Hyperplasia of smooth muscle contributes to the thickening of the intestinal wall that is characteristic of inflammation, but the mechanisms of growth control are unknown. Nitric oxide (NO) from enteric neurons expressing neuronal NO synthase (nNOS) might normally inhibit intestinal smooth muscle cell (ISMC) growth, and this was tested in vitro. In ISMC from the circular smooth muscle of the adult rat colon, chemical NO donors inhibited [3H]thymidine uptake in response to FCS, reducing this to baseline without toxicity. This effect was inhibited by the guanylyl cyclase inhibitor ODQ and potentiated by the phosphodiesterase-5 inhibitor zaprinast. Inhibition was mimicked by 8-bromo (8-Br)-cGMP, and ELISA measurements showed increased levels of cGMP but not cAMP in response to sodium nitroprusside. However, 8-Br-cAMP and cilostamide also showed inhibitory actions, suggesting an additional role for cAMP. Via a coculture model of ISMC and myenteric neurons, immunocytochemistry and image analysis showed that innervation reduced bromodeoxyuridine uptake by ISMC. Specific blockers of nNOS (7-NI, NAAN) significantly increased [3H]thymidine uptake in response to a standard stimulus, showing that nNOS activity normally inhibits ISMC growth. In vivo, nNOS axon number was reduced threefold by day 1 of trinitrobenzene sulfonic acid-induced rat colitis, preceding the hyperplasia of ISMC described earlier in this model. We conclude that NO can inhibit ISMC growth primarily via a cGMP-dependent mechanism. Functional evidence that NO derived from nNOS causes inhibition of ISMC growth in vitro predicts that the loss of nNOS expression in colitis contributes to ISMC hyperplasia in vivo.
- growth inhibition
- intestine nitric oxide
- neuronal nitric oxide synthase
inflammation causes a characteristic thickening of the intestinal wall, but the cellular mechanisms of this are poorly understood. In studies of two animal models of intestinal inflammation, we found that hyperplasia of the intestinal smooth muscle cells (ISMC) was the primary factor responsible, with a modest additional contribution of hypertrophy, or enlargement of the ISMC and a minor contribution of increased extracellular matrix (7, 18, 25). The increased number of ISMC was not reversed upon resolution of inflammation (22), and the greater muscle mass is a continuing challenge to the neurogenic and myogenic control systems that regulate intestinal motility.
Hyperplasia of the ISMC occurred in experimental models despite contrasting immunological profiles [Th1 in trinitrobenzene sulfonic acid (TNBS)-colitis and Th2 in Trichinella-induced jejunitis (14, 15)]. Elsewhere, growth of smooth muscle cells occurs in vascular and airway disease and likely reflects the combined events of the appearance of growth-stimulating factors such as PDGF and IGF, as well as the loss or inhibition of endogenous growth-suppressive mechanisms (15, 19, 32).
Considerable research into the mechanisms of growth control of vascular smooth muscle has demonstrated the importance of tonic inhibition of hyperplasia by nitric oxide (NO) from endothelial NO synthase (eNOS) (1, 32). Damage to the endothelium and impairment of eNOS activity results in vascular smooth muscle hyperplasia due to the exposure to mitogenic serum factors and inflammatory factors in the absence of NO. In contrast, the factors that regulate the growth of ISMC in the normal and inflamed conditions are less clear, with evidence for an inhibitory role of the neuropeptide VIP in cultured rabbit colonic smooth muscle cells (41) as well as for IL-1 on rat ileal smooth muscle in organ culture (30). These factors may influence the responsiveness to growth factors such as IGF-I (21).
Earlier, we hypothesized that damage to the enteric nervous system (ENS) might be correlated with the onset of ISMC hyperplasia, since neuronal loss and axonal damage were early events in colitis that preceded ISMC growth (34). In vitro, the presence of enteric neurons suppressed the growth response of ISMC and maintained the contractile smooth muscle phenotype that was otherwise compromised by hyperplasia (6). In the TNBS-induced model of colitis in the rat, the number of ISMC increased rapidly after initial neuronal damage but then stabilized at the time of peak axon proliferation within the smooth muscle layers (22). This did not represent the simple repair of damage to ISMC, since there was a 2′-fold increase in total ISMC number over control following colitis. Furthermore, this was associated with restoration of the innervation density (i.e., the ratio of axons to ISMC) that exactly matched the control value, arguing for a closely regulated relationship between innervation and smooth muscle growth.
