Gianrico Farrugia MD, Division of Gastroenterology and Hepatology, Mayo Clinic, 200 First Street SW, Rochester, MN 55905, USA. Tel: +1 507 284 4695; fax: +1 507 284 0266; e-mail: firstname.lastname@example.org
Abstract The objective of this study was to determine the distribution of enteric nerves and interstitial cells of Cajal (ICC) in the normal human appendix and in type 1 diabetes. Appendixes were collected from patients with type 1 diabetes and from non-diabetic controls. Volumes of nerves and ICC were determined using 3-D reconstruction and neuronal nitric oxide synthase (nNOS) expressing neurons were counted. Enteric ganglia were found in the myenteric plexus region and within the longitudinal muscle. ICC were found throughout the muscle layers. In diabetes, c-Kit positive ICC volumes were significantly reduced as were nNOS expressing neurons. In conclusion, we describe the distribution of ICC and enteric nerves in health and in diabetes. The data also suggest that the human appendix, a readily available source of human tissue, may be useful model for the study of motility disorders.
Diabetic gastroenteropathy is a motor and sensory disorder of the gastrointestinal tract that occurs in a subset of patients with diabetes. Both type 1 and type 2 diabetics can be affected.1,2 The stomach is often affected in patients with diabetic gastroenteropathy, leading to delayed gastric emptying, known as diabetic gastroparesis, or in some cases to accelerated gastric emptying.3,4 Diabetic gastroenteropathy can be a severe disease and affects predominantly young adults resulting in considerable long-term morbidity.5 The symptom profile of diabetic gastroenteropathy includes nausea, vomiting, early satiety, constipation, diarrhoea, abdominal pain and bloating. These symptoms may occur alone or in combination.6 The diverse nature of the symptoms suggests a multifactorial cause for diabetic gastroenteropathy.
The control of gastrointestinal motility is regulated by several cell types including smooth muscle cells, endocrine cells, interstitial cells of Cajal (ICC), extrinsic and intrinsic neurons. ICC are important regulators of gastrointestinal motility as they generate the pacemaker signals,7,8 relay neuronal input to smooth muscle cells,9,10 set the smooth muscle membrane potential gradient,11 and also function as mechanosensors.12 Diabetic gastroenteropathy is often the result of damage to several cell types. There is evidence for the presence of an extrinsic (autonomic) and intrinsic (enteric) neuropathy, evidence for the role hyperglycaemia can play in development of cellular damage and also evidence for the role of autoimmune deregulation13 in diabetes. There is increasing evidence pointing towards disrupted or disordered ICC and loss of neuronal nitric oxide synthase (nNOS) playing a central role in the pathophysiology of diabetic gastroenteropathy. This hypothesis has strong support in the experimental literature however, human evidence for it remains sparse. A limitation of human studies has been the inability to collect and process specimens prospectively due to the scarcity of appropriate human tissue. The non-inflamed appendix is an organ that is often removed during abdominal surgery for other indications to prevent future complications. Thus, the appendix is also removed by some transplant centres during pancreatic transplant for treatment of diabetes. The appendix may therefore serve as a useful source of both normal and diabetic human tissue that can be collected prospectively to study diabetic gastroenteropathy. However, it is unknown whether ICC are present in the human appendix. Further, the effects of diabetes on ICC and on the appendiceal enteric nervous system have not been investigated systematically. The aim of this study was therefore to prospectively collect non-inflamed appendixes from both non-diabetic and diabetic patients and determine the normal distribution of ICC and changes, if present, in ICC and enteric neurons in diabetic patients.
Materials and methods
Procurement of human tissues was approved by the Institutional Review Boards of the Mayo Clinic and the Cleveland Clinic. Appendixes from patients with type 1 diabetes were obtained from six patients undergoing pancreatic transplant surgery at the Cleveland Clinic. All patients had long-standing diabetes for at least 22 years (Table 1). Control appendices from non-diabetic patients were obtained from six patients at the Mayo Clinic who were undergoing abdominal surgery and had their appendix removed prophylactically (Table 1). We also used sections from previously banked jejunal tissue from a patient who underwent gastric-bypass tissue to compare the distribution of enteric nerves in the appendix with the jejunum. These sections were stained as outlined below for the appendix.
