Branched Short‑Chain Fatty Acid Isovaleric Acid Causes Colonic Smooth Muscle Relaxation via cAMP/PKA Pathway
Bryan A. Blakeney1 · Molly S. Crowe1 · Sunila Mahavadi1 · Karnam S. Murthy1 · John R. Grider1
Abstract
Background Isovaleric acid (IVA) is a 5-carbon branched-chain fatty acid present in fermented foods and produced in the colon by bacterial fermentation of leucine. We previously reported that the shorter, straight-chain fatty acids acetate, pro- pionate and butyrate differentially affect colonic motility; however, the effect of branched-chain fatty acids on gut smooth muscle and motility is unknown.
Aims To determine the effect of IVA on contractility of colonic smooth muscle.
Methods Murine colonic segments were placed in a longitudinal orientation in organ baths in Krebs buffer and fastened to force transducers. Segments were contracted with acetylcholine (ACh), and the effects of IVA on ACh-induced contraction were measured in the absence and presence of tetrodotoxin (TTx) or inhibitors of nitric oxide synthase [L-N-nitroarginine (L-NNA)] or adenylate cyclase (SQ22536). The effect of IVA on ACh-induced contraction was also measured in isolated muscle cells in the presence or absence of SQ22536 or protein kinase A (PKA) inhibitor (H-89). Direct activation of PKA was measured in isolated muscle cells.
Results In colonic segments, ACh-induced contraction was inhibited by IVA in a concentration-dependent fashion; the IVA
response was not affected by TTx or L-NNA but inhibited by SQ22536. Similarly, in isolated colonic muscle cells, ACh- induced contraction was inhibited by IVA in a concentration-dependent fashion and the effect blocked by SQ22536 and H-89. IVA also increased PKA activity in isolated smooth muscle cells.
Conclusions The branched-chain fatty acid IVA acts directly on colonic smooth muscle and causes muscle relaxation via
the PKA pathway.
Introduction
Fatty acids are carboxylic acids with an alkyl tail of vary- ing carbon length. Short-chain fatty acids (SCFA) have fatty acids with lengths of 2 to 6 carbons, medium-chain fatty acids (MCFA) have lengths of 7 to 12 carbons, and long- chain fatty acids (LCFA) have lengths of 13 to 21 carbons. Branched-chain fatty acids (BCFA) possess a fork in the carbon chain of at least a methyl group or longer.The primary source of dietary SCFA is fermentation by the resident bacteria of the colon of hydrolysis-resistant starches which pass unabsorbed through the small intes- tine [1, 2]. Short-chain fatty acids are present in significant concentrations in feces with actual concentrations ranging widely between individuals based on diet, length along the colon, pH, and microbiota status [1–6]. In general, totalconcentrations of fatty acids in feces are in the mM range, whereas concentrations in the blood are in the high µM to low mM level. The main SCFA in both feces and blood are the straight-chain fatty acids acetate, butyrate, and propion- ate. Recent interest has been directed at these SCFA because of their potential role in gut function, glucose homeosta- sis, appetite, regulation of metabolism, inflammation and immune competence, as well as tumorigenesis and colon cancers.
These have been the topic of many recent reviews [3, 5, 7–12].In contrast to SCFA, BCFA are produced when pro- teins pass through the small intestine unabsorbed and the branched amino acids valine, isoleucine, and leucine are, respectively, fermented to isobutyric acid, 2-methylbutyric acid, and isovaleric acid (IVA) [5, 6, 13–16]. In addition, BCFA are derived from ingested foods, especially dairy and beef products [17, 18]. While SCFA have been studied in detail and considered beneficial, BCFA and the products of protein fermentation are not well characterized and are generally considered deleterious to the gut. Even though pro- duced in the colon or via ingested foods, IVA is one of the most prominent fatty acids in human blood. Although much less is known about blood levels of BCFA and because con- centrations vary based on the diet, methods, time period, and sampling in portal, hepatic, and peripheral blood, significant concentrations have been measured. In fasting patients in Denmark, Jakobsdottir et al. [19] found that IVA concentra- tion in venous blood from control subjects ranked second at 40 μM, following acetic acid (245 μM), and at higher levels than propionic acid and the other prevalent BCFA isobutyric acid (13 μM). Blood levels of IVA can rise dramatically in pathological conditions, such as isovaleric acidemia where a genetic defect in the gene for isovaleryl-CoA dehydrogenase produces either a reduced efficiency or ineffective enzyme leading to a buildup of IVA [20]. In one reported case study from Japan, stable patients had IVA levels of up to 78 μM as compared to 6 μM in controls and up to 7960 μM during crisis [21].
