Human colon cancer cells and primary colon cancer silence the gene coding for LDH (lactate dehydrogenase)-B and up-regulate the gene coding for LDH-A, resulting in effective conversion of pyruvate into lactate. This is associated with markedly reduced levels of pyruvate in cancer cells compared with non-malignant cells. The silencing of LDH-B in cancer cells occurs via DNA methylation, with involvement of the DNMTs (DNA methyltransferases) DNMT1 and DNMT3b. Colon cancer is also associated with the expression of pyruvate kinase M2, a splice variant with low catalytic activity. We have shown recently that pyruvate is an inhibitor of HDACs (histone deacetylases). Here we show that pyruvate is a specific inhibitor of HDAC1 and HDAC3. Lactate has no effect on any of the HDACs examined. Colon cancer cells exhibit increased HDAC activity compared with non-malignant cells. HDAC1 and HDAC3 are up-regulated in colon cancer cells and in primary colon cancer, and siRNA (small interfering RNA)-mediated silencing of HDAC1 and HDAC3 in colon cancer cells induces apoptosis. Colon cancer cells silence SLC5A8, the gene coding for a Na+-coupled pyruvate transporter. Heterologous expression of SLC5A8 in the human colon cancer cell line SW480 leads to inhibition of HDAC activity when cultured in the presence of pyruvate. This process is associated with an increase in intracellular levels of pyruvate, increase in the acetylation status of histone H4, and enhanced cell death. These studies show that cancer cells effectively maintain low levels of pyruvate to prevent inhibition of HDAC1/HDAC3 and thereby to evade cell death.
Enhanced glucose uptake and glycolysis with suppressed oxidative metabolism of pyruvate in mitochondria are a hallmark of most tumour cells [1–6]. This occurs in the presence of adequate oxygen supply. The switching to “aerobic glycolysis” from oxidative phosphorylation as the primary source of ATP forms the central core of the Warburg hypothesis . According to this hypothesis, the above-mentioned metabolic switching is the underlying cause of tumourigenesis. Even though the cause/effect relationship between aerobic glycolysis and tumour formation as proposed by Warburg  is still being debated, it is widely believed that most tumour cells derive a major portion of metabolic energy from glycolysis rather than from mitochondrial oxidation. Activation of glycolysis in tumour cells with decreased mitochondrial function obligates the conversion of pyruvate into lactate to regenerate NAD+, which is consumed at the level of glyceraldehyde-3-phosphate dehydrogenase. If pyruvate is not converted into lactate in the cytoplasm, and NADH is not oxidized to NAD+ in mitochondria, glycolysis cannot continue. Therefore, it makes sense that tumour cells robustly convert pyruvate into lactate.
Enhanced glycolysis/suppressed mitochondrial function certainly play a role in tumour progression. Since optimal mitochondrial respiration is dependent on adequate oxygen supply, suppression of mitochondrial function may lead the tumour cells to behave as if they are in hypoxia (pseudohypoxia). Heightened hypoxic response such as stabilization of HIF-1α (hypoxia-inducible factor-1α) with subsequent up-regulation of glucose uptake, glycolysis and vascular endothelial growth factor will promote tumour progression through alternative sources of metabolic energy (i.e. glycolysis instead of oxidative phosphorylation), as well as neovascularization and nutrient supply [4,8]. This is supported by the findings that mutations in the citric acid cycle enzymes, succinate dehydrogenase and fumarate hydratase, are associated with cancer and with concomitant increase in HIF-1α levels [9–11]. Lactate produced by the tumour cells may also have its own effects related to tumour progression and metastasis .
Conversion of pyruvate into lactate is obligatory for aerobic glycolysis when mitochondrial function is defective. This reaction is mediated by LDH (lactate dehydrogenase). There are two LDH subunits coded by two separate genes, LDH-A and LDH-B. These two subunits associate to form five different isoforms of tetrameric LDH: LDH1 (LDH-B4), LDH2 (LDH-B3A1), LDH3 (LDH-B2A2), LDH4 (LDH-B1A3), and LDH5 (LDH-A4). LDH1 is effective in the conversion of lactate into pyruvate, whereas LDH5 is effective in the conversion of pyruvate into lactate. Since tumour cells robustly convert pyruvate into lactate, one would expect a differential expression of LDH-A and LDH-B in these cells. Available evidence indicates that LDH-A is up-regulated and LDH-B is silenced in a variety of cancers [13–18]. In addition, tumour-associated fibroblasts and endothelial cells exhibit a complementary LDH-A/LDH-B phenotype, with high LDH-B and low LDH-A , facilitating metabolic utilization of tumour-generated lactate to support proliferation and growth of these cells.
