The linkage of methionine addiction, overmethylation of histone H3 lysines and malignancy demonstrated when cancer cells revert to methionine-independence

Methionine addiction is a fundamental and general hallmark of cancer and is an area of current intense interest. Methionine addiction results from the overuse of methionine by cancer cells for excess transmethylation reactions. In order to identify excess transmethylation reactions in cancer and further understand the basis of methionine addiction, we compared the histone H3 lysine-methylation status and malignancy between methionine-addicted cancer cells and their methionine-independent revertants which have regained the ability to grow on low levels of methionine or independently of exogenous methionine. The levels of trimethylated histone H3 lysine marks were reduced in methionine-independent revertants compared to parental cancer cells in vitro. Tumorigenicity and experimental metastatic potential in nude mice were also highly reduced in the methionine-independent revertants compared to the parental cells. Our present results demonstrate that overmethylation of histone H3 lysines is linked with methionine addiction of cancer and to malignancy which suggests a possible causal relationship.


Introduction
Methionine addiction is a fundamental and general hallmark of cancer discovered by one of us (RMH) [1][2][3][4] and is an area of current intense interest [5][6][7] . Methionine addiction is characterized by a requirement for exogenous methionine for growth by cancer cells even though the methionine-addicted cancer cells synthesize normal or excess amounts of methionine [1][2][3][4] . Methionine restriction (MR) by either methionine-free medium 8 or by a low-methionine diet 9 or by methioninase 10 selectively arrests cancer cells in the late S/G 2 phase of the cell cycle, but not normal cells, where the cancer cells become sensitive to cytotoxic chemotherapy 11 . Methionine addiction results from the overuse of methionine by cancer cells for excess transmethylation reactions 12 , termed the Hoffman-effect, analogous to the Warburg effect for glucose overuse by cancer cells 13,14 . Methionine addiction is tightly linked to other hallmarks of cancer 13,15 and is thought possibly the very basis of malignancy itself 16 .
Although we have long since known methionine is overused for transmethylation reactions cancer cells 12,17 , we have poorly understood the fate of at least a significant amount of the excess methyl groups that were transferred. Our previous studies have shown that histone H3 lysine marks are overmethylated in cancer cells compared to normal cells 18,19 and that the histone H3 lysine overmethylation is unstable during MR of methionine-addicted cancer cells which arrests their proliferation. In contrast the lesser amount of histone H3 lysine methylation is stable in normal cells under MR, which unlike cancer cells do not arrest their proliferation 18 . These results suggested that histone H3 lysine overmethylation may be related to methionine addiction.
In the present study, we report that methionine-independent revertants, selected in methionine-free medium containing homocysteine or low methionine from 5 methionine-addicted cancer cells, lose over-methylation at histone H3 lysine and malignancy compared to methionine-addicted cancer cells. Importantly, methionine-independent revertants lose their requirement for exogenous methionine. Our results indicate that histone H3 lysine overmethylation is linked with methionine addiction and malignancy, suggesting a possible causal relationship.

Cancer cells selected to grow in low-methionine media revert from methionine-addiction to methionine independence.
We used recombinant methioninase (rMETase) to deplete methionine in vitro. The level of methionine in the medium was rapidly decreased to less than 30% within 1 hour after addition of 1 U/ml of rMETase (Fig. 1B).
To compare the methionine requirement of methionine-addicted parental cancer cells and low-methionine-requirement revertants derived from the parental cells, we then evaluated their cell proliferation kinetics under methionine restriction effected by rMETase. The proliferation of parental methionine-addicted cancer cells arrested within 72-96 hours of methionine restriction. In contrast, methionine-low-requirement revertants were able to continuously proliferate under methionine restriction, similar to normal cells 13,15,17,20 (Fig.   1C). We also evaluated the cell proliferation of low methionine-requirement revertatnts in methionine-free medium containing homocysteine or methylthioadenosine (MTA).
Low-methionine-requirement revertants proliferated in medium which methionine was replaced by either homocysteine or MTA similar to normal cells (Fig. 1D). In contrast, parental methionine-addicted cancer cells arrested in either replacement medium. These results indicate that the revertants have lost their methionine addiction and requirement for 6 exogenous methionine.

Methionine synthase and methylthioadenosine phosphorylase are upregulated in methionine-independent revertants.
To determine the differences in methionine metabolism in methionine-addicted cancer cells and methionine-independent revertants, we examined changes in protein expression levels for enzymes involved in methionine biosynthesis in parental methionine-addicted HCT 116 and H460 cells and their respective methionine-independent revertants ( Fig. 2A). We found the methionine synthase (MTR) was modestly upregulated in methionine-independent revertants of each cell line and that methylthioadenosine phosphorylase (MTAP) was strongly upregulated in the HCT 116 revertant cells and moderately upregulated in the H460 revertant cells (Fig. 2B).