In cases of bacterial overgrowth in small intes- tine, total SCFA concentrations can range from 210 μM to 12 mM and the levels of IVA can range from 2 µM to nearly 700 μM. Consistent with their production in the colon by fermentation, IVA is present in feces in significant amounts. Studies in human feces indicate that while the generation of SCFA predominates in the proximal colon, the contributions of BCFA increase from 17% of the total in proximal colon to 38% in distal colon [3, 5]. IVA alone represents approxi- mately 3.5–5% of SCFA in feces [22–24]. Specifically, esti- mates range from 0.8 to 22.0 mmol/kg feces (wet weight) in different studies and depending on diet [6, 25–28]. The significant concentrations of IVA in colon and blood raise the possibility of its intraluminal and/or extraluminal effects. The effect of SCFA on mucosal health and function has been studied more extensively than the effects on motilityalthough several studies suggest that SCFA increase colonic motility and secretion via stimulation of neural- and/or sero- tonin-mediated pathways [29–35]. Direct contractile effects of SCFA on rat colonic smooth muscle have been reported; however, little is known of the effects of BCFA on colonic smooth muscle [36]. The present study examined the effects of IVA, the most prevalent BCFA, on colonic smooth muscle strips and isolated cells. The results show that IVA causes a concentration-dependent inhibition of ACh-evoked contrac- tion (i.e., relaxation) of smooth muscle from the longitudinal muscle layer, which is mediated by activation of PKA.
Methods
Materials
Chemicals not specified were acquired from Sigma- Aldrich (St. Louis, MO). Tetrodotoxin, NG-nitro-L-arginine (L-NNA), SQ22536, and H-89 were obtained from Toc- ris Bio-Techne (Minneapolis, MN). Isovaleric acid was obtained from Sigma-Aldrich and dissolved in Krebs buffer (pH 7.2). Eagle’s essential amino acid mixture was obtained from Thermo Fisher, (Waltham, MA). Adenosine 5′-triphos- phate [γ-32P] 6000 Ci/mmol was acquired from PerkinElmer (Boston, MA). Kemptide was acquired from Cayman Chem- ical (Ann Arbor, MI).
Animals
Mus musculus C57BL/6J were purchased from Jackson Lab- oratories (Bar Harbor, ME) and housed in the AAALAC- accredited Division of Animal Resources Facility, Virginia Commonwealth University (VCU). Both male and female mice of 6-8 weeks age were used in this study. As no statis- tical significance in responses was detected between colons from the sexes, the data were combined. Mice were eutha- nized by CO2 asphyxiation under protocols approved by VCU Institutional Animal Care and Use Committee, and all procedures were in their accordance.
Measurement of Contraction and Relaxation in Colonic Segments
The colon was removed, the mesentery trimmed, and the lumen flushed with Krebs buffer warmed to 37 °C. Krebs buffer was prepared as follows: 118 mM NaCl, 4.75 mM KCl, 1.19 mM KH2PO4, 1.2 mM MgSO4, 2.54 mM CaCl2 and 25 mM NaHCO3, 11 mM glucose. Krebs buffer was bubbled with 95% O2 and 5% CO2 and adjusted to pH 7.4 before use as a control solution or as thebathing solution. Segments of approximately 4 mm in length were cut from the colon and tied at both ends with surgical silk suture without occluding the lumen. One end of suture was attached to a glass hook which was submerged in Krebs buffer in a 5-ml water-jacketed organ bath maintained at 37 °C (Rad- noti, Monrovia, CA). The silk suture on the opposite end of the colonic segment was attached to a model FT03C Force Transducer (Grass Technologies, Quincy, MA) which was connected to a Powerlabs 8/35 with Octal Bridge Amp (ADInstruments, Colorado Springs, CO). Continuous data were collected and analyzed with LabChart 8 (ADInstru- ments) software. The colon tension was adjusted to an initial tension of 1.0 g and the segment equilibrated for 60 min. The Krebs buffer was replaced at 15-min intervals three times before the initial experiment. Isovaleric acid (IVA) was added in single concentrations, and the bath was rinsed with at least three bath volumes of Krebs buffer at 15-min intervals between each addition of IVA.
Following equilibration, the viability of the strips was tested by contraction with acetylcholine (ACh) (10 µM). The effects of IVA at different concentrations were tested by pre- contracting the strip with ACh. When the contraction to ACh was stable (about 10 min after addition), IVA was added and the response measured. Vehicle controls for Krebs buffer, citrate buffer, DMSO and water had no significant effect. In some segments, the effects of 50 mM IVA were com- pared to equal concentrations of acetic acid and propionic acid measured in the same colonic preparation. To test the site and mechanism of action of IVA, some colon segments were pretreated with the neural toxin tetrodotoxin (TTx; 10 µM in citrate buffer), the nitric oxide synthase inhibitor L-NNA (100 µM in water), or the adenylate cyclase inhibitor SQ22536 (10 and 500 µM in DMSO). All inhibitors were added before ACh and were in contact with the segment during measurement of the effect of IVA. The response to IVA was calculated as the decrease in tension from the stable ACh-induced contraction. The mean tension for a 60 s period in the presence of ACh alone imme- diately prior to addition of IVA was compared to the mean tension for a 30 s period measured at peak decrease in tone after addition of IVA. The decrease in tone was measured in grams, and the percent relaxation induced by IVA was calculated as the percent decrease from the maximum ACh level of tone.