The specific neural factors or neurotransmitters responsible for suppression of ISMC growth in vivo or in vitro are unclear. However, neurons that express neuronal NO synthase (nNOS) are a major subset within the ENS and represent ∼30–40% of all ENS neurons (9). These nitrergic neurons mediate the relaxation of ISMC that is necessary for appropriate motility, with each neuron influencing multiple cellular targets. Therefore, we propose that neuronal NO could also exert an inhibitory influence on the growth response of ISMC. Accordingly, we tested the effect of NO on ISMC in vitro and examined the role of nNOS innervation on hyperplasia of ISMC in vitro and in vivo. We concluded that NO inhibits ISMC growth primarily via a cGMP-dependent mechanism and found functional evidence that NO derived from nNOS caused inhibition of ISMC growth in vitro.
Adult male Sprague-Dawley rats (200–250 g) were obtained from Charles River and housed in pairs in microfilter-isolated cages for at least 5 days prior to use, with free access to food and water. Colitis was induced by instillation of 500 μl of 200 mM trinitrobenzene sulfonic acid (TNBS; Fluka) dissolved in 50% ethanol into the colon 8 cm distal to the anus while rats were under light anesthesia without prior fasting as described earlier (45). Control animals received either no treatment or an equal volume of 50% ethanol alone. All procedures received prior approval from the University Animal Care Committee of Queen's University.
Cultured adult ISMC.
To obtain primary cultures of ISMC, strips of circular smooth muscle were obtained by microdissection, after removal of mucosa and submucosa from segments of the adult colon. These were enzymatically dissociated to yield a suspension of single viable cells as described previously (45, 46). Briefly, muscle strips were placed in a HEPES-buffered digestion solution containing papain (0.5 mg/ml; Sigma), BSA (1 mg/ml), dl-dithiothreitol (1 μM; Sigma), and collagenase type-F (0.5 mg/ml; Sigma) and incubated at 5°C for 2 h, at room temperature for 20 min, and finally at 31°C for 10 min. The solution containing the circular smooth muscle was then poured over a 200-μm nylon mesh filter, rinsed to remove excess enzyme solution, and the tissue resuspended with gentle trituration to produce a suspension of single ISMC. The cell suspension was plated on 60-mm tissue culture dishes (Sarstedt) in DMEM (GIBCO) with 10% FCS and allowed to grow to confluence. Cell nature was verified by inspection and immunocytochemistry for smooth muscle phenotype. Exclusion of myenteric neurons and glia from the cultures was verified by the absence of immunoreactivity for the neuronal marker HuD and the glial marker GFAP (glial fibrillary acidic protein).
Cocultures of myenteric neurons and ISMC.
Primary cultures of enteric neurons and smooth muscle were obtained by dissection and enzymatic dissociation of the smooth muscle layer with enclosed myenteric plexus from the intestine of 3- to 10-day-old male Sprague-Dawley rats as recently described (23).
For the analysis of growth response, confluent cultures of adult ISMC or cocultures were dissociated and plated onto 24-well collagen-coated plates. These were maintained in DMEM with 5% FCS for 24–72 h before serum withdrawal by washing with DMEM alone; 72 h later, the medium was replaced with either DMEM alone (baseline condition) or medium containing FCS and/or pharmacological agents or their carrier solutions. In some cases, drugs were added 20 min before addition of the FCS stimulus, and 20 h later, [3H]thymidine was added for 4 h, followed by processing for scintillation counting (6). For immunocytochemistry, cell suspensions were plated similarly onto circular collagen-coated glass coverslips placed within the wells of 24-well plates.
Vascular smooth muscle cells.
Cultured aortic smooth muscle cells were obtained by enzymatic dissociation of the aorta from young adult rats, by using 0.25% trypsin and 0.1 mg/ml collagenase for 1 h at 37°C, after removal of the adventitia and endothelium. Cells were grown to confluency on 60-mm culture plates in DMEM with 5% FCS and subcultured into 35-mm plates for replicate assays for the expression of inducible NO synthase (iNOS) by Western blotting.
Cyclic nucleotide assay.
Assay of the cellular content of cGMP and cAMP in ISMC was attempted with ELISA kits from R&D Systems (Minneapolis, MN) and Cayman Chemical (Ann Arbor, MI). Consistent results were obtained with the latter's thiocholine-based assay system and are reported here. ISMC were obtained as above and subcultured into 60-mm tissue culture dishes in replicate for assay. NO donors were dissolved into an aliquot of the culture medium (4 ml per dish total) and returned to the culture to obtain the final dilution; 30 min later, each culture dish was washed briefly with phosphate-buffered saline and assayed for cGMP and cAMP content in simultaneous, parallel ELISA assays according to manufacturer's instructions without acetylation. One dish of each cohort was used for determination of ISMC number. Each experiment consisted of a single cohort of identical cultures, and at least three different ISMC lines were assayed. Data were analyzed per manufacturer's instructions and expressed as fold increase over untreated control.