Overmethylation of histone H3 lysine marks is strongly decreased in methionine-independent revertants.
We then compared the methylation status of histone H3 lysine marks in methionine-addicted parental cells and methionine-independent revertants. Parental cancer cells and methionine-independent revertants were cultured in normal methionine-containing medium for 96 hours and histones were extracted. The overall level of lysine methylation of histone H3 was decreased in the methionine-independent revertants compared to parental methionine-addicted cells even in the presence of methionine (Fig. 2C). The levels of specific methylation marks including H3K4me3, H3K9me3, H3K27me3, H3K36me3 and H3K79me3 were also decreased in the methionine-independent revertants compared to parental methionine-addicted cells in the presence of the methionine (Fig. 2C). The levels of pan-methyl lysine of H3, H3K4me3, H3K9me3, H3K27me3, H3K36me3 and H3K79me3 7 were much lower in HCT 116-R cells than in HCT 116 cells as shown by immunoblotting (Fig. 2C). However, the expression levels of several histone methyltransferase enzymes which methylates H3K9me3 did not change in the methionine-independent revertants (Fig. 2D).
These results show that methionine addiction and histone H3 lysine-mark overmethylation are linked to decreased use of methionine for histone H3 lysine methylation.

Methionine-independent revertants lose malignancy.
To compare the malignancy of parental methionine-addicted cancer cells and methionine-independent revertants derived from them, the tumorigenicity of the parental and revertant cells was compared in subcutaneous xenograft mouse models. Methionine-addicted parental HCT 116 cells and their methionine-independent revertants (HCT 116-R) were compared for their ability to form tumors in nude mice. The mean tumor volume was significantly lower in HCT 116-R tumors than HCT 116 tumors after injection of 1 × 10 6 cells in nude mice (P = 0.0085), and only half of 10 mice formed tumors in HCT 116-R compared to all mice with HCT 116 (Fig. 3A). While 8 out of 10 mice injected with 5 × 10 5 parental methionine-addicted HCT 116 cells formed tumors, no mice injected with 5 × 10 5 HCT 116-R cells formed tumors.
The metastatic potential of HCT 116-GFP expressing cells and HCT116-R-GFP expressing cells was also compared after spleen injection the two cell types in nude mice. The parental HCT 116-GFP cells formed significantly more experimental liver metastasis compared to revertant HCT-116-R-GFP cells (p = 0.011) (Fig. 3B). These results show that methionine-independent revertants have significantly reduced malignancy.
Immunohistochemistry staining showed that the number of Ki-67-positive cells was significantly lower in the HCT 116-R subcutaneous tumors compared to the parental HCT 8 116 subcutaneous tumors (n = 4, P < 0.0001) indicating reduced in vivo proliferation of the methionine-independent revertants (Fig. 3C). The tumorigenicity, metastasis and cell proliferation results indicate that methionine-independent revertants are much less malignant than their methionine-addicted parents.
The malignancy decreases in methionine-independent revertants compared to parental methionine-addicted cancer cells is linked with a decrease in histone H3 lysine methylation as described above. The results of the present study and our previous study 21 suggest that overmethylation of histone H3 lysine marks linked to both methionine-addiction and malignancy and their relationship may be causal.

RNA sequencing of methionine-addicted parent and methionine-independent revertant colon cancer and lung cancer cell lines show no large scale difference in gene expression.
To determine if large-scale changes in gene expression are involved in the reversion of cancer cell lines from methionine addiction to methionine-independence and loss of tumorigenicity, we compared mRNA transcript levels in parent and revertant cells from the H4CT 116 cell line and the H460 cell line. In Table 1, we show genes with transcription levels that were significantly elevated or depressed more than two-fold between the parent and revertant cell lines. Of the some 19,000 human genes measured for expression, only a small number met these criteria. For the HCT 116 parental cells, we found 14 down-regulated at least two-fold and 7 genes that were up-regulated by at least two-fold compared to HCT 116 revertant cells (Table 1A). The corresponding numbers for the H460 cells were 80 genes down-regulated and 14 up-regulated genes in the revertants compared to parental cells (Table   1B). We found no common genes whose expression changed two fold or more between the lung-cancer and colon-cancer derived cell lines. These results suggest that there may be no common transcriptional pathway that would account for the reversion of methionine-addicted 9 cancer cells to methionine-independence, or the methods used were not sensitive enough to detect such common genes. With the possible exception of the FOLR1 gene encoding the alpha folate receptor, none of these genes with a two-fold or more change appeared to be involved in methionine metabolism.

