Journal of Applied Physiology

Physiology of sulfide in the rat colon: use of bismuth to assess colonic sulfide production

Michael D. Levitt, John Springfield, Julie Furne, Thomas Koenig, Fabrizis L. Suarez


Colonic bacteria produce hydrogen sulfide, a toxic compound postulated to play a pathogenetic role in ulcerative colitis. Colonic sulfide exposure has previously been assessed via measurements of fecal sulfide concentration. However, we found that <1% of fecal sulfide of rats was free, the remainder being bound in soluble and insoluble complexes. Thus fecal sulfide concentrations may reflect sulfide binding capacity rather than the toxic potential of feces. We utilized bismuth subnitrate to quantitate intracolonic sulfide release based on observations that bismuth1) avidly binds sulfide; 2) quantitatively releases bound sulfide when acidified; and 3) does not influence fecal sulfide production by fecal homogenates. Rats ingesting bismuth subnitrate excreted 350 ± 18 μmol/day of fecal sulfide compared with 9 ± 1 μmol/day in control rats. Thus the colon normally absorbs ∼340 μmol of sulfide daily, a quantity that would produce local and systemic injury if not efficiently detoxified by the colonic mucosa. Studies utilizing bismuth should help to clarify the factors influencing sulfide production in the human colon.

  • hydrogen sulfide
  • colonic mucosal injury
  • bismuth subnitrate

fecal bacteria release hydrogen sulfide (H2S) via the metabolism of inorganic sulfate and a variety of organic sulfur-containing compounds such as mucin and cysteine (8). Although H2S is known to have a 50% lethal dose comparable to that of cyanide (19), there is limited information concerning the physiology of this gas in the colon. Recently, interest in fecal sulfide production has been stimulated by the observation that exposure of the colonic mucosa to sulfide inhibits its ability to metabolize butyrate (17, 18), the preferred fuel of the mucosa (15). The finding of a similar defect in butyrate oxidation in the colonic mucosa of subjects with ulcerative colitis (UC) (1) led to the proposal that sulfide toxicity could play a pathogenetic role in this condition.

Previous investigators have attempted to determine whether the colon of subjects with UC is exposed to excessive sulfide via comparison of sulfide concentrations in feces obtained from controls and subjects with UC. Two studies (3, 6) showed moderate (20–40%) but statistically significant increases in fecal sulfide in UC. A third study (14) showed no significant increase in total fecal sulfide in UC, although free sulfate was significantly increased. The most comprehensive study (11) found no difference in sulfide concentrations and concluded that subjects with UC do not produce excessive quantities of sulfide.

The interpretation of fecal sulfide measurements is complicated by the ability of sulfide to exist in multiple forms. Sulfide tightly binds to a variety of fecal compounds, yielding insoluble and soluble complexes. Unbound sulfide theoretically can exist as S−2, HS, or H2S. The acidic dissociation constants of these forms of sulfide are 14 and 6.9 for S−2 ↔ HS and HS ↔ H2S, respectively (13). Therefore, at the pH of feces (usually 6.0–7.0), there is negligible S−2, and at any moment at least 50% of unbound sulfide is in the form of H2S. Because the gut mucosa is extremely permeable to H2S (21), the extrafecal concentration of H2S is maintained at a very low level relative to feces, allowing free sulfide to rapidly leave feces in the form of highly diffusible H2S. In this case, fecal sulfide concentration would primarily reflect the sulfide binding capacity of feces rather than sulfide production rate in the colon.

Our previous observations (22) suggested that bismuth, via its ability to bind sulfide, could provide a novel tool with which to measure sulfide release in the colon. The present report describes studies carried out in rats in which bismuth was employed to obtain the first quantitative measurements of sulfide release in the colon.



All studies were carried out by using 300- to 400-g male Sprague-Dawley rats. The basic diet of the rats was standard chow (Harlan Teklad, Madison, WI) ingested ad libitum. In some experiments, bismuth subnitrate was added to the chow. The quantity of food ingested was monitored daily, and the weight of the rats was recorded weekly.