Tumour cells also differ from normal cells in the expression of pyruvate kinase, the enzyme responsible for the generation of pyruvate in glycolysis. The splice variant PKM2 (muscle-specific pyruvate kinase 2) is expressed specifically in tumour cells in the dimeric form with low catalytic activity . Most recent studies have shown that this tumour-specific isoform is a phosphotyrosine-binding protein  and that silencing of PKM2 and forced expression of PKM1 in cancer cells switch the metabolic phenotype to one similar to that of normal cells . Glutaminolysis is another metabolic pathway that is highly active in tumour cells, generating lactate, with pyruvate as the intermediate . Thus pyruvate is at the intersection of the metabolic switching that is characteristic of tumour cells, but little is known on the relevance of pyruvate to cancer. We recently reported that SLC5A8, a Na+-coupled transporter for pyruvate and other monocarboxylates, triggers tumour cell apoptosis through pyruvate-dependent inhibition of HDACs (histone deacetylases) . This phenomenon is related to the entry of extracellular pyruvate into tumour cells. The expression of SLC5A8 is silenced in a variety of cancers [25–27], suggesting that tumour cells prevent the entry of extracellular pyruvate to avoid pyruvate-induced cell death. In the present study, we investigated the effects of the reciprocal regulation of LDH-A and LDH-B on intracellular levels of pyruvate in colon cancer cells and its significance in the light of our recent finding that pyruvate is an HDAC inhibitor and a tumour suppressor . These studies have generated new and important data which propel pyruvate to the forefront of cancer biology as a critical metabolite and have clinical and therapeutic implications.
MATERIALS AND METHODS
Collection of normal and cancer tissue specimens from human colon
This study received the approval of the Medical College of Georgia Institutional Human Assurance Committee. The research was carried out in accordance with the Declaration of Helsinki (2000) of the World Medical Association. Adult patients with colorectal adenocarcinoma (n=18), without a history of prior chemoradiation, were included in this study after obtaining their informed consent. A pathologist harvested normal colorectal epithelium and tissue from the luminal surface of the colorectal cancer from the freshly resected surgical specimens. Portions of the tissues (0.3 to 0.5 g from each site) were processed for total RNA extraction using TRIzol® reagent (Invitrogen Life Technologies). Some of these tissue specimens have been used previously for the expression analysis of the amino acid transporter ATB0,+ .
Colon cell lines
Two non-malignant colonic epithelial cell lines (NCM460 and CCD841) and nine malignant colonic epithelial cell lines (SW480, SW620, KM12C, KM12L4, HT29, HCT116, Colo201, Colo205 and Ls174T) were used in the present study. The wild-type HCT116 cell line, which is positive for DNMTs (DNA methyltransferases) (DNMT+/+) and the isogenic cell lines with homologous deletion of DNMT1 (DNMT1−/−), DNMT3b (DNMT3b−/−), DNMT1 plus DNMT3b (DKO) were provided by Dr. Bert Vogelstein (Johns Hopkins University, Balitimore, MD, U.S.A.).
RT–PCR (reverse transcription–PCR)
RNA prepared from colon tissue specimens and cell lines was used for semi-quantitative RT–PCR. RNA (2 μg) was reverse transcribed into cDNA using the GeneAmp PCR system (Roche). HPRT1 (hypoxanthine phosphoribosyltransferase 1) mRNA was used as the internal control.
This was done for LDH-A, LDH-B, various isoforms of HDAC, histones H3 and H4, and their acetylated forms. Cell lysates were prepared by sonication in 10 mM Tris/HCl buffer (pH 7.6), supplemented with a cocktail of protease inhibitors (50 mM NaF, 0.2 mM vanadate, 1 mM PMSF, 5 μg/ml aprotinin, 1 μg/ml pepstatin and 2 μg/ml leupeptin) and 1% Triton X-100. Proteins were size-fractionated on to SDS/PAGE gels and then transferred on to Protran nitrocellulose membranes (Schliecher & Schuell). Membranes were blocked with BSA, treated with primary antibody at 4 °C overnight, followed by treatment with appropriate secondary antibody conjugated to horseradish peroxidase. The antigen/antibody reaction was detected by Enhanced Chemiluminescence SuperSignal Western System (Amersham). Primary antibodies were obtained from the following sources: LDH-A (Santa Cruz), LDH-B (Sigma–Aldrich), HDACs 1–5 (Sigma–Aldrich or BioVision), and histones H3 and H4, and their acetylated forms (Santa Cruz or Upstate Biotechnology).