Changes in expression of methionine metabolism genes in methionine-addicted parental cancer cells and methionine-independent revertants.
We then examined the changes in gene expression for genes involved in the biosynthesis of methionine, its conversion to S-adenosylmethionine, and the use of S-adenosylmethionine for polyamine biosynthesis and transmethylation reactions, as well as genes encoding enzymes involved in demethylation ( Figure 2B; Table 2; Supplemental Table   1). These genes included methionine synthase and related proteins, S-adenosylmethionine synthetases, enzymes and proteins of the folate cycle that provide methyl groups for methionine biosynthesis, methionine salvage and polyamine biosynthetic proteins, and enzymes involved in the metabolism of the product of all methyltransferases, S-adenosylhomocysteine. Additionally, we examined the expression of genes for 182 putative and established S-adenosylmethionine-dependent methyltransferases, catalyzing methylation of protein, RNA, DNA, and small molecules as well as genes for 17 putative and established demethylases of protein, DNA, and RNA. We found very few changes in the expression of these genes when we compared the parents and revertants of the HCT 116 and H460 cancer cells. The most interesting statistically-significant change was an almost two-fold increase in the expression of the MAT2A gene encoding the catalytic subunit of the S-adenosylmethionine synthetase in the H460 revertant (p = 0.01) (Table 2A). However, the expression of this gene was decreased in the HCT 116 revertant cells. These results suggest that there are multiple pathways that can lead to reversion to methionine-independence.
1 0 Additionally, we should point out that changes in mRNA levels need not necessary be correlated with changes in the activities of the protein, since translational and posttranslational controls may play larger roles than transcriptional controls in determining the activities of some enzymes. For example, when we looked at the protein levels of MAT2A by immunoblotting in Figure 2B, we observed no change in the methionine-independent revertants of either of the methionine-addicted cancer cell lines.
As noted above, we also found a statistically-significant change in the expression of the gene for the alpha folate receptor in the HCT 116 revertants (~two-fold decrease; p = 0.02) and the gene encoding an enzyme in the trans-sulfuration pathway from homocysteine to cysteine (cystathionine gamma lyase, which had a 1.7 fold decrease in the H460 revertants; p = 0.03). However, neither change was common to the revertants of both colon-cancer and lung-cancer cell lines (Supplemental Table 1).
We noted statistically-significant decreases in the expression of the genes in the revertant lines from both the HCT 116 and H460 cells for two protein lysine demethylases, KDM6B in the H460 revertants (1.9 fold, p = 0.04) and KDM7A in the HCT 116 revertants (1.5 fold; p = 0.02) (Table 2B). However, from the immunoblotting results in Figure 2D, we observed decreases in the level of histone H3 trimethylation at this site in both the colon cancer and lung cancer revertnants, indicating that the decrease in levels of histone H3 lysine methylation in the methionine-independent revertant is not due to demethylases but possibly to decreased S-adenosylmethioinine utilization.
The present results suggest that there are few significant changes in gene expression and associated with reversion to a methionine-independent phenotype 1 1

Discussion
Methionine dependence of cancer has been known since 1959 when Sugimura showed that tumors in rats slowed their growth when methionine was removed from their diet compared to when other amino acids were removed from the diet 22 . It was not until the early 1970s that methionine dependence was found in cultured cancer cells 23,24 and it was initially claimed that methionine dependence was due to reduced ability of the cancer cells to synthesize methionine from homocysteine 24  note that we observed that as little as 1μM methionine stimulated the growth of methionine-addicted cancer cells to initiate rapid proliferation on methionine-free homocysteine-containing medium in which they were arrested. 8 . These results give rise to speculation that exogenous methionine may have a special use in methionine-addicted cancer cells. Future metabolomics experiments may give us more information on the fate of exogenous methionine in methionine-addicted cancer cells compared to methionine-independent revertants.
We previously showed that an increase in the rate of synthesis of methionine from homocysteine was not necessary for reversion to methionine independence 20 , further demonstrating that decreased methionine synthetic capacity was not the basis of methionine addiction. Then we found that despite the large amounts of endogenous methionine made by cancer cells, under restriction of exogenous methionine the cancer cells had very low levels of free methionine, low levels of S-adenosylmethionine (SAM) and a low ratio of SAM to 1 2 S-adenosylhomocysteine (SAH) compared to normal cells and methionine-independent revertants [1][2][3][4]15,20 . This was explained next when we observed that all cancer cells tested had elevated rates of transmethylation compared to normal cells which used excess amounts of methionine and SAM and depleted their cellular pools 2,3,12 . We then understood that methionine addiction was due to overuse of methionine for transmethylation 12 . This was confirmed when it was demonstrated that overall transmethylation was strongly reduced in methionine-independent revertants derived from the methionine-addicted cancer cells 16,17 .
The phenomenon of methionine-addicted cancer cells is termed the Hoffman-effect 14 .
Tam's group in 2019 7 , confirmed our results of 44 years previously 1   In the present study we did not detect a difference of the expression level of histone methyltransferases between methionine-addicted cancer cells and methionine-independent revertants by immunoblotting or RNA expression. In addition, the MAT2A protein level is not changed in the revertants. There are previous studies which report that high expression level of histone lysine methyltransferases and MAT2A are markers of malignancy 7,37,38 .
However, our present study shows that methionine-independent revertants lose malignancy without detected changes in the expression levels of those enzymes.  40 . Other molecules, such as various RNAs, may also be overmethylated in 1 5 cancer cells 41,42 . Further experiments are necessary to account for all the transferred methyl groups in cancers and the difference in comparison to normal cells.
In summary, our results suggest that overmethylation of histone H3 lysines is linked with methionine addiction and malignancy of cancer cells, and possibly necessary for both fundamental hallmarks of cancer (Fig. 4). Methionine addiction is being widely recognized as critical for malignancy 43 and may provide a universal target for methionine-restriction cancer therapy, such as with methioninase which depletes methionine in tumors rapidly and shown to be broadly effective against all cancer types tested including the clinic 44 ,.