Incubation vessels.

In a previous study (20), we found that H2S rapidly reacted with glass and most plastic surfaces but was quite stable in polypropylene containers. All incubations were carried out in polypropylene syringes sealed with stopcocks.

H2S release by incubated fecal specimens.

The plunger was removed from a 20-ml preweighed syringe fitted with a stopcock. Rats were induced to defecate via palpation of the lower abdomen, and fecal samples were collected directly into a preweighed syringe, which had been prewarmed to 37°C. The plunger was immediately reinserted, the gas space was flushed with N2, and the stopcock was then sealed with 20 ml of N2 in the syringe. The syringes were immediately placed in a 37°C incubator, and 0.3-ml gas samples were removed at intervals over 1 h for measurements of H2S release into the gas phase. At the end of the experiment, the syringes were weighed, and fecal weight was determined by difference. The rate of release of H2S per gram wet weight of feces was calculated from the concentration of H2S accumulating in the syringe, the gas space of the syringe, and the fecal weight.

Free vs. bound sulfide content of feces.

Free and bound fecal sulfide concentrations were initially determined by using a technique described in the literature in which an alkaline fecal homogenate is centrifuged and sulfide concentrations are measured in the precipitate and the supernatant (2). (All sulfide is ionized at the high pH of this solution, preventing H2S escape into the atmosphere during the homogenization process.) A freshly passed fecal sample was collected in 10 ml in 0.5% NaOH. The sample was disrupted via vortexing and then centrifuged at 45,000g for 10 min. The precipitate was washed with an additional 10 ml of 0.5% NaOH, and the sulfide concentrations of the precipitate and the supernatant plus wash were determined.

Free sulfide concentration also was determined via analysis of the concentration of H2S in a gas space equilibrated with a fecal homogenate. The accuracy of technique requires inhibition of all ongoing H2S production during the homogenization and equilibration process, and this inhibition was obtained via two methods. In both methods, the plunger was removed from a preweighed 50-ml syringe fitted with a stopcock, and a freshly passed fecal sample was collected directly into the barrel of the syringe. In one set of experiments (n = 5), the syringe contained 10 ml of ice-cold 0.5% NaOH, to which had been added 0.15 M sodium phosphate (which served as a buffer when the sample was titrated to pH 6.7). In the second set of experiments (n = 5), the fecal sample was collected into a syringe barrel that contained 10 ml of ice-cold 0.15 M phosphate buffer (pH 6.7) containing ethanol at a concentration of 20%. The syringe plungers were immediately replaced, 40 ml of air were added to the syringe, and the syringe was sealed with a stopcock. While maintained at ice-water temperature, the feces were disrupted via vigorous agitation of the syringe by using a vortex mixer. The pH of the alkaline homogenate was then reduced to 6.7 via the addition of ∼0.2 ml of 50% HCl injected via a needle inserted through the stopcock. (The precise quantity of HCl required for this neutralization was predetermined via titration of an aliquot of the homogenate.) The fecal homogenates (in both sets of experiments) were then incubated at 38°C with constant agitation and frequent additional brisk stirring with the use of a vortex mixer. Gas samples were removed at 30 min for H2S analysis, the homogenate was analyzed for sulfide, and the free-to-total sulfide ratio was determined. Calculation of the free sulfide concentration of the homogenate from the H2S concentration observed in the headspace requires knowledge of the distribution of free sulfide between the gas and liquid phases. To this end, 10 μl of varying concentrations of sodium sulfide dissolved in 0.5 N NaOH were added to 50-ml syringes containing 10 ml of 0.15 M phosphate buffer (pH 6.7) or 10 ml of the buffer containing 20% ethanol. The quantity of NaOH added (10 μl) had no measurable effect on the pH of the buffer. The syringes, which contained 40 ml of air, were incubated at 37°C and vigorously mixed at frequent intervals by using a vortex mixer. After full equilibration between the gas and liquid, the H2S concentration of the gas space was determined, and the ratio of H2S in the gas space to sulfide in the buffer was determined.

