Hydrogen sulfide (H₂S) and hypoxia inhibit salmonid gastrointestinal motility: evidence for H₂S as an oxygen sensor

Recent studies have shown that exogenous hydrogen sulfide (H₂S) initiates pharmacophysiological responses in most, if not all, organ systems (Li et al., 2011; Olson, 2011). These studies have been instrumental in the hypothesis that hydrogen sulfide is a biologically relevant signaling molecule thereby joining nitric oxide (NO) and carbon monoxide (CO) as ‘gasotransmitters’ (Wang, 2002).

H₂S is of particular interest in the gastrointestinal (GI) tract as it is both produced by GI tissues and generated in large quantities by bacterial flora in the lumen of the gut (Blachier et al., 2010; Hosoki et al., 1997; Linden et al., 2008; Martin et al., 2010; Teague et al., 2002; Wallace, 2010). Both pathological and physiological attributes have been ascribed to H₂S in the GI tract (Wallace, 2010). H₂S has been suggested to contribute to a variety of intestinal disorders including ulcerative colitis, inflammatory bowel disease and colorectal cancer (Attene-Ramos et al., 2010; Medani et al., 2010; Rowan et al., 2009). Conversely, H₂S has been shown to have anti-inflammatory and antinociceptive actions in the GI tract and H₂S ‘releasing’ compounds have been added to non-steroidal anti-inflammatory drugs (NSAIDs) to prevent gastric irritation and promote healing of ulcers (Wallace, 2010; Wallace et al., 2007). H₂S has even been shown to serve as an energy source for ATP production by mucosal epithelial cells (Blachier et al., 2010; Bouillaud and Blachier, 2011; Goubern et al., 2007; Lagoutte et al., 2010).

Exogenous H₂S has been observed to affect both GI smooth muscle and the enteric nervous system. H₂S directly relaxes GI smooth muscle in the mouse stomach and colon (Dhaese and Lefebvre, 2009; Dhaese et al., 2010) and rat ileum (Nagao et al., 2010). In the guinea-pig stomach, ‘low doses’ of H₂S (0.1–0.3 mmol l^–1) enhance resting tension and slightly reduce contractile amplitude, whereas at higher concentrations H₂S suppresses the amplitude of spontaneous contractions (Zhao et al., 2009). The mechanism of H₂S-mediated relaxation is poorly understood and most common mediators of muscle relaxation have in fact been ruled out. In the mouse, H₂S-induced relaxations are independent of potassium (K⁺) channels, NO, cAMP and cGMP pathways, and nerve inhibition, and do not involve membrane ATPase, calcium (Ca²⁺) channels, internal calcium stores or Rho-kinase; H₂S may, or may not, act via activation of myosin light chain phosphatase (Dhaese and Lefebvre, 2009; Dhaese et al., 2010). In the rat ileum, H₂S-mediated relaxations are independent of intrinsic, enteric and visceral afferent nerves, NO and both ATP-(K_ATP) and calcium-sensitive (K_Ca) potassium channels (Nagao et al., 2010). Low-dose activation of guinea-pig gastric smooth muscle may be mediated via inhibition of voltage-gated K⁺ channels and high-dose inhibition may be mediated via K_ATP channel activation (Zhao et al., 2009). In the mouse intestine, H₂S inhibits pacemaker amplitude and frequency of interstitial cells of Cajal at high concentrations (0.5–1.0 mmol l^–1), but may slightly stimulate at low concentrations (50–100 mmol l^–1); these effects appear to be via modulation of intracellular calcium (Parajuli et al., 2010). H₂S also stimulates proliferation of interstitial cells of Cajal (Huang et al., 2010).

Linden and colleagues critically evaluated the criteria for H₂S as a biologically relevant gasotransmitter in the GI tract (Linden et al., 2010). They concluded that, while there is considerable evidence that H₂S is generated by GI tissues and that H₂S can affect a variety of GI functions, H₂S cannot yet be considered an endogenous signaling molecule because, among other things, there is no evidence to date to show that the production of H₂S by GI tissue is regulated.

