Sulphide-induced energy deficiency in colonic cells is prevented by glucose but not by butyrate


Dr M. C. Eggo, Department of Medicine, University of Birmingham Medical School, Birmingham, B15 2TT, UK. E-mail:



In ulcerative colitis, hydrogen sulphide is postulated to impair colonocyte butyrate metabolism, leading to cellular energy deficiency and dysfunction.


To determine the effects of sulphide exposure on butyrate metabolism and adenosine triphosphate levels of HT29 colonic epithelial cancer cells, and to establish whether energy deficiency can be prevented by increased butyrate concentrations or the presence of glucose.


HT29 cells were maintained in medium containing 3 mM butyrate, 5 mM glucose, or both substrates. Oxidation rates were measured by 14CO2 release from 14C-labelled substrates. Cellular adenosine triphosphate was assayed using the luciferin/luciferase chemiluminescent method. The effects of sulphide (0–5 mM) on substrate oxidation and adenosine triphosphate levels and of increasing butyrate concentration (0–30 mM) with sulphide were observed.


HT29 cells showed similar energy substrate usage to primary colonocyte cultures. Sulphide exposure inhibited butyrate oxidation and led to a reduction in cellular adenosine triphosphate. This fall was prevented by co-incubation with glucose, but not by increasing concentrations of butyrate.


HT29 cells utilize butyrate as an energy substrate and represent a useful in vitro model of the effects of sulphide on colonocytes. Sulphide inhibits butyrate oxidation and leads to demonstrable energy deficiency, prevented by the presence of glucose but not by increased butyrate concentrations.


This study examines the energy deficiency hypothesis for the aetiology of ulcerative colitis.

The short chain fatty acid, butyrate, is produced within the colonic lumen by anaerobic bacterial fermentation. It is readily absorbed, and may provide around 70% of the energy requirement of the colonic mucosa.1, 2 Butyrate is the most readily utilized fuel available to colonocytes, as shown by a far lower Km value (the substrate concentration producing 50% of the maximal velocity for the reaction) than for other short chain fatty acids or glucose.3, 4 Defective metabolism of butyrate was first proposed as an aetiological factor in the inflammatory bowel disease ulcerative colitis by Roediger in 1980,5 and is postulated to lead to energy deficiency at the cellular level, cellular dysfunction and hence to the changes of colitis. Reduced metabolism of butyrate in colitic compared with normal mucosa has since been demonstrated both in vitro6 and in vivo7 in human tissue, and in animal models.8 Although attempts have been made to calculate theoretical adenosine triphosphate (ATP) values from substrate metabolism, energy deficiency has never been directly demonstrated.

Because no significant deficiency in the activity of enzymes involved in butyrate β-oxidation has been found in ulcerative colitis patients,9 another factor must be responsible for the defective butyrate oxidation. Animal models suggest that a bacterial toxin may be a more likely cause, as pre-treatment with metronidazole can prevent dextran sodium sulphate-induced colitis in mice.10 Animals maintained in germ-free conditions also fail to develop colitis, showing that the presence of bacteria is essential. Hydrogen sulphide has been advocated as a luminal toxin in ulcerative colitis.11 Exposure of normal colonocyte cultures to sulphide reproduces the defect in butyrate metabolism seen in colonocytes of colitic patients12, 13 by inhibiting the enzyme butyryl co-enzyme A dehydrogenase.14

Sulphate reducing bacteria within the colonic lumen use dietary sulphite ions as terminal electron acceptors, helping to maintain the luminal fermentation balance and anaerobic environment.15 The end product of this reaction is the highly toxic compound hydrogen sulphide. Sulphate reducing bacteria are universally carried by, and are seen in larger numbers in, ulcerative colitis patients.16, 17 Sulphide levels in the colonic lumen are likely to be related to the numbers of sulphate reducing bacteria carried, as well as to the concentrations of metabolizable substrates, and are higher in the distal colon18 where features of ulcerative colitis first appear. Actual concentrations of hydrogen sulphide within the colonic lumen in vivo are not known. Faecal sulphide has been measured in several studies and may be higher in patients with ulcerative colitis,16, 19, 20 but, due to the volatile nature of hydrogen sulphide, these figures may not be representative of those within the colonic lumen.