Similar methodology was also used to measure the free H2S concentration of fecal homogenates. Ten milliliters of an alkaline fecal homogenate (∼0.3 g of feces in 10 ml of 0.5 N NaOH) were transferred to a 50-ml syringe, and 40 ml of air were added to the syringe. The quantity of HCl required to reduce the pH of the homogenate to 6.7 was added via a stopcock to the homogenate, the stopcock was sealed, and the homogenate was vigorously stirred during incubation at 37°C. An aliquot of the gas space was then removed for H2S analysis, and the homogenate was assayed for total sulfide content.

Fecal H2S release during bismuth subnitrate administration.

Calculations based on a previous study indicated that the quantity of bismuth subsalicylate required to bind all fecal sulfide would deliver quantities of salicylate that were near the predicted toxic level for rats. Therefore, bismuth subnitrate was substituted for bismuth subsalicylate in the present experiments. In preliminary studies, bismuth subnitrate was added to rat chow in varying concentrations. After 1 wk on this diet, H2S release by fecal samples was determined during incubation at 37°C as described above. The dosage required to reduce sulfide release to <2% of that observed with feces of control animals was 5 g of bismuth subnitrate per 100 g of chow (data not shown).

Influence of bismuth subnitrate on sulfide production by fecal homogenates.

Under an atmosphere of N2, freshly passed feces were homogenized in 0.1 M PO4 buffer (1 g feces to 2 ml buffer). Aliquots (0.5 ml) of this homogenate were transferred to 20-ml syringes, and 4–10 mg of bismuth subnitrate were added to some of the syringes. After either 2, 4, or 24 h of incubation at 38°C (n = 5 for each time period), the quantity of H2S in the gas space was determined. The total sulfide content of the gas space plus the homogenate was then determined by adding 2 ml of 2% zinc acetate to the syringe (which bound all free H2S) followed by analysis of the liquid phase for sulfide.

Fecal sulfide content of bismuth-treated rats.

Rats (5 in each group) were fed chow or chow plus bismuth subnitrate (6.25 g/100 g chow). After 1 wk, individually passed fecal samples or 24-h collections of fecal samples were homogenized in 2% zinc acetate, and the homogenates were analyzed for sulfide content.

Cecal H2S concentration and sulfide content of cecal contents.

The volume of H2S in the cecum of five animals was determined at about 10 AM after a nocturnal period when the rats had been allowed to eat ad libitum. Under pentobarbital anesthesia, a midline laparotomy was performed. The cecum was identified, and a ligature was loosely placed around the terminal ileum. The right colon was ligated at its juncture with the cecum. A 240 PE catheter was inserted into the terminal ileum. The catheter was advanced into the cecum, and the ligature was tightened to prevent leakage. A 5-ml bolus of nitrogen was injected into the cecum, which was gently palpated to induce mixing. After a 30-s period of equilibration, an aliquot of cecal gas was removed for analysis for H2S. The sulfide content of the fecal contents of cecum was determined by flushing all material from the cecum with 30 ml of 2% zinc acetate. This flushed material was quantitatively collected and analyzed for sulfide concentration.

Fate of H2S instilled into the colon.

In two rats, under pentobarbital anesthesia, the cecum was isolated between ligatures as described above, and H2 35S (∼1 μCi) was instilled into the cecum. Urine collected for 1 h was analyzed by using an HPLC technique that separated thiosulfate from sulfate.

Analytic techniques.