Table 1. Demographics of the patients
Duration of diabetes (years)
A piece from each diabetic appendix was immediately snap frozen in the operating room and shipped on dry ice to the Mayo Clinic. Another piece from each diabetic appendix was fixed in 10% formalin and processed for paraffin embedding. Pieces of non-diabetic appendix were immediately placed in ice-cold minimal essential medium and within 1 h were fixed overnight in 4% paraformaldehyde solution, rinsed in phosphate-buffered saline (PBS), incubated overnight at 4 °C in PBS containing 30% sucrose and then frozen in isopentance cooled to −50 °C. Serial full thickness transverse sections, 12-μm thick, were cut from the frozen pieces of diabetic and non-diabetic tissues using a cryostat. Every third section was collected and thaw-mounted onto glass microscope slides until a total of 20 sections were collected from each tissue. The slides were next air-dried at room temperature and used immediately, or stored at −20 °C for up to 1 month before using. Six sections (10-μm thick) were also cut from each paraffin-embedded diabetic appendix and collected onto glass microscope slides. Each section contained the entire cross-sectional area of the appendix.
ICC and nerves were detected by an indirect immunofluorescence technique using a rabbit polyclonal anti-c-Kit antibody (1 : 200; Medical & Biological Laboratories, Nagoya, Japan) and a rabbit polyclonal anti-PGP 9.5 antibody (1 : 1000; Biogenesis, Kingston, NH, USA), respectively, as previously described.14 Three sections from each tissue were stained for ICC and three sections from each were stained for PGP 9.5. The sections of fresh frozen diabetic appendices were fixed in 4% paraformaldehyde at 4 °C for 20 min. before immunostaining. PGP 9.5 immunostaining was also performed on three paraffin sections from each diabetic appendix. The sections were deparaffinized and rehydrated before immunostaining.
NADPH-diaphorase staining for nNOS
NADPH-diaphorase (NADPH-d) histochemistry, a marker for nNOS-containing neurons in the human gastrointestinal tract14,15 was performed on three cryostat sections from each diabetic and non-diabetic appendix. The diabetic appendix sections were fixed in 4% paraformaldehyde for 20 min. prior to NADPH-d staining. To stain for NADPH-d, sections were rinsed in PBS, rinsed in 0.1 mol L−1 Tris, pH 7.6 containing 0.3% Triton X-100, incubated at 37 °C for 10–15 min. in the same solution containing 0.2 mg mL−1 of nitroblue tetrazolium and 0.5 mg mL−1 of β-NADPH (Sigma Chemicals, St Louis, MO, USA) and then rinsed in PBS and coverslipped.
Analysis was performed by an investigator who was blinded as to the status of the patient from which the tissue was obtained. NADPH-d stained sections were examined by bright field microscopy. The number of NADPH-d stained neuron cell bodies was counted in each section. Immunostained sections were examined with by epifluorescence and laser scanning confocal microscopy (LSM 510; Zeiss, Thornwood, NY, USA). ICC and nerves were quantitated from confocal microscope images as previously described.16,17 The sections were divided into four quadrants and a randomly selected area of the longitudinal muscle layer and an area of the circular muscle layer in every quadrant were imaged by confocal microscopy using a X40 water immersion objective. In each area, a stack of 10–15 optical slices (230 μm by 230 μm in x–y dimension) was made at 0.8-μm intervals through the thickness (z-axis) of the section. The volumes of c-Kit immunopositive ICC and PGP 9.5 immunopositive nerves were calculated using ANALYZE format software (Mayo Foundation, Rochester, MN, USA) as previously described in detail.18 Mast cells, also c-Kit positive, were identified based on their characteristic size and shape and excluded from the analysis of c-Kit volumes. For each section, the ICC and PGP 9.5 volumes in the quadrants were averaged for the longitudinal and for the circular muscle layers.