Measurement of pyruvate and lactate levels in cultured cells
The levels of pyruvate and lactate were measured using a fluorescence-based assay (BioVision). For the pyruvate assay, subconfluent cells were collected and washed three times with PBS. Cells were then lysed in pyruvate assay buffer and the clear lysate was collected by centrifugation at 50000 g for 15 min at 4 °C. Lysate (50 μl) was aliquoted into each well in a 96-well plate in triplicate. The reaction assay mixture was prepared according to the manufacturer's instructions, and 50 μl of this mixture was then added to each well. In this assay, pyruvate oxidation generates a fluorescent signal which is measured in a fluorimeter at an excitation wavelength of 535 nm and an emission wavelength of 590 nm. The procedure was exactly the same for measurement of lactate levels, except that the assay mixture contained the enzyme for lactate oxidation, generating a fluorescent signal.
Measurement of HDAC activity
The measurement of HDAC activity in lysates from colonic cell lines or immunoprecipitates was carried out using a commercially available assay kit (Cayman Chemical Company). The activity of recombinant human HDAC isoforms was also measured using the same assay kit. The recombinant HDAC isoforms were purchased from Cayman Chemical Company. Cell lysate protein (50 μg) or recombinant HDAC (10 ng) was incubated with or without butyrate, pyruvate or lactate, and the reaction was initiated by the addition of HDAC substrate. The enzyme activity was monitored using a fluorimetric assay. The immunoprecipitates were prepared from the malignant human colon cancer cell line SW480 as follows. Sub-confluent cells were washed twice and collected in PBS. Cells were solubilized with 25 mM Tris/HCl buffer (pH 7.5) containing 100 mM NaCl, 1 μl/ml of 2-mer-captoethanol and protease inhibitor cocktail (Roche). Cell lysate was incubated on ice for 20 min followed by sonication, and then centrifuged at 14000 g for 15 min. Soluble proteins (1 mg) were incubated with anti-HDAC1, anti-HDAC2 and anti-HDAC3 antibodies (Sigma–Aldrich) individually for 3 h at 4 °C. The resultant immunocomplexes were collected using Protein A/G plus-agarose beads (Santa Cruz Biotechnology) at 4 °C overnight. Beads were then washed three times with RIPA buffer (Sigma–Aldrich) and resuspended in 700 μl of HDAC assay buffer. These immunoprecipitates (50 μl) were used for measurement of HDAC activity.
Flow cytometric analysis of apoptosis
Cells were transiently transfected with either pcDNA3.1 or SLC5A8 cDNA, and cultured in the presence or absence of pyruvate (1 mM) for 48 h. Cells were then fixed in 50% ethanol, treated with 0.1% sodium citrate, 1 mg/ml RNase A and 50 μg/ml propidium iodide, and then subjected to FACS (FACS Caliber, Becton Dickinson) analysis. For siRNA (small interfering RNA) studies, CCD841 (a non-malignant human colon cell line) and SW480 (a malignant human colon cell line) cells were seeded in 6-well plates and cultured in RPMI 1640 medium. After 24 h, cells were transfected with HDAC1 siRNA, HDAC2 siRNA or HDAC3 siRNA [Santa Cruz Biotechnology; catalogue numbers: sc-29343 for HDAC1 siRNA (h); sc-29345 for HDAC2 siRNA (h); sc-35538 for HDAC3 siRNA (h)] according to the manufacturer's instructions. Scrambled siRNA (Santa Cruz Biotechnology; catalogue number: sc-37007) was used as a negative control. After 48 h, cells were collected and processed for the FACS analysis.