The aim of this study was to determine the effects of sulphide on ATP concentration and butyrate metabolism in colonic epithelial cells, and to investigate whether any resulting energy deficiency could be prevented by increasing butyrate concentrations (as clinical evidence suggests that increasing luminal butyrate may be advantageous in ulcerative colitis patients). 21, 22 Our initial aim was to examine both primary cultures of colonocytes and established colonic cell lines. Failure to detect ATP in primary cultures, despite numerous attempts, led us to focus on the HT29 human colonic cancer cell line, which was originally established in permanent culture from a primary human colonic adenocarcinoma. Sulphide effects were examined in HT29 cells grown in the absence of butyrate, and in HT29 cells acclimatized to butyrate in the culture medium.



HT29 cells were obtained from the European Collection of Cell Cultures. Dulbecco’s modified eagle medium and heat-inactivated foetal bovine serum were obtained from Gibco BRL, Life Technologies (Paisley, UK). Tissue culture plastics and polypropylene tubes were obtained from Scitech (Japan). Radiolabelled 1-14C-butyrate (0.1 mCi/mL) and U-14C- glucose (0.1 mCi/mL) were obtained from DuPont New England Nuclear (Boston, USA). Filter papers were obtained from Whatman International Ltd. (Maidstone, UK). Perchloric acid was obtained from Acros Organics (New Jersey, USA). Scintillation fluid (Optiphase ‘Hi-safe 3’) was obtained from Fisher Chemicals (Loughborough, UK). Unlabelled sodium butyrate and D-glucose, sodium sulphide and materials for the ATP assay were obtained from Sigma-Aldrich Company Ltd. (Poole, UK). Protein assay reagents were obtained from Bio-Rad Laboratories Ltd. (Hemel Hempstead, UK).

HT29 cell culture

HT29 cells were routinely cultured on tissue culture plastic in Dulbecco’s modified eagle medium containing glucose (1000 mg/L), glutamine (862 mg/L), pyruvate (110 mg/L) and amino acids, supplemented with 10% heat-inactivated foetal bovine serum (standard medium), and were screened at intervals to exclude Mycoplasma infection. Cells were used in assays when approximately 90% confluent. A subset of HT29 cells was acclimatized for a period of 9 weeks in standard medium with 3 mM butyrate. This represents a minimum period of six passages, by which time a stable population of slowly dividing cells exhibiting a differentiated phenotype was established.

HT29 energy substrate usage

The estimation of energy substrate oxidation was based on the production of 14CO2 from 14C-labelled substrates. HT29 cells were passaged using 0.25% trypsin in phosphate-buffered saline/ethylenediaminetetra-acetic acid, and re-suspended in a total volume of 1 mL medium in polypropylene tubes. Tubes contained Dulbecco’s modified eagle medium with amino acids, 10 ng/mL insulin and 3 mM butyrate, 5 mM glucose or both 3 mM butyrate and 5 mM glucose. This concentration of glucose is within the normal physiological range. Butyrate was used at 3 mM as this is within the optimal concentration range for HT29 cell butyrate oxidation (P. Aujla 2000, unpublished data). Radiolabelled substrate (0.25 μCi of either butyrate or glucose) was added to each tube. Tubes were incubated at 37 °C, 5% CO2 for 3 h. At the end of this period, filter papers soaked in 2 M sodium hydroxide solution were placed inside the lids of the tubes, and 100 μL of 1 M perchloric acid was injected through the lids to acidify the medium and liberate the radiolabelled 14CO2. Tubes were incubated at 37 °C for a further 1 h to allow entrapment of 14CO2 on the filter papers, which were then removed to scintillation vials containing 4 mL Optiphase ‘Hi-safe 3’ scintillation fluid. Disintegrations per minute (dpm) were counted on an LKB scintillation counter, and used to calculate the total amount of 14CO2 produced. Counts were corrected for non-specific activity due to volatility of the radiolabelled substrates.

The total 14CO2 values were corrected for the amount of protein in each tube, protein being assayed by a modified Lowry method.23

Sulphide effects on HT29 cell metabolism

Assays were performed as described above by adding sodium sulphide solution to tubes to give final sulphide concentrations in the range 0–5 mM. The total 14CO2 levels were calculated and corrected for protein levels in the same way.