Immunocytochemistry and Histochemistry
Immunocytochemistry was used to detect the expression of neural markers, markers of smooth muscle phenotype or GFAP in sections of normal or inflamed colon, in cultures of adult circular colonic smooth muscle and in primary cocultures of enteric neurons (Table 1). Segments of colon were fixed in neutral buffered formalin for 24 h, dehydrated in ethanol, and embedded in paraffin before sectioning at 4 μm. Cocultures were fixed in formalin for 20 min, followed by washing in PBS, except for detection of α-smooth muscle-specific actin, desmin, and smoothelin, which required fixation in ethanol. Dual labeling for α-1 soluble guanylyl cyclase (sGC) and β-III-tubulin required fixation in methanol-acetone (50:50). Dewaxed sections or cocultures were blocked in 0.2% Tween-20 in PBS containing 1% goat serum for 1 h before overnight incubation [sections at room temperature; cultured cells at 4°C with primary antibody in antibody diluting fluid (ADF; DAKO)]. Sections or cultures were then incubated at room temperature with species-specific ALEXA-labeled secondary antibodies (1/1,000 in ADF; Molecular Probes). Nuclei were stained nonspecifically with 0.1 μg/ml 4′,6-diamidino-2-phenylindole (DAPI; Invitrogen). Staining was visualized by fluorescence microscopy (Olympus) with digital image capture (ImagePro 6.0, MediaCybernetics).
For detection of NO synthase activity by histochemistry, cultures were washed, fixed with neutral buffered formalin for 10 min, and incubated in a solution of 1 mM NADPH (Sigma) and 0.25 mM nitroblue tetrazolium in Tris-buffered saline (50 mM Tris·HCl with 0.2% Triton-X, pH 7.4) for 1 h at 37°C.
For detection of ISMC hyperplasia in vitro, cultures were incubated with 1 μM bromodeoxyuridine (BrdU) for 3 h, followed by routine fixation and washing. Optimal detection of BrdU uptake required treatment with 1 M HCl for 20 min, followed by incubation with anti-BrdU (1:50; DAKO) overnight at room temperature. For dual-labeling with anti-BrdU and anti-nNOS, processing to detect BrdU uptake was completed with incubation in red fluorescent secondary antibodies (1:1,000 Alexa-555 anti-mouse; Molecular Probes) before immunocytochemistry for nNOS was carried out, with secondary detection using Alexa-488 goat anti-rabbit antibody (Molecular Probes).
Detection of Apoptotic Cell Death
To detect apoptotic smooth muscle cells in vivo, sections of rat ileum and control or inflamed rat colon were treated for detection of TdT-mediated dUTP nick-end labeling (TUNEL) according to manufacturer's instructions (Promega) or for immunocytochemistry for the expression of cleaved caspase-3 (1:100 AB9661; Cell Signaling). For TUNEL staining, a negative control was the omission of TdT enzyme on adjacent slides and the positive control was control rat small intestine. Control or NO-donor-treated cultures of smooth muscle cells were processed similarly as above. As a positive control, some cultures were treated with doxorubicin to promote apoptotic cell death (e.g., Ref. 39). In all cases, labeling with the nuclear fluorochrome DAPI was used to establish cell number as well as assess nuclear morphology.
For immunoblotting, cells were lysed in buffer containing 10 mM Tris·HCl pH 8.0, 150 mM NaCl, 5 mM EDTA pH 8.0, 2% Triton X-100, 0.5% sodium deoxycholate, 10 μM leupeptin, and 1 mM phenylmethylsulfonyl fluoride and stored at −70°C. Aliquots (20 μg per lane) were resolved by 12% SDS-PAGE and transferred to polyvinylidene difluoride membrane by use of a semidry transfer apparatus (Bio-Rad). The membrane was blocked in 5% nonfat milk in Tris-buffered saline containing 0.2% Tween-20, then incubated with primary antibodies (Table 1). This was followed by incubation with appropriate horseradish peroxidase-linked secondary antibodies (1:20,000; Pierce) and visualization with use of a chemiluminescent substrate (Super Signal, Pierce).
The density of nNOS-expressing axons in the circular smooth muscle layer of the normal or inflamed colon was determined in five nonadjacent microscope fields in three nonadjacent sections from each animal, using a ×40 objective (numerical aperture 0.85) and digital image capture (Roper Scientific) and an image analysis program (ImagePro Plus 6.0; Media Cybernetics).
Innervation and ISMC hyperplasia.
