Immunohistochemical analysis of myenteric ganglia and interstitial cells of Cajal in ulcerative colitis

Abstract Ulcerative colitis (UC) is an inflammatory bowel disease with alterations of colonic motility, which influence clinical symptoms. Although morpho-functional abnormalities in the enteric nervous system have been suggested, in UC patients scarce attention has been paid to possible changes in the cells that control colonic motility, including myenteric neurons, glial cells and interstitial cells of Cajal (ICC). This study evaluated the neural-glial components of myenteric ganglia and ICC in the colonic neuromuscular compartment of UC patients by quantitative immunohistochemical analysis. Full-thickness archival samples of the left colon were collected from 10 patients with UC (5 males, 5 females; age range 45–62 years) who underwent elective bowel resection. The colonic neuromuscular compartment was evaluated immunohistochemically in paraffin cross-sections. The distribution and number of neurons, glial cells and ICC were assessed by anti-HuC/D, -S100β and -c-Kit antibodies, respectively. Data were compared with findings on archival samples of normal left colon from 10 sex- and age-matched control patients, who underwent surgery for uncomplicated colon cancer. Compared to controls, patients with UC showed: (i) reduced density of myenteric HuC/D+ neurons and S100β+ glial cells, with a loss over 61% and 38%, respectively, and increased glial cell/neuron ratio; (ii) ICC decrease in the whole neuromuscular compartment. The quantitative variations of myenteric neuro-glial cells and ICC indicate considerable alterations of the colonic neuromuscular compartment in the setting of mucosal inflammation associated with UC, and provide a morphological basis for better understanding the motor abnormalities often observed in UC patients.


Introduction
Ulcerative colitis (UC) is a chronic inflammatory bowel disorder associated with functional changes, including abnormal colonic motility and transit, which may contribute to abdominal pain and diarrhoea [1][2][3] experienced by patients with active disease or in remission [4]. These motor disturbances are suggestive of alter-ations of colonic neuromuscular components, including enteric neurons, interstitial cells of Cajal (ICC) and smooth muscle cells, whose structural and functional integrity are essential requirements to ensure physiological patterns of coordinated motor activity in the gut [5][6][7][8].
Although colonic dysmotility in UC patients has been well established, scarce attention has been paid to underlying alterations of the anatomical structures involved in the control of large bowel neuromuscular activity. Furthermore, heterogeneous methodological approaches, differing mainly in terms of histological techniques (e.g. orientation and sectioning, ganglionic cell counting, etc.), have been previously employed in evaluating the enteric nervous system (ENS)  Although some efforts have been previously made to obtain reliable quantitative estimations of ganglionic cells and ICC, careful morphological examinations and development of standardized protocols are still particularly required in the field of gastrointestinal neuromuscular pathology, in order to overcome the heterogeneity of available data [9][10][11]. Based on the above considerations, and following the recommendations issued by the International Working Group on Gastrointestinal Neuromuscular Disease [9,10], we designed the present study, which was conducted on left colonic samples from UC patients and control patients, with the purpose of performing an accurate and standardized quantitative immunohistochemical analysis of the neuralglial components of myenteric ganglia and ICC populations in this gut region. Gaining knowledge in this setting is critical for a better definition of mechanisms underlying colonic dysfunction in patients with UC.

Patients and tissue samples
The study was carried out on full-thickness archival samples of left (descending and sigmoid) colon obtained from 10 patients (5 males, 5 females; age range 45-62 years) with UC, who had undergone elective bowel resection due to left-sided colitis from the dentate line to the splenic flexure, lasting over 5 years. All patients had been scheduled for surgical intervention owing to a persistent condition of refractoriness to immunosuppressant therapy and/or steroid dependence. The investigation was focused on the left colon for two reasons: (1) normative values from otherwise normal left-side colon have been previously published by our group [12]; (2) to minimize inter-individual variability when comparing data from different segments of the colon. Care was taken to select areas including teniae with macroscopic involvement: the external surface appeared normal or slightly contracted, the mucosa diffusely congested, granular and haemorrhagic with ulcers linearly distributed in particular at level of the attachment of teniae. Archival colonic samples from ten patients (5 males, 5 females; age range 42-60 years), who had undergone surgery for uncomplicated left colon cancer and without previous history of abdominal surgery, inflammatory bowel disease or intestinal obstruction, served as controls. Control samples were also selected from areas including teniae at least 10 cm away from any macroscopically visible lesion. Because the study was performed on archival material, no individual patient identification was involved, and no study-driven clinical intervention was performed, a simplified Institutional Review Board approval was obtained.
Routinely fixed and processed full-thickness colonic samples were serially cross-sectioned to obtain 10 m-thick sections with circular layer and myenteric ganglia cut longitudinally. Serial sections, 1/18 sections for a distance of 180 m to prevent counting the same neuron in adjacent sections, were evaluated by two independent observers for each selected colonic specimen. Before use, slides were deparaffinized, rehydrated and processed for routine haematoxylin and eosin staining or immunohistochemistry. The morphological features of the colon were assessed on haematoxylin and eosin stained specimens by the pathologists (D.C., M.C.): the microscopic analysis of samples obtained from UC patients confirmed the presence of classical mucosal/submucosal lesions consistent with UC as well as the occurrence of plexitis/ganglionitis in the myenteric plexus of all UC patients according to previously published data [13,14]. A normal morphology was detected in colonic samples from all control patients.

Immunohistochemistry
Myenteric neurons, glial cells and ICC were identified by anti-HuC/D, anti-S100␤ and anti-c-Kit antibodies, respectively. Antibody and reagent details are reported in Table 1

Identification and quantitative analysis of ganglion cells
To detect myenteric neurons and glial cells, a double immunolabelling with anti-HuC/D and -S100␤ antibodies was performed as previously described [12]. Briefly, after heat-induced epitope retrieval and non-specific binding block, the following reagents were applied sequentially: primary antibodies (anti-HuC/D and anti-S100␤), detection kit LSAB ® ϩ system-HRP and chromogen substrate, which consisted of DAB and nickel enhanced-DAB for HuC/D and S100␤, respectively.
Five cross-sections from one colonic specimen serially cut per patient (1/18 sections) were examined as previously reported [12]: the first section processed for immunohistochemistry was the 18th section after the haematoxylin and eosin slice examined by the pathologists. For each section, 30 non-adjacent randomly selected, microscopic fields, taken along the myenteric ganglionic ridge, were examined in order to count immunolabelled cells and estimate the respective ganglionic areas, which were expressed as a percentage of the respective microscopic field area. Samples were examined at 400ϫ magnification by a light microscope equipped with a digital camera (DFC480; Leica, Cambridge, UK), and ganglionic areas were estimated by highlighting them using an Image Analysis System Quantimet, Qwinv 500 plus (Cambridge Instruments, Cambridge, UK). A ganglionic area of at least 1.4 mm 2 was examined for each patient and it was expressed as percentage of the total neuromuscular tissue area examined (at least 10.35 mm 2 /each patient).
Neurons were counted when their nuclei and/or perikarya displayed HuC/D ϩ signals, in order to avoid underestimation of neuronal number. For each patient, the number of HuC/D ϩ neurons and S100␤ ϩ cells as well as the respective ganglionic area percentage were expressed as sums (⌺s) of values obtained from examination of five serial sections (30 fields/section). Cell density (cell number/mm 2 ) and glial cell/neuron ratio were calculated from the respective ⌺s values. Data from different patients or control patients were also averaged and expressed as mean Ϯ S.D.

Statistical analysis
Linear regression analysis, performed by commercial software (GraphPad Prism ® , version 3.0; GraphPad Software, Inc., San Diego, CA, USA), was employed to analyse correlation between ganglionic areas and the respective cell counts estimated for each slice within each patient and in different patients. Student's t-test for paired data (two tailed) was performed to assess the statistical differences between groups. A P-value Ͻ0.05 was considered statistically significant.

Routine light microscopy
In control patients, neurons and associated glia formed a homogeneous cluster of cells aggregated in well visualizable and readily detectable myenteric ganglia showing normal appearance (Fig. 1A). In the colon of UC patients, myenteric ganglia showed several empty lacunar spaces, associated with depletion of neuronal cell bodies, and some ganglia were 'ectopic', being embedded within the CM or LM layers, outside from their typical location (i.e. the MP) [16]. Signs of myenteric ganglionitis/ plexitis were observed in all UC samples and consisted of an inflammatory infiltrate (Ͻ4 cells) including eosinophils, lymphocytes or, occasionally, plasma cells adjacent to or within myenteric ganglia or nerve bundles (Fig. 1B).

c-Kit ؉ cells: ICC and mast cells A marked c-Kit immunoreaction was detected in the ICC of neuromuscular layer and mast cells (used as internal controls) (Fig. 3A). In colonic specimens from control patients, c-Kit immunoreactive ICC appeared as dendritic cell bodies typically located within the CM (ICC-CM) and LM (ICC-LM) layers as well as in the MP (ICC-MP) (Figs 3, 4A, D, G and 5A). Intramuscular ICC were spindle-shaped cells with long bipolar processes mostly running parallel to the axis of smooth muscle cells (Fig. 3B). In the intermuscular region, ICC-MP appeared as multipolar cells and formed a network encasing the myenteric ganglia (Figs 3C and 4D). In colonic samples from UC patients, the overall morphological appearance and distribution pattern of ICC were markedly altered: the c-Kit ϩ network was rarefied and often disrupted in each layer, and most of ICC displayed considerable morphological alterations with blunted and shortened branching (Figs 4B, E, H and 5B).
Only the c-Kit ϩ spindle-and/or dendritic-shaped cells with a nuclear fast red counterstained nucleus (Fig. 4) (Table 3). Quantitative analysis of the area covered by ICC was performed on c-Kit immunostained slides without counterstaining (Fig. 5A, B). A significant decrease in ICC area was observed in UC patients, in whom ICC covered an area corresponding to 0.62 Ϯ 0.18%, as compared to 3.49 Ϯ 0.94% detected in control patients (Fig. 5C). [1,3]. This carefully regulated neural/myogenic network is markedly altered in the setting of bowel inflammation, resulting in gut dismotility [5,17]. However, despite the increasing incidence of UC and its debilitating symptoms, few studies have evaluated the occurrence of morphological alterations in the anatomical structures regulating colonic motility. Previous histopathological data suggested that the ENS and ICC can be targeted by the inflammatory disease in patients with UC [14,17,18]. However, there is also evidence suggesting that a primary damage to the enteric glia can cause inflammatory bowel disease, leading to secondary alterations of enteric neurons [19]. The present study was performed to examine the cell composition of myenteric ganglia and ICC distribution in the colon of UC patients in accordance with recently identified technical and methodological criteria of crucial importance for a correct interpretation of neuronal-glia-ICC data [9,10].

Enteric neurons and glial cells, together with ICC, represent the main regulators of motor functions in the gut wall, ensuring coordinated patterns of smooth muscle activity
Significant quantitative changes in both myenteric ganglia and ICC were detected by our analysis. Myenteric ganglia were markedly altered in all colonic samples from UC patients. In accordance with previous reports [13], there was a ganglionic vacuolization associated with depletion of neurons and glial cells and a minimal inflammatory infiltration. In particular, the density of HuC/D ϩ neurons was significantly decreased by about 61% and S100␤ ϩ glial cells by about 38%. These data are at variance with those by Villanacci et al. [18], who showed that in UC the number of myenteric neurons did not change significantly. However, this discrepancy can be explained by several reasons: (1) the enrolled patients were younger than those included in our study (i.e. 35 patients (A, D, G) and UC patients (B, E, H). In colonic specimens from UC patients, few ICC are detectable and they appear markedly altered: most of them exhibit blunted and shortened branching (B, E, H). was referred to microscopic field areas. In contrast, in our study, HuC/D was preferred over other pan-neuronal markers, as it labels selectively the neural cell bodies, thus increasing the accuracy of quantitative analysis by avoiding confounding factors such as neuropil and nerve endings [10,12,[20][21][22], and it has been included among the neuronal markers recommended for immunohistochemical studies of gastrointestinal neuromuscu-lar diseases [9]. Despite these advantages, HuC/D is a marker which stains one single protein, and this limitation may result in inflated estimates of enteric neuronal loss, particularly in aged tissues, as suggested by Phillips et al. [23], who studied enteric neurons from aged rats. In accordance with these findings, different degrees of HuC/D labelling were recorded also in myenteric neurons of the patients evaluated both in the present study and a previous one from our group [12]. Although the sensitivity of HuC/D immunostaining, extensively tested together with the panneuronal cuprolic blue, is considered to be equivalent to cuprolic blue for enteric neuronal quantification [12,21,22,24], we cannot certainly rule out that some neurons failed to express the HuC/D protein due to aging. According to the present findings, the glia/neuron ratio ranged from 3.72 to 4.19, a value lower than that estimated in previous reports (5.9-8.7) [25][26][27]. Again, such apparent discrepancies might depend on different criteria for cell counts. The present analysis included also neurons with HuC/D ϩ cytoplasm without a visible nucleus and glial cells specifically labelled for S100␤ antigen, whereas in previous studies ganglionic cells were identified either as neurons when clearly nucleated, or as glial cells when endowed with small basophil nuclei.

Fig. 4 Evaluation of distribution, morphology (A, B, D, E, G, H) and counting (C, F, I) of c-Kit immunolabelled and nuclear fast red-counterstained ICC (black) at the level of CM (ICC-CM) and LM (ICC-LM) layers, and the myenteric plexus region (ICC-MP) of full-thickness human left colon in control
In our study, correlations between myenteric cell counts and the respective ganglionic areas were found for each control patient, but not for all UC patients. In particular, significant correlations were obtained only in two patients for neurons and seven patients for glial cells. In addition, the overall regression analysis of cell density for each group yielded highly significant values for neurons and glial cells in the control group, whereas few significant values, in particular for neurons, were obtained in UC patients. Taken together, these data suggest that neuronal rather than glial cell density was more affected in the UC patients herein examined. The present findings raise the important question of whether the neuronal loss in UC involves indiscriminately all myenteric neurons or selectively affects specific subsets. In a previous study on UC, Neunlist et al. [28] found a neurochemical phenotype shift from acetylcholine to substance P. However, when studying trinitrobenzene sulphonate (TNBS)-induced colitis in guinea-pigs, Linden et al. [29] observed an indiscriminate loss of myenteric neurons. Overall, this is a very important issue, which has not been addressed in the present study and remains open to future investigations. ENS abnormalities, underlying gut motility disorders, are often associated with altered density, distribution and/or morphology of ICC [30]. Accordingly, we found marked morphological and quantitative alterations of ICC in all UC patients. Indeed, all ICC types (ICC-CM, ICC-LM and ICC-MP) were significantly decreased by over 50%, as compared to control patients. These results, which are at variance with previous reports showing an increased number of intramuscular ICC and unchanged ICC-MP in UC colon [14,18] displayed blunted and shortened branching. These morphological alterations, together with reduction in the number of these cells, might contribute to a malfunctioning network. An attempt to assess ICC structural abnormalities in the colon of UC patients has been previously made by Rumessen [8] who, however, focused his evaluations only on submuscular ICC. Using electron microscopy, this author found that ICC had large intracytoplasmatic bodies. Briefly, our data do not allow explaining whether ICC abnormalities in UC patients result from direct effects of inflammatory mediators [14,17,18], secondary changes reflecting primary glial damage [19] or inflammatory-related remodelling processes, thereby leading to alterations in the mechanical properties of colonic wall.
Whether mast cells play also an active role in the pathophysiology of inflammation in UC remains to be determined. This is a relevant issue, because in the human gut mast cells are involved in the maintenance of intramuscular ICC viability, through paracrine production of the stem cell factor, which acts as a ligand for the c-Kit receptor [31]. In addition, mast cells can contribute to fibrogenesis and neuroplasticity [32] and possible functional interactions between mast cells and ICC have been suggested in intestinal muscularis externa [33]. However, these cells have been also found to reduce myenteric neuron survival in culture [34]. In the present study, we observed increased numbers of mast cells within the colonic neuromuscular compartment. These findings are in line with previous studies showing a significantly increased mast cell density in the colonic muscle layer of UC patients [32].
In conclusion, the present findings provide a thorough morphological and quantitative analysis showing changes in myenteric neurons, glial cells and ICC in the left colon of patients with UC. Because these cells play critical roles in the regulation of colonic functions, including motility, our data on enteric neuronalglia-ICC network abnormalities shed light on the mechanisms underlying gut dysfunctions observed in the subset of UC patients with long-lasting chronic disease.