1α expression further controls SRC expression through HIF1α‐dependent transcriptional regulation (Fig 4). Transcriptional co‐regulator SRC is one of the abundantly deregulated oncogenes among the landscape of genetic alterations that drive aggressive metastatic tumors 35, 36. Based on the previous findings, our results demonstrate that IDH1 K224 deacetylation regulates the metastasis of CRC by affecting α‐KG‐HIF1α‐SRC axis. Overall, our study revealed an important mechanism of IDH1 regulation via SIRT2 deacetylation, thus providing a potential strategy for colorectal cancer therapy by targeting the IDH1 K224 residue or α‐KG‐HIF1α‐SRC axis.

Materials and Methods

Cell lines and human tissue specimens

Human colorectal cancer cell lines SW480, SW620, HT29, HCT116, and LoVo were purchased from the American Type Culture Collection (ATCC), and NCM460 cells from INCELL Corporation (USA). Cells were cultured at 37°C in a humidified environment containing 5% CO₂ in RPMI‐1640 (HT29, LoVo, and HCT116) and Leibovitz's L‐15 medium (SW480 and SW620), respectively. They were supplemented with 10% fetal bovine serum (FBS, Gibco), 100 U/ml penicillin (Sigma‐Aldrich), and 100 μg/ml streptomycin (Sigma‐Aldrich). All cell lines were tested for mycoplasma contamination and authenticated using short tandem repeat profiling. The same batch of cells was thawed every 1–2 months.

Human tissue specimens were collected preoperatively in subjects undergoing coloproctectomy and hepatectomy, according to the National Comprehensive Cancer Network (NCCN) guidelines for colon/rectal carcinoma. The samples were kept at −80°C until use.

Written informed consent was provided by all patients before sample collection. The local Research Ethics Committee of Peking University People's Hospital approved this study.

Cell transfection or infection

The cDNA of IDH1 and SIRT2 was amplified and cloned into pcDNA3.1 vector. IDH1‐mutant constructs were generated using a KOD Plus Mutagenesis Kit (TOYOBO). All expression constructs were checked by Sanger sequencing. Transfection was carried out using Lipofectamine 3000 (Invitrogen). SIRT2 and IDH1 plasmids were used at 50 nM for transient transfection. Lentiviral vectors (GeneChem Co. Ltd) expressing IDH1, IDH1‐mutant, or the negative control sequence were employed to infect CRC cells for in vivo experiments.

Western blot assay

Protein was extracted from cancer cells with lysis buffer. Lysates were denatured with SDS sample buffer at 100°C for 10 min, separated in 8–12% polyacrylamide gels, and then transferred to nitrocellulose blotting (NC) membranes (Millipore). Membranes were blocked with 5% non‐fat milk powder in TBST buffer and then incubated with primary antibodies overnight at 4°C. Membranes were probed with corresponding secondary antibodies at room temperature. Band signals were visualized using enhanced chemiluminescence (ECL; Pierce, Rockford, IL), exposed to a ChemiDocTM XRSC System (Bio‐Rad); the band density was evaluated by Bio‐Rad Quantity One software.

The anti‐IDH1 K224Ac antibody was synthesized by PTM Biolabs lnc. (China), which was suitable for Western blot and immunohistochemistry staining detection.

IDH1 enzyme activity assay

Flag‐IDH1 was overexpressed in cells; proteins were immunoprecipitated by M2 beads and eluted using 250 μg/ml Flag peptide. IDH1 enzyme activity was evaluated by IDH Activity Assay Kit (Sigma‐Aldrich). The eluent was added to a reaction buffer containing IDH assay buffer, developer and NAD⁺ according to the manufacturer's instructions. The change in absorbance at 450 nm was measured using a Hitachi F‐4600 fluorescence spectrophotometer. The Michaelis constant (Km) values of the wild‐type and mutant enzymes for NADP⁺ were measured by fixing the isocitrate concentration with varying cofactor concentrations. Apparent maximum velocity (V_max) and K_m values were calculated by nonlinear regression using Prism 6.0 (Prism, San Diego, CA, USA). All kinetic parameters were recorded from at least three measurements.

In vivo experiments

BALB/c nude mice (6 weeks old, purchased from Beijing Vital River Laboratories, China) were housed in an environment with temperature of 22 ± 1°C, relative humidity of 50 ± 1%, and a light/dark cycle of 12/12 h. All animal studies (including the mice euthanasia procedure) were done in compliance with the regulations and guidelines of Peking University People's Hospital institutional animal care and conducted according to the AAALAC and the IACUC guidelines.

IDH1 knockdown HCT116 or SW620 cell lines were constructed using CRISPR/Cas9 methods. IDH1 sgRNA CRISPR/Cas9 All‐in‐One Lentivector was purchased from Applied Biological Materials (Canada). Cells were infected using IDH1 wild‐type or IDH1 K224R lentiviruses and screened using puromycin.