Differential expression of LDH-A and LDH-B in colon cancer and its relevance to intracellular levels of pyruvate and lactate
Pyruvate is generated in tumour cells principally by glycolysis and glutaminolysis. The levels of pyruvate in these cells are controlled not only by these two metabolic pathways but also by the cells' ability to convert pyruvate into lactate. Since LDH1, a tetramer of LDH-B, preferentially functions in the conversion of lactate into pyruvate, whereas LDH5, a tetramer of LDH-A, preferentially functions in the conversion of pyruvate into lactate, we anticipated that the differential expression of LDH-A and LDH-B in normal cells compared with cancer cells would be a key determinant of the intracellular levels of pyruvate. Therefore, we examined the expression of these two genes in primary colon cancer and in colon cancer cell lines. With paired specimens of normal and cancer colon tissues, we found the expression of LDH-B to be markedly silenced in cancer, along with up-regulation of LDH-A (Figure 1A). Of the 18 paired tissue samples evaluated, the silencing of LDH-B was seen in 16 cases and the up-regulation of LDH-A was seen in all cases. The decrease in the expression of LDH-B mRNA in cancer tissues compared with normal tissues was ∼5-fold and the increase in the expression of LDH-A mRNA was ∼6-fold. We also compared the expression levels of these two genes between non-malignant and malignant human colonic cell lines. The non-malignant cell lines NCM460 and CCD841 expressed high levels of LDH-B and low levels of LDH-A. This was evident both at mRNA level and protein level (Figure 1B). In contrast, the expression of LDH-B was lower and that of LDH-A was higher in nine different malignant cell lines. We then measured the intracellular levels of pyruvate and lactate in these cells. The levels of pyruvate were markedly reduced in cancer cell lines compared with non-malignant cell lines, the concentration in the former being only about 20% of the concentration in the latter (Figure 1C). On the contrary, the reverse was true with lactate. Non-malignant cells had 4-fold lower levels of lactate compared with malignant cells (Figure 1D).
Correlation between the differential expression of LDH-A/LDH-B and intracellular levels of lactate/pyruvate in primary human colon cancer and in colon cancer cell lines
Role of DNA methylation in cancer-associated silencing of LDH-B
Even though the silencing of LDH-B has been demonstrated in a variety of tumours, the underlying molecular mechanism has not been investigated in detail. In a recent report, Leiblich et al.  showed that LDH-B is silenced by promoter hypermethylation in prostate cancer. Here we examined the role of DNA methylation in the silencing of LDH-B in colon cancer cell lines. The expression of LDH-B in non-malignant cell lines, which constitutively express this gene at high levels, was not affected by treatment with 5-azadeoxycytidine, an inhibitor of DNA methylation (Figure 2A). But in malignant cell lines, which express very low levels of LDH-B, treatment with 5-azadeoxycytidine induced the expression of LDH-B, detectable at the level of both mRNA and protein. Procainamide, a specific inhibitor of DNMT1 , was unable to induce the expression of LDH-B, suggesting that DNMT1 alone is insufficient to silence LDH-B. Studies with wild-type HCT116 cell line and isogenic cell lines with deletion of DNMT1, DNMT3b or both showed that LDH-B expression was induced only in the cell line which lacked both DNMT1 and DNMT3b (Figure 2B). These data demonstrate for the first time the involvement of specific DNMTs in the silencing of LDH-B in cancer cells. We then correlated the intracellular levels of pyruvate in HCT116 cells with the expression of LDH-B (Figure 2C). In the double knockout cell line (DKO), in which both DNMT1 and DNMT3b have been deleted with consequent expression of LDH-B, intracellular pyruvate levels were several-fold higher than in the wild-type cell line. Interestingly, the DNMT1−/− cell line had significantly higher levels of pyruvate compared with the wild-type cell line, even though there was no expression of LDH-B. This was principally due to the induction of the expression of SLC5A8, a Na+-coupled transporter for pyruvate. We have demonstrated that DNMT1 alone is sufficient to cause the silencing of SLC5A8 . Since the cells were cultured in the presence of pyruvate, the induction of SLC5A8 in these cells leads to an increase in intracellular pyruvate levels by facilitating the entry of pyruvate from the medium into the cells. This was not the case with DNMT3b−/− cells which do not express SLC5A8. The intracellular levels of pyruvate correlated inversely with HDAC activity in these cell lines (Figure 2D). Wild-type cells and the DNMT3b−/− cells, which had low levels of pyruvate, had high HDAC activity. In contrast, DNMT1−/− and DKO cells, which had high levels of pyruvate, had low HDAC activity. This agrees with our recent findings that pyruvate is an HDAC inhibitor .