Measurement of Contraction and Relaxation in Dispersed Colonic Muscle Cells
Smooth muscle cells were isolated from colon after dissec- tion by gently scraping off the mucosa and placing strips of colon in HEPES buffer medium containing 25 mM HEPES, 120 mM NaCl, 4 mM KCl, 2.6 mM KH2PO4, 0.6 mM MgCl2, 14 mM glucose, and 2.1% Eagle’s essential amino acid mixture [37, 38]. After washing the strips with medium, tissue was incubated for 30 min at 31 °C in 10 mL of buffer with 0.1% collagenase II and 0.1% soybean trypsin inhibitor. The tissue was washed with collagenase-free medium and resuspended in collagenase-free medium for an additional 30 min to allow spontaneous dispersion of cells. Dispersed muscle cells were filtered through 500 μm Nitex mesh. A cell suspension containing 104 muscle cells/mL was treated with 1 μM ACh for 1 min to induce contraction. Separate suspensions of cells were incubated with IVA (1–100 µM) for 1 and 5 min before addition of ACh, and then cell lengths were measured. The percent decrease in cell length in response to ACh was calculated in the presence or absence of IVA (1–100 µM). In some experiments, cells were incubated for 10 min with SQ22536 (1 and 10 µM) or cAMP-dependent protein kinase (PKA) inhibitor H-89 (10 µM) prior to the addition of ACh and IVA. After treat- ments, cells were fixed with 1% acrolein and cell lengths were measured by scanning micrometry (Lasico, Los Ange- les, CA). A decrease in cell length in response to ACh was considered as contraction, and inhibition of contraction by IVA was considered as relaxation. A minimum of 50 cells were counted in each treatment group, and separate animals were used in each experiment.
Measurement of PKA Activation in Dispersed Colonic Muscle Cells
Colonic smooth muscle cells were isolated as previously described [37, 38]. The cell suspensions were then treated with 1 μM vasoactive intestinal peptide (VIP) or 10 or 100 μM IVA for 5 min. VIP was used as a control peptide well known to activate the cAMP/PKA signaling pathway in gut smooth muscle [37, 38]. The reaction was stopped by quick centrifugation, and the pellet was suspended in a lysis buffer consisting of 1% NP-40, 0.5% deoxycholic acid, 0.1% SDS, 10 mM sodium pyrophosphate, 150 mM NaCl, 50 mM Tris hydrochloride pH 7.4, and 2% 50X BD BaculoGold pro- tease inhibitor cocktail (Sigma, St Louis, MO). The lysate was then centrifuged at 10,600 g and the supernatant saved and frozen at − 80 °C overnight. To measure PKA activity, 40 μL of cell lysate was mixed with 40 μL of PKA reaction buffer including 50 mM Tris hydrochloride pH 7.4, 12.5 mM MgCl2, 25 mM KCl, 25 μM DTT, 150 μM kemptide, and 250 μM ATP and [32P]ATP (6000 Ci/mmol). Reactions were performed in the presence (total activity) or absence (stimulated activity) of 5 µM cAMP. Non-specific activity was measured in the presence of 10 µM PKA inhibitor (myristoylated PKI) and was sub- tracted from total and stimulated activities. The samples were incubated at 37 °C for 15 min, and then 45 μL of sam- ple absorbed onto Whatman Grade P81 filter paper. The filter paper was rinsed five times at 20-min intervals with 75 mM H3PO4 and dried overnight. The samples were then placed in a scintillation vial in 5 mL of ScintiSafe (Fisher Scientific, Waltham, MA) and γ-32P measured with a PerkinElmer Tri- Carb 2810 TR Liquid Scintillation Analyzer. PKA activity was calculated and represented as a ratio of counts per min- ute (CPM) in the absence or presence of cAMP [38].
Immunofluorescence Staining and Imaging
Colonic segments were fixed in 4% paraformaldehyde in PBS for 2 h. For antigen retrieval, they were submerged in a solution of 10 mM sodium citrate and 0.05% Tween 20 at pH 6.0 overnight at 4 °C, boiled for 3 min with gentle stirring, followed by submersion in 30% sucrose in PBS overnight at 4 °C. Tissue was embedded in OCT compound (Tissue-Tek, VWR, Radnor, PA) and frozen at − 80 °C, and 10 μm sec- tions were cut with a cryo-microtome. Sections were dried for 1.5 h at room temperature and kept at 4 °C. Sections were washed four times with PBS-Tween 0.5%, twice with PBS, and then incubated 30 min with 5% normal goat serum (Jackson ImmunoResearch, West Grove, PA) in PBS. Sec- tions were then incubated overnight with 1:200 OR51E1 antibody (Thermo PA5-35298) in 5% normal goat serum or in the absence of primary antibody as a control at 4 °C. Slides were washed again four times with PBS-Tween 0.5%, twice with PBS, and then incubated 1 h at room temperature with goat anti-rabbit Alexafluor 594 1:100 in 2.5% normal goat serum in PBS, washed four times with PBS-Tween 0.5% and twice with PBS before covering with Fluoroshield DAPI (Abcam, Cambridge, MA). Slides were imaged with a Zeiss Imager Z1 controlled with ZEN software and using an EXFO X-Cite Series 120 fluorescence illumination source. Slides were imaged with 400 ms exposure for rhodamine filter, 200 ms for DAPI fil- ter. Channels were combined using ImageJ164 (NIH) with FIJI [39], and contrast and brightness kept constant between images.