An alternative to regulating H₂S production to control H₂S signaling would be to actively regulate H₂S metabolism in the face of constitutive H₂S production. We have previously shown that bovine lung tissue, cultured pulmonary arterial smooth muscle cells and isolated mitochondria inactivate H₂S by a P_O2-dependent process at physiologically relevant P_O2 values (Olson et al., 2010). This balance between constitutive H₂S production and O₂-dependent inactivation in cells forms the basis for the hypothesis that H₂S metabolism is the crucial step coupling hypoxia to tissue response, i.e. the ‘O₂ sensor’ (Olson and Whitfield, 2010). Evidence for this mechanism has been found in systemic and respiratory vascular smooth muscle (Olson et al., 2006; Olson et al., 2008a; Olson et al., 2010), urinary bladder smooth muscle (Dombkowski et al., 2006), chromaffin cells (Perry et al., 2009) and O₂-sensing chemoreceptor cells (Li et al., 2010; Olson et al., 2008b; Peng et al., 2010; Telezhkin et al., 2009; Telezhkin et al., 2010).

The GI tract is an intriguing candidate for O₂ regulation of H₂S signaling. Not only is H₂S produced by GI tissue and luminal bacteria but also both tissue and bacterial metabolism can affect P_O2. In the mouse, for example, tissue extraction and bacterial O₂ consumption lower luminal P_O2 from nearly 60 mmHg in the stomach to 11 mmHg in the small intestine and ultimately down to 3 mmHg at the sigmoid–rectal junction (He et al., 1999). A decrease in the tissue perfusion/metabolism ratio, the hallmark of ischemic bowel disease, can halve these values within 5 min (He et al., 1999). These P_O2 values are sufficiently low to inhibit H₂S oxidation and tip the scale in the intracellular environment from net H₂S consumption to H₂S production (Olson et al., 2010).

Although the effects of H₂S on the mammalian GI tract have been examined in some detail, information on H₂S effects in the GI tract of non-mammalian vertebrates is lacking. The present study was designed to examine the effects of this signaling molecule on the GI tract of fish as they are perhaps more intensively studied than any other non-mammalian vertebrate and they provide a broad phylogenetic and evolutionary perspective of vertebrate development. These studies also allowed us to examine the hypothesis that H₂S is involved in O₂ sensing in the GI tract. We first examined the effects of H₂S and hypoxia on the intrinsic rhythmicity and contractile activity of esophagus, stomach, gall bladder, and circular and longitudinal GI smooth muscle of the rainbow trout (Oncorhynchus mykiss). Next, we compared the intestinal responses from rainbow trout with those of a much smaller and technically more manageable salmonid, the juvenile coho salmon (Oncorhynchus kisutsch). We then used the coho intestine to examine the hypothesis that H₂S mediates the hypoxic responses by directly comparing the effects of these two stimuli on mechanical and/or enteric communication in the intestine and by examining the effect of inhibitors of H₂S biosynthesis and a sulfide donor as a precursor of H₂S synthesis on the hypoxic responses. Confirmation of the inverse relationship between H₂S and O₂ was obtained by direct measurement, in real time, of intestinal H₂S production and its O₂-dependent consumption using polarographic H₂S and O₂ electrodes.

Rainbow trout [O. mykiss, Kamloops strain (Jordan 1892), 400–800 g] of either sex were purchased from local hatcheries (Harrietta Hills Trout Farm, Harrietta, MI, USA, and Sweetwater Springs Fish Farm, Peru, IN, USA) and kept in circulating 2000 l tanks containing well-water at 12–15°C, aerated with filtered room air, and exposed to 12 h:12 h light:dark cycles. The fish were fed a maintenance diet of commercial trout pellets (Purina, St Louis, MO, USA). The trout were stunned by a blow to the head and segments of the esophagus, stomach, gall bladder, anterior intestine and posterior intestine were dissected out and placed in cold (∼4°C) Cortland buffered saline (for composition, see below). Anterior intestinal segments were taken from a 2 cm portion of intestine starting ∼1 cm distal to the pyloric cecae. Posterior intestinal segments, 2 cm long, were taken 1 cm from the vent. The dissected tissues were cleaned and placed in fresh Cortland buffered saline at 4°C until use. All segments were used for experimentation within 24 h of removal.

Coho salmon [O. kisutch (Walbaum 1792), 10–20 cm, ∼10–30 g] juveniles were obtained from the Bodine State Fish Hatchery, Mishawaka, IN, USA, and kept in a 500 l rectangular tank containing continuously flowing well-water at 12–15°C for several weeks prior to experimentation. Coho were stunned by a blow to the head and the intestines were removed and prepared for myography as described above.

All procedures followed NIH guidelines and were approved by the local Institutional Animal Care and Use Committee (IACUC).

Circular smooth muscle rings (∼5 mm long) from the trout esophagus, stomach, gall bladder, anterior intestine and posterior intestine, and from the coho anterior intestine and posterior intestine were mounted on 280 mm diameter stainless steel wire hooks through the tissue lumen and suspended in 5 ml water-jacketed smooth muscle baths filled with 14°C Cortland buffered saline and aerated with room air. Segments from both the trout and coho anterior and posterior intestine were also mounted in a longitudinal direction by puncturing the intestinal walls with the same hooks. The cylindrical structure of the intestinal segments was not compromised in either setup. The bottom hooks were stationary; the upper hooks were connected to Grass model FT03C force-displacement transducers (Grass Instruments, West Warwick, RI, USA). Tension was measured on a Grass Model 7E or 7F polygraph (Grass Instruments). Polygraph sensitivity was calibrated prior to the beginning of each experiment and was able to detect changes as small as 5 μN. Data were archived on a PC computer at 1 Hz using Softwire software (Measurement Computing, Middleboro, MA, USA). The chart recorders and software were calibrated for zero tension and 2 mN loads prior to each experiment.

Baseline (resting) tension of ∼5–7 mN for trout gall bladders and ∼8–1.2 mN for the remaining tissues was applied and continuously adjusted for at least 1 h prior to experimentation as the tissues exhibited substantial stress relaxation. Gall bladders and intestinal segments from both fish established baseline resting loads ranging from 2 to 6 mN, and trout stomach and esophagus maintained baseline tension from ∼4 to 8 mN. All tissues were allowed to establish baseline activity in an undisturbed state for a minimum of 1 h prior to the beginning of experimentation. In the absence of external stimuli, trout gall bladder rings and circular and longitudinal intestinal smooth muscle from both fish usually (>90%) exhibited large spontaneous contractions that were approximately 2–5 times resting load. Spontaneous contractions were less frequently observed (<30%) in circular smooth muscle from trout stomach and esophagus. Tissues that did not exhibit spontaneous activity or stretch-induced relaxation were not examined.

Cumulative H₂S concentration–response profiles were obtained for all tissues in the absence of other stimulation. NaHS was used in initial experiments whereas Na₂S was used in later studies as it was more readily available and has fewer sulfur impurities (Doeller et al., 2005). Both salts form HS^– and H₂S when dissolved and we have not noticed any obvious differences in the effects produced by these salts (Dombkowski et al., 2006). In these studies ‘H₂S’ refers to total free sulfide, i.e. dissolved H₂S plus HS^– based on the molarities of the dissolved salts.

H₂S concentration–response profiles were obtained from circular smooth muscle of trout esophagus, stomach and gall bladder, as well as from circular and longitudinal smooth muscle of both trout and coho intestine. Concentration–response profiles were also obtained from trout esophagus, stomach, gall bladder and circular smooth muscle and coho intestinal longitudinal smooth muscle pre-contracted with 10 mmol l^–1 carbamylcholine chloride (carbachol).

The effects of hypoxia, produced by gassing with 100% N₂, were examined in otherwise unstimulated and carbachol (10 mmol l^–1) pre-contracted anterior and posterior circular and longitudinal intestinal smooth muscle from both trout and coho. Unstimulated or carbachol pre-contracted intestinal segments were exposed to either two or three 20 min bouts of hypoxia, separated by 20 min of aeration with room air.

Two protocols were employed to examine the role of H₂S in hypoxic responses. In the first, the effects of hypoxia were examined in coho longitudinal intestinal smooth muscle in the presence of inhibitors of H₂S biosynthesis. The cystathionine γ-lyase (CSE) inhibitor propargylglycine (PPG, 10 mmol l^–1) or the cystathionine β-synthase (CBS) inhibitor amino-oxyacetic acid (AOA, 1 mmol l^–1) was added to the tissues 30 min prior to hypoxia. In another group of experiments, the tissues were exposed to the inhibitor for 30 min then pre-contracted with 10 mmol l^–1 carbachol and hypoxia was administered when the carbachol response plateaued. In a third group of experiments the sulfide donor cysteine (0.1, 1 or 10 mmol l^–1) was added 30 min prior to the 10 mmol l^–1 carbachol pre-contraction and subsequent hypoxia.

H₂S production by homogenized trout intestine was measured in real time using an amperometric (polarographic) H₂S sensor constructed in the laboratory (Whitfield et al., 2008). Approximately 1 g of anterior intestine was homogenized, on ice, in 9 ml of Cortland buffered saline and gassed on a rotary tonometer with humidified 100% nitrogen to remove the oxygen. A 1.5 ml sample of homogenate was placed in a closed chamber with ports for simultaneous measurement of oxygen with a standard Clark-type electrode and H₂S. H₂S production was measured in the absence and in the presence of 1 and 10 mmol l^–1 cysteine. Room air was administered through an injection port and the relationship between P_O2 and P_H2S recorded. The relationship between P_H2S and H₂S concentration was derived from a standard curve using Na₂S; the rate of H₂S production was obtained from the slope of the H₂S concentration as a function of time.

Text files were converted to .acq files for analysis with Biopac Lab Pro (AcqKnowledge, Biopac Systems Inc., Goleta, CA, USA). Four different parameters were captured for each tissue segment using this software: (1) baseline tension, defined by the average tension in between spontaneous contractions; (2) spontaneous contraction amplitude, defined by the difference between the highest and lowest tension measured during each cycle; (3) spontaneous contraction frequency, if present; and (4) total contractile activity, determined as the area under the contractile curve (AUC) from zero tension, which essentially captures baseline tension and spontaneous activity combined. For comparison, all data were averaged over a 600 s (10 min) time frame immediately before treatment and during the final 600 s (10 min) of each respective H₂S concentration or hypoxic exposure. Data presented are all from the initial exposure to hypoxia. For all H₂S concentration–response experiments on carbachol pre-contracted segments, only the total contractile activity was examined for comparison with otherwise unstimulated tissues. All four parameters were analyzed in the coho intestinal 1 mmol l^–1 Na₂S bolus experiments and hypoxic experiments with PPG, AOA and cysteine. Esophageal and gastric rings infrequently displayed spontaneous contractions; therefore, only baseline tension and total contractile activity were examined for these two tissues. Because H₂S and hypoxia reduced all parameters, the H₂S and hypoxic responses are presented as a percentage of pretreatment values. The H₂S concentration–response curves were truncated at 1 mmol l^–1 NaHS or Na₂S as higher concentrations are quite likely irrelevant physiologically. The effective concentration for half-maximal response (EC₅₀) was calculated using TableCurve® (Jandel Corp., Chicago, IL, USA).

In several instances the relative responses in different intestinal segments to H₂S or hypoxia were not significantly different and these data were combined. In coho, only two significant differences were found between anterior and posterior intestinal segments. The amplitude of spontaneous contractions in both anterior circular and anterior longitudinal smooth muscle was nearly twice the amplitude observed in posterior tissues (Fig. 1). This was likely due to the degree of muscularity, which was considerably greater in the anterior segments. Because the relative (as a percentage of control) responses of these segments to either H₂S or hypoxia were not significantly different, the percentage change in anterior and posterior intestinal data were combined for EC₅₀ calculations and for comparisons between circular and longitudinal smooth muscle preparations. In trout, H₂S and hypoxia produced a significantly greater relaxation of baseline tension in posterior longitudinal intestinal smooth muscle than in anterior longitudinal intestinal smooth muscle (Fig. 2). However, all of the other responses to H₂S and hypoxia expressed as a percentage of pretreatment values (total contractile activity, spontaneous contraction amplitude and frequency) were not significantly different and these data were pooled for calculation of EC₅₀ values and for comparisons between circular and longitudinal intestinal smooth muscle. Student's t-tests and one-way ANOVA were used for comparisons between groups with SigmaStat® (Jandel Corp.). Results are presented as means ± s.e.m. Significance was assumed at P≤0.05.

Unless otherwise stated all chemicals were purchased from Sigma-Aldrich (St Louis, MO, USA). Na₂S and NaHS were purchased from Fisher Scientific (Pittsburgh, PA, USA). Cortland buffered saline (pH 7.8) was: 124 mmol l^–1 NaCl, 3 mmol l^–1 KCl, 1.1 mmol l^–1 MgSO₄·7H₂O, 2 mmol l^–1 CaCl₂·2H₂O, 5.55 mmol l^–1 glucose, 12 mmol l^–1 NaHCO₃, 0.09 mmol l^–1 NaH₂PO₄ and 1.8 mmol l^–1 Na₂HPO₄. NaHS and Na₂S stock solutions were made fresh daily under nitrogen and used within 8 h of preparation. By convention, H₂S is used hereafter to indicate either Na₂S or NaHS.

There was no significant deterioration in any of the measured parameters in any of the control tissues for up to 4 h (not shown). All experimental protocols were completed in 4 h or less.

The effects of H₂S on total contractile activity (AUC), baseline tone, spontaneous contraction amplitude and frequency of spontaneous contractions for trout GI tissues are shown in Table 1 and in Figs 3, 4, 5. Longitudinal intestinal smooth muscle was the most sensitive to H₂S, with significant decreases in all but spontaneous contraction amplitude occurring at 1 mmol l^–1. In other tissues half of the measured parameters were significantly reduced at 10 mmol l^–1 H₂S and all parameters were significantly reduced at 100 mmol l^–1 H₂S. The effects of high concentrations of H₂S (100 mmol l^–1 to 1 mmol l^–1) were often significantly greater in longitudinal smooth muscle than they were in circular smooth muscle. In a number of tissues, but especially noticeable in longitudinal smooth muscle of trout intestine (Fig. 4), application of 1 mmol l^–1 H₂S produced an initial increase in baseline tension and amplitude of spontaneous contractions which was then followed by a decrease in baseline tension and spontaneous contractions. Application of 1 mmol l^–1 H₂S produced one or several spontaneous contractions in anterior intestinal circular smooth muscle but did not increase baseline tension. H₂S also produced a concentration-dependent relaxation of 10 mmol l^–1 carbachol pre-contracted esophagus and anterior intestinal circular smooth muscle (Fig. 5), as well as stomach and gall bladder (not shown). Spontaneous activity and baseline tension were restored when the tissues were rinsed with H₂S-free buffer.

The EC₅₀ values for the H₂S effect on contractile activity (AUC) for both trout and coho tissues are shown in Table 2. Longitudinal intestinal smooth muscle preparations from both fish were significantly more sensitive to H₂S than the respective circular smooth muscle preparations; there were no other significant differences between or within species, tissues or unstimulated vs pre-contracted preparations.

The effects of hypoxia (gassing with 100% N₂) on AUC, spontaneous contraction amplitude, baseline tone and frequency in trout circular and longitudinal intestinal smooth muscle are shown in Table 1 and Fig. 6. The effects of hypoxia were similar, if not identical, to those produced by H₂S. Hypoxia significantly reduced all parameters in circular and longitudinal intestinal smooth muscle. Hypoxia was significantly more efficacious in decreasing AUC in longitudinal smooth muscle than it was in circular smooth muscle and, although not significant, there was a tendency for a greater effect on spontaneous contraction amplitude, baseline tone and frequency as well. In longitudinal smooth muscle of trout intestine, the onset of hypoxia produced an initial increase in baseline tension and spontaneous contraction amplitude followed by a subsequent decrease in these parameters (Fig. 6). These responses were strikingly similar to those produced by 1 mmol l^–1 H₂S (Fig. 4). Spontaneous activity returned and baseline tension partially recovered when the tissues were aerated with room air.

As in trout, H₂S caused a concentration-dependent reduction in AUC, spontaneous contraction amplitude, baseline tone and frequency of spontaneous contractions in coho circular and longitudinal intestinal smooth muscle (Table 3, Fig. 7). Longitudinal intestinal smooth muscle was also more sensitive to H₂S than circular smooth muscle (Table 2). Application of high concentrations (100 mmol l^–1 to 1 mmol l^–1) of H₂S often resulted in a transient burst of contractile activity followed by a stable relaxation and essentially complete loss of spontaneous activity (Fig. 7). Spontaneous activity and baseline tension were restored when the tissues were rinsed with H₂S-free buffer.

Carbachol (10 mmol l^–1) increased AUC, spontaneous contraction amplitude and frequency of spontaneous contractions in longitudinal smooth muscle of the coho intestine (Table 4). Pretreatment with the CSE inhibitor PPG had no effect on any of the variables, whereas pretreatment with the CBS inhibitor AOA significantly reduced AUC. Addition of H₂S (1 mmol l^–1) significantly decreased the carbachol-induced effect on AUC, spontaneous contraction amplitude and frequency of spontaneous contractions (Table 4).

Hypoxia (N₂) significantly reduced AUC, spontaneous contraction amplitude, baseline tone and frequency of spontaneous contractions in otherwise unstimulated coho anterior and posterior, circular and longitudinal smooth muscle (Table 5, Fig. 8). As shown in Fig. 8, the onset of hypoxia often resulted in a transient burst of contractile activity, followed by a prolonged relaxation that persisted until normal aeration was restored. This response was similar to the effects of elevated H₂S (Fig. 7). The effects of hypoxia on AUC and baseline tone were significantly more pronounced in longitudinal than in circular intestinal smooth muscle. There were no significant differences between hypoxia and 1 mmol l^–1 H₂S for any measured parameter in either circular or longitudinal smooth muscle.

Incubation with the CSE inhibitor PPG had no effect on the hypoxic relaxation or inhibition of spontaneous contractions in unstimulated coho longitudinal intestinal smooth muscle, whereas incubation with the CBS inhibitor AOA significantly inhibited the hypoxic effect on AUC, spontaneous contraction amplitude and frequency (Table 5). In carbachol-precontracted longitudinal smooth muscle, PPG significantly inhibited the hypoxic effect on AUC, spontaneous contraction amplitude and baseline tone while AOA inhibited spontaneous contraction amplitude and frequency (Fig. 9).

Prior addition of 0.1 mmol l^–1 cysteine did not affect the hypoxic response of carbachol pre-contracted intestinal strips whereas 1 and 10 mmol l^–1 cysteine augmented the hypoxic effect on AUC, baseline tone and frequency of spontaneous contractions (Fig. 9, Table 5). The effects of 1 and 10 mmol l^–1 of cysteine on the hypoxic response were not significantly different.

Addition of 1 mmol l^–1 cysteine to deoxygenated, homogenized trout intestine slightly stimulated H₂S production (not shown). Subsequent addition of 10 mmol l^–1 cysteine greatly increased the rate of H₂S formation (10.5±6.3 mmol min^–1 g^–1 wet mass, N=5 fish). H₂S production quickly reverted to net H₂S consumption upon addition of room air to the reaction chamber. After the oxygen was removed, presumably by tissue consumption, H₂S production returned (Fig. 10).

Our findings show that exogenous H₂S relaxes intestinal smooth muscle and inhibits spontaneous contractions in multiple areas of the fish GI tract, indicative of effects on either contractile activity of GI smooth muscle or enteric signaling, or both. Longitudinal smooth muscle of intestine from both trout and coho was consistently more sensitive to H₂S than circular smooth muscle, which may indicate differential regulation of function. Further evidence that H₂S is the oxygen-sensitive couple linking tissue hypoxia to physiological responses was provided by observations that: (1) hypoxic responses of the GI tract exactly mimicked those produced by H₂S, (2) hypoxic responses were inhibited by inhibitors of H₂S biosynthesis and were augmented by the sulfide donor cysteine, and (3) H₂S production by the GI tract is stimulated by cysteine and inversely related to oxygen availability.

Our study shows that the effects of H₂S on AUC, spontaneous contractions, baseline tone and rhythmicity in the GI tract of fish are qualitatively, if not quantitatively, identical to those observed in a number of mammalian preparations. Salmonid tissues, however, appear to be more sensitive to H₂S.

Although few studies have provided detailed concentration–response characterization of the effects of H₂S in the mammalian GI tract, most studies suggest that the threshold for a response is 10 mmol l^–1 or above and EC₅₀ values (either provided by the authors or estimated from their data) are usually well above 300 mmol l^–1. In the mouse gastric fundus pre-contracted with prostaglandin F2α (PGF_2α), relaxation was first observed at 10 mmol l^–1 and EC₅₀ values ranged from 350 to 429 mmol l^–1 (Dhaese and Lefebvre, 2009). In the guinea-pig stomach, 10 mmol l^–1 H₂S significantly enhanced resting tension and slightly reduced the contractile amplitude, whereas 300 mmol l^–1 H₂S suppressed spontaneous contractions (Zhao et al., 2009). EC₅₀ values estimated from the inhibitory effect of H₂S appeared to be ∼400 mmol l^–1. Relaxation of the mouse distal colon was observed at 100 mmol l^–1 H₂S, although this was the lowest concentration tested, and the EC₅₀ was 444 mmol l^–1 (Dhaese et al., 2010). In longitudinal muscle from rat ileum, 10 mmol l^–1 H₂S reduced AUC, average amplitude of spontaneous activity and baseline tone by 7%, 3% and 6%, respectively (Nagao et al., 2010). These parameters were reduced 7%, 3% and 5% by 100 mmol l^–1 H₂S (Nagao et al., 2010). As the responses to 10 and 100 mmol l^–1 H₂S were not different, this does not appear to be a concentration-dependent response (each tissue served as its own control; time-matched controls were not employed, making it impossible to determine whether these effects were due to slight deterioration or stabilization of the preparation). However, 500 mmol l^–1 H₂S reduced these parameters by 60%, 20% and 32%, suggesting that the threshold may be above 100 mmol l^–1. Based on the responses at 100, 500 and 1000 mmol l^–1 (Nagao et al., 2010), the EC₅₀ appears to be ∼500 mmol l^–1. Similarly high concentrations (500 mmol l^–1 H₂S) were necessary to affect pacemaker currents in interstitial cells of Cajal from mouse small intestine (Parajuli et al., 2010).

Significant changes in either motility or frequency in the trout and coho GI tract were observed between 1 and 10 mmol l^–1 H₂S and the EC₅₀ values for tissue response was between 78 and 171 mmol l^–1 (overall mean was 120 mmol l^–1). These concentrations are around 5-fold (or more) lower than those commonly necessary to achieve physiological responses in mammalian preparations. This suggests that the GI tract of the fish is exquisitely sensitive to this signaling molecule. However, the reason for this difference is not clear. Tissue (body) temperature could influence H₂S metabolism or kinetics of downstream mediators, although one would expect that the intrinsically lower metabolic rate of exothermic fish would decrease responsivness, not increase it. It is also possible that there is not as much background H₂S from bacterial production of H₂S in the lumen of the fish GI tract as there is in mammals. To our knowledge bacterial H₂S production has not been measured in the fish GI tract.

It is not clear whether the effects of H₂S produced in this study (or others) are ‘physiological’ responses. This is due in large part to uncertainties surrounding: (1) the concentration of endogenous H₂S in and around cells, (2) the actual concentration of exogenous H₂S at the effector site, and (3) whether tissue H₂S concentration can be regulated by physiologically relevant stimuli.

Although there are many reports of plasma and blood H₂S concentration between 30 and 300 mmol l^–1, these may be experimental artifacts (reviewed in Olson, 2009). In fact, H₂S may not even exist in the circulation (Whitfield et al., 2008). Recent measurements have also shown that the concentration of H₂S in a variety of tissues appears to be well below 1 mmol l^–1 (Furne et al., 2008). Furthermore, as H₂S is rapidly oxidized at H₂S concentrations below 20 mmol l^–1, yet higher concentrations of H₂S inhibit mitochondrial cytochrome oxidase, and hence oxidative phosphorylation (Lagoutte et al., 2010), one could argue that H₂S most likely does not exist in appreciable amounts in the GI tract and that it would be physiologically disadvantageous to allow H₂S concentrations to even approach 20 mmol l^–1. Passive loss of H₂S from experimental apparatus due to volatility could be another confounding factor in the discrepancy between endogenous and physiologically effective exogenous H₂S concentrations. We (DeLeon et al., 2011) recently observed that the half-time for H₂S in a smooth muscle myograph of the type used in the present study was under 5 min, and this was due to volatilization not oxidation of H₂S. More accurate assessment of EC₅₀ values awaits continuous monitoring of H₂S concentrations throughout an experiment.

The distal GI tract, however, may be exposed to high H₂S concentrations as a result of bacterial H₂S generation in the lumen. But here it is difficult to envisage how lumenal H₂S production could have a signaling function. It is more likely that bacterially generated H₂S is a potentially toxic metabolite to be disposed of or, perhaps, used as an energy source (Blachier et al., 2010). Collectively, this suggests that if H₂S is a physiologically relevant signal, then it acts in a paracrine or more probably autocrine manner, and it is most likely of tissue, not lumenal, origin. Exact details of H₂S signaling activities await detailed measurement of intracellular H₂S concentrations.

Rapid oxidation of H₂S by the colon has been well documented (see Furne et al., 2001; Lagoutte et al., 2010; Levitt et al., 1999). This supports the argument that not only are tissue H₂S concentrations minimal but also removal of H₂S requires oxygen. The corollary to this is that an increase in H₂S is predicated upon the reciprocal decrease in O₂. Thus, the inverse relationship between H₂S and O₂ may be the physiological couple involved in H₂S signaling.

As described above, we have proposed that O₂-dependent metabolism of H₂S serves as an O₂ ‘sensor’ in a variety of mammalian and non-mammalian tissues (Dombkowski et al., 2006; Olson et al., 2006; Olson et al., 2008a; Olson et al., 2008b; Olson et al., 2010; Whitfield et al., 2008) (reviewed in Olson and Whitfield, 2010). Our hypothesis is based on the following observations: (1) H₂S and hypoxia produce essentially identical responses in a variety of tissues, (2) tissues enzymatically generate H₂S, (3) tissue production of H₂S, or the concentration of exogenous H₂S, is inversely related to oxygen availability, i.e. P_O2, (4) inhibitors of H₂S synthesis inhibit hypoxic responses, and (5) potential sulfide donors augment the hypoxic responses. Recent studies on the mammalian carotid body (Li et al., 2010; Peng et al., 2010; Telezhkin et al., 2009; Telezhkin et al., 2010) and kidney (Beltowski, 2010) have supported this hypothesis and our assertion that this is a phylogenetically ancient and ubiquitous O₂-sensing mechanism. The present study extends these observations to the fish GI tract and supports the hypothesis that O₂-dependent H₂S metabolism is a fundamental oxygen-sensing mechanism.

In the fish GI tract essentially all of the effects of hypoxia are similar, if not identical, to those produced by H₂S. Both stimuli relax visceral smooth muscle, inhibit spontaneous contractions and decrease their frequency. Furthermore, the magnitude of the effects of these two stimuli is also similar and they vary from tissue to tissue. Both hypoxia and H₂S are more efficacious in longitudinal than in circular intestinal smooth muscle and, within a preparation (longitudinal or circular smooth muscle), the magnitude of inhibition produced by these two stimuli is virtually identical. Even more striking is the transient stimulatory effect observed in both trout and coho intestine at the onset of hypoxia or H₂S application. This is reminiscent of the signature biphasic hypoxic constriction of rat pulmonary arteries that is uniquely mimicked by exogenous H₂S (Olson et al., 2006).

Fig. 10 clearly shows that the trout intestine generates H₂S in the presence of cysteine. This is consistent with the demonstrated production of H₂S by CBS and CSE in the mouse colon (Linden et al., 2008). By simultaneously measuring P_O2, we were also able to determine that H₂S was only generated in the absence of O₂ or, at the very least, at very low P_O2 values. Furthermore, the reversal from net H₂S production by intestinal tissue to net H₂S consumption when oxygen was injected into the intestinal homogenate supports our previous observations in other tissues that H₂S production and/or tissue concentrations are inversely related to oxygen availability.

Given that the effects of hypoxia and H₂S on the trout and coho GI tract were virtually identical, and that the GI tissue produces H₂S when hypoxic, one would expect that inhibition of H₂S biosynthesis would inhibit the hypoxic response and that an increase in H₂S biosynthesis by addition of the sulfide donor cysteine would enhance the hypoxic response. This is precisely what we observed (Fig. 9). Interestingly, however, the hypoxic response was equally inhibited by inhibition of CSE with PPG and of CBS with AOA. The effects of PPG on the hypoxic response appeared to be direct, whereas AOA may have an indirect effect on the hypoxic response as it also inhibited the carbachol contraction independent of hypoxia. It remains to be determined whether H₂S generated by CBS has other effects on the intestine that are independent of the hypoxic response.

The present results show that H₂S inhibits GI activity and provide considerable evidence that this may be the mechanism through which the hypoxic response is expressed. This raises the question, what is the physiological significance of either H₂S or hypoxia in GI function? As both of these stimuli would presumably decrease the GI motility and transit time through the gut, they may serve to also decrease oxygen demand by GI tissues in the face of an oxygen deficit or perhaps permit more thorough processing of lumenal contents. Alternatively, these responses may be an evolutionary relic. As H₂S and hypoxia have similar effects on vascular smooth muscle in a broad range of vessels, and on smooth muscle in the urinary bladder, and as shown here in a variety of GI tissues, it may be that this response was at one time employed in a different signaling capacity during the early evolution of eukaryotes. However, as tissues became more specialized it was no longer linked to a specific physiological function. Clearly, this remains to be investigated.

The authors wish to thank D. Meunick, B. Bell and the staff at the Richard Clay Bodine State Fish Hatchery, Indiana Department of Natural Resources, for their help in obtaining coho tissues.