Involvement of DNMTs in the silencing of LDH-B in colon cancer cell lines
Differential expression of the pyruvate kinase isoforms PKM1 and PKM2 in colon cancer
The concentration of pyruvate in cancer cells is controlled not only by the expression pattern of LDH-A and LDH-B isoforms but also by pyruvate kinase. Pyruvate is generated in cancer cells by glycolysis as well as by glutaminolysis. There is evidence that cancer cells use the carbon atoms present in glucose for the synthesis of nucleotides, fatty acids, lipids, and non-essential amino acids, and that glutamine is metabolized preferentially into lactate via pyruvate . For the utilization of glucose carbon in biosynthetic processes, the conversion of phosphoenolpyruvate into pyruvate must be decreased. Recent studies have shown that cancer cells express pyruvate kinase primarily in the form of PKM2 dimer which possesses low catalytic activity [20–22]. We confirmed the cancer-associated expression of the PKM2 variant with primary colon cancer and colon cancer cell lines (Figure 3). PKM1 was expressed at high levels in normal colon tissues and in non-malignant colon cell lines, whereas PKM2 was expressed at high levels in colon cancer and in colon cancer cell lines. In paired samples of primary tumour tissues and normal colon tissues, the cancer-associated decrease in PKM1 expression was 7-fold and the cancer-associated increase in PKM2 expression was 3-fold. These findings suggest that the differential expression of PKM isoforms also contributes to the lower levels of pyruvate in colon cancer cells.
Semi-quantitative RT–PCR analysis of the steady-state levels of pyruvate kinase PKM1 and PKM2
Isoform selectivity for the inhibition of HDACs by pyruvate
We have demonstrated previously that pyruvate is an inhibitor of HDAC ; however, the isoform specificity of this inhibition was not known. Therefore, we examined the effects of pyruvate on the activity of different isoforms of human recombinant HDACs. We used butyrate as a positive control. Butyrate inhibited HDAC1 and HDAC3, but had no effect on HDAC2, HDAC5, HDAC6, HDAC8 and HDAC9 (Figure 4A). Pyruvate showed similar isoform specificity. It also inhibited only HDAC1 and HDAC3 (Figure 4B). On the other hand, lactate had no effect on any of the HDAC isoforms. Butyrate and pyruvate had comparable inhibitor potency (Figures 4C and 4D). HDAC1 was inhibited more potently than HDAC3 by these compounds. The IC50 values for butyrate and pyruvate for the inhibition of HDAC1 were 20±4 and 24±5 μM respectively. The corresponding values for the inhibition of HDAC3 were 75±10 and 80±15 μM. We confirmed this isoform selectivity using endogenous HDAC1 and HDAC3 expressed in the colon cancer cell line SW480. These isoforms were immunoprecipitated from SW480 cell lysates with specific antibodies, and the immunocomplexes were then used as the source of enzyme activity. HDAC1 and HDAC3 were inhibited by butyrate and pyruvate, but not by lactate (Figure 4E). In contrast, HDAC2 was not affected by any of these compounds.
Isoform specificity of HDAC inhibition by pyruvate
Up-regulation of HDAC1 and HDAC3 in colon cancer
Data from wild-type and isogenic HCT116 cell lines indicated an inverse relationship between cellular pyruvate levels and HDAC activity (Figures 2C and 2D). To confirm these findings, we measured HDAC activity in non-malignant and malignant colon cell lines for which data on cellular levels of pyruvate are available in the present study. We already showed that the non-malignant cell lines had much higher levels of pyruvate compared with malignant cell lines (Figure 1C). As expected from the ability of pyruvate to inhibit HDAC, the HDAC activity was much lower in non-malignant cell lines than in malignant cell lines (Figure 5A). However, since pyruvate is a specific inhibitor of HDAC1 and HDAC3, the postulated inverse relationship would be expected only if the higher HDAC activity in cancer cell lines than in non-malignant cell lines is predominantly due to increased expression of these two HDAC isoforms in cancer cells. Therefore, we monitored the expression pattern of HDAC isoforms in non-malignant and malignant colon cell lines (Figures 5B and 5C). While the expression levels of HDAC2, HDAC4 and HDAC5 are similar in non-malignant and malignant cell lines, there was a clear difference in the expression of HDAC1 and HDAC3 between these two groups of cell lines. The expression of these two HDAC isoforms was higher in malignant cell lines compared with non-malignant cell lines. These data suggest that the increase in HDAC activity in cancer cells is primarily due to an increase in the expression of the pyruvate-sensitive isoforms HDAC1 and HDAC3. To confirm that this is also true in primary colon cancer, we compared the expression of HDAC1 and HDAC3 between normal colon tissue specimens and paired colon cancer specimens (Figure 5D). The expression of these two isoforms was higher in cancer tissues compared with adjacent normal tissues (∼3-fold for HDAC1 and ∼2-fold for HDAC3), corroborating the findings in colon cell lines. The expression of HDAC5 was also increased slightly but significantly in cancer tissues compared with normal tissues (∼70% increase), whereas there was no difference in the expression levels of HDAC2 and HDAC4.