Statistical Analysis
In colon segment experiments, IVA-induced relaxations were calculated as decrease in tone from maximum ACh- induced increases in tone. Effects of inhibitors on IVA- induced relaxation were compared to responses in the absence of inhibitors. In isolated muscle cell experiments, the decrease in cell length from control response to ACh was compared to the response in the presence of ACh plus IVA alone (representing relaxation) and in the presence of inhibitors. Values are reported as mean ± standard error (segments) or ± σ to indicate standard deviation (isolated cells). Statistically significant difference was determined by one-way ANOVA, N-way ANOVA, or repeated measures ANOVA as applicable followed by Bonferroni test post hoc when multiple groups were compared, paired t test when only two treatments within same segment were compared, and unpaired t test when two different groups of segments were compared. Results were considered significant at P < 0.05. Statistical tests were carried out using MATLAB R2018a with the Statistics and Machine Learning Toolbox version 11.3.
Results
Effect of IVA on ACh‑Induced Contraction in Colonic Segments
As noted earlier, no statistically significant differences in response to ACh or IVA were noted between colons from male and female mice so data from mice of both sexes were combined. Similarly, when responses from proximal, middle, and distal section of the colon were compared, no statisti- cally significant differences between responses of the regions were noted. Thus, data from colonic segments of males and females and from all regions were combined and hereafter referred to as colonic segments or colonic responses. In colonic segments, ACh (10 µM) induced contrac- tion, which peaked rapidly but then maintained a stable plateau phase for a prolonged period lasting over 10 min. The response to IVA was tested by addition of single con- centrations of IVA at 10 min after addition of ACh. The mean increase in tension in response to ACh at this point was 1.02 ± 0.05 g above basal tone. Addition of IVA at the sustained plateau contraction consistently caused relaxa- tion (i.e., inhibited ACh-induced contraction). The relaxa- tion phase began almost immediately after the addition of IVA, and peak relaxation usually occurred within a minute (Fig. 1). The decrease in tension partially recovered over a 20-min period and was reversed to basal levels by repeated washing with fresh Krebs buffer. The relaxation in response to IVA was concentration-dependent over the range tested (Fig. 2).
IVA caused a 22 ± 1% relaxation at 10 mM, a 40 ± 4% relaxation at Fig. 1 Effect of IVA on acetylcholine-induced contraction. Representative tracings illustrating contraction in response to acetylcholine (10 µM) and the inhibition of contraction upon application of IVA (50 mM) Fig. 2 Concentration-dependent relaxation by IVA in colonic seg- ments. Colonic segments were equilibrated in an organ bath. The seg- ments were allowed to equilibrate at 1 g tension for 1 h before the addition of acetylcholine (10 µM). During the sustained phase of ACh-induced contraction, segments were treated with different con- centrations of IVA (10–50 mM). IVA inhibited ACh-induced contrac- tion in a concentration-dependent fashion. Results are expressed as percent of inhibition of ACh-induced contraction by IVA. Values are means ± SEM of 35–50 experiments 20 mM, a 49 ± 4% relaxation at 30 mM, and a relaxation of 52 ± 3% at 50 mM.Straight‑Chain Fatty Acids Compared to IVAon ACh‑Induced Contraction in Colonic SegmentsTo determine whether the relaxation mediated by the branched-chain fatty acid IVA differs from that of the shorter unbranched fatty acids common in stool and blood, we tested 50 mM propionic or acetic acid on colon seg- ments and compared them to the effect of 50 mM IVA in the same colonic segment (Fig. 3). IVA-induced relaxation of ACh-precontracted segments was 68 ± 6%, and the same segments relaxed 50 ± 8% and 51 ± 8% in response to ace- tic acid and propionic acid, respectively. Thus, IVA caused a greater relaxation than either of the straight-chain fatty acids (n = 6–13; P < 0.05 for each straight-chain FA vs IVA).