All data analysis was carried out on the raw non-normalized data. Statistical analysis was performed using anova. A P-value of <0.05 was considered significant. All data for these experiments are expressed as the mean ± SEM.
Table 1 summarizes the patient data from which the diabetic appendices were obtained. All had insulin requiring type I diabetes with a mean duration of 27 years and all had at least one complication of the disease. Five of the six had gastrointestinal symptoms and in one the presence or absence of gastrointestinal symptoms was not documented. The normal appendices were age (within 10 years) and sex matched and none had any gastrointestinal symptoms at the time of surgery.
Distribution of nerve cell bodies and fibres in the normal appendix
Fig. 1 shows the muscle layers of the human appendix as shown by haematoxylin and eosin staining of a cross-section through the appendix. As previously described, the appendix contained both a longitudinal and circular muscle layer.19 The mean thickness of the circular muscle layer was 484 ± 13 μm and of the longitudinal muscle layer, 379 ± 11 μm. Immunostaining for PGP 9.5 showed the presence of nerve cell bodies in myenteric ganglia between the longitudinal and circular muscle layers. However, in contrast to other regions of the human gastrointestinal tract, the myenteric plexus of the appendix was less well-defined (Fig. 2). Unlike the human stomach, small intestine (Fig. 2) or colon, the appendix contained numerous ganglia (defined as containing two or more neurons) in the longitudinal muscle layer (Fig. 2). Also, individual neurons were present in both the circular and longitudinal muscle layers, as previously reported.20 Similar to the rest of the human gastrointestinal tract, nerve fibres were numerous in both muscle layers (Fig. 2). A deep muscular plexus (present in the small intestine) and a submucosal plexus (present in the colon) were not identified in the appendix. NADPH-d showed that nNOS expressing neurons were present in the myenteric plexus as well as in the circular and longitudinal muscle layers (Fig. 3, Table 2). nNOS-containing nerves were also present in both muscle layers (Fig. 3).
Table 2. NADPH-diaphorase positive neurons in diabetic and control appendixes (36 cryostat sections from six appendixes per group)
NADPH-d positive neurons/section
Control (mean ± SEM)
Diabetic (mean ± SEM)
NADPH-d, NADPH-diaphorase. *P < 0.001.
0.4 ± 0.2
0.1 ± 0.08
23.2 ± 2.4
9.3 ± 1.3*
Longitudinal muscle ganglia
21.9 ± 2.6
10.1 ± 2.1*
Circular muscle (single neurons)
13.6 ± 1.8
5.1 ± 1.1*
Longitudinal muscle (single neurons)
0.5 ± 0.2
0.1 ± 0.07
59.7 ± 5.3
24.4 ± 3.9*
Distribution of ICC in the normal appendix
ICC were present in both the circular and longitudinal muscle layers. The distribution of intramuscular c-Kit positive ICC was similar to the small intestine and colon. The per cent volume occupied by the intramuscular ICC was greater in the circular compared to the longitudinal muscle layer (Fig. 3). In contrast to the body of the human stomach, the small intestine and colon, there was neither a well-defined myenteric network of ICC nor a deep muscular or submucosal plexus of ICC in the appendix.
Distribution of nerve cell bodies and fibres in the diabetic appendix
The mean thickness of the circular muscle layer was 433 ± 15 μm and of the longitudinal 340 ± 12 μm. Both layers were less thick (P < 0.05) than in the non-diabetic appendix. Similar to the findings in the normal appendix, neurons were present within both the longitudinal and circular muscle layers and in the myenteric region. Given the lack of a well-defined myenteric plexus region, with ganglia and individual neuronal cell bodies present in both circular and longitudinal muscle layers, the volume of PGP 9.5 positive structures was initially determined for the full thickness of the muscle wall. No difference was noted between diabetic appendixes and controls (7.72 ± 0.84%vs 9.09 ± 0.91% PGP 9.5 positive neuronal structures by volume, P > 0.05). When the circular muscle layer and the longitudinal muscle layer were analysed separately, no difference was noted for the circular muscle layer from diabetic appendixes compared to controls (6.26 ± 0.79%vs 6.91 ± 0.74% PGP 9.5 positive neuronal structures, P > 0.05) but a significant decrease in PGP 9.5 positive neuronal structures was present in the longitudinal muscle of diabetic appendixes compared to controls (1.45 ± 0.22%vs 2.18 ± 0.26% PGP 9.5 positive neuronal structures, P < 0.05). The number of nNOS-containing neurons in the diabetic appendixes was significantly decreased compared to controls for both myenteric ganglia and ganglia within the longitudinal muscle layer (Table 2). Similarly, the number of solitary nNOS-containing neurons in the longitudinal and circular muscle layers was decreased in the diabetic appendixes (Table 2).