Cell culture
The HCT 116 human colon cancer cell line and H460 human lung cancer cell line were used in the present study. HCT 116 cells were stably transduced to express green fluorescent protein (GFP) as previously described 45 . All cell lines were obtained from the American Type Culture Collection (Manassas, VA, USA) and were routinely checked for mycoplasma. Cells were maintained in DMEM supplemented with 10% fetal bovine serum (FBS) and 100 IU/ml penicillin/streptomycin.

Selection and establishment of methionine-independent revertant cancer cells
The H460 and HCT 116 parental cell lines were cultured in normal methionine-containing medium with 1 U/ml of rMETase for more than one month as a first selection and surviving cells were isolated. These surviving cells were cultured in normal methionine-containing medium with 5 U/ml rMETase for more than 2 weeks as a second selection and surviving cells were isolated. The surviving cells isolated after the second selection were termed H460-R and HCT 116-R respectively.

rMETase activity assay in vitro
Medium with 1 U/ml of rMETase was incubated, and the methionine levels in the medium were measured with an HPLC (Hitachi L-6200A Intelligent pump; Hitachi, Ltd., Tokyo, Japan) before and after added rMETase as described previously 48 . dialyzed fetal bovine serum. Cell proliferation was measured using the Cell Counting Kit-8 (Dojindo, Kumamoto, Japan) after 0, 1, 3, 5 and 7 days of medium change.

Immunoblotting
Extraction of histone and protein from cells and tumors and immunoblotting were performed as described 18,36 . Antibodies used are listed in Supplemental table 2 Immunoreactive proteins were visualized with Clarity Western ECL Substrate (Bio-Rad

Comparison of in vivo tumorgenicity of methionine-addicted parental cancer cells and methionine-independent revertants in a subcutaneous mouse model
Two different doses of HCT 116 and HCT 116-R cells (5 × 10 5 or 1 × 10 6 cells / 100 μ l PBS each) and were injected subcutaneously into the flanks of nude mice. Each group comprised ten mice. The mice with HCT 116 or HCT 116-R tumors were sacrificed on day 1 8

Comparison of ability of methionine-addicted parental cancer cells and methionine-independent revertants to form experimental liver-metastasis
HCT 116-GFP and HCT 116-R-GFP (1 × 10 6 cells / 100 μ l PBS) were injected to the spleen of nude mice. Each group comprised three mice. The mice were sacrificed on day 42 and fluorescence intensity was measured at termination with the UVP ChemStudio (Analytik Jena US LLC) 45 .

H & E staining and immunohistochemistry
Hematoxylin and eosin (H&E) staining and immunohistochemical staining (IHC) were performed as described 36

Next generation sequencing
Cells (1 × 10 6 ) were pelleted and frozen at -80 °C and provided to GENEWIZ (South Plainfield, NJ, USA) for RNA sequencing. RNA was extracted from the cells and were sequenced by HiSeq sequencing system (Illumina, San Diego, CA, USA).

Statistical analyses
All statistical analyses were performed with JMP PRO ver. 15.0.0 (SAS Institute, Cary, NC, USA). Student t-test was used to compare between groups for animal studies. Bar graphs show the mean, and error bars express standard error of the mean. A probability value of P < 0.05 is defined as statistically significant.     trimethylated histone H3 lysine marks in methionine-addicted cancer cells and methionine-independent revertants grown in methionine-containing medium. (D) Immunoblot of H3K9 methyltransferases in methionine-addicted cancer cells and methionine-independent revertants grown in methionine-containing medium. trimethyl-histone H3 marks / total H3 in the tumors formed from HCT 116 and HCT 116-R (mean ± SEM, n = 3). Table 1. Genes with expression levels that are significantly different between methionine-addicted cancer cells and methionine-independent revertants. Supplemental Table 1