The mice were randomly assigned to four groups (four mice per group). Two groups were injected subcutaneously with SW620 cells, and two were treated with cells in splenic subcapsule. Subcutaneous tumors were assessed at 4‐day intervals for volume determination as V = 0.5 × L (length) × W² (width). The animals were euthanized 40 or 50 days after cell inoculation. Consequently, liver metastases, tumor volume, and weight were analyzed.

ChIP assay

The ChIP assays were performed using a ChIP kit (Millipore). Briefly, 1 × 10⁷ SW620 cells and 5 μg anti‐HIF1α antibody were used for ChIP experiment. Immune globulin G was used as a negative control.

Cell cycle and apoptosis analysis

To assess cell cycle distribution, the cells were stained using the BD Cycletest™ plus DNA reagent kit (BD Biosciences) as suggested by the manufacturer. To detect apoptosis, transfected cells (72 h) were incubated with the Alexa FluorR488 Annexin V/Dead Cell Apoptosis Kit (Invitrogen). Data analysis was carried out with FlowJo V7 (Tree Star, USA) on a BD Biosciences flow cytometer.

Cell invasion and migration assay

The invasive and migration abilities of CRC cells were evaluated using transwell assay and wound healing assay accordingly. Briefly, for transwell assay, cells were seeded into upper chamber with media containing 0.1% FBS while 30% FBS were placed in the lower chamber (24‐well plates, 8‐μm pore size, Corning). Consequently, invasive cells were stained, counted, and photographed under a microscope. For cell migration assay, a linear wound was made by scratching the surface of the plates with a yellow sterile pipette tip. After being incubated for several hours in 1% FBS cell culture media, the width of wound gaps was photographed and measured.

Co‐immunoprecipitation analysis (co‐IP)

Cells were lysed in BC100 lysis buffer [20 mM Tris–HCl (pH 7.9), 100 mM NaCl, 0.2% NP‐40, 20% glycerol] with protease inhibitor cocktail and 1 mM dithiothreitol (Sigma‐Aldrich) and 1 mM phenylmethyl sulfonyl fluoride (Sigma‐Aldrich). The lysates were incubated with anti‐IDH1 or anti‐SIRT2 primary antibody at 4°C overnight. Combined proteins were eluted with 0.1 M glycine (pH 2.5) and neutralized with 1 M Tris buffer. Elution proteins were evaluated by Western blot.

Protein purification and GST pull‐down assay

Cells were harvested and incubated in Flag lysis buffer [50 mM Tris–HCl (pH 7.9), 137 mM NaCl, 1% Triton X‐100, 0.2% sarkosyl, 1 mM NaF, 1 mM Na₃VO₄, 10% glycerol] with regular protease inhibitor cocktail and 1 mM dithiothreitol and 1 mM phenylmethyl sulfonyl fluoride. Flag‐M2 affinity gel (Sigma‐Aldrich) was carried out in first immunoprecipitation. GST‐SIRT2 were expressed and purified from Rosetta bacterial cells and bound to a GST‐agarose column (Novagen). When incubation of Flag‐IDH1 and GST‐SIRT2 was completed, beads were washed and eluted by BC100 buffer. Finally, the elution was analyzed by Western blot.

Luciferase assay

HCT116 cells in 24‐well plates were co‐transfected with luciferase vectors containing the SRC promoter sequences with or without HIF1α luciferase reporter plasmids. After 48 h, luciferase activity was measured by the Dual‐Luciferase Reporter Assay System (Promega). Data are presented as ratios between firefly and renilla activity.

Statistical analysis

All results were expressed as mean ± SD and analyzed using the SPSS 20.0 software (SPSS, Chicago, IL, USA). Differences between groups were evaluated with a two‐tailed unpaired Student's t‐test or a one‐way ANOVA or a two‐way ANOVA test according to different conditions. The relationship between IDH1 acetylation expression and clinicopathologic features of CRC was analyzed using the Pearson chi‐square test. The overall survival was analyzed by the log‐rank test using the Kaplan–Meier method. A P‐value of < 0.05 was considered statistically significant.

Author contributions

Conception and design: BW, JL, ZS. Development of methodology: BW, YY, XY, BL, ZW. Acquisition of data (provided animals, provided facilities, etc.): BW, YY, SC, KJ, WZ, HJ. Analysis and interpretation of data (e.g., statistical analysis, computational analysis): BW, XY, BL, KJ, HM, SW. Writing, review, and/or revision of the manuscript: BW, YY, XY, ZW, HM, PP, JL, ZS. Administrative, technical, or material support (i.e., reporting or organizing data): XY, ZW, BL, SC, PP, JL. Study supervision: SW, JL, ZS. Other (provides cell lines and plasmids): WZ, HJ, BL.

Conflict of interest

The authors declare that they have no conflict of interest.

Supporting information

Expanded View Figures PDF

Table EV1

Table EV2

Table EV3

Table EV4

Dataset EV1

Review Process File

EMBO Reports (2020) 21: e48183