HDAC activity and expression pattern of HDAC isoforms in non-malignant and malignant colon cell lines and in primary human colon cancer and adjacent normal colon tissue
Cancer cell-specific apoptosis by inhibition of HDAC1 and HDAC3
SLC5A8 is a tumour suppressor gene that is silenced in colon cancer and in colon cancer cell lines . We and others have shown that SLC5A8 is a Na+-coupled transporter for monocarboxylates, including short-chain fatty acids, pyruvate and lactate [32–34]. In normal intestinal tract, the transporter is expressed in the lumen-facing apical membrane of epithelial cells [35,36]. However, when transformed into cancer, these epithelial cells lose their polarity and the transporter would have access to metabolites present in the circulation. Pyruvate is present in blood at significant levels (∼100 μM) . It is likely that colon cancer cells silence SLC5A8 to prevent the transporter-mediated entry of circulating pyruvate into the cells. Thus, the silencing of SLC5A8 would complement the differential expression of LDH-A and LDH-B in cancer cells to maintain low intracellular levels of pyruvate as a means to avoid inhibition of HDAC1 and HDAC3. Then ectopic expression of SLC5A8 in colon cancer cell lines should increase the cellular levels of pyruvate and reduce HDAC activity when cultured in a medium containing pyruvate. To examine this, we transfected the non-malignant cell line CCD841 and the malignant cell line SW480 with either vector alone or SLC5A8 cDNA and cultured the cells in a pyruvate-containing medium for 48 h. CCD841 cells express SLC5A8 constitutively, whereas the gene is silenced in SW480 cells. We found that the cellular levels of pyruvate in CCD841 cells were high and not altered in response to overexpression of the transporter (Figure 6A). In contrast, pyruvate levels in SW480 cells were relatively lower, but the levels increased markedly with the expression of the transporter (Figure 6A). The changes in pyruvate levels mirrored the changes in HDAC activity reciprocally. The HDAC activity was low in CCD841 cells, and the activity was not influenced by ectopic expression of SLC5A8 (Figure 6B). This was expected because pyruvate levels were not altered under these conditions. In contrast, the HDAC activity in SW480 cells was higher than in CCD841 cells, and the activity was reduced significantly with ectopic expression of SLC5A8 (Figure 6B). This corroborated with the increase in pyruvate levels in SW480 cells. We then examined the effects of the changes in pyruvate levels and HDAC activity in these cell lines on apoptosis. With CCD841 cells, there was very little apoptosis when transfected with either vector alone or SLC5A8 cDNA irrespective of whether or not the cells were cultured in the presence of pyruvate (Figure 6C). With SW480 cells also, there was very little apoptosis when transfected with either vector alone or SLC5A8 cDNA, but only when cultured in the absence of pyruvate. When cultured in the presence of pyruvate, vector-transfected cells did not undergo apoptosis, whereas massive cell death occurred in SLC5A8-expressing cells (Figure 6C). We confirmed the changes in HDAC activity in CCD841 and SW480 cells under these conditions by analysing the acetylation status of histones H3 and H4 (Figure 6D). The acetylation status of histones H3 and H4 was high in CCD841 cells irrespective of whether the cells were transfected with vector alone or SLC5A8 cDNA and whether the cells were cultured in the presence or absence of pyruvate. In contrast, the acetylation status of histone H4, in particular the acetylation of H4-Lys12 and H4-Lys16, was much lower in vector-transfected and SLC5A8 cDNA-transfected SW480 cells than that in CCD841 cells when the cells were cultured in the absence of pyruvate. But the acetylation status of histone H4 increased in SLC5A8-expressing SW480 cells when cultured in the presence of pyruvate. There was no change in the acetylation status of histone H3. These results show that elevation of pyruvate levels in colon cancer cells through SLC5A8-mediated transport leads to HDAC inhibition, enhances the acetylation status of histone H4, and consequently induces apoptosis. In contrast, non-malignant cells do not undergo changes in the acetylation status of histone H4 under these conditions and hence there is no apoptosis in these cells.