This trend is the same when comparing decrease in tone measured in grams force. For ACh-contracted segments, the relaxation was 0.9 ± 0.1 g for IVA treatment, 0.5 ± 0.1 g for acetic acid treatment, and 0.6 ± 0.1 g for propionic acid (P < 0.05 for each versus IVA).Effect of TTx, L‑NNA and SQ22536 on IVA‑Induced Relaxation in Colonic SegmentsTo determine the effect of neuronal activity as potentially mediating the effects of IVA, colon segments were pre- treated with the neuronal conduction blocker 10 μM TTx forFig. 3 Comparison of relaxation of colon segments induced by50 mM IVA, acetic acid, and propionic acid. Colonic segments were equilibrated in an organ bath. The segments were allowed to equilibrate at 1 g tension for 1 h before the addition of acetylcholine (10 µM). Initial IVA-induced relaxation of ACh-contracted segments was 68 ± 6% (n = 13). After washout of IVA, ACh-contracted colon segments relaxed 50 ± 8% or 51 ± 8% in response to acetic acid (n = 7) or propionic acid (n = 6), respectively. Values are means ± SEM;*P < 0.0530 min prior to the addition of 10 µM ACh and 50 mM IVA. Neither TTx nor its citric acid vehicle control had a signifi- cant effect on the IVA-induced relaxation. In the same strips, the relaxation response to IVA was 41 ± 6% in the absence of TTx and 57 ± 7% in the presence of TTx (P > 0.05; n = 5). The decrease in tone in response to IVA expressed as grams force was also similar in the presence and absence of TTx (0.37 ± 0.06 g relaxation and 0.34 ± 0.07 g relaxation in the absence and presence of TTx, respectively, P > 0.05).
This suggests that the effect of IVA is direct on smooth muscle rather than mediated through neuronal mechanisms.To determine the involvement of nitric oxide on the relax- ant effects of IVA, colon segments were pretreated with the NOS inhibitor 100 μM L-NNA for at least 25 min prior to the addition of IVA and ACh. L-NNA did not signifi- cantly affect relaxation in response to IVA. In the absence to L-NNA, IVA caused a 53 ± 5% relaxation of ACh-induced contraction and a 70 ± 9% relaxation of ACh-induced con- traction in the presence of L-NNA (P > 0.05; n = 6). The decrease in tone in response to IVA expressed as grams force was also similar in the presence and absence of L-NNA (0.41 ± 0.07 g relaxation and 0.48 ± 0.13 g relaxation in the presence and absence of L-NNA, respectively, P > 0.05). Consistent with the effect of TTx, the lack of a statistically significant effect of L-NNA suggests that NO did not medi- ate the inhibitory effect of IVA on colonic muscle.To examine the involvement of GαS-coupled pathways in IVA-mediated relaxation, colonic segments were pre- treated with a selective adenylyl cyclase inhibitor SQ22536(500 μM). Treatment with SQ22536 caused an inhibition inIVA-induced relaxation from that obtained in the absence of SQ22536; however, the difference in relaxation was not statistically significant (48 ± 4% relaxation of ACh-induced contraction in the absence of SQ22536 vs 37 ± 3% relaxation in the presence of SQ22536 n = 4, P > 0.05). In these strips, the vehicle, DMSO, caused a slight increase in the relaxation in response to IVA. When this is taken into account, the inhi- bition by SQ22536 was significant (53 ± 3% in the presence of the DMSO vehicle control vs 37 ± 3% in the presence of SQ22536, n = 4; P < 0.05). This suggested that the response to IVA might be mediated by activation of adenylyl cyclase and activation of the cAMP/PKA signaling pathway. This notion was tested further in isolated smooth muscle cells.Effect of IVA on ACh‑Induced Contraction in Colonic Smooth Muscle CellsThe lack of effect of L-NNA or TTx on IVA-induced relaxa- tion in colonic segments suggests a direct effect of IVA on smooth muscle cells.
To test this notion, the effect of IVA on ACh-induced contraction was examined in isolated colonic muscle cells. Muscle cells were treated with different con- centrations of IVA (1 µM to 100 µM) for 1 or 5 min before the addition for ACh (1 µM). Addition of ACh caused a significant 37% ± σ 6% decrease in cell length (control cell length 61 ± σ 4 µm, in the presence of ACh 38 ± σ 4 µm). Pretreatment of muscle cells with IVA for either 1 or 5 min inhibited ACh-induced contraction in a concentration- dependent fashion (Fig. 4). Inhibition of contraction by IVAwas referred to as relaxation and represented as percent inhi- bition of ACh contraction. With 1 min of pretreatment with IVA, ACh-induced decrease in cell length was 30 ± σ 8% (or 25% relaxation) at 1 µm IVA, 20 ± σ 6% (or 50% relaxation)at 10 µM IVA, and 10 ± σ 6% (or 75% relaxation) at 100 µM IVA (P < 0.01). The relaxation of isolated muscle cells after pretreatment with IVA for 5 min before addition of ACh was similar to the results obtained with 1 min pretreatment (data not shown).Effect of SQ22536 and H‑89 on IVA‑Induced Relaxation in Colonic Smooth Muscle CellsStudies in colonic segments suggested that the response to IVA was mediated by activation of adenylyl cyclase and production of cAMP/PKA. To further examine the involve- ment of GαS-coupled cAMP/PKA pathway in IVA-medi- ated relaxation, isolated muscle cells were incubated with SQ22536 (1 and 10 µM) or PKA inhibitor H-89 (10 µM) prior to the addition of IVA (10 and 100 µM) and ACh (1 µM) (Fig. 5). The decrease in cell length in response to ACh and ACh plus IVA was measured and the percent relax- ation calculated. Experiments were repeated in the presence and absence of SQ22536 and H-89.