Fecal sulfide was measured by using a previously described technique (5). The plunger was removed from a 20-ml syringe fitted with a stopcock, and a 0.5-ml aliquot of a fecal homogenate was added to the syringe. (The fecal homogenate had been previously treated with either zinc acetate or 0.5% NaOH; hence sulfide was bound to zinc or was ionized and not volatile.) The plunger was then reinserted into the syringe such that there was 19 ml of air in the syringe. (Recovery studies showed similar sulfide recovery with air vs. N2 in the gas space.) The fecal homogenate was then acidified by the addition of 0.5 ml of concentrated HCl contained in a 1-ml syringe, which was injected via a needle inserted through the stopcock attached to the 20-ml syringe. The needle was immediately withdrawn, and the stopcock was closed. After 30 min of agitation, a 0.3-ml aliquot of the gas space was analyzed for H2S concentration via gas chromatography. Because H2S has a relatively high water-to-gas solubility, a minor correction is necessary for H2S that remains dissolved in the acidified homogenate. Over a wide range of sulfide concentrations (tracing either H2S or H2 35S), acidification resulted in 80 ± 2% recovery of sulfide as H2S in the gas phase, and 20% remained in the homogenate by using the gas and liquid volumes employed above. Thus the total sulfide content of the 0.5 ml of homogenate was calculated as followsTotal sulfide/0.5ml =[H2S]in gas space×19ml volume of gas space/0.80 where brackets denote concentration. The total sulfide concentration of feces was calculated from the above measurement and the feces-to-buffer ratio in the homogenate (usually ∼1 g feces per 15 ml of homogenate).

The free sulfide content determined from the concentration of H2S in a 40-ml gas space equilibrating with 10 ml of pH 6.7 fecal homogenate was calculated as followsTotal free sulfide/10ml homogenate=quantity ofH2Sin40ml gas space+free sulfide in homogenate With the use of sulfide standards, the ratio of H2S in 40 ml of gas to sulfide in 10 ml of pH 6.7 buffer was observed to average 1:1.5 over a wide range of sulfide concentrations.

The free sulfide content of feces homogenized in 20% ethanol in buffer (pH 6.7) was calculated in a similar fashion to that described above, with the exception that the distribution of H2S in gas-to-pH 6.7 buffer-ethanol mixture was 1:1.6.


The labeled metabolites excreted in the urine after instillation of H2 35S into the cecum of two rats were identified by HPLC (solvent delivery module LC-6A and Chromatopac data processor model C-R3A, Shimadzu, Kyoto, Japan) run at 2 ml/min and 2,000 lbs. pressure, using a 4 × 250 mm anion ion exchange column (IonPac AS16, Dionex, Salt Lake City, UT) and a conductivity monitor (Amersham Pharmacia Biotech, Piscataway, NJ) for mass measurements (9). The eluent was 20 mM NaOH. A 200-μl volume was injected onto the HPLC column, and 2.0-ml fractions were collected in individual scintillation vials. The radioactivity of these fractions was determined by scintillation counting, and the identity of the35S-labeled compounds was determined via comparison of their retention times to the retention times of authentic sulfate and thiosulfate.


H2S release by fecal specimens.

The rate of release of H2S by unmanipulated fecal specimens incubated at 37°C is shown in Fig. 1. The rate of release of H2S was constant over 1 h and averaged 6.3 ± 1.3 nmol · g−1 · min−1.

Fig. 1.

Release of hydrogen sulfide by freshly passed rat feces incubated at 37°C for 1 h. Values are means ± SE.

Total and free sulfide concentrations of fecal specimens of rats on a chow diet.

The concentrations of sulfide observed in the supernatant and precipitate fractions of an alkaline fecal homogenate averaged 0.11 ± 0.004 and 0.46 ± 0.02 μmol/g feces, respectively, with an apparent free-to-bound ratio of 1:4.2. When the supernatant was titrated to pH 6.7 and equilibrated with a gas space (n= 3), the free and bound concentrations averaged 0.0043 μmol/g and 0.107 μmol/ml, respectively, yielding a mean free-to-bound ratio of 1:23 in the supernatant. Given that a mean of 1/5.2 of the total sulfide of homogenates was in the supernatant, the free sulfide of the supernatant represented roughly 1/120 of the total fecal sulfide content.