Relevance of intracellular pyruvate levels to HDAC activity, histone acetylation, and apoptosis in the non-malignant colon cell line CCD841 and in the colon cancer cell line SW480
We confirmed the relevance of HDAC1/HDAC3 inhibition to cancer cell-specific apoptosis using siRNA-mediated silencing of these isoforms in SW480 cells (Figure 6E). Transfection of the non-malignant colon cell line CCD841 with siRNAs specific for the three isoforms of HDAC (HDAC1, 2 and 3) did not have any effect. In contrast, there was a differential effect of these siRNAs on the malignant colon cell line SW480. Transfection of these cells with HDAC1 siRNA or HDAC3 siRNA induced apoptosis, whereas transfection with HDAC2 siRNA had no effect. Scrambled siRNA was used as a negative control in these experiments. These studies show that silencing of HDAC1 and HDAC3 leads to apoptosis in a cancer cell-specific manner.
The most important findings in the present study are: (i) intracellular levels of pyruvate in colon cancer cells are much lower than those in non-malignant cells; (ii) the lower levels of pyruvate in colon cancer cells are the result of not only the differential expression of enzymes involved in the generation and metabolism of pyruvate, namely increased expression of LDH-A and PKM2, and decreased expression of LDH-B and PKM1, but also the decreased expression of SLC5A8, a Na+-coupled active transporter for pyruvate; (iii) the silencing of LDH-B in colon cancer cells involves DNMT1 and DNMT3b; (iv) pyruvate is a specific inhibitor of the HDAC isoforms HDAC1 and HDAC3; (v) colon cancer cells have much higher HDAC activity than non-malignant cells, primarily due to increased expression of HDAC1 and HDAC3; (vi) elevation of intracellulal levels of pyruvate leads to apoptosis specifically in colon cancer cells without having any effect in non-malignant cells; and (vii) siRNA-mediated silencing of HDAC1 and HDAC3 causes apoptosis specifically in cancer cells. These findings are novel and have immense clinical and therapeutic significance. It is well known that cancer cells are principally lactate producers. But this was presumed to be mostly due to mass action, where increased generation of pyruvate in glycolysis results in increased production of lactate. Cancer cells have enhanced glycolysis, the end product of which is pyruvate. Since mitochondrial function is down-regulated in cancer, what do the cancer cells do with this glycolytic end product? These cells induce LDH-A and silence LDH-B so as to convert pyruvate into lactate. The underlying notion here is that increased generation of lactate in cancer cells is the direct result of increased pyruvate levels. But the present study suggests that this is not true. Our results show that the increased lactate production in cancer cells is not simply the result of ‘more pyruvate means more lactate’. Cancer cells generate lactate purposely to reduce the intracellular levels of pyruvate. The conversion of pyruvate into lactate is not the only mechanism by which cancer cells manage to keep the pyurvate levels low. These cells also differentially express the pyruvate kinase splice variants PKM1 and PKM2 such that the activity of pyruvate kinase, which converts phosphoenolpyruvate into pyruvate, is low. In addition, the expression of SLC5A8, the gene coding for the Na+-coupled pyruvate transporter, is silenced in cancer cells, which decreases the entry of blood-borne pyruvate into cancer cells. Collectively, these processes maintain the intracellular levels of pyruvate low in cancer cells. To our knowledge, this is the first report describing this unique phenomenon of significantly reduced pyruvate levels in cancer cells.
Even though the silencing of LDH-B in cancer has been demonstrated in several studies [13–18], the underlying mechanism is just beginning to be understood. A recent study has demonstrated that the silencing of LDH-B in prostate cancer involves promoter hypermethylation . Here we show a similar mechanism in colon cancer cells. Treatment of colon cancer cells with 5′-azadeoxycytidine, a pan-inhibitor of DNMTs, induces the expression of LDH-B. However, procainamide, a specific inhibitor of DNMT1, has no effect, suggesting that inhibition of DNMT1 alone is not sufficient to induce the re-expression of the gene. Additional studies with isogenic HCT116 cell lines show that both DNMT1 and DNMT3b are involved in the silencing of the gene. This is in contrast with the mechanism associated with the silencing of SLC5A8 in these cells, where inhibition of DNMT1 alone is enough to induce the re-expression of the transporter .