Interstitial cells of Cajal in the diabetic appendix
The volume of c-Kit positive ICC in the diabetic appendix was significantly reduced compared to the normal appendix (Fig. 4). The volume of c-Kit positive ICC in the circular muscle of the normal appendix expressed as a percentage of the total volume was 3.4 ± 0.2% compared to 2.47 ± 0.15% in the appendix from diabetic patients (12 stacks per sample, n = 6, P < 0.01, Fig. 5). The volume of c-Kit positive ICC in the longitudinal muscle of the normal appendix expressed as a percentage of the total volume was 2.59 ± 0.16% (Table 3). This was not statistically different when compared to values obtained from appendixes from diabetic patients (2.34 ± 0.15%, P > 0.05, Table 3).
Table 3. Per cent volume occupied by interstitial cells of Cajal (expressed as percentage of total volume of tissue) in sections of control and diabetic human appendices
*P < 0.01.
2.59 ± 0.16
2.34 ± 0.15
3.40 ± 0.20
2.47 ± 0.15*
In this study, we found both similarities and differences between the distribution of ganglia and individual neurons in the human appendix and the rest of the human gastrointestinal tract. The two main plexi present in the rest of the human gastrointestinal tract, the myenteric and submucosal plexi were both present, however, there were also several ganglia present in the longitudinal muscle layer and individual neurons present in both muscle layers. The distribution of nerve fibres in the circular and longitudinal muscle layers was similar to other regions of the human gastrointestinal tract. The presence of ganglia outside of the boundary between the circular and longitudinal muscle has been previously reported in both the non-English and, more recently, in the English literature.20 The distribution of ICC also had both similarities and differences with the rest of the human gastrointestinal tract. In the human colon ICC form two networks, the myenteric network and the submucosal network, and are also distributed in the circular and longitudinal muscle layers as single intramuscular ICC and along septa.17 While ICC were distributed in the muscle layers, with a greater number of ICC in the circular muscle layer, as in the rest of the human gastrointestinal tract,17 no distinct myenteric or submucosal networks of ICC were present, rather the density of ICC was fairly uniform across the muscle thickness including in the myenteric plexus region.
Another finding of this study was the loss of nNOS expressing neurons and c-Kit positive ICC in the human appendix taken from patients with severe type 1 diabetes. There is increasing evidence for several factors contributing to the development of symptoms associated with diabetic gastroenteropathy. A number of studies have shown vagal nerve abnormalities in animal models of chronic diabetes.21–24 Smooth muscle abnormalities have also been reported in both humans and in mouse and rat models of chronic diabetes. In severe diabetes, in both humans and small animal models a loss of smooth muscle cells and increased fibrosis was noted.25,26 In this regard, it was interesting to note that the muscle layers of the appendix from the patients with severe type 1 diabetes were thinner than from controls. Two cell types that have received increasing attention in the past years are nNOS expressing enteric nerves and ICC with several studies showing a decrease in nNOS expression or function associated with diabetes and or gastric emptying.16,27–33 Loss of nNOS expression in enteric nerves in diabetic gastroenteropathy may not necessarily mean that the enteric nerves themselves are also lost. A study in non-obese diabetic streptozotocin models of diabetes in mice, showed that development of diabetes was associated in a marked reduction in pyloric nNOS protein and mRNA. However, nNOS expression and pyloric function were restored to normal levels by insulin treatment suggesting that enteric nerves were not lost.34 In contrast, an increase in apoptosis of colonic myenteric neurons, dorsal root ganglion, and vagus nodose ganglion has been noted in streptozotocin-induced diabetic rats35 as well as a relatively specific loss of nitrergic neurons through apoptosis in the proximal colon of mice.36 While we did not look for apoptosis of enteric nerves, the data in this report suggest that both neuronal loss as well as loss of nNOS expression without loss of enteric nerves may occur in the diabetic appendix. There was a two- to fourfold reduction of nNOS-containing enteric neurons in the appendix from diabetic patients compared to controls, yet there was no difference in the volume of PGP 9.5 positive structures (nerve cell bodies and fibres) for the whole thickness of the muscle wall. This suggests that there was loss of expression of nNOS without neuronal loss. In the longitudinal muscle layer, the volume of PGP 9.5 positive structures was significantly decreased by about 33% compared with a more than 50% loss of nNOS expressing neurons suggesting that in the longitudinal muscle layer there may be both loss of nNOS expression and loss of enteric neurons.
Loss of or disrupted ICC networks have also been described in humans with diabetic gastroenteropathy and in animal models of diabetes.16,33,37–39 Most of the data from humans are retrospective, utilizing previously collected tissue from pathology data banks. While this approach often results in important advances in our understanding of a particular disease, it does have several limitations when applied to the study of ICC distribution and number. ICC are neither distributed uniformly in the gastrointestinal tract nor distributed uniformly in a particular organ. Therefore, the region that is chosen for study needs to be carefully matched to controls, often difficult when the area resected is not well documented, or when appropriately located control tissue is not available. Secondly, pathology tissue banks almost exclusively consist of paraffin-embedded tissue. While adequate for the study of enteric nerves, the study of ICC requires reactivation of the c-Kit antigen to visualize ICC. In the best of conditions, this results in about a 50% visualization of ICC when compared to non-paraffin fixation methods (G. Farrugia, unpublished observation). Thirdly, the time from resection to fixation is often variable. ICC are easily damaged by ischaemia/hypoxia and therefore minimization of this time period is essential when coming to conclusions on the number and distribution of ICC in human disease. This of course cannot be controlled in retrospective studies. Use of the appendix circumvents some of these limitations. The appendix is often removed, making it widely available and therefore prospective studies can be planned and the collected appendix processed quickly in the appropriate solutions and fixatives. Also, the distribution of both ICC and enteric nerves is uniform along the circumference and length of the appendix allowing appropriate comparisons to control tissue. There are however also limitations to the use of the appendix as a model for the rest of the gastrointestinal tract. These include the distribution of ICC, the lack of information on the electrical properties of the appendix and the fact that disease processes may affect the rest of the gastrointestinal tract but not the appendix. From the data presented in this report, it however does appear that in diabetes, at least in type 1 diabetes, a similar loss of nNOS-containing neurons and c-Kit positive ICC is present as in the rest of the gastrointestinal tract.
In summary, the distribution of enteric nerves in the human appendix had both similarities and differences to the rest of the gastrointestinal tract. Myenteric ganglia were present, however, they were less well-defined into myenteric and submucous regions compared to the stomach, small intestine or colon. Unlike the situation in other regions of the gastrointestinal tract, ganglia were numerous in the longitudinal muscle layer of the appendix. The distribution of ICC in the muscle layers of the human appendix was similar to that of other regions of the gut but there was not a well-defined network of ICC either at the myenteric region of the appendix or at the submucosal border. The volume of ICC was significantly decreased in the diabetic appendix compared to non-diabetic tissues. The number of nNOS-containing nerve cell bodies was also significantly decreased in the appendices of diabetic compared to non-diabetic patients. The appendix may serve as readily available source of human tissue and a useful model for the study of diabetic gastroenteropathy.
We thank Kristy Zodrow for secretarial assistance and Gary Stoltz for technical assistance. Supported by NIH grants DK68055 and DK57061.