ATP assays

HT29 cells were cultured in 24-well tissue culture plates, and allowed to grow in standard medium until approximately 90% confluent. The medium was then changed to Dulbecco’s modified eagle medium containing amino acids, 10 ng/mL insulin and 3 mM butyrate, 5 mM glucose or both 3 mM butyrate and 5 mM glucose (serum-free). Cells were incubated for 16 h prior to the addition of sodium sulphide solution (final concentrations of sulphide, 0–5 mM). Cultures were incubated for 3 h, after which cells were assayed for ATP and protein concentrations. A chemiluminescent method utilizing firefly luciferase24 was used to measure ATP levels.

Statistical methods

Quadruplicate samples were used for all assays. Data are expressed as the mean ± standard error of the mean (± S.E.M.). One-way analysis of variance (ANOVA) was used to make comparisons between groups in all cases, except for the results of HT29 cell substrate usage experiments and comparison between conditions for the varied butyrate concentration experiments, when an unpaired two-tailed t-test was performed. In cases in which ANOVA produced a statistically significant result (P < 0.05), Dunnett’s multiple comparison test was used to further compare groups with the control (no sulphide) group.


HT29 cell energy substrate usage

HT29 cells oxidize significantly more butyrate than glucose (Figure 1). 14CO2 production from 3 mM butyrate was 0.827 ± 0.075 nmol/min/mg protein, compared with only 0.189 ± 0.024 nmol/min/mg protein from 5 mM glucose. This corresponds to 0.207 ± 0.019 nmol/min/mg protein of butyrate oxidized, compared with only 0.032 ± 0.004 nmol/min/mg protein for glucose.

Figure 1.

 Oxidation rates for butyrate and glucose by HT29 cells. Oxidation rates (nmol substrate.min/mg protein) according to substrate by HT29 cells in serum-free culture with 3 mM butyrate, 5 mM glucose or both substrates (n=3).

When cells were incubated in medium containing both 3 mM butyrate and 5 mM glucose, no significant difference was seen in the rate of oxidation of either substrate: 14CO2 production rates were 0.799 ± 0.059 and 0.158 ± 0.025 nmol/min/mg protein for butyrate and glucose, respectively. Oxidation rates for cells conditioned to butyrate for a prolonged period were not significantly different from those for HT29 cells cultured in standard medium.

Effect of sulphide on HT29 metabolism

Co-incubation with 3 mM butyrate and hydrogen sulphide (0.1–5 mM) significantly reduced the production of 14CO2 from radiolabelled butyrate in a dose-dependent manner (Figure 2). Incubation of HT29 cells with 5 mM glucose and the same concentration of sulphide produced no significant fall in the rate of 14CO2 production from 14C-glucose, although a downward trend was noted at 2.5 and 5 mM sulphide (data not shown). Butyrate, but not glucose, oxidation rates were similarly inhibited by sulphide in HT29 cells conditioned in 3 mM butyrate.

Figure 2.

 Effects of sulphide exposure on butyrate oxidation and adenosine triphosphate (ATP) levels. (A) Butyrate oxidation, as measured by 14CO2 production (mean ± S.E.M.), by HT29 cells in serum-free medium containing 3 mM butyrate and sulphide concentrations in the range 0–5 mM (n=4) (*P < 0.05, **P < 0.01, all values compared with no sulphide group). (B) μg ATP/mg protein (mean ± S.E.M.) for HT29 cells incubated with 3 mM butyrate and 0–5 mM hydrogen sulphide (n=5) (**P < 0.01, all values compared with no sulphide group).

Effect of sulphide on ATP levels

In cells cultured in 3 mM butyrate, hydrogen sulphide at concentrations above 0.5 mM led to a significant fall in cellular ATP levels as shown in Figure 2. Cells cultured in 5 mM glucose did not show this fall in cellular ATP until much higher sulphide concentrations (as shown in Figure 3), the reduction not quite reaching significance at 5 mM hydrogen sulphide (P=0.08). HT29 cells cultured in both 3 mM butyrate and 5 mM glucose showed no decrease in cellular ATP levels compared with control values for concentrations of sulphide up to 5 mM (Figure 3). A significant fall was then seen at 5 mM sulphide (P < 0.01).

Figure 3.

 Effect of sulphide on adenosine triphosphate (ATP) levels according to substrate(s) present. μg ATP/mg protein (mean ± S.E.M.) for HT29 cells co-incubated with 3 mM butyrate, 5 mM glucose or both substrates, and 0–5 mM hydrogen sulphide (n=5) (**P < 0.01, all values compared with no sulphide group).

