Polo-like kinase 1 (PLK1)-dependent phosphorylation of methylenetetrahydrofolate reductase (MTHFR) regulates replication via histone methylation

ABSTRACT

Methylenetetrahydrofolate reductase (MTHFR) is a key enzyme regulating the folate cycle and its genetic variations have been associated with various human diseases. Previously we identified that MTHFR is phosphorylated by cyclin-dependent kinase 1 (CDK1) at T34 and MTHFR underlies heterochromatin maintenance marked by H3K9me3 levels. Herein we demonstrate that pT34 creates a binding motif that docks MTHFR to the polo-binding domain (PBD) of polo-like kinase 1 (PLK1), a fundamental kinase that orchestrates many cell cycle events. We show that PLK1 phosphorylates MTHFR at T549 in vitro and in vivo. Further, we uncovered a role of MTHFR in replication. First, MTHFR depletion increased the fraction of cells in S phase. This defect could not be rescued by siRNA resistant plasmids harboring T549A, but could be restored by overproduction of Suv4–20H2, the H4K20 methyltransferase. Moreover, siMTHFR attenuated H4K20me3 levels, which could be rescued by Suv4–20H2 overproduction. More importantly, we also investigated MTHFR-E429A, the protein product of an MTHFR single nucleotide variant. MTHFR-E429A overexpression also increased S phase cells and decreased H4K20me3 levels, and it is linked to a poor glioma prognosis in the Chinese population. Collectively, we have unveiled a vital role of PLK1-dependent phosphorylation of MTHFR in replication via histone methylation, and implicate folate metabolism with glioma.

KEYWORDS:

• cell cycle
• serine/threonine protein kinase
• S-adenosylmethionine (SAM)
• protein phosphorylation
• histone modification
• DNA replication
• MTHFR
• Plk1

Introduction

One-carbon metabolism, comprising folate and methionine metabolism, sits at the heart of cancer biology. Into the one-carbon pathway enter glucose and amino acids, and out of it one expects a wealth of components used for processes encompassing all aspects of life: biosynthesis, genome integrity and epigenetics.¹ The folate cycle begins with folic acid, which, upon cell entry, is reduced to tetrahydrofolate (THF). THF, in turn, is modified to 5,10-methylene-THF (me-THF). This is when methylenetetrahydrofolate reductase (MTHFR) comes into play, and reduces me-THF to 5-methyltetrahydrofolate (mTHF). It is exactly mTHF that is at the hub of the folate cycle and the methionine cycle. mTHF is demethylated to release a carbon, which is taken up by homocysteine and thus generating methionine. Methionine, then, is adenylated to make S-adenosylmethionine (SAM), which is the ultimate methyl donor for methylation reactions in the cell. Thus, carbon units are transferred between the folate cycle and the methionine cycle. Its innate cyclic nature makes it ideal to be an integrator of nutrient status, cell growth and human diseases.¹

Among all the consequences of one-carbon metabolism, of particular interest, is epigenetics.² Epigenetics communicates nutrient status outside to cellular physiology inside, and dictates cellular homeostasis maintenance or chronicle disease conditions. Histone methylation, an indispensable part of epigenetics, involves transferring the methyl group from SAM to histones by histone methyltransferases (HMTs), resulting in mono-, di-, or trimethylated lysines or arginines. Recent years have witnessed a surge of interest in H4K20 methylation, whose monomethylated form is catalyzed by SET8 (also named PR-SET7 or SETD8) and di- and tri-methylated forms are catalyzed by SUV4–20H1 and SUV4–20H2.³ Interestingly, H4K20me3, a hallmark of silenced heterochromatins, co-localizes with H3K9me3 at centromeres,⁴ telomeres,⁵ and pericentric heterochromatin regions.⁶ In fact, it has been proposed that the heterochromatin marker H3K9me3 brings its reader heterochromatin protein 1 (HP1), which subsequently recruits SUV4–20H2, thus generating H4K20me3 at the heterochromatin regions.⁷

There has been increasing awareness of the link between H4K20 methylation levels and cell cycle control, DNA damage response and DNA replication.⁷ DNA replication is the intricate process of precise genome duplication. It needs to be carefully monitored, because replication failure could be a threat to genome integrity. Cells depleted of Set8, the HMT of H4K20me1, accumulate in S phase.^8-10 Set8 is normally degraded in S phase. When a non-degradable SET8 is introduced into cells, H4K20me3 aberrantly accumulates in S phase,¹¹ correlating with aberrant re-replication phenotypes.¹² These data suggest it is not methylation per se, but the precise timing of methylation that matters. Knockdown of SUV4–20H2, the essential enzyme that generates H4K20me3, also culminated in more S phase cell fractions.¹³ Forced targeting of Set8 to genomic foci aberrantly recruits the origin of replication complex (ORC) in an H4K20me3 and SUV4–20H2-dependent fashion,¹² suggesting that H4K20me3 governs the replication pathway.

Perhaps most enlightening of all, is the finding that the replication protein machinery actually harbors readers of H4K20 methylation. During G1 phase of the cell cycle at the replication origins, a legion of the pre-replication complex (pre-RC) assembles, enlisting ORC, the minichromosome maintenance complex (MCM), Cdc6 and Cdt1.¹⁴ The bromo-adjacent homology (BAH) domain of ORC1 reads H4K20me2,¹⁵ while the WD repeat domain of ORCA/ORWD1 reads H4K20me3.^12,16 Indeed, knockout of both SUV4–20H1 and SUV4–20H2 in mice is perinatally lethal.¹⁷ And ORCA associates with late-firing origins during G1 to regulate heterochromatin replication.¹⁸ It is thus conceivable that H4K20me2/3 stabilizes the ORC complex, the disruption of which delays cell division and proliferation.

MTHFR has long been the culprit for a swathe of human diseases, for instance, occlusive vascular disease, neural tube defects and colon cancer.¹⁹ Targeting MTHFR, methotrexate and pemetrexed have been approved for multiple cancers.²⁰ Two most intensively studied pathogenic single nucleotide variants (SNVs) are C667T and A1298C.²¹ A role of MTHFR in cell cycle control has never been suspected, although its multitude of phosphorylation events caught attention a decade ago.²² We reported before that T34 of MTHFR is phosphorylated by cyclin-dependent kinase 1 (CDK1), the crucial kinase that orchestrates cell cycle progression, and found that MTHFR maintains centromeric heterochromatin by modulating H3K9me3 levels.¹⁹ Here we further demonstrate that pT34 actually creates a docking motif for the polo-binding domain (PBD) of polo-like kinase 1 (PLK1), another master kinase of cell division.²³ PLK1 in turn phosphorylates MTHFR at T549, thus affecting S phase progression via H4K20me3. Importantly, we show that E429A, encoded by A1286C (chr1:11854476:G:70 COSM3735923), is implicated in the prognosis of Chinese glioma patients, and might as well follow the same route by affecting the S phase and H4K20me3. We propose that the cell cycle role of MTHFR intersects histone methylation levels and tumorigenesis.

Results

Association between MTHFR and PLK1

Previously, we identified that Cdk1 phosphorylates MTHFR at T34 in vitro and in vivo.¹⁹ A closer look surrounding T34 (Fig. 1A) reminded us of the motif (S-phospho S/T-P) that binds to the PBD of PLK1.²⁴ We also performed mass spectrometry (MS) analysis using HeLa cells extracts transiently expressing Flag-MTHFR, and PLK1 was identified in the anti-Flag immunoprecipitates (data not shown). Naturally, we tested the interaction between MTHFR and PLK1. HeLa cells extracts treated with Noc or mock treated were subject to IP assays using anti-MTHFR antibodies, and PLK1 was identified in the IP complexes (Fig. 1B). Then Flag-MTHFR and HA-PLK1 were co-transfected into HeLa cells, and reciprocal coIP interaction was identified between these two proteins (Fig. 1 C-D). Moreover, recombinant GST-PLK1 efficiently pulled-down His-MTHFR (Fig. 1E). Collectively, these data suggest that PLK1 directly interacts with MTHFR.

Figure 1. MTHFR interacts with PLK1. (A) A comparison of PLK1 polo-box domain (PBD) binding motif and various MTHFR homologues at T34. (B) MTHFR and PLK1 coIP. HeLa cells were synchronized in mitosis by thymidine-nocodazole block, then subject to IP by indicated antibodies. IgG immunoprecipitates were used as a negative control. (C-D) Epitope-tagged MTHFR and PLK1 were transfected into HeLa cells and subject to IP experiments with the antibodies indicated. (E-F) GST pull-down assays. Binding of GST-PLK1-FL with recombinant MTHFR was tested (E). The asterisk indicates a non-specific band. Binding between GST-PLK1-PBD and overexpressed MTHFR WT or T34A was examined in (F). (G) Domain structures of MTHFR. The boundaries of MTHFR-N or C fragments and mutations used in this study are indicated. (H) PLK1-kinase domain (KDN) was transfected into HeLa cells, together with MTHFR-FL, N or C constructs. Then coIP experiments were performed using antibodies indicated.

Then we investigated whether pT34 is essential for the interaction between MTHFR and PLK1. When we mutated T34 into phospho-deficient Ala and transfected wild-type (WT) and T34A of MTHFR into HeLa cells, PLK1-PBD failed to pull-down T34A (Fig. 1F), suggesting that the phosphorylation of T34 is crucial for MTHFR to dock onto the PBD of PLK1. Since MTHFR consists of an N-terminal catalytic domain and a C-terminal regulatory domain¹⁹ (Fig. 1G), we examined which domain interacts with PLK1. Full-length (FL), N and C-termini of MTHFR were transfected into HeLa cells, together with the kinase domain (KDN) of PLK1. Only the C-terminus of MTHFR efficiently coIPed with PLK1-KDN (Fig. 1H). Taken together, these data suggest that CDK1-induced T34 phosphorylation docks MTHFR onto PLK1 PBD, then PLK1-KDN binds with the MTHFR C-terminus.

Phosphorylation of MTHFR by PLK1

According to the established paradigm of PLK1 activation, PLK1 KDN will phosphorylate the substrate after it docks to the PBD domain.²⁵ Therefore, we tested whether PLK1 phosphorylates MTHFR by an in vitro kinase (IVK) assay. Since only the C-terminus binds with PLK1-KDN, we purified the C-terminus of MTHFR from E. coli, and incubated it with recombinant PLK1. PLK1 phosphorylated the MTHFR C-terminus efficiently (Fig. 2A). We also tried to purify the N-terminal catalytic domain of MTHFR, but it does not express sufficiently, as previously reported.²⁶

Figure 2. PLK1 phosphorylates MTHFR at T549. (A) An in vitro kinase (IVK) assay using recombinant PLK1 and MTHFR-C in kinase buffers containing ³²P-ATP. Coomassie blue (CBB) staining showed input MTHFR- C proteins. (B) T549 of MTHFR is phosphorylated by PLK1 in vitro. From this collision-induced dissociation spectrum, a phosphorylated peptide ENI(pT)NAPELQPNAVTWGIFPGR of MTHFR was identified. “b” and “y” ion series represent fragment ions containing the N- and C-termini of the peptide, respectively. (C) A comparison of T549 and the PLK1 phosphorylation consensus motif. The residues in red fit the consensus motif. (D)An IP-kinase assay with the MTHFR pT549 antibody. HeLa cells were transfected with Flag-PLK1 WT or kinase dead (KD), and then the anti-Flag immunoprecipitates were incubated with recombinant GST-MTHFR-WT or T549A. (E) MTHFR is phosphorylated at T549 in vivo. HeLa cells were transfected with HA-MTHFR-WT or T549A, treated with BI2536 (a PLK1 inhibitor) or without, then the total lysates were subject to IB with indicated antibodies.

The IVK products were then analyzed by liquid chromatography-tandem mass spectrometry (LC-MS/MS), and only one phosphorylation site, T549, was identified (Fig. 2B).We then constructed the phospho-deficient T549A mutant, and recombinant MTHFR-T549A failed to be phosphorylated in the IVK assay (Fig. 2A). Moreover, T549 also conforms to PLK1 consensus site and is conserved among various species (Fig. 2C). The result indicated that T549 was the major phosphorylation site.

