Quantification of subclasses of human colonic myenteric neurons by immunoreactivity to Hu, choline acetyltransferase and nitric oxide synthase


Dr D. A. Wattchow, Department of Surgery, Flinders Medical Centre, Bedford Park, 5042, SA, Australia.
Tel: 08 8204 4253; fax: 08 8204 5843; e-mail: david.wattchow@flinders.edu.au


Abstract  An accurate method to count human enteric neurons is essential to develop a comprehensive account of the classes of nerve cells responsible for gut function and dysfunction. The majority of cells in the enteric nervous system utilize acetyl choline, or nitric oxide, or a combination of these, as neurotransmitters. Antisera raised against the RNA-binding protein Hu, were used to identify nerve cell bodies in whole mounts of the myenteric plexus of human colon, and then were utilized to analyse cells immunoreactive for combinations of choline acetyltransferase and nitric oxide synthase. Antisera to Hu provided a reliable means to count apparently all enteric nerve cell bodies, revealing 10% more cell bodies than labelling with neuron specific enolase, and no labelling of glial cells as revealed by S100. ChAT+/NOS− neurons accounted for 48% (±3%) of myenteric neurons and ChAT−/NOS+ neurons accounted for 43% (±2.5%). ChAT+/NOS+ neurons comprised 4% (±0.5) of the total number of neurons, and a novel class of small ChAT−/NOS− neurons, making up 5% (±0.9%) of all cells, was described for the first time.


The cell bodies of the different functional classes of enteric neurons (sensory, motor, vasomotor, secretomotor and interneurons) are intermingled in the submucous and myenteric ganglia of the gut wall. Considerable work in laboratory animals, and in humans, indicates that these functional subclasses can be distinguished, by the chemicals that they contain.1 A full, quantitative account has been developed for the guinea pig small intestine,2,3 and substantial descriptions of neurons exist for the guinea pig stomach,4 and large intestine,5 rat small and large intestine,6,7 and mouse small and large intestine.8 Whilst considerable progress has been made in identifying functional classes in the human myenteric plexus9–15 and submucous plexus,13,16 the accurate proportions of these classes remains to be determined.

To determine such proportions requires a reliable marker for all neurons against which other markers can be counted. A number of pan-neuronal markers have been used to label enteric neurons, including nicotinamide adenine dinucleotide (NADH),17 neuron specific enolase (NSE),18 and protein gene product (PGP) 9.5.19,20 NSE, and particularly PGP 9.5 antisera label large numbers of nerve fibres within ganglia, which often obscures the outlines of nerve cell bodies, making them hard to identify.21,22 A ‘nerve cell body antiserum’ that labelled cell bodies with minimal labelling of fibres has been used in the guinea pig gut,2,23 but in our experience labels human enteric neurones poorly. Cuprolinic blue has been suggested as an alternative method for labelling all neurons22,24–26 but is not readily combined with immunohistochemistry.

Anti-Hu antibodies were originally isolated from patients with malignancies, such as small cell lung carcinoma, who developed para-neoplastic peripheral neuropathies or encephalomyelitides.27,28 These antibodies bind specifically to Hu antigens, which are RNA-binding proteins present in neuronal cells.29,30 Anti-Hu labelling has been employed in guinea pigs,31,32 and reveals neuronal cell bodies with very few nerve cell processes,31 thus offering the prospect of being an ideal marker for human enteric neurons.

The aim of this study was to examine the utility of anti-Hu antisera in labelling all nerve cell bodies in whole mounts of human myenteric plexus, and determine the proportion of nerve cells reactive for the synthetic enzymes of acetylcholine and nitric oxide.

Materials and methods

Tissue collection

Specimens of fresh colon were taken from 10 patients who were undergoing surgery for colorectal adenocarcinoma (five women and five men: median age, 68 years: age range, 52–83 years). The specimen was taken from the uninvolved resection margin and was transported to the laboratory in oxygenated Kreb's solution. There were three specimens of transverse colon, five specimens of descending colon and two specimens of sigmoid colon. All patients signed a written consent form agreeing to donation of tissue for the purpose of research and experiments were performed according to the Declaration of Helsinki. The Flinders Medical Centre Committee on Clinical Investigation approved the use of human colon for these experiments.

Tissue preparation

From each specimen, an intertaenial segment of colon measuring 2 × 2 cm was fixed fresh, and another similar segment was cultured for 3 days to enhance immunoreactivity for the markers in neurons.

Fixation and dissection

Tissue was fixed by pinning out under tension in a Sylgard lined Petri dish (Dow Corning Corp., Midland, MI, USA) and immersing overnight in modified Zamboni's fixative (2% paraformaldehyde and 0.2% saturated picric acid in 0.1 mol L−1 phosphate buffer; pH 7.2) at 4 °C. After clearing in dimethyl sulfoxide (10 min immersion, repeated three times), tissue was washed in phosphate buffered saline (phosphate buffered saline (PBS); 0.2 mol L−1 sodiumphosphate buffer, pH 7.2) and a whole mount of the myenteric plexus and longitudinal muscle was prepared by removing the mucosa, submucosa and circular muscle with the aid of a dissecting microscope (the submucosal ganglia were not analysed in these experiments). All preparations were soaked in 100% glycerol for 2 days to aid penetration of antibodies prior to application of antisera.

Organotypic culture

The tissue was pinned out in a Sylgard lined Petri dish in sterile Kreb's solution at room temperature. The mucosa was dissected away from the submucosa in the plane between the muscularis mucosa and submucous plexus. After rinsing in sterile Krebs solution, the specimens were then covered with culture medium (DME/F12; Sigma Chemical Co., St. Louis, MO, USA) containing 10% fetal bovine serum, 100 IU mL−1 penicillin, 100 μg mL−1 streptomycin, 20 μg mL−1 gentamicin and 2.5 μg mL−1 fungizone (Cytosystems, Castle Hill, NSW, Australia) and incubated on a rocking tray at 37 °C in humidified air with 5% CO2, for 3 days. Medium was changed daily and after 3 days, the specimens were fixed as described above.


To compare labelling of myenteric neurons in fresh and cultured preparations antisera to NSE and Hu were applied for 2 days at room temperature then washed repeatedly in PBS. They were then incubated for a further 2 days in secondary antisera (donkey antimouse CY3 and donkey antirabbit fluorescein isothiocyanate; Jackson Immunoresearch Laboratories Inc., West Grove, PA, USA) (Table 1), then washed repeatedly in PBS and mounted in bicarbonate-buffered glycerol (pH 8.6).

Table 1.   Primary antisera
AntigenAntiserum codeHostDilutionReference
  1. ChAT, choline acetyltransferase; NOS, nitric oxide synthase; NSE, neuron specific enolase.

NSE129 DakoSheep1 : 50012,59
HuA21271 Molecular ProbesMouse1 : 100031
S100S-2644 SigmaRabbit1 : 100063
ChATYeboah donated by Prof. SchemannRabbit1 : 100012
NOSK205 donated by Prof. EmsonSheep1 : 100064

To analyse for any possible labelling of glial cells by anti-Hu, two fresh fixed and two cultured preparations were double labelled with antisera to Hu and S100 protein.

Four cultured specimens [as immunoreactivity for choline acetyltransferase (ChAT) is enhanced by culturing12] were triple labelled by incubation with antisera to nitric oxide synthase (NOS), ChAT and Hu for 2 days then washed repeatedly in PBS. Secondary antisera were then applied (donkey antirabbit CY3, donkey antimouse fluorescine isothiocyanate (FITC) and donkey antisheep CY5; Jackson Immunoresearch Laboratories Inc.) for a further 2 days at room temperature. For details of primary antibodies, sources and concentrations, see Table 1.


Preparations were viewed under a BX 50 epifluorescence microscope (Olympus, Tokyo, Japan) equipped with the appropriate filters to distinguish CY3, CY5 and FITC fluorophores. The autofluorescence of lipofuscins was readily distinguishable from immunohistochemical labelling due to its non-selectivity in the filters. At high magnification, neurons in the myenteric ganglia were counted by focusing through the full thickness of the myenteric plexus. Images were recorded using a modified version of the image analysis programme developed by the National Institute of Health (NIH Image 1.61; National Institutes of Health, Bethesda, MD, USA). In each preparation, using epifluorescence microscopy, 500 neurons were counted and their immunohistochemistry recorded using Hu – labelled cells as reference.

The preparations were also viewed with a Biorad 1024 confocal microscope (Biorad Laboratories, Hercules, CA, USA) to determine the size distribution of classes of nerve cell bodies. The area of cell bodies in the plane of the myenteric plexus were measured in NIH Image, for 50 myenteric nerve cell bodies in each of four preparations for ChAT+/NOS− and NOS+/ChAT− neurons; and for 10 nerve cell bodies for ChAT+/NOS+ and ChAT−/NOS− (being much less frequent). Measurements were made for each cell body at its maximum diameter, determined by focusing through the myenteric plexus. The measurements taken were the major axis at the cell body's maximum diameter and the minor axis perpendicular to the first measurement. Statistics are given as a mean ± SEM.


Anti-Hu antibody

The anti-Hu antibody labelled myenteric neuronal somata with homogenous staining of the cytoplasm, variable labelling of nuclei, and no staining of nucleoli. Different sizes and shapes of cell bodies were recognizable but detail of the nerve cell processes was limited. In some preparations it was possible to see a few nerve fibres in the myenteric plexus and also within the tertiary plexus, however these were not bright and did not obscure any detail of nerve cell bodies. Few Hu-immunoreactive nerve fibres were observed within the ganglia. Staining was generally very consistent, with similar intensity of labelling in fresh fixed and cultured tissue.

Double labelling with anti-Hu and NSE

Immunoreactivity for NSE was more variable in intensity than anti-Hu, both between preparations and within the same myenteric plexus preparation. Myenteric nerve cell bodies were often difficult to distinguish from numerous nerve fibres and varicosities, which were also labelled (Fig. 1B). In the most intensely labelled, fresh-fixed NSE preparations, 90% (SEM, 8%; n = 4) of neurons labelled with anti-Hu, were also immunoreactive for NSE. A mean of 0.01% (SEM, 0.3%, two neurons out of 2010; n = 4) of myenteric neurons showed NSE immunoreactivity only and 10% (SEM, 7.8%; n = 4) of myenteric neurons were immunoreactive for anti-Hu, but were not distinguishable by NSE immunoreactivity. In cultured tissue, a similar proportion of myenteric neurons were immunoreactive for NSE (mean 91%, SEM, 6%; n = 4), 10% of neurons were only immunoreactive for anti-Hu and, no myenteric neurons were labelled by NSE only. