Abstract
The aim of this study was to compare immunoreactivities for substance P with other enteric neuropeptides and GAP-43, a general marker for enteric nerves, in normal human colon and in different stages of ulcerative colitis. Tissue samples from normal colon and regions of ulcerative colitis colon were obtained at surgery and immunostained for substance P, vasoactive intestinal polypeptide (VIP), somatostatin, calcitonin gene-related peptide (CGRP), enkephalin, galanin, GAP-43, and neuron-specific enolase (NSE). Visual examination and semiquantitative analysis revealed a clear increase in the immunoreactivity for substance P in ulcerative colitis, whereas no differences were observed in the distribution of the other peptides. Therefore, quantitative analysis was performed only for substance P immunoreactivity in the lamina propria, circular muscle layer, and myenteric ganglia. In the lamina propria, the score of total intensity of substance P immunoreactivity was 0.55 ± 0.15 (mean ± SEM) in normal colon, 1.30 ± 0.35 (p = 0.087) in least affected colon, and 2.22 ± 0.28 (p < 0.001) in moderately affected colon, whereas no significant differences were observed in immunoreactivities for GAP-43. Similar results were obtained for the mean substance P- or GAP-43-immunoreactive area. In the circular muscle layer, the number, density, total intensity, and perimeter of substance P- and GAP-43-immunoreactive fibers were essentially similar in normal colon, and in mild or moderately affected colon. We conclude that ulcerative colitis does not change the density of gut innervation as a whole. However, the density of substance P-containing nerves is specifically increased, probably due to increased peptide synthesis leading to better visibility of the fibers.
Ulcerative colitis (UC) is an inflammatory ulcerating process in the mucosa of colon, usually characterized by successive exacerbations and remissions of variable intensity and duration. The precise etiology of UC is unknown. Previous studies have revealed that substance P concentration is increased in inflamed colon mucosa of UC patients (Koch et al. 1987; Goldin et al. 1989). Quantitative histochemical studies indicated that this change is due to increased number of substance P-immunoreactive nerve fibers in the lamina propria (Keränen et al. 1995; Watanabe et al. 1998). It has remained unclear whether UC increases the density of enteric innervation in general and whether the increase in substance P occurs through increased intracellular peptide level, sprouting of nerve fibers, or increased number of ganglion neurons.
The growth-associated protein-43 (GAP-43) is expressed by nerve fibers under conditions of embryonic growth and axon regeneration. However, GAP-43 is localized abundantly in the autonomic neurons and nerve fibers even in the mature human intestine (Sharkey et al. 1990; Vento and Soinila 1999). The expression of GAP-43 in mature intestine fits with the idea of plasticity of the enteric nervous system and represents the special nature of enteric nerves. Our recent quantitative comparison of GAP-43, neuron-specific enolase (NSE), and protein gene product 9.5 revealed that GAP-43 is the most sensitive marker for nerve fibers and NSE for neuronal somata (Vento and Soinila 1999). We now apply these markers on specimens of human UC colon to further characterize the morphological changes in the enteric nervous system as a whole and to determine whether they are specific for substance P-containing innervation.
Materials and Methods
Tissue Specimens
Specimens of normal and diseased human gut were obtained immediately after intestinal resection. The permission for human studies was granted by the Ethical Committee of the Helsinki University Central Hospital.
Specimens of Normal Colon. Specimens were taken from eight patients (one woman and seven men) undergoing resection of colon or rectum for treatment of neoplasia. The mean age of the patients was 67 years (range 38–83). None of the patients had bowel obstruction or other additional colon disease. A single whole-wall specimen was obtained from a region at the farthest possible distance from the tumor.
Specimens of Ulcerative Colitis. The diagnosis of UC was based on history, clinical examination, endoscopic examination, and histopathological findings of inflammatory cell infiltrates in the lamina propria, crypt abscesses, and paucity of goblet cells. Specimens of colon were taken from the resected bowel of 10 patients with UC undergoing proctocolectomy (six women and four men). This material represents a completely new group of patients compared to that analyzed in our previous study (Keränen et al. 1995). The mean age was 31 years (range 22–46 years). The UC patients underwent surgery for failed conservative treatment or for side effects of corticosteroids. One patient underwent surgery for moderate dysplasia and primary sclerosing cholangitis. None of the patients were operated on for fulminant colitis. All patients received 5-amino-salicylic acid treatment orally; 7/10 patients also received oral corticosteroids. In addition to the corticosteroid therapy, one patient received azathioprine. The mean duration of disease was 7 years (range 1–15 years). Whole-wall specimens of colon were taken from the least affected, moderately affected, and ulcerated regions with destroyed epithelium. The specimens were stained for routine histological examination. In least affected UC colon, routine histological examination indicated no changes in some of these colon specimens, whereas in others epithelial proliferation suggestive of regeneration as well as signs of chronic inflammation were observed. The moderately affected region of UC colon showed ulceration of mucosa in all specimens. The mucosal thickness varied and the total number of glands was clearly reduced. Lamina propria contained a large number of lymphocytes and plasma cells, occasionally neutrophils and crypt abscesses. The epithelium showed signs of regenerative atypia. No inflammatory changes were observed in the muscle layer. In severe lesions the mucosa had been destroyed and masses of leukocytes or necrotic tissue could be seen over the muscle layer. Occasionally some deformed, regenerative low villous profiles could be found.
Immunohistochemistry
Immediately after resection the specimens were immersed in 4% paraformaldehyde in PBS, pH 7.4, for 8 hr and transferred into 20% sucrose in PBS. The specimens were embedded in freezing gel. Cryostat sections of 10 μm were cut on chrome–alum–gelatin-coated glass slides. Endogenous peroxidase activity was blocked with 0.3% hydrogen peroxide in methanol for 5 min at room temperature (RT). For immunohistochemistry the sections were incubated with 5% normal swine or goat serum, depending on the host species of the secondary antiserum, for 20 min at RT and with the primary antiserum diluted in PBS overnight at 4C. The dilutions were 1:500 for mouse anti-GAP-43 (Boehringer; Mannheim, Germany), 1:100 for rat anti-substance P (Chemicon; Temecula, CA), 1:8000 for rabbit anti-NSE (Chemicon), 1:500 for rabbit vasoactive intestinal peptide (VIP), enkephalin, somatostatin, or galanin (Incstar; Stillwater, MN), and calcitonin gene-related peptide (CGRP) (Amersham; Poole, UK) antisera. Preincubation with 0.1 μM solution of substance P totally abolished all immunoreactivity. The specificity of the other antibodies used has been characterized elsewhere: anti-GAP-43, clone 91E12 (Schreyer and Skene 1991), anti-NSE, batch AB 951 (Thompson et al. 1983). Characterization of immunoreactivity for VIP, somatostatin, galanin and enkephalin, and CGRP has been reported in our previous studies (Szabat et al. 1994; Vento et al. 1998). After the primary incubation specimens were stained according to the avidin-biotin protocol using a Vector ABC kit. For some specimens, fluorescein-conjugated swine anti-rabbit IgG (DAKO F205; 1:200 in PBS) was used as the secondary antibody.
Semiquantitative analysis of VIP-, enkephalin-, somatostatin-, galanin-, CGRP-, or substance P-immunoreactive nerve fibers was performed for the lamina propria, muscularis mucosae, submucosa, circular and longitudinal muscle layers, and submucous and myenteric ganglia. The estimation of the density of immunoreactive nerve fibers was expressed as – (none), + (sparse), ++ (moderate), or +++ (dense).
Morphometric Analysis
Morphometric analysis was performed on seven normal colon specimens, eight least affected, and eight moderately affected UC colon. To obtained maximal standardization, the specimens were processed simultaneously for immunohistochemistry. Likewise, digitization was carried out in a single session. To eliminate unevenness of the illumination, background (illuminated field without a specimen) was subtracted from each image. The threshold value for significant immunoreactivity was empirically determined from the intensity histogram over the measured area by testing different values for comparison with the original view through the microscope. This was performed for both normal and diseased specimens. The differences between threshold values obtained for different specimens were approximately 2% at the most, regardless of whether the specimens were controls or UC specimens. The threshold values for GAP-43 and substance P stainings were determined separately, and the same threshold value was used for all images stained for each antibody.
Lamina Propria. From regions containing longitudinal profiles of the villi, three sections were selected according to a random set of numbers and digitized under standardized circumstances through a video camera connected to the microscope and digitizing board (PC Vision Plus; Image Technology, Woburn, MA). The images were analyzed using Sigma Scan Pro 4.0 software (SPSS Science; Erkrath, Germany). The area measured was determined by the epithelial basement membrane of the villus and the muscularis mucosae. The pixels representing values below the threshold value were electronically removed. The total intensity and area of substance P and GAP-43 immunoreactivity/measured area, i.e., the sum of intensity values or the number of all pixels exceeding the threshold divided by the total number of pixels in the measured area, was calculated. The values obtained from the three sections of each specimen were averaged.
Circular Muscle Layer. From the sections stained for GAP-43 or substance P and containing transverse sections of nerve fibers, three sections were randomly selected, stored in the computer memory, and threshold values determined as described above. To obtain an estimate of nerve fiber density, the number of nerve profiles per unit area and the total area represented by immunoreactive pixels in the measured area were determined. To obtain an estimate of the intensity of immunoreactivity, the total intensity, i.e., the sum of intensity values of all profiles exceeding the background, was calculated. To obtain an estimate of the thickness of the nerve fibers, the mean perimeter of immunoreactive nerve fiber profiles was measured. For each profile, shape factor SF was calculated using the formula SF=4π × area/(perimeter)2. This parameter indicates how circular a profile is, the value for a circle being 1.0 and that of a line approaching 0. To exclude oblique sections, only those profiles were accepted of which the SF was 0.4 or greater. This limit was obtained by measuring a large number of profiles in an optimal cross-section through the gut wall, which revealed that shape factors of cross-sectioned nerve profiles fall between 0.4 and 1.0. The values obtained from the three sections of each specimen were averaged.
Myenteric Ganglion. One of five sections of each specimen stained for NSE was selected and photographed under standardized conditions so that the whole ganglion chain profile of the section was included. The photographs were digitized with a Hewlett Packard Scanjet IIcx scanner. The total (cumulative) area of ganglion profiles in each section was calculated. The value was divided by the length of the ganglion chain in the section (Figure 1). Average intensity of substance P immunoreactivity of the ganglion profiles was measured for each specimen.
The average of each parameter described above for each patient group was calculated and data expressed as the mean ± SEM, and statistically analyzed using the Student's t-test and ANOVA or Kruskal-Wallis ANOVA on Ranks. Correlation between the degree of UC activity and the total intensity or total area of GAP-43 and substance P-immunoreactive pixels was tested by Spearman's correlation test.
Results
Distribution of GAP-43 Immunoreactivity
Normal Colon. In the lamina propria, many GAP-43-immunoreactive nerve fibers were distributed throughout the villous cores and formed a dense network at the base of the villi (Figure 2A). Muscularis mucosae and the submucous layer contained a network of GAP-43-immunoreactive fibers and the submucous plexus included a few GAP-43-immunoreactive neurons. In the muscle layer, abundant GAP-43-immunoreactive nerve fibers were seen in both the circular and longitudinal layers. The sarcolemma lacked GAP-43 immunoreactivity. In the myenteric plexus, a dense network of GAP-43-immunoreactive fibers surrounded the neurons.
Measurement of cumulative NSE-immunoreactive area. The area of each NSE-immunoreactive ganglion (G) was measured and the sum value was divided by the length of the myenteric ganglion chain in the section (L).
Least and Moderately Affected Regions of Ulcerative Colitis. The distribution of GAP-43 immunoreactive fibers in the lamina propria was similar to that in the normal colon (Figures 2B and 2C). However, the very severely affected lesions, which contained only destroyed villous profiles or no villi at all, showed very few fragmented or no GAP-immunoreactive fibers. The staining also revealed many GAP-43-immunoreactive fibers in the thickened muscularis mucosae. In the circular muscle layer, the GAP-43-immunoreactive fibers seemed to be more numerous in moderately affected regions of colon than in normal colon or in least affected colon. In the ganglia, no essential differences in GAP-43 immunoreactivity were observed between normal colon and UC.
GAP-43 immunoreactivity in lamina propria. (
Quantitative Measurements of GAP-43 Immunoreactivity
In the lamina propria, the total intensity of GAP-43 immunoreactivity was 4.79 ± 0.99 in normal colon, 4.64 ± 0.78 in least affected UC colon, and 5.26 ± 0.90 in moderately affected UC colon (Figure 3A). The total area of GAP-43 immunoreactivity was 64 ± 12 in normal colon, 60 ± 8 in least affected UC colon, and 68 ± 1 in moderately affected UC colon. There were no significant differences between the groups (Figure 3B).
Morphometry of substance P and GAP-43 immunoreactivity in lamina propria (mean ± SEM). (
In the circular muscle layer, the number of GAP-43-immunoreactive nerve fiber profiles per unit area was 102 ± 9 in normal colon, 99 ± 10 in least affected UC colon, and 133 ± 18 in moderately affected UC colon (Figure 4A). The differences between the groups were nonsignificant. The total area of GAP-43-immunoreactive fiber profiles was 165 ± 35 in normal colon, 149 ± 25 in least affected UC colon, and 183 ± 30 in moderately affected UC colon (Figure 4B). These values did not significantly differ from each other. Total intensity of GAP-43 immunoreactivity was 2691 ± 592 in normal colon, 2428 ± 428 in least affected UC colon, and 2945 ± 503 in moderately affected UC colon (Figure 4C). The differences between the groups were nonsignificant. Neither were there significant differences between the groups in the mean perimeter of a single nerve profile (14.4 ±1.3 for normal colon, 14.3 ±1.2 for least affected UC colon, and 13.6 ± 0.9 for moderately affected UC colon) (Figure 4D).
Morphometry of GAP-43 and substance P in circular muscle layer (mean ± SEM). (
Distribution of Substance P Immunoreactivity
Normal Colon. In the lamina propria, some thin varicose substance P-immunoreactive fibers extending through the villus were regularly found (Figure 5A). Some fibers in the muscularis mucosae were immunoreactive for substance P. In the submucosa, some thick substance P-immunoreactive fiber bundles, as well as thin substance P-immunoreactive fibers around blood vessels, were revealed. Several small submucous ganglia included substance P-immunoreactive neurons. Both muscle layers contained thin substance P-immunoreactive nerve fibers, which ran parallel to the muscle fibers. The myenteric plexus contained some substance P-immunoreactive neurons and quite a few substance P-immunoreactive fibers.
Least Affected Region of UC Colon. On visual estimation, the number of substance P-immunoreactive fibers in the lamina propria was greater in UC than in normal colon (Figure 5B), whereas no such difference was observed in the muscularis mucosae, submucosa, the muscle layer, or the submucous or myenteric plexuses.
Moderately Affected Region of UC Colon. The number of substance P-immunoreactive nerve fibers in the lamina propria was markedly increased compared to the two other groups (Figure 5C). Furthermore, the intensity of nerve fibers seemed to be higher. The thickened muscularis mucosae contained many fragmented fibers. However, the very severely affected lesions, which contained only destroyed villous profiles or no villi at all, showed very few fragmented or no substance P-immunoreactive fibers. Distribution of substance P immunoreactivity in the submucous layer, the muscle layers, and the submucous and myenteric ganglia was similar to that of normal colon.
Substance P immunoreactivity in lamina propria. (
Quantitative Measurements of Substance P Immunoreactivity
Total intensity of substance P-immunoreactive nerve fibers in the lamina propria was 0.55 ± 0.15 in the normal colon, 1.30 ± 0.35 in least affected UC colon (p = 0.087), and 2.22 ± 0.28 in moderately affected UC colon (Figure 3A) (p < 0.001). The values for least affected UC colon and that for moderately affected UC colon were 136% and 304% higher than the control value. The value for moderately affected UC colon was 71% higher than that for least affected UC colon. Total area of substance P-immunoreactive nerve fibers was 8 ± 2 in the normal colon, 17 ± 5 in least affected UC colon (p = 0.12), and 30 ± 4 in moderately affected UC colon (p < 0.001). The values for least affected UC colon and for moderately affected UC colon were 114% and 280% higher than the control value. The value of moderately affected UC colon was 76% higher than that for least affected UC colon. The difference between least affected and moderately affected UC colon was significant (p < 0.05) (Figure 3B). The area and total intensity of substance P immunoreactivity correlate with the degree of UC (Spearman's correlation test p < 0.001).
In the circular muscle layer, the number of substance P-immunoreactive nerve fibers per unit area was 17 ± 2 in normal colon, 15 ± 3 in least affected colon, and 25 ± 6 in moderately affected UC colon; the differences were nonsignificant (Figure 4A). The total area of SP-immunoreactive nerve fibers was 9.68 ± 1.53 in normal colon, 9.16 ± 2.98 in least affected colon, and 11.85 ± 2.77 in moderately affected colon (Figure 4B). The differences were nonsignificant. The total intensity of substance P-immunoreactive fibers was 134 ± 22 in normal colon, 128 ± 44 in least affected colon, and 157 ± 38 in moderately affected colon, the differences being nonsignificant (Figure 4C). The average perimeter of a single nerve fiber profile was 8.1 ± 0.6 in normal colon, 8.1 ± 0.5 in least affected colon, and 7.3 ± 0.5 in moderately affected colon (n.s.) (Figure 4D).
The average intensity of substance P immunoreactivity in myenteric ganglion was 75.8 ± 1.5 in normal colon, 78.1 ± 1.2 in least affected colon, and 77.1 ± 1.2 in moderately affected colon. The differences between the groups are nonsignificant.
Distribution of substance P, VIP, CGRP, somatostatin, enkephalin, and galanin in normal and UC colon specimens was estimated by visual examination and semiquantitative analysis. The substance P immunoreactivity was increased in UC colon, whereas no difference between control and UC colon was observed for the other peptides. Therefore the quantitative analysis was limited to substance P.
The average cumulative area of NSE-immunoreactive profile of the myenteric ganglia per unit perimeter was 1.39 ± 0.21 in normal colon, 1.86 ± 0.17 in least affected colon, and 1.88 ± 0.15 in moderately affected colon (Figure 6). The differences between the groups were nonsignificant.
Discussion
Previous studies from our (Keränen et al. 1995) and other laboratories (Watanabe et al. 1998) have demonstrated an increased density of substance P-immunoreactive innervation in mild to moderate UC. The present data were obtained from a novel set of specimens and thus confirm the previous results. Furthermore, owing to improved morphometric methodology, we were able to demonstrate that the increase in substance P immunoreactivity is proportional to the degree of inflammation when control specimens and least and moderately affected UC specimens are compared. The effect of UC on substance P immunoreactivity may be even greater than indicated by the present material, because the control patients did not receive anti-inflammatory medication. In consecutive sections stained for GAP-43 or any of the other neuropeptides studied, no corresponding differences were observed. Because GAP-43 is the most sensitive general marker of nerve fibers in the human gut (Vento and Soinila 1999), our results suggest that the increase in nerve density is specific for substance P-containing nerve fibers rather than representing general neuronal sprouting. The density of GAP-43-immunoreactive nerve fibers remaining unchanged in control and UC specimens speaks against the possibility that the increase in substance P fibers is due to tissue shrinkage caused by UC.
Cumulative area of NSE-immunoreactive ganglion profiles of myenteric plexus per unit perimeter (mean ± SEM).
The age difference between the control and UC patients is an unavoidable problem both in the present and all previous studies comparing whole-wall specimens (Bishop et al. 1980; Koch et al. 1987; Mantyh et al. 1989, 1995; Kimura et al. 1994; Keränen et al. 1995; Singer et al. 1996; Watanabe et al. 1998). This is because the control specimens are obtained from the peripheral (histologically normal) regions of resected gut of cancer patients, who are always considerably older than UC patients. The difference raises the possibility that an age-related decrease in the density of gut innervation might be responsible for the greater number of substance P-immunoreactive nerve fibers in UC specimens reported in our previous study (Keränen et al. 1995). Although the number of myenteric neurons decreases with age (Gomes et al. 1997), no information is available on the density of nerve fibers. Significant age-related change in the nerve density should be revealed by the ideal general neuronal marker GAP-43 (Vento and Soinila 1999). Consequently, unchanged density of GAP-43-immunoreactive nerves speaks against the possibility that age would explain a greater density of substance P-immunoreactive nerves in UC patients. This conclusion is further supported by the observation that no changes in substance P or GAP-43 were seen in the innervation of the muscular layer.
The density of substance P-immunoreactive nerve fibers decreases in severe UC (Kimura et al. 1994; Watanabe et al. 1998). Similar changes occur in VIP-immunoreactive nerves (Kubota et al. 1992; Kimura et al. 1994). This is in line with the present observation on lack of substance P-immunoreactive nerves in severely affected UC specimens. Lack of substance P-containing nerves correlated with subtotal disappearance of GAP-43-immunoreactive nerves in the lamina propria. In the same specimens, no reduction in substance P- or GAP-43-immunoreactive nerve fibers were observed by visual examination in the muscle layers. Taken together, these findings suggest that disappearance or a strong decrease in neuropeptide-containing nerves of the lamina propria in severe UC is simply due to complete destruction of the mucosal innervation at the extreme stage of the disease.
No previous reports have been published on GAP-43 in UC. This protein is the best available marker of human gut nerves (Vento and Soinila 1999). In addition to being a general neuronal marker, GAP-43 expression is induced on nerve sprouting (Benowitz and Routtenberg 1997). Our observation that GAP-43 immunoreactivity remained unchanged suggests that mild to moderate UC does not induce general proliferation of the enteric nerves. Therefore, the increase in substance P immunoreactivity in the lamina propria is likely a result of increased synthesis of substance P, so that more substance P-immunoreactive fibers become visible.
Several parameters describing the innervation of the circular muscle layer, such as the density and thickness of substance P-immunoreactive nerves or the intensity of the immunoreaction, remained unchanged in UC compared to controls. The number of substance P-containing nerves of the circular muscle is not affected by extrinsic denervation (Holzer et al. 1980; Costa et al. 1981; Malmfors et al. 1981; Ekblad et al. 1987), indicating that these nerves are exclusively derived from gut ganglia. Our finding therefore suggests that no marked changes in the intrinsic nervous system are induced by UC. This conclusion is substantiated by lack of changes in specimens stained for GAP-43, which reveals the entire nerve fiber population (Vento and Soinila 1999). Therefore, UC does not appear to induce significant quantitative changes in muscular innervation. This contrasts with previous findings on thickened nerve fibers in the muscle layer of Crohn's disease patients (Bishop et al. 1980; Sjölund et al 1983) suggesting some differences in the pathogenesis of these two states of inflammatory bowel disease.
The values for mean nerve fiber perimeter were greater in GAP-43-stained specimens compared to substance P-stained ones. This is due to the fact that substance P is localized primarily in storage vesicles that accumulate in terminal varicosities, resulting in a pearl chain-like staining pattern. In contrast, GAP-43 is a neuronal membrane protein and consequently the staining results in an evenly thick, fibrous appearance, which is accentuated by GAP-43 staining of nonmyelinating Schwann cells and satellite cell of the enteric nerves (Curtis et al. 1992).
To examine whether UC is associated with increased size of myenteric plexus ganglia, we measured the cumulative area of ganglion profiles per unit perimeter in sections stained for NSE, an ideal marker for neuronal somata (Vento and Soinila 1999). The need for these measurements was suggested by previous studies reporting a two- to threefold increase in the number of myenteric ganglion cells in both Crohn's disease (Davis et al. 1955) and UC (Storsteen et al. 1953). In our study, no evidence for increased ganglion size was obtained. The present method does not exclude the possibility of increased number of glial cells.
Substance P-immunoreactive fibers in the enteric nervous system originate in both the intrinsic (gut ganglia) and extrinsic neurons (primary sensory neurons in the dorsal root ganglia). The intrinsic neurons are quantitatively the most important source of the substance P-containing neurons projecting to mucosal and submucosal blood vessels, mucosal epithelium, and the muscle layers, and providing interganglionic connections. The extrinsic neurons connect to the gut via sympathetic (splanchnic and hypogastric) and sacral parasympathetic (pelvic) nerves and project to the submucosal and mucosal blood vessels (for review see Holzer and Holzer–Petsche 1997). The present observations raise the question of which component is responsible for the increased number of substance P fibers. Although unchanged substance P immunoreactivity in the gut ganglia does not rule out the possibility of increased synthesis, this is unlikely because immunoreactivity for substance P in the circular muscle, another target of intrinsic neurons, was unaffected by UC. A more likely possibility is that the fibers showing increased substance P immunoreactivity are extrinsic sensory fibers. Other inflammatory processes, such as psoriasis, arthritis, and asthma involve increased substance P immunoreactivity of sensory nerves (Levine et al. 1985; Naukkarinen et al. 1989; Ollerenshaw et al. 1991). Unfortunately, we could not address this question directly in our material because the neurochemical indentification of sensory nerves, i.e., co-localization of substance P and CGRP, does not hold for human gut (Gattuso et al. 1996; Sjölund et al. 1997).
In conclusion, we consider that the increased density of substance P fibers in UC is due to increased peptide synthesis and consequently to increased visibility of the fibers by immunohistochemistry. Sprouting caused by UC cannot be excluded, although it is unlikely because it should show as an increased immunoreactivity of GAP-43 (Benowitz and Routtenberg 1997). Other possible mechanisms include increased axon transport, redistribution of substance P within the terminal, or modulation by neutral endopeptidase (Sturiale et al. 1999).
Footnotes
Acknowledgements
Supported by grants from the Foundation of Gastroenterological Diseases and the Sigrid Jusélius Foundation.
We wish to thank Mr H. Mustonen, MSc, for expertise in statistical analysis, Ms H. Wennäkoski for expert technical assistance, and Mr R. Karppinen for photography.