We proceeded to prepare the phospho-specific antibody targeting pT549. In an IP-kinase assay (Fig. 2D), Flag-PLK1 immunoprecipitates efficiently phosphorylated recombinant MTHFR, but the PLK1-kinase dead (KD) did not, nor when MTHFR-T549A was used as the substrate, as detected by the pT549 antibodies. Then we examined whether the phosphorylation of T549 occurs in vivo. HeLa cells were transfected HA-MTHFR WT or T549A, treated or not treated with BI2536, an inhibitor of PLK1 (Fig. 2E). Cell extracts were then subject to IB with pT549 antibodies, and a crisp band was only detected in the HA-MTHFR-WT lane without BI treatment, suggesting that PLK1-dependent phosphorylation of MTHFR T549 occurs in vivo.

MTHFR-depletion increases S phase cells

As PLK1 is a key kinase that orchestrates many mitotic event, including the recently identified process of replication,²⁵ we sought to examine whether MTHFR depletion is relevant to cell cycle control. MTHFR siRNA efficiently knocked down MTHFR proteins (Fig. 3A). MTHFR-depleted and control HeLa cells were analyzed by flow cytometry using propidium iodide (PI) staining (Fig. 3B). MTHFR depletion readily increased G1/S cells from ∼20% to ∼40% (Fig. 3B), indicating that S phase cells accumulated in these cells.

Figure 3. MTHFR depleted cells accumulate in the S phase. HeLa cells were transfected with siMTHFR or control siRNA (A) and then analyzed by flow cytometry after PI staining (B). The percentage of cells in different cell cycle stages were shown in (B).The result is representative of 3 independent experiments. (C) Cells were pulsed with BrdU for 30 min, then stained with DAPI after BrdU staining, and examined by fluorescence microscopy. Scale bar, 10 µm. The percentage of BrdU positive cells were indicated in (D). More than 250 cells were counted in each experiment. Histograms represent mean ± SD of 3 independent experiments (p = 0.021< 0.05). Cells were pulse-labeled with BrdU as in (C), then measured for DNA synthesis rate and DNA content by flow cytometry. Quantitation was shown in (E) (p = 0.043 < 0.05).

To further confirm that S phase cells did accumulate, we pulsed the cells with 5-bromo-2′-deoxyuridine (BrdU), then stained them with standard BrdU protocols (Fig. 3C). As BrdU incorporation only occurs in the S-phase of the cell cycle, nuclear BrdU staining indicates S phase cells. As shown in Fig. 3D, MTHFR-depleted cells increased ∼1.5-fold of BrdU positive cells as determined by immunofluorescence. We then used BrdU-labeled cells for flow cytometry analysis (Fig. 3E). More siMTHFR cells were positive for BrdU staining, compared with WT (Fig. 3E). These results indicated that more MTHFR-depleted cells were actively replicating their DNA compared with WT.

Then we sought after the possibility whether S phase accumulation was due to checkpoint activation. We examined Chk1 phosphorylation in MTHFR-depletion cells, and Chk1 was not activated (data not shown), suggesting that cell cycle progression did not arrest.

MTHFR(1286A > C) correlates with poor glioma survival rates in Han Chinese populations

Recently, RNA-seq of 272 Han Chinese gliomas revealed a novel, recurrent fusion transcript involving the PTPRZ1 and MET gene (ZM) that was found in 15% of secondary glioblastoma (sGBMs).²⁷ The same analysis also unveiled a great many SNVs associated with both low grade gliomas and glioblastomas. Survival analysis revealed that the 7-year cumulative recurrence-free survival in sGBM patients carrying MTHFR(1286A>C)(chr1:11854476:G:70 COSM3735923) heterozygosity (AC:63 patients; CC: 8 patients) was significantly worse than those with WT MTHFR (p = 0.0068, Fig. 4A). As MTHFR(1286A>C) encodes MTHFR-E429A, we also incorporated it in our following studies.

Figure 4. MTHFR(1286A>C) correlates with a poor prognosis in Chinese glioma patients. Kaplan–Meier curves estimate the recurrence-free survival rates.

Cells harboring T549A and E429A failed to rescue the S phase accumulation defects

We then examined whether the phosphorylation of MTHFR by PLK1 contributes to S phase accumulation. We performed rescue assays using HA-MTHFR-res, T549A-res and E429A-res plasmids. These plasmids were transfected into HeLa cells and the resultant stable transfectants were treated with siMTHFR, as indicated in Fig. 5A. Endogenous MTHFR was attenuated substantially (decreased ∼50%), while the exogenous HA-MTHFR-res, T549A-res and E429A-res were stably expressed, as shown by IB (Fig. 5A).

Figure 5. T549A and E429A failed to rescue the S phase defects of MTHFR-depleted cells. (A) Western blots showing stable transfectants harboring rescue, rescue-T549A and rescue-E429A plasmids. (B) Cells in (A) were analyzed by BrdU staining and PI as described in materials and methods, and then analyzed by flow cytometry. (C) Quantitation of results in (B). Histograms represent mean ± SD of 3 independent experiments. Asterisks indicate significant differences from Lane 1 (p values are: p1–2 = 0.0003, p1–3 = 0.42, p1–4 = 0.024, p1–5 = 0.017). (D) Cells were transfected with control vectors, or MTHFR-WT, E429A plasmids. Then extracts were collected and analyzed by IB with the antibodies indicated.

These cells were then pulsed with BrdU and subject to flow cytometry analysis (Fig. 5B). MTHFR depletion increased BrdU incorporation. Re-expression of HA-MTHFR-res restored the defects (Fig. 5B). The T549A-res and E429A plasmids, however, did not efficiently recover the S phase defects. The results were then subject to t-test (Fig. 5C), and there is significant difference among WT, and T549A-res, E429A-res plasmids (Fig. 5B-C). Thus, the rescue experiments indicate that pT549 and E429 are essential for MTHFR to participate in S phase progression.

Since MTHFR-T549A and E429A manifested similar cell cycle defects, we examined T549 phosphorylation levels in cells haboring E429A. Cells were transfected with MTHFR-WT and E429A plasmids, then extracts were collected and analyzed by Western analysis (Fig. 5D). Indeed, pT549 was abolished in E429 cells, suggesting a common mechanism between these two mutants.

MTHFR-depletion dampens H4K20me3 levels

Previously we have demonstrated that MTHFR depletion results in relaxation of heterochromatin and a decrease of H3K9me3 levels.¹⁹ Another marker of heterochromatin, H4K20me3, is mostly enriched in quiescent chromatin, and colocalizes with H3K9me3⁷. Most importantly, knockdown of Suv4–20H1 and Suv4–20H2, the two enzymes that create H4K20me3, results in an increased fraction of cells in S phase.¹³ Thus, we examined H4K20me3 levels by ChIP in MTHFR-depleted cells. And we found that siMTHFR decreased H4K20me3 levels across a panel of pericentric heterochromatin loci (Fig. 6A). Moreover, re-expression of WT-res plasmids could efficiently rescue the H4K20me3 levels, but not expression of the T549A or E429A mutants (Fig. 6B). Taken together, the results suggest that MTHFR might exert its impact on cell cycle progression via H4K20me3.

Figure 6. T549 phosphorylation of MTHFR influences H4K20me3 levels, a hallmark of heterochromatin. (A) HeLa cells depleted of MTHFR or control were analyzed by ChIP assays using H4K20me3 antibodies across a panel of pericentromeric heterochromatin regions. (B) HeLa cells were stably transfected with vector, rescue, rescue-T549A or E429A plasmids, and then analyzed by ChIP. qPCR data are presented as means ± SD of 3 independent experiments.

SUV4–20H2 suppresses the S phase accumulation and H4K20me3 defects

Genetically, if the S phase defects were caused by H4K20me3 defects, then overexpression of SUV4–20H2, the enzyme that converts H4K20me2 to H4K20me3, should rescue both defects. To test the hypothesis, we overexpressed HA-SUV4–20H2 in MTHFR-depleted cells (Fig. 7A), and indeed H4K20me3 increased in MTHFR-depleted cells by ChIP analysis (Fig. 7B). When the cell cycle profiles were measured by BrdU staining followed by flow cytometry, cells accumulating in S cells were significantly decreased (Fig. 7C), thus unveiling the mechanistic link between MTHFR and cell cycle progression.

Figure 7. Overproduction of the H4K20 trimethyltransferase Suv4–20H2 rescues S phase and H4K20me3 levels in MTHFR-depleted cells. (A) IB indicated Suv4–20H2 overproduction in MTHFR-depleted cells. (B) ChIP analysis of the cells in A. qPCR data are presented as means ± SD of 3 independent experiments. (C) The cells in A were stained with BrdU and analyzed by flow cytometry, showing that Suv4–20H4 overproduction rescues the S phase accumulation defects in MTHFR-depleted cells. The asterisk indicates significant differences from Lane 1 (p values are: p1–2 = 0.002, p1–3 = 0.05).

Discussion

In a continued endeavor to study MTHFR, we identified Plk1 as its interacting partner. MTHFR is phosphorylated by CDK1 at T34, and thus targeted to the Plk1 PBD domain and in turn phosphorylated by Plk1 at T549. We show here that both MTHFR-T549A and E429A exhibit S phase accumulation as well as H4K20me3 level defects, both of which could be restored by SUV4–20H2 overproduction. As E429A glioma patients manifested poorer survival rates in Han Chinese and also decreased pT549 levels, we thus propose a model where MTHFR links the heterochromatin landscape, cell cycle profile and tumor progression.

Plk1's role in mitosis has long been appreciated, in centrosome duplication, mitotic Golgi dispersal, mitotic entry, chromosome segregation and cytokinesis.²⁸ Plk1 also governs replication, the inkling of which came as early as 1996, when Cdc5, the yeast homolog of human Plk1, was reported to phosphorylate Dbf4 in vitro, one of the DNA replication initiation proteins.²⁹ Then the coupling of Plk1 and replication became more obvious, as Plk1 is found to interact with and phosphorylate several components of the replication machinery: Mcm2 and 7,^30,31 Orc2³² and the histone acetylase Hbo1.³³ And it came to no surprise that long-term Plk1 knockdown slowed S phase progression.³⁴ By showing that Plk1-phosphorylated MTHFR ensures a steady level of H4K20me3, our study adds to the already versatile role of Plk1 during the cell cycle.

The crosstalk between master cell cycle kinases and the epigenetics machinery is not few and far between. Suv39H1, the HMT specifically for H3K9me3, another heterochromatin marker, is phosphorylated by CDK2 during replication and then dissociates from chromatin, thus affecting heterochromatin replication timing.³⁵ Set8, the HMT for H4K20me1, is phosphorylated by CDK1 during mitosis, followed by chromatin dissociation, dephosphorylation and subsequent ubiquitination.³⁶ Plk1 has also been implicated in activating Haspin, the atypical mitotic kinase responsible for H3pT3 and subsequent targeting of Aurora B to centromeres.^37,38

Our results are in line with the report that knockdown SUV4–20H2, the HMT of H4K20me3, culminated in increase of cells in S phase.¹³ It is conceivable that compromised H4K20me3 may destabilize the chromatin bound ORC complex and thus causing S phase defects.^12,39 Alternatively, H4K20me3 may execute its function via Rb. It has been known that H4K20me3 localization is dependent on Rb.⁴⁰ And recently it was shown that overexpression of SUV4–20H2 rescued the cohesin and H4K20me3 defects of Rb-depleted cells.⁴¹ Since replication defects could very well lead to cohesin defects,⁴² our data suggest that cohesin phenotypes in Rb-devoid cells might be secondary to replication.

As we reported before, MTHFR also affects H3K9me3 levels.¹⁹ Intriguingly, KDM4D, the H3K9me3 demethylase, is recruited to replication origins by ORC and MCM complexes, which further enlist Cdc45, PCNA and polymerases.⁴³ It was postulated that H3K9me3 levels were attenuated by Kdm4d to facilitate formation of the pre-initiation complex. Concomitantly, SUMOylation of ORC2 is also involved in recruiting KDM5A, the demethylase to convert H3K4me3 to H3K4me2, the ablation of which caused re-replication of the heterochromatin.⁴⁴ Hence epigenetics may fine-tune replication through multiple channels.

It is possible that some methylation marks are more sensitive to SAM shortage than others. MTHFR depletion does not affect H3K9me1 or H3K9me2¹⁹. The budding yeast homolog of MTHFR is Met13. Deletion of MET13 specifically reduced H3K4me3 and H3pT11, but not H3K4me1/2 or H3K36me3, suggesting that some types of chromatin modifications are more sensitive to serine derived from glucose.⁴⁵ H3K9 methylation was not examined due to lack of the catalytic enzyme in S. cerevisiae (Dr. Shanshan Li, personal communication).

A link between histone modification and metabolism could be conserved.⁴⁶ The yeast SESAME (Serine-responsive SAM-containing Metabolic Enzyme) Complex comprises Pyk1 (the yeast homolog of Pyruvate kinase M2 [PKM2]), Serine metabolic enzymes, SAM synthetases and an acetyl-CoA synthetase.⁴⁵ SESAME senses glycolysis and serine metabolism and thus modulates the crosstalk between H3K4me3 and H3pT11.⁴⁵ Threonine depletion in mouse embryonic stem cells decreased SAM and H3K4me3 levels, culminating in increased differentiation.⁴⁷ Congruently, methionine deprivation in human pluripotent stem cells decreases SAM levels, rapidly reduces H3K4me3 levels, modestly dampens DNA methylation levels and potentiates cellular differentiation.⁴⁸ Hence the cycle of glucose and one-carbon metabolism might be interconnected with cell cycle and cell fate.

Our results do not exclude the possibility that the cell cycle profile change might be contributed by DNA methylation alteration induced by MTHFR depletion. As reported previously, DNA methyltransferase 1 (DNMT1) knockdown triggered an intra-S-arrest of DNA replication.⁴⁹ siDNMT1 activates the genotoxic stress checkpoint kinase cascade, including Chk1 and Chk2.⁵⁰ In MTHFR-depleted cells, however, we did not observe Chk1 activation (data not shown), suggesting the mechanism might be different.

Glioma is particularly prone to metabolic disruptions, in particular, one-carbon metabolism.⁵¹ Moreover, glioma cancer incidents are imbued with ample cases of histone alterations. In a tour de force study by Lewis et al.,⁵² it was found that pediatric diffuse intrinsic pontine glioma (DIPG) samples containing the K27M mutation in histone H3 impaired H3K27me3 levels. Expression levels of enhancer of zeste homolog-2 (EZH2), the regulator of H3K27me3, can be used to stratify low- and high-grade astrocytic tumors.⁵³ Strikingly, small molecule EZH2 inhibitors could effectively abolish tumor growths in H3K27M mutant DIPG mouse models.⁵⁴ Hence, the possibility of targeting epigenetics for brain tumor therapeutics could be limitless.

H4K20me3 only constitutes less than 2% of cellular H4,⁷ yet ripples across the metabolic network could cause undesired repercussions in cell cycle regulation, epigenetics and diseases states. We posit that Plk1 assists MTHFR in ascertaining H4K20me3 levels, so that a faithful and unerring replication will ensue.

Experimental procedures

Antibodies, plasmids, and cell culture

HeLa cells were grown in DMEM (Hyclone, UT, USA) containing 10% fetal bovine serum and 1% penicillin-streptomycin at 37°C in a 5% CO₂ incubator.

Rabbit polyclonal antibodies against MTHFR were purchased from Bethyl Laboratories Inc. (Montgomery, TX, USA, Cat. # A303–624A). Anti-PLK1 antibodies were from Santa Cruz (Cat. # sc17783). Antibodies against MTHFR phospho-T549 (pT549-Ab) were raised in rabbits using the sequence of CGENI(pT)NAPEL and manufactured by Beijing B&M Biotech Co., Ltd. MTHFR-T549A and T549D plasmids were generated using specific primers (sequences available upon request) following the manufacturer's instructions (QuickChange II, Stratagene). MTHFR siRNA sequences and plasmids were described previously.¹⁹

In vitro kinase assays (IVK)

The PLK1 IVK assay was performed as described previously.⁵⁵ Briefly, recombinant PLK1 kinase (R & D systems) was incubated with purified GST-MTHFR with 10 mM HEPES (pH 7.5), 50 mM NaCl, 2 mM MgCl₂, 1 mM Dithiothreitol, 1 mM EGTA, 1 mM ATP or 1 μCi of γ-[³²P]ATP. After 20 min at 30°C, reactions were stopped by the sample buffer. Protein samples were separated by SDS-PAGE and phosphate incorporation was determined by PhosphorImager. Liquid chromatography-tandem mass spectrometry (LC-MS/MS) analysis was performed as described.⁵⁶

Flow cytometry and cell cycle analysis

Flow cytometry analysis was performed as described.⁵⁷ Cells were collected by trypsinization and washed once with PBS, fixed with cold 75% ethanol and stored at 4°C overnight. Fixed cells were blocked with 2% BSA in PBST (PBS with 0.1% Tween 20) for 0.5 h at room temperature, and incubated with PBS containing propidium iodide (PI) (8 μg/ml) and RNAse A (1 μg/ml) at 37°C for 0.5 h, and then subjected to flow cytometry analysis.

For 5-bromo-2′-deoxyuridine (BrdU) incorporation, cells were fixed in 70% ethanol at 4°C for 1 h. BrdU incorporation was detected by using FITC-linked anti-BrdU (PharMingen) antibodies according to the manufacturer's protocols (HCl denaturation method). After washing, cells were resuspended in PI, then subject to flow cytometry analysis. Cells stained with BrdU were also examined with indirect immunofluorescence as described before .⁵⁷

Immunoprecipitation (IP) and immunoblotting (IB) assays

IP and IB experiments were performed as described before.⁵⁵ The following primary antibodies were used for IB: anti-β-actin (1:10000), anti-HA (1:1000), and anti-FLAG M2 (Sigma) (1:1000), anti-MTHFR (1:1000), anti-Plk1 (1:1000), anti-MTHFR-pT549 (1:100). Peroxidase-conjugated secondary antibodies were from JacksonImmuno Research. BI2356 (PLK1 inhibitor) was used at 1 μm for 2 hours. Nocodazole (Noc) was used at 100 ng/ml for 16 hours. Blotted proteins were visualized using the ECL detection system (Amersham). Signals were detected by a LAS-4000, and quantitatively analyzed by densitometry using the Multi Gauge software (Fujifilm). All western blots were repeated for at least 3 times.

Chromatin Immunoprecipitation (ChIP)

HeLa cells were cross-linked for 10 min with formaldehyde (final concentration 1%), and then stopped by glycine, followed by cell lysis and sonication. Cross-linked chromatin was incubated with corresponding antibodies at 4°C overnight. Cross-linking was reversed at 4°C for 2 h with the resultant purified DNA analyzed by quantitative polymerase chain reaction (qPCR) using primer sets for target heterochromatic loci. Primers sets were described before.³⁵

Abbreviations

Plk1	=	polo-like kinase 1
MTHFR	=	Methylenetetrahydrofolate reductase
IP	=	Immunoprecipitation
IB	=	Immunoblotting
IVK	=	In vitro kinase assays
CDK1	=	cyclin-dependent kinase 1
PBD	=	the polo-binding domain
SAM	=	S-adenosylmethionine

Disclosure of potential conflict of interest

The authors declare that they have no conflicts of interest with the contents of this article.

Author contributions

J. L. wrote the manuscript; J. L., X. X. and W-G. Z. designed the project and analyzed the data; Y. D. and M-Q. D. performed MS analysis; K.Y. and X-D. S analyzed the glioma data; X. L., S. N., Q. G. and B. Z. performed all the other experiments. All authors reviewed and approved the manuscript.

Acknowledgments

We thank Dr. Huiqiang Lou for support and members of the Xu laboratory for comments on the manuscript. This work was supported by the National Natural Science Foundation under Grant 31461143012 and 31130017 (to X.X.); the 973 projects under Grant 2013CB911002 and 2015CB910601 (to X. X.); the Beijing Nova Program Interdisciplinary Cooperation Project under Grant Z161100004916042 (to J. L.); and the CNU Interdisciplinary Project (to J. L.).