The H2S concentration in a gas space equilibrated with feces initially homogenized in NaOH and then titrated to pH 6.7 averaged 0.37 parts/million (ppm), yielding a calculated free sulfide concentration of only 0.0027 ± 0.0028 μmol/g feces and a free-to-total sulfide ratio of 1:159. The H2S concentration in the gas space over feces homogenized in 20% ethanol averaged 0.19 ppm, indicating a free sulfide concentration of 0.0030 ± 0.0002 μmol/g and a free-to-total ratio of 1:244.

Influence of bismuth on H2S release and sulfide production by fecal homogenates.

Figure 2 shows the accumulation of free sulfide (calculated from H2S release) and total sulfide of fecal homogenates incubated for 2, 4, or 24 h in the presence or absence of bismuth subnitrate. In contrast to the copious release of H2S by the control homogenates, negligible H2S was released by the bismuth-treated samples. However, the total sulfide observed with the bismuth-treated homogenates at each time point was not significantly different from that of the untreated homogenates (P > 0.3 for each of the three time points).

Fig. 2.

Free, bound, and total sulfide concentration ([sulfide]) in rat fecal homogenates incubated for 2, 4, and 24 h, in the presence of (open bars) or absence of (solid bars) bismuth subnitrate. Note that negligible free hydrogen sulfide was released in the presence of bismuth, but the total sulfide production was similar in the bismuth-treated and untreated preparations. Values are means ± SE.

Fecal sulfide during bismuth subnitrate administration.

The fecal sulfide concentration of bismuth subnitrate-treated animals averaged 23 ± 1.5 μmol/g. The 24-h fecal sulfide output of bismuth subnitrate-treated animals averaged 350 ± 18.0 μmol/day, a value that was 39 times higher than the 9.0 ± 1.0 μmol/day observed in control animals.

Influence of bismuth subnitrate administration on cecal H2S and sulfide.

The concentrations of H2S in cecal gas of control and bismuth subnitrate-treated animals averaged 1,148 ± 203 and 0.72 ± 0.36 ppm, respectively. The total sulfide content of cecal contents averaged 1.5 ± 0.21 and 17 ± 1.2 μmol/g in control and bismuth subnitrate-treated animals, respectively.

Urinary products resulting from H235S metabolism.

After instillation of H2 35S into the cecum, HPLC analysis of urine showed that virtually all 35S eluted with sulfate.


The distribution of fecal sulfide between free and bound forms has been previously determined by homogenizing feces in a highly alkaline solution to prevent the loss of volatile H2S into the atmosphere (2, 11). After centrifugation, the sulfide concentrations observed in the precipitate and supernatant were assumed to represent bound and free fractions, respectively. A free-to-bound ratio of roughly 1:3 has been reported (2, 11), and, with this technique, a similar value was observed in the present study.

We assessed free fecal sulfide concentration via measurements of the H2S concentration in a gas space equilibrated with a fecal homogenate maintained at a physiological pH of 6.7 and 37°C. This technique assumes that free sulfide equilibrates with H2S, which, in turn, equilibrates with the gas space. Because fecal bacteria produce and release copious H2S immediately on passage (see Fig. 1), this methodology required inhibition of H2S production during the homogenization and equilibration processes. This was achieved in one set of studies via collection and homogenization of feces in cold 0.5 M NaOH, the pH of which is bactericidal. After titration back to pH 6.7, H2S concentration was determined at 37°C in an equilibrated gas space, and free sulfide concentration was calculated from the known distribution of free sulfide between gas and pH 6.7 buffer. Only trivial amounts of H2S entered the gas phase, and the calculated free sulfide concentration was only 0.0027 μmol/g, ∼1/160 of the total sulfide concentration of the feces (0.42 μmol/g). In a second set of studies, feces were homogenized in pH 6.7 buffer plus 20% ethanol to inhibit bacterial H2S production. Free sulfide concentrations were again calculated to be very low (mean 0.0030 μmol/g), with a mean free-to-total sulfide concentration ratio of 1:244. Thus previous measurements, which assumed that all nonprecipitating sulfide was “free,” appear to have grossly overestimated the fraction of fecal sulfide that exists in an unbound state.