SQ22536 inhibited relaxation in response to both 10 and 100 µM IVA. Relaxa- tion was 45 ± σ 9% and 71 ± σ 6% at 10 and 100 µM IVA, respectively, whereas in the presence of SQ22536 (10 µM) relaxation was 26 ± σ 9% and 3 ± σ 6% at 10 and 100 µM IVA, respectively (Fig. 5). The inhibition of relaxation byIVA Concentration (µM)Fig. 4 Concentration-dependent relaxation by IVA in isolated colonic muscle cells. Relaxation of muscle cells was measured as decrease in maximal cell contraction induced by ACh. Muscle cells were treated with ACh (1 µM) in the absence or presence of pretreatment with dif- ferent concentrations (1–100 µM) of IVA. IVA inhibited (i.e., caused relaxation) ACh-induced contraction in a concentration-dependent fashion. Relaxation was expressed as percent inhibition of contrac- tion. Values are means ± SDFig. 5 Effect of SQ22536 and H-89 on IVA-induced relaxation in colonic muscle cells. Relaxation of muscle cells was measured as decrease in maximal cell contraction induced by ACh. Muscle cells were treated with ACh (1 µM) alone or in the presence of IVA (10 and 100 µM). In some experiments, cells were incubated with SQ22536 (10 µM) or H-89 (10 µM) prior to the addition of IVA. IVA-inhibited contraction (i.e., caused relaxation) in response to ACh. Relaxation was expressed as percent inhibition of contraction. IVA-induced relaxation was suppressed by both SQ22536 and H-89. Values are means ± SD of four experiments. **P < 0.01SQ22536 was dose-dependent as inhibition of relaxation was decreased with 1 µM SQ22536. The 100 µM IVA relaxation was 71 ± σ 6% in the absence of SQ22536, and in the pres- ence of 1 µM SQ22536 was 23 ± σ 7% and 3 ± σ 6% in the presence of 10 µM SQ22536 (P < 0.01). H-89 also inhibited relaxation in response to both 10 and 100 µM IVA. In the presence of H-89 (10 µM), relaxation was 3 ± σ 11% and 13 ± σ 7% at 10 and 100 µM of IVA, respectively (P < 0.01). Consistent with data obtained in the colonic segments, the results in isolated muscle cells suggest that IVA-induced relaxation was mediated by Gαs/cAMP/PKA pathway.
Effect of IVA on PKA Activation
The effect of the PKA inhibitor H-89, and the adenylate cyclase inhibitor SQ22536, on cell length implies an involvement of the cAMP/PKA-mediated pathway in medi- ating relaxation by IVA. This notion was examined by direct measurements of PKA activity in response to IVA and vas- oactive intestinal peptide (VIP), a ligand for Gαs-coupled VPAC2 receptors in gastrointestinal smooth muscle [41]. Both 10 and 100 µM IVA caused a significant increase in PKA activity in colonic smooth muscle cells. Cells treated with 10 μM showed 6.5 ± σ 1.0 more PKA activity and 100 μM IVA showed 5.0 ± σ 0.5 times more PKA activity than the untreated cells (1.0 ± σ 0.01), compared to 1 μM VIP which showed 7.3 ± σ 0.4 times more PKA activity than the untreated cells (Fig. 6). These results suggest that activation of PKA by IVA is involved in mediating relaxa- tion response and is consistent with the inhibition of IVA- induced relaxation by the PKA inhibitor H-89. Fig. 6 IVA-induced PKA activation in colonic muscle cells. PKA activity in smooth muscle cells was measured by an in vitro kinase assay by the phosphorylation of PKA substrate kemptide using [32P] ATP. PKA activity was measured as described in the “Methods” sec- tion and represented as relative PKA activity [ratio of counts per minute in the absence of added cAMP (stimulated activity) or in the presence of added cAMP (total activity)]. Values are means ± SD;
*P < 0.05
IVA‑Sensitive Receptor Olfr558 (OR51E1 Homolog) on Colon Smooth Muscle
Relaxation of smooth muscle with IVA via a cAMP/PKA pathway would indicate the presence of a G protein-coupled receptor sensitive to IVA which activates adenylate cyclase in a Gαs-like manner. Murine receptor Olfr558, homolo- gous to human receptor OR51E1, is a Gαolf-linked receptor, similar to Gαs in that it activates adenylate cyclase (specifi- cally, ACIII). IVA is an activating ligand for OR51E1 while shorter unbranched fatty acids such as propionic and acetic are not [40, 41]. Immunofluorescent staining of mouse colon for Olfr558 shows staining for the receptor on the longitudi- nal smooth muscle on the outer layer of the colon, as well as some staining in the circular layer of smooth muscle (Fig. 7). Control stains without the primary antibody did not show specific staining of the secondary antibody in the smooth muscle layers (Fig. 7).
Discussion
In the present study, we have demonstrated that the branched-chain fatty acid isovaleric acid causes relaxation of colonic smooth muscle. We have also identified that relaxa- tion was mediated via cAMP/PKA pathway raising the pos- sibility that IVA interacts with receptors coupled to Gαs-like protein. The evidence for the notion that IVA acts directly on smooth muscle cells was based on the following results. (1) IVA causes relaxation in a concentration-dependent fashion by inhibiting ACh-induced muscle contraction in colonic segments; (2) IVA-induced relaxation in colonic segments was not affected by L-NNA that blocks neuronal nitric oxide synthase activity and generation of nitric oxide, a key inhibi- tory neurotransmitter in the colon; and (3) IVA-induced relaxation in colonic segments was also not affected by TTx that blocks neural activation to release neurotransmitters. A direct effect of IVA on smooth muscle was confirmed by demonstrating that IVA causes relaxation in a concentration- dependent fashion by inhibiting ACh-induced decrease in length in isolated muscle cells. This preparation is devoid of enteroendocrine cells, neurons, and interstitial cell of Cajal (ICC) allowing the examination of the direct effect on smooth muscle cells and avoiding the confounding effects of other cell types [37]. Activation of PKA and/or protein kinase G (PKG) plays a key role in smooth muscle relaxa- tion. Lack of effect of L-NNA suggests a PKG-independent pathway probably involving PKA. The involvement of PKA pathway in relaxation was tested by H-89, which blocks pro- tein kinase (PKA) by binding to the ATP-binding site in the PKA catalytic site, and by direct measurement of PKA activ- ity in response to IVA in smooth muscle cells. IVA caused a significant increase in PKA activity and the stimulatory Fig. 7 Fluorescent staining of mouse colon for Olfr558 (OR51E1). Cross sections (10 μm thick) of mouse colon demonstrate staining for Olfr558 (Red) in the longitudinal layers of smooth muscle, with some noticeable staining in the circular muscle and epithelium (a).
Sections without primary antibody (b) did not show any noticeable fluorescence for the secondary antibody. Blue indicates nuclear DAPI staining. Magnification is ×400 effect is similar to that of VIP, a known activator of Gαs- coupled VPAC2 receptors in gastrointestinal smooth muscle [38]. In addition, IVA-induced relaxation was blocked by H-89 suggesting activation of PKA by IVA causes muscle relaxation. The similarity of action to VIP which is known to induce relaxation via the Gαs/cAMP/PKA pathway and the ability of the adenylate cyclase inhibitor SQ22536 to strongly inhibit IVA-induced relaxation support the notion that the Gαs/cAMP/PKA mediates the response to IVA in the present study. However, it should be noted that other non- canonical mechanism may also activate PKA and cannot be ruled out as alternative or additive pathways activated by IVA in colonic smooth muscle [42–47].
An essential step in smooth muscle contraction in response to contractile agonists such as ACh is phospho- rylation of the 20-kDa regulatory light chain of myosin II (MLC20) by Ca2+/calmodulin-dependent myosin light-chain kinase (MLCK) [37]. The increase in Ca2+ required for acti- vation of MLCK is derived either by release of Ca2+ from the sarcoplasmic reticulum or influx of Ca2+ from the extra- cellular medium via Ca2+ channels. MLC20 phosphoryla- tion is sustained even after the termination of initial raise in Ca2+ by inhibition of MLC phosphatase (MLCP) which normally terminates contractions by dephosphorylation of MLC20. Each of these steps plays a role in initiating or sus- taining contractions (reviewed in [37]).
The activation of PKA produced by IVA as evidenced by the effects of H-89 on relaxation and measured directly by PKA substrate phos- phorylation in the present study could cause relaxation of ACh-induced contraction by several mechanisms at the level of the smooth muscle cell. PKA could reduce the increase in intracellular Ca2+ levels by (a) inhibiting IP3 formation and/ or IP3-dependent Ca2+ release, (b) stimulating Ca2+ uptake into the stores and/or stimulating Ca2+ efflux, and (c) inhib- iting Ca2+ influx. In addition, PKA could inhibit muscle contraction by stimulating MLCP activity, thereby reducing phosphorylation of MLC20. Although the exact mechanism by which activation of PKA pathway in response to IVA mediates muscle relaxation has not been determined in the present study, mechanisms have been demonstrated for other relaxant agents such as VIP, which activates Gαs-coupled cAMP/PKA pathway (reviewed in [37]) [48, 49]. The relaxation of colonic smooth muscle by IVA could be mediated additionally through activation of K+ channels. Previous studies have shown the BCFA valproic acid acti- vated TREK-1, a TWIK-related K+ channel. Activation of K+ channels causes hyperpolarization leading to inhibition of Ca2+ influx. Expression of TREK-1 in the smooth muscle and inhibition of contraction by valproic in rat and guinea pig stomach and tenia coli have been reported [50–52]. IVA has also been shown to activate KCNQ2-4 K+ channels, which are present on murine gastrointestinal smooth muscle [53–56]. The potential role of these K+ channels in addi- tionally mediating the relaxation effects of IVA in smooth muscle of the mouse colon remains to be determined.