Effect of increasing butyrate concentration

Increasing the concentration of butyrate available in the cell culture medium up to 30 mM did not prevent the fall in cellular ATP levels on addition of sulphide (Table 1).

Table 1.   Increasing butyrate concentration does not prevent the adenosine triphosphate (ATP) reduction produced by sulphide exposure Thumbnail image of


This study shows that glucose, but not butyrate, can protect HT29 colonic cancer cells against energy deficiency induced by exposure to sulphide. When cells were cultured with butyrate alone, exposure to relatively low concentrations of hydrogen sulphide likely to be seen in vivo led to a fall in cellular ATP levels. Oxidation of butyrate was inhibited in these cells by sulphide at a similar concentration to that previously described for primary cultures of rat and human colonocytes.13 The effect of sulphide on butyrate oxidation was only responsible in part for the energy deficiency, however, as, in the absence of added energy producing substrate, sulphide exposure still led to a significant fall in ATP levels (Table 1). This is likely to be due to the inhibitory effect of sulphide on cytochrome c oxidase, downstream of glycolysis and β-oxidation.25, 26 Although sulphide exposure produced a relative ‘energy deficiency’ if butyrate was present as the only energy substrate, this effect was prevented by the co-incubation with physiological levels of glucose (Figure 3). We deduce that, if the glucose level in ulcerative colitic mucosa remains normal, it is unlikely that sulphide inhibition of butyrate metabolism will lead to energy deficiency within the mucosa.

Our initial intention was to examine sulphide effects on ATP levels in primary cultures of human colonocytes obtained from surgical resection specimens. However, despite numerous attempts, and modifications to the method of primary culture, we were unable to detect significant levels of ATP in these cells. Colonocytes in primary culture are generally mechanically or chemically stripped from the extracellular matrix, a procedure likely to result in the triggering of apoptotic mechanisms. This, together with the failure to detect ATP in such cells in this study, may explain why they tend to be short lived in culture.

Although routinely cultured in the absence of butyrate, HT29 colonic cells can oxidize significant quantities of this substrate, the oxidation rate for butyrate being far in excess of that for glucose. The rates of butyrate oxidation found in HT29 cells are very similar to published data for human colonocyte primary cultures.1, 5 Rates of glucose oxidation are relatively low in HT29 cells, but this is likely to be coupled with a high rate of glycolysis as seen in primary colonocyte cultures.1 When both butyrate and glucose were present, no inhibition of glucose oxidation as a result of the presence of butyrate was seen. This suggests that the oxidation of the two substrates is non-competitive, and disagrees with the observations made by Roediger;1 however, the concentrations of butyrate and glucose used in his paper were far higher than those in our study. The ATP levels seen in cells incubated with butyrate were far lower than the levels observed in cells cultured with glucose, despite the higher oxidation rate of butyrate. However, only the net ATP concentration can be measured — the result of production and utilization by the cell. Butyrate is known to induce manifold changes at the molecular level in cells, and this may result in an increased utilization of ATP and hence a reduced net ATP level. Sulphide exposure was seen to reduce net levels of ATP in a dose-dependent manner. HT29 cells conditioned in butyrate-containing medium for a minimum of six passages showed a more differentiated phenotype, but no significant change was found in their rate of butyrate oxidation or in the effects of sulphide. We propose that the HT29 human colonic cancer cell line represents a reasonable model for the effect of sulphide on colonocytes.

In HT29 colonic cells, the inhibitory effect of sulphide on ATP levels was not overcome by increasing butyrate concentrations, suggesting a non-competitive interaction between sulphide and butyrate metabolism. This observation is consistent with the mixed results obtained in trials of butyrate enemas for the treatment of ulcerative colitis.21, 22, 27–29

In conclusion, if these in vitro results can be extrapolated to colonic epithelial cells in vivo, it appears likely that hydrogen sulphide inhibits butyrate metabolism in colonocytes, and produces a fall in ATP generated from this substrate. However, if physiological levels of glucose are present within the colonic mucosa, no overall energy deficiency due to sulphide effects on butyrate metabolism seems likely.


We thank the Royal College of Surgeons of England for providing funding for this project via a Research Fellowship for Sarah Hulin.