Cocultures were stained for synaptosome-associated protein of 25,000 daltons (SNAP-25) and BrdU for investigation of the effect of innervation on ISMC hyperplasia. In nonadjacent fields in each culture, the approximate area within that field occupied by axon arborization was determined from SNAP-25 labeling (×20 objective, numerical aperture 0.70); in practice, neuronal clustering allowed irregular but roughly circular outlines to be drawn which fell at most 10 μm distal to any axon. After switching fluorescence filters to view BrdU labeling, the number of BrdU-positive cells within that area was recorded. To examine the correlation between nNOS innervation and ISMC hyperplasia, the number of ISMC with BrdU-positive nuclei that were contacted by nNOS-positive axons was manually determined in dual-labeled images (nNOS and BrdU; ×235 final magnification). The control value, representing the likelihood of random intersection, was derived from the number of intersections of randomly placed lines (of total length equal to the nNOS axons) with BrdU-labeled ISMC.
[3H]thymidine was obtained from Amersham Biosciences. S-nitroso-N-acetyl-d,l-penicillamine (SNAP), sodium nitroprusside (SNP), 8-bromoguanosine 3′:5′cyclic monophosphate (8-Br-cGMP), and 8-bromoadenosine 3′:5′cyclic monophosphate (8-Br-cAMP), were from Sigma (St. Louis, MO). The selective inhibitors of nNOS 7-nitroindazole (7-NI) and N-[(4S)-4-amino-5-[2-aminoethyl](amino)pentyl]-N′-nitroguanidine (NAAN; A5727) were also from Sigma, and 2,2′(hydroxy nitrosohydrazino)bisethanamine (DETA-NONO) was obtained from Calbiochem (VWR CANLAB, Mississauga, ON, Canada). Interleukin-1 (IL-1) was from Peprotech, and vasoactive intestinal polypeptide (VIP) was obtained from Sigma.
Data were analyzed by the Student-Newman-Keuls multiple-comparison test or single-factor ANOVA, with P < 0.05 considered to be statistically significant.
Adult male rats treated with TNBS per rectum showed signs of colitis within 48 h, displaying diarrhea, bloody stool, and weight loss as described elsewhere (e.g., Refs. 15, 45), with tissue samples showing elevated MPO values (Fig. 1). In contrast, animals receiving the ethanol control solution showed no signs of illness. Immunocytochemistry for PCNA expression, a marker of cell division, in sections of the mid-descending colon showed positive labeling in the basal mucosa in both control and inflamed animals as expected, but also showed the onset of positive labeling in the smooth muscle layers by 12 h post-TNBS (Fig. 1, B and C). This evidence for smooth muscle cell mitosis occurred without detectable evidence for apoptotic cell death in either the circular or longitudinal muscle layers, as assessed by TUNEL labeling (Fig. 1D) and immunocytochemistry for cleaved caspase-3 expression (not shown). Positive controls for each technique came from the distal tips of ileal mucosa as well as the presence of TUNEL-positive cells in the mucosa of inflamed colon (Fig. 1D).
The evidence for the early onset of smooth muscle growth without detectable apoptotic death, occurring at a time when neuronal death and axonal damage has been shown earlier (22, 34), led us to pursue the potential for neuronal regulation of ISMC growth in vitro.
NO and Growth of ISMC
Pure cultures of ISMC were obtained from strips of the circular smooth muscle of the mid-descending colon of adult rats as described earlier (45, 46). The majority of the freshly isolated ISMC attached and adapted to culture, developed a typical bipolar spindle- or ribbon-shape, and proceeded to grow to confluence by ∼7–10 days. Immunocytochemistry and Western blotting of cultured ISMC showed the expression of abundant smooth muscle-specific α-actin, smoothelin, and desmin, which are markers of the smooth muscle phenotype (e.g., Ref. 29), and the typical appearance of smoothelin expression is shown in Fig. 2. Immunocytochemistry for GFAP was consistently negative, excluding the presence of glial cells. The presence of sGC, the target enzyme of NO, was verified in cultured ISMC by Western blotting and immunocytochemistry (not shown) for the α1- and β2-isoforms of sGC. Both isoforms were present in cells confirmed as ISMC, and Western blotting showed similar levels of expression in vitro as in vivo (Fig. 2B).
Primary cultures of ISMC were passaged once and then subcultured into multiwell plates for determination of growth responses by [3H]thymidine uptake. The response to fetal calf serum (FCS) was maximal at ≥7.5% FCS (not shown), and subsequent studies used the intermediate-level stimulus of 2.5% FCS. Similar results were obtained when circular smooth muscle cells were plated onto laminin or fibronectin-coated surfaces, with no significant differences among baseline or stimulated responses (data not shown). To test the effect of NO on the growth response of ISMC, representatives from each of three classes of NO donors were applied 20 min before the addition of 2.5% FCS, and the effect on [3H]thymidine uptake expressed as percent of the untreated control. In these, the average baseline value was 3,300 ± 525 dpm/well (n = 24), and this was increased by 2.5% FCS by to 34,300 ± 5,250 dpm (n = 24), representing a 13.2 ± 1.9-fold (n = 24) increase. Figure 3 shows the outcomes of concentration-dependent reduction in growth response, where both SNP and DETA-NONO reduced the uptake of [3H]thymidine to ∼50% of untreated control at 50 μM (P < 0.05) whereas SNAP was effective over a narrower range of concentrations. The highest concentration of each donor reduced the [3H]thymidine uptake to baseline levels without cytotoxicity (i.e., exclusion of Trypan blue and propidium iodide was maintained; not shown). In addition, investigation of apoptotic cell death in ISMC cultures treated with NO donors showed no positive labeling for either TUNEL or cleaved caspase-3, although labeling was detected in cultures exposed to doxorubicin as a positive control (Fig. 3D).
In addition, we investigated the possibility of endogenous iNOS activity in vitro that might originate from dissociation and culturing as described for vascular smooth muscle cells (2). Western blotting for the presence of iNOS in ISMC was consistently negative, despite its successful induction in aortic smooth muscle cells following exposure to IL-1 (50 ng/ml, 18 h; Fig. 3E). Furthermore, pretreatment of ISMC cultures with the specific iNOS blocker L-N6-(1-iminoethyl)-lysine (1 μM or 100 μM) did not affect their subsequent response to 2.5% FCS (Fig. 3F), supporting the conclusion that there is no detectable role for this enzyme in ISMC in vitro.
NO and cyclic nucleotides.
Although these experiments established the potential for NO to inhibit ISMC hyperplasia, there is also evidence for the growth-inhibitory effect of VIP on rabbit intestinal smooth muscle (41) as well as its corelease from nitrergic inhibitory nerves in vivo (43). However, examination of the effect of VIP over a wide concentration range showed no effect on the subsequent response of ISMC to FCS (Fig. 4A).
The suppression of growth of vascular smooth muscle cells by NO typically involves the elevation of intracellular cGMP, and in intestinal smooth muscle tissue, NO acts via elevated cGMP to cause relaxation (12, 24, 28). Therefore, we tested the potential for 8-Br-cGMP, a membrane-permeable analog of cGMP, to mimic the actions of the NO donors on ISMC in vitro. Figure 4B shows that 8-Br-cGMP caused significant reduction of [3H]thymidine uptake in response to FCS, over a range of 1–500 μM (P < 0.05). Addition of the sGC inhibitor 1H-oxadiazolo-quinoxalin-1-one (ODQ) significantly reduced the inhibitory action of SNP, indicating a role for cGMP synthesis in its action (Fig. 4C), whereas ODQ alone had no significant effect on either basal or FCS-stimulated growth responses. Although inhibition of the cGMP-specific phosphodiesterase PDE5 by zaprinast (25 μM) reduced the FCS response to 72 ± 11% (n = 4) of untreated control, the combination of zaprinast with SNP caused a strong concentration-dependent inhibition of growth, significantly greater than that of SNP alone (Fig. 4D). All concentrations of inhibitors were similar to those used in published studies of pharmacological experiments on vascular smooth muscle cells.
Recent evidence implicates the actions of both cGMP and cAMP in some NO-mediated events in kidney, heart, and vascular smooth muscle (4, 26, 38). The nonspecific phosphodiesterase (PDE) inhibitor IBMX strongly potentiated the action of SNP on ISMC (Fig. 5A). However, application of 8-Br-cAMP at the same concentrations as used with 8-Br-cGMP (Fig. 4) achieved significantly less inhibition of the FCS-induced growth response (Fig. 5B), supporting a dominant role for the cGMP-based pathway. Nonetheless, there was significant inhibition of growth with 500 μM 8-Br-cAMP, evidence for a contributory role of cAMP in growth control of ISMC. The possibility of nonspecific activation of sGC by 8-Br-cAMP was tested by addition of ODQ (50 μM), but there was no difference in the inhibition of the growth response to 2.5% FCS with 8-Br-cAMP (100 μM) with ODQ pretreatment [79.9 ± 4.0% (n = 3) of the FCS response] or without [79.5 ± 9.3% (n = 3)].
To test whether NO activated a cAMP signaling pathway via cGMP-mediated inhibition of PDE3 (responsible for cAMP breakdown), ISMC were treated with the PDE3 inhibitor cilostamide prior to exposure to SNP and FCS, or FCS alone. Cilostamide (25 μM) reduced the growth response of ISMC to 76.3 ± 3.0% (n = 8) of FCS alone. In combination with SNP, cilostamide showed strong, concentration-dependent potentiation of its inhibition of the growth response. At the low level of 10 μM, SNP reduced the growth response to 74.2 ± 4.3% (n = 16) of FCS alone, whereas the addition of cilostamide further decreased this response to 61.2 ± 6.4% (n = 10) and 42.2 ± 7.5% (n = 8) of FCS alone, at 25 μM and 50 μM, respectively. No effect with lower doses of cilostamide was seen [1 and 10 μM cilostamide plus SNP; 103 ± 4.3% (n = 8) and 100 ± 2.9% (n = 8) of SNP alone, respectively].
These experiments showed an important role for elevation of cyclic nucleotides in growth inhibition and supported a cGMP-dominated process of growth inhibition by NO. To further explore this, we directly measured the effects of NO donors on the intracellular levels of cAMP and cGMP in ISMC exposed to NO donors. The amounts of cAMP and cGMP were determined by separate ELISA assay run in parallel, by using ISMC from three different cell lines at 30 min after addition of DETA-NONO, SNP, or SNAP (50 μM). In each case, the NO donors stimulated a consistent increase in cGMP while causing no measurable elevation of cAMP (Fig. 5C). Furthermore, additional experiments showed that SNP caused the sustained elevation of cGMP through 120 min postaddition (Fig. 5D). Overall, we concluded that NO inhibited ISMC growth primarily through elevation of cGMP.
Neural regulation of ISMC growth in vitro.
Earlier, we showed that coculturing myenteric neurons with ISMC could suppress growth and promote differentiation of the ISMC, but the mechanism was unknown (6). We examined the possibility that nNOS-positive neurons could accomplish this via a NO-based mechanism, using the same coculture model of neonatal ISMC, myenteric neurons, and glia. In these primary cultures, the ISMC formed a complex multilayered culture and were readily detected by the expression of actin (not shown) and desmin (Fig. 6). The use of NADPH diaphorase histochemistry to detect NOS activity showed an extensive network of NOS-positive myenteric neurons, axons, and varicosities distributed among the ISMC (Fig. 6C). Western blotting of cellular lysates from cocultures identified both α-1 sGC and β-1 sGC (not shown), and dual-label immunocytochemistry detected α-1 sGC within both ISMC and axons (Fig. 6A). Immunocytochemistry confirmed that there was a large population of nNOS-expressing neurons with extensive axonal processes running among desmin-positive ISMC (Fig. 6D), thus establishing a culture model of NO-generating neural elements and potentially NO-responsive ISMC.
To examine the effect of innervation on the growth of ISMC in cocultures, immunocytochemistry was used to detect the uptake of BrdU. Positive labeling for BrdU, evidence of the S phase of mitosis, was correlated with the distribution of the axonal marker SNAP-25 and showed that the normally abundant distribution of BrdU-positive ISMC was reduced in areas of extensive axonal presence (Fig. 7, A and B). Comparison between areas with and without innervation showed that innervation was associated with a significant reduction in the number of BrdU-positive nuclei (Fig. 7C), implying a growth-inhibitory effect.
To examine the neural basis of this effect, we first determined the representation of nNOS neurons in vivo and in vitro. nNOS is abundant within the ENS in several species, but the proportion has not been studied in rat colon. Dual-label immunocytochemistry showed that 37 ± 3% (n = 3 rats) of neurons identified by anti-HuD immunocytochemistry were also positive for nNOS expression, and a similar proportion of nNOS-expressing neurons was found within the cocultures (34 ± 3%; n = 4), suggesting that nNOS activity could be involved in growth inhibition in vitro. In cocultures stained to detect BrdU uptake and then dual labeled for the expression of nNOS, we found that contact with nNOS-positive axons reduced the number of BrdU-positive ISMC nuclei compared with the statistical control of random intersection, implying that axonal contact by nitrergic axons conveyed a growth-suppressive effect (Fig. 7, D and E).
This indirect evidence of an nNOS-associated mechanism of growth control of ISMC was investigated further by assessing the effect of pharmacological inhibition of nNOS on the ISMC growth response. Either 7-NI or NAAN, considered to be specific inhibitors of nNOS (5, 16), was applied to cocultures 1 h prior to the growth stimulus of 2.5% FCS, and the uptake of [3H]thymidine was assessed. Over a concentration range of 0.1–10 μM, we detected significant enhancement of [3H]thymidine uptake, implying that tonic inhibition of growth by nNOS neurons was normally present (Fig. 7F). Each agent caused a significant stimulation of uptake of [3H]thymidine over 2.5% FCS alone, by ∼10% (NAAN) or 25% (7-NI). Concentrations <0.1 μM were ineffective (i.e., outcomes similar to FCS alone) and concentrations ≥100 μM were either ineffective or inhibitory, suggesting the onset of toxicity at higher concentrations. In each case, the application of drug vehicle alone (7-NI, DMSO, NAAN, saline) had no action. Thus it appears that neural signaling via NO normally suppresses the ISMC growth response in vitro.
nNOS innervation in colitis.
In the TNBS model of colitis in the rat, damage to the ENS occurred early (34), prior to the marked hyperplasia of ISMC on day 4 of inflammation (22). The evidence for nNOS-mediated regulation of ISMC hyperplasia in vitro led us to test whether colitis also caused damage to nitrergic innervation, which might contribute to subsequent ISMC hyperplasia.
The distribution of nNOS was examined in cross-sections of the mid-descending colon at early time points in the course of TNBS-induced colitis. In control animals, nNOS-positive neuronal cell bodies were present in the myenteric plexus and nNOS-positive axons were abundant within the smooth muscle layers (e.g., Fig. 8A). However, this changed rapidly with colitis (Fig. 8B), and quantification of the number of nNOS-positive axons in the circular smooth muscle showed a substantial decrease early in colitis (Fig. 8, B and C) that preceded the hyperplasia of ISMC in these animals. Axon number was unchanged by 12 h post-TNBS but dropped to one third of control levels by day 1 (Fig. 8C), in advance of the ISMC hyperplasia that occurs by days 2-4, as reported earlier (18, 22). Since nNOS immunoreactivity was retained in the cell bodies of the myenteric neurons (Fig. 8B) and the number of nNOS-positive axons showed a trend to return to control levels by day 2 (Fig. 8C), inflammation may impair the expression of nNOS within damaged, but surviving, neurons. This supports the hypothesis that innervation of ISMC via NO contributes to the mechanisms that maintain the contractile, nonmitotic phenotype in vitro and in vivo.
In TNBS-colitis, hyperplasia of ISMC gave rise to a threefold increase in smooth muscle cells in the circular layer by day 6 and this did not reverse with resolution of inflammation (22). The obstructive muscle bulk and potentially unregulated contractile activity may contribute directly to the pathology of gastrointestinal disease, as it does in atherosclerosis and airway disease, but the factors that maintain the differentiated, nondividing state of the ISMC are virtually unknown. In the vascular system, the endothelium maintains the smooth muscle in the nonproliferative state via NO derived from eNOS. Endothelial damage both removes growth control and exposes the vascular smooth muscle cell to mitogenic serum factors (reviewed in Ref. 32), and the experimental restoration of eNOS activity can inhibit formation of the neointima in damaged blood vessels in vivo (36). Here, we have identified a novel neural mechanism of regulation of growth of ISMC, where nNOS-derived NO and cGMP elevation tend to suppress ISMC growth. Our research suggests that inflammation of the intestine damages this mechanism, which may promote the growth response of the ISMC.
Although NO is well recognized to inhibit the growth of vascular and airway smooth muscle (e.g., Refs. 11, 17; reviewed in Ref. 1), this has not been previously addressed in ISMC. Using several classes of NO donor chemicals, we showed the concentration-dependent inhibition of ISMC growth in vitro without cytotoxicity. These NO donors were effective at concentrations similar to those shown effective in inhibiting the growth of vascular smooth muscle cells in vitro, suggesting a parallel mechanism. Other potential contributors to the regulation of growth of ISMC include VIP (41), which is coreleased with NO from inhibitory innervation (43), and NO derived from iNOS, which might appear in isolated ISMC in vitro. However, neither the addition of VIP nor the pharmacological inhibition of iNOS affected the growth response of ISMC. In addition, there was no detectable expression of iNOS even with IL-1 stimulation, which upregulated iNOS expression in the aortic smooth muscle cells that were examined in parallel (e.g., Ref. 10; reviewed in Ref. 37). Although some evidence suggested that IL-1 induced the expression of iNOS in intestinal smooth muscle, these experiments employed intact tissue where the resident macrophages could also contribute to that signal (30). Overall, this emphasizes the important differences can exist among smooth muscle types according to origin and species.
Three lines of evidence support a major role for the elevation of cGMP in response to NO: the membrane-permeable analog of cGMP mimicked the inhibition of growth, whereas ODQ, a specific inhibitor of sGC, reversed this effect. In addition, inhibition of the breakdown of cGMP by the PDE5 inhibitor zaprinast strongly potentiated the actions of SNP. Most telling, direct measurement of cGMP by ELISA established its sustained elevation with NO donors. However, some contribution of altered cAMP levels is also possible since there was a significant inhibitory effect with higher levels of 8-Br-cAMP, even though altered levels of cAMP were not detected after NO donor addition. In NO-mediated growth control, this may be secondary to the elevation of cGMP, since inhibition of PDE3 with cilostamide potentiated the action of SNP. Overall, these multiple lines of evidence established that the mitotic response of ISMC can be regulated by NO.
Neurons expressing nNOS are abundant within the ENS in several species including the rat, where these neurons influence their target cells in the smooth muscle layer by arborization of axons with numerous varicosities (sites of neurotransmitter release) along their length. We hypothesized that NO, already established as the major inhibitory neurotransmitter for intestinal smooth muscle, could also exert a significant inhibitory influence on ISMC growth. This is a possible explanation of earlier work in vitro showing an enhanced growth response of ISMC when myenteric neurons were removed from cocultures (6).
In this coculture model of myenteric neurons and ISMC, proximity to nNOS-positive axons was associated with a reduced incidence of mitosis in individual ISMC. As well, inhibition of nNOS activity by relatively selective inhibitors led to a significant enhancement of the growth response of ISMC to FCS. The potential contribution of eNOS is unclear, with evidence for eNOS mRNA but not protein in ISMC (40). Overall, the data indicate that NO mediates a growth-suppressive effect of nNOS-expressing axons on ISMC in close contact in vitro.
Our data suggest that a disruption of nNOS function in vivo might be associated with subsequent ISMC hyperplasia. The effect of intestinal inflammation on nNOS expression in the human intestine is virtually unknown beyond evidence of disruption in ulcerative colitis patient tissues (44), and general evidence of inflammation affecting the myenteric ganglia in Crohn's disease (31). However, animal models show impairment of both nNOS expression and function in colitis (13, 18, 27, 33) with direct visualization of reduced NO production by myenteric neurons in colitis (20). Furthermore, loss of nNOS worsened colitis in knockout mice, suggesting a possible protective function (3), whereas the viability of the control nNOS knockout mice shows the presence of compensatory mechanisms.
Numerous studies have examined the outcomes of selective and nonselective inhibition of NOS isoforms in intestinal inflammation, but very few have examined ISMC hyperplasia as an outcome. For example, our earlier use of the NOS blocker NG-nitro-l-arginine methyl ester (l-NAME) in colitis showed significant reduction of ISMC hyperplasia (18), but the actions of l-NAME include iNOS inhibition and thus a potentially anti-inflammatory effect, which prevents a clear interpretation of the effect on ISMC growth.
Therefore, we used immunocytochemistry to show that TNBS colitis caused a threefold loss in nNOS-immunoreactive axons within the circular smooth muscle layer by 24 h. This is a key step in the association of nNOS activity with maintenance of the normal ISMC phenotype, since comparison with previous work places the loss of nNOS expression clearly in advance of the increase in ISMC number by days 4-6 (22). This raises the possibility that neuronal NO normally exerts a significant inhibitory influence on ISMC growth in vivo, with damage then releasing the ISMC to respond to mitogenic factors associated with inflammation. The subsequent reversal toward control levels of nNOS-expressing axons by day 2 showed that nNOS axons can display rapid recovery following damage. Indeed, this may be a selective effect on the nNOS subpopulation, since axon loss reaches a maximum subsequently to this, by day 4 (22). The restoration of nNOS expression may be critical for reestablishing neural control of ISMC and the resumption of the nondividing contractile phenotype.
It is very unlikely that neural influence is the sole determinant of the growth response of ISMC in either the normal or inflamed intestine. The inflammatory milieu can affect many aspects of the ISMC directly or indirectly, such as receptor expression, the mechanics of Ca2+ elevation, and contractility at the cellular level (45, 46), and this could also include the responsiveness of ISMC to NO. For example, the NO-guanylyl cyclase-cGMP pathway is impaired in pathological states of vascular smooth muscle and in experimental colitis, possibly via desensitization of sGC (8, 35, 42). Furthermore, there is evidence of transiently decreased expression of protein kinase G in actively proliferating neointimal smooth muscle cells, early in the response to vessel wall injury (2). This may represent an additional aspect of inflammation-associated changes to the smooth muscle cell that facilitates phenotypic modulation or growth and could potentiate the growth response in vivo; it is unknown whether similar events occur in ISMC.
We conclude that nitrergic innervation of ISMC normally contributes to the suppression of a growth response via NO-mediated elevation of cGMP. Disruption of the neural production of NO via nNOS leads to an enhancement of growth of ISMC in vitro, and the association between nNOS axon loss and subsequent hyperplasia of ISMC in experimental colitis suggests that this may also be an important homeostatic regulatory factor in vivo.
This work was supported by the Crohn's and Colitis Foundation of Canada (M. G. Blennerhassett). A.-M. Pelletier received a summer scholarship from the Canadian Association of Gastroenterology.
No conflicts of interest are declared by the author(s).
We thank T. R. Blennerhassett for carrying out the immunocytochemistry of nNOS expression.
- Copyright © 2010 the American Physiological Society