Observations from published studies may need to be interpreted in light of the finding that >99% of fecal sulfide is bound. Exposure of colonocytes to a sulfide concentration of 0.2 mM has been found to produce detectable alterations in the metabolism of butyrate, the preferred energy source of the colonic mucosa (15). Given that sulfide concentrations of >1 mM are observed in normal human feces (3, 6, 11), the potential for sulfide toxicity is seemingly present in healthy subjects. However, studies of sulfide toxicity (both in vitro and in vivo) have employed solutions in which all sulfide was free (16, 17). Because bound sulfide presumably has little or no toxic potential, it is not surprising that the concentration of sulfide in feces of healthy controls may far exceed the levels found to be toxic to the mucosa in experimental studies.

Sulfide binding also has implications for studies in which fecal sulfide measurements were used to assess the contribution of dietary components to sulfide production in the colon. For example, the high fecal sulfide concentration observed with meat ingestion has been interpreted to indicate that sulfur-containing compounds in meat provide an important substrate for fecal sulfide production (10). However, as discussed, sulfide production normally far exceeds binding capacity, and free sulfide is rapidly lost from the fecal stream. Thus the sulfide concentration of feces is seemingly determined more by binding capacity than by sulfide production. Whereas meat ingestion may well increase fecal sulfide production, it seems likely that the high fecal sulfide induced by this diet reflects its ability to provide fecal sulfide binders as opposed to its ability to increase fecal sulfide production.

Lastly, multiple studies (1, 3, 6, 11, 14) have attempted to assess the exposure of the colon to sulfide via measurements of fecal sulfide concentrations. The most extensive study (11) concluded that “colonic luminal sulfide is not elevated in ulcerative colitis” based on the finding of comparable sulfide concentrations in feces obtained from controls and subjects with UC. It seems likely that these measurements primarily assessed fecal sulfide binding capacity, not free-sulfide concentration, and these studies shed little light on the colonic exposure to sulfide.

Given that the sulfide binding capacity of material passed per rectum is saturated, the rate of release of H2S from the fecal material should reflect quantitatively the production rate of this gas. Whereas our laboratory found increased rates of H2S release by incubated feces from subjects with UC (8), this result must be interpreted with caution. Rapid colonic transit might yield a fecal sample rich in substrates for sulfide producing reactions, which would enhance sulfide production during incubation. Although perhaps indicative of an increased rectal exposure to sulfide, one could speculate that this high production outside the body was associated with a reduced sulfide release in the colon. Thus it seems unlikely that any measurement reported to date accurately reflects the quantity of sulfide released in situ in the colon, the crucial measurement required to link alterations of sulfide production to colonic disease states.

Because a previous study showed that sulfide was very efficiently metabolized to thiosulfate by the colonic mucosa (9), we initially evaluated the possibility that urinary thiosulfate excretion might serve as a measure of the exposure of the colon to sulfide. However, after intracecal instillation of H2 35S in the rat, virtually all urinary 35S was excreted in the form of sulfate, indicating rapid conversion of thiosulfate to sulfate in the rat. Because sulfate derived from thiosulfate is swamped by sulfate from other sources, urinary sulfate measurements provide no insight into colonic sulfide exposure.

Sulfide release in the colon also could be quantitated if there were a compound that bound all fecal sulfide in a measurable form but did not appreciably alter bacterial sulfide production. Our previous studies with bismuth subsalicylate suggested that bismuth might serve as such a compound (22). The addition of sufficient bismuth subsalicylate to fecal homogenates virtually eliminated the usual copious release of H2S observed with incubation at 37°C, and feces of rats and human subjects ingesting bismuth subsalicylate released almost no H2S. Because the quantity of bismuth subsalicylate required to inhibit fecal sulfide release in rats delivered near toxic quantities of salicylate, bismuth subnitrate was employed in the studies described in the present report.

For bismuth subnitrate to serve as a tool to accurately measure sulfide release in the colon, it needed to be shown that 1) sulfide can be stoichiometrically released in a measurable form from bismuth sulfide and 2) bismuth subnitrate does not appreciably influence sulfide production by the colonic bacteria. Studies in which known quantities of sulfide were added to bismuth subnitrate showed that subsequent treatment with concentrated HCl resulted in near-complete recovery of the sulfide as H2S. The influence of bismuth subnitrate on sulfide production by fecal bacteria was studied in fecal homogenates. As shown in Fig. 2, total sulfide present in fecal homogenates after 2, 4, and 24 h of incubation was similar in bismuth-treated and control fecal homogenates; the virtually complete inhibition of H2S release into the gas space of bismuth-treated feces was quantitatively accounted for by sulfide binding in these homogenates. Evidence that this efficient binding of sulfide by bismuth occurs throughout the colon (as well as in feces) was provided by the findings that cecal H2S concentrations were 1,000 times less in bismuth-treated animals vs. controls and that this reduction in H2S release was associated with a much higher sulfide concentration in cecal contents.

We concluded from the above studies that bismuth administration provides a means with which to quantitate the physiological exposure of the colon to sulfide. Rats were administered a dose of bismuth subnitrate sufficient to produce a >98% reduction in H2S release of feces when incubated for 1 h at 38°C. The 24-h fecal output of sulfide averaged 350 ± 18 μmol/24 h in bismuth-treated rats, a value roughly 39 times greater than the 9 ± 1 μmol/day output observed in controls. Thus >95% of the sulfide produced in the colon leaves the feces and is absorbed by the mucosa, and <5% is excreted in feces.

The colonic sulfide production observed in the present study would have appreciable toxic potential if it reached the systemic circulation. For example, the 50% lethal concentration for atmospheric H2S is 444 ppm for a 1-h exposure in the rat (12). Given an alveolar ventilation in rats of ∼4,000 ml/h (7), this pulmonary exposure yields a dose of 80 μmol over 1 h. The present study indicates that sulfide is absorbed from the rat colon at a rate of roughly 350 μmol/day or 15 μmol/h. Thus the rat is constantly exposed to intracolonically produced sulfide at a rate roughly one-fifth of that which is lethal over 1 h when the route of administration results in systemic exposure. Tissue damage (both locally in the colon and systemically) is prevented by an efficient detoxification mechanism in the colonic mucosa that converts H2S to thiosulfate (9). This system is a relatively specific adaptation of the colonic mucosa, which converts H2S to thiosulfate ∼20 times more rapidly than does the small bowel or gastric mucosa (4).

Although speculative, a very high production of sulfide in the colon possibly could overwhelm the mucosal detoxification system with resultant damage to the colonic mucosa. Support for this speculation requires quantitation of sulfide production in the human colon. Bismuth subsalicylate seemingly provides a tool with which to obtain this measurement. Whereas bismuth administered in an absorbable form is neurotoxic, when complexed with subnitrate or subsalicylate, bismuth is very poorly absorbed. Bismuth subsalicylate is sold as an over-the-counter preparation in the US, and complexes with subnitrate, citrate, and subgallate are used in other parts of the world. Ingestion of a commonly recommended dose of bismuth subsalicylate by human volunteers resulted in the near complete elimination of H2S release by fecal material (22). Thus analysis of feces obtained during administration of bismuth subsalicylate should provide a quantitative measure of sulfide release in the colon. Application of this methodology should make it possible to investigate the factors that influence sulfide production in the human colon and determine whether excessive colonic sulfide release plays a role in disease states.


This study was supported in part by General Medical Research funds from the US Department of Veteran Affairs, and National Institute of Diabetes and Digestive and Kidney Diseases Grant RO1-DK-13093.


  • Address for reprint requests and other correspondence: M. D. Levitt, Research Product Line Director, Minneapolis Veterans Affairs Medical Center (151), 1 Veterans Dr., Minneapolis, MN 55417 (E-mail:Michael.Levitt{at}

  • The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

  • First published December 14, 2001;10.1152/japplphysiol.00907.2001


View Abstract