The metabolic rationale underlying the conversion of pyruvate into lactate in cancer cells is understandable in terms of regeneration of NAD+ and maintenance of enhanced glycolytic rate under conditions of decreased mitochondrial function. But what is the need for cancer cells to maintain reduced intracellular levels of pyruvate? The findings of the present study provide an answer to this question. While pyruvate is an energy-rich nutrient necessary for growth in non-malignant cells, this metabolite is a tumour suppressor and an inducer of apoptosis in cancer cells. The tumour-suppressive function of pyruvate is related to its ability to inhibit HDAC1 and HDAC3. We have already shown in a previous study that pyruvate is an HDAC inhibitor and a tumour suppressor . Here we describe for the first time the isoform specificity of pyruvate inhibition of HDAC. It is interesting and important to note that HDAC1 and HDAC3, which are inhibitable by pyruvate, are the two isoforms which are up-regulated in cancer cells. The elevation of HDAC activity is presumably necessary for the cancer cells to maintain their malignant phenotype. Therefore, cancer cells must maintain the low intracellular levels of pyruvate, lest HDAC1 and HDAC3 will be inhibited and cell growth prevented by enhanced apoptosis. This is supported by the findings of the present study, where we show marked induction of apoptosis in cancer cells when the intracellular levels of pyruvate were forced to increase through ectopic expression of the Na+-coupled pyruvate transporter SLC5A8. This is confirmed further by the induction of apoptosis specifically in colon cancer cells by siRNA-mediated silencing of HDAC1 and HDAC3.
We stumbled upon the tumour-suppressive function of pyruvate when we were investigating the significance of the silencing of SLC5A8 in cancer. The silencing of this gene was first reported in colon cancer . This led us to postulate that SLC5A8 may be a transporter for the short-chain fatty acid butyrate, which is an HDAC inhibitor, a tumour suppressor, and a bacterial fermentation product generated in the colonic lumen. Subsequent studies from our laboratory and from others have shown that SLC5A8 is indeed a Na+-coupled transporter for butyrate and other short-chain fatty acids such as acetate and propionate [32–34]. Interestingly, the transporter is expressed not only in the colon, where its role in butyrate transport may provide the reason for its expression, but also in the kidney, where there is no possible relevance of butyrate transport. This led us to hypothesize that SLC5A8 in the kidney may be responsible for the reabsorption of lactate, which is a monocarboxylate similar to butyrate. Subsequent studies from our laboratory have shown that this is indeed true [39,40]. Detailed investigations of the substrate specificity of SLC5A8 followed, which showed that the transporter is able to transport a variety of other monocarboxylates, including pyruvate [32,39], nicotinate , β-hydroxybutyrate  and monocarboxylate drugs . During this time, studies from other laboratories have demonstrated that the silencing of SLC5A8 is not unique to colon cancer. The gene is silenced in thyroid cancer [43–45], stomach cancer  and brain cancer . This was puzzling because these non-colonic tissues are not exposed to butyrate. The butyrate connection to the tumour-suppressive function of SLC5A8 is not relevant to non-colonic tissues, such as the thyroid, stomach and brain. This suggested that some other endogenous metabolite commonly found in circulation must be a substrate for SLC5A8 and function as a tumour suppressor similar to butyrate. This rationale led us to the discovery that pyruvate, a high affinity substrate for SLC5A8, is an HDAC inhibitor and a tumour suppressor . Since then, the silencing of SLC5A8 has been shown to occur in other non-colonic tissues such as the mammary gland , pancreas  and prostate , extending the relevance of the pyruvate/SLC5A8 connection to cancer in a broader range of tissues.
The studies reported here demonstrate that the silencing of SLC5A8 is not the only means used by cancer cells to avoid pyruvate-induced HDAC inhibition and cell death. These cells have an elaborate mechanism to maintain low intracellular levels of pyruvate. It includes differential expression of LDH isoforms and pyruvate kinase splice variants. These studies suggest that elevation of intracellular levels of pyruvate may provide a potential therapeutic strategy in the treatment of cancer. Possible means to achieve this goal include induction of SLC5A8 and LDH-B expression with the use of DNA methylase inhibitors, and inhibition of LDH-A. These therapeutic strategies may be applicable to cancer treatment in a wide variety of tissues because the differential expression of LDH-A and LDH-B and the silencing of SLC5A8 appear to be a common phenomenon in cancer.
This research received no specific grant from any funding agency in the public, commercial or not-for-profit sectors.