Recently, several members of a different class of recep- tors, the olfactory receptors (Olfr), have been suggested as a possible mediator of SCFA and BCFA responses in a variety of tissues [40, 41, 57–59]. The Olfr78 has been demonstrated in renal juxtaglomerular apparatus and smooth muscle cells of the small resistance vessels of the kidney where they have been postulated to regulate renin secretion and vasodilata- tion in response to SCFA produced by colonic microbiota [59]. The olfactory receptor Olfr558 has been shown to be sensitive to a number of medium- and short-chain fatty acids. Recent studies demonstrated the presence of Olfr558 in enterochromaffin cells where it induces serotonin release [40]. Notably, Olfr558 was shown to be most responsive to IVA (EC50: 8.9 µM) and to a much lesser extent isobu- tyrate and not responsive to the SCFAs acetate, butyrate, or propionate. The effect is mediated by activation of Gαolf and activation of adenylate cyclase III, activating the cAMP/ PKA signaling pathway. We demonstrate the presence of the Olfr588 on colonic muscle cells (Fig. 7). However, the putative antagonist of this receptor [41], the BCFA 2-eth- ylhexanoic acid, caused relaxation itself of colonic smooth muscle in our study (data not shown). Thus, although the olfactory receptors are potential mediators of the response to IVA, we were unable to determine whether they mediated the relaxation of colonic smooth muscle by IVA.
While little is known of the mechanisms of response to BCFA in the gut, the responses to SCFAs (mainly acetate, propionate, and butyrate) are mediated through the fatty acid receptor 2 (FFA2) and/or 3 (FFA3) formerly known as GPR43 and GPR41, respectively. These receptors are expressed in a variety of cells in the gut including enteroen- docrine cells, adipocytes, pancreatic cells, immune cells, and enteric neurons; however, their expression in smooth muscle has not been studied [4, 5, 7–10, 30–35]. Similarly, their role in mediating the response to BCFA has not been studied in any of these cell types in any detail. It is unlikely that either of these FFA receptors mediates the relaxant effects of IVA in mouse colonic smooth muscle in this study because of their well-known signaling mechanisms. FFA2 is coupled to Gαi/o and/or Gαq/11 subunits of the G protein, and FFA3 is coupled to Gαi. Thus, activation of FFA2 or 3 would lead to an increase in Ca2+ and/or decrease in cAMP genera- tion and result in contraction of smooth muscle cells rather than the increase in PKA activation in smooth muscle and the relaxation we have observed. In addition, recent stud- ies in proximal colon demonstrate that FFA3 receptors are located on enteric neurons and that FFA3-specific agonists inhibit circular muscle contraction by inhibiting neuronal ACh
release [35]. This infers that FFA3-mediated effects on circular smooth muscle are neuronal, not directly on smooth muscle cells.
As with other microbial products, the site of action of SCFA is likely to be multidimensional. SCFA have been shown to act intraluminally to release hormones and par- acrine agents from enteroendocrine cells which then acti- vate enteric neurons as well as enter the circulation. It is increasingly being recognized that microbial agents can also have effects after uptake through mucosa into vasculature. For example, recent studies suggest that GLP-1 secretion from the proximal intestine in response to FFA1 agonists is the result of vascular rather than luminal receptor activa- tion [60]. A role of IVA from stool has also been correlated with depression in human, suggesting that vascular-mediated effects may be more widespread than expected [61]. Con- sidering that BCFA are also produced intraluminally and ingested in the diet, it is also likely that they have both intra- luminal effects as evidenced in their ability to release sero- tonin from enterochromaffin cells and extraluminal vascular effects. The use of nonperfused segmental colonic prepara- tions and isolated smooth muscle cells in the current study strongly indicate a direct smooth muscle effect. This does not exclude an indirect effect mediated by release of endo- crine or paracrine agents. The effects of released endocrine and paracrine agents may be more evident in preparations of intact colon designed to investigate colonic motility pat- terns rather than direct contraction and relaxation of colonic smooth muscle.
In summary, the present study demonstrates a direct effect of the branched-chain fatty acid, isovaleric acid, on smooth muscle cells of the longitudinal layer of the mouse colon. The effect is the result of activation of a PKA signaling cas- cade, which is well known to mediate the relaxation of gut smooth muscle by a variety of other neurotransmitters and SQ22536 hormones that are coupled to similar signaling cascade.
Acknowledgments This work was supported by grants DK34153 (JR Grider) and DK28300 and DK15564 (KS Murthy) from the National Institute of Diabetes and Kidney Diseases.
Compliance with ethical standards
Conflict of interest The authors declare that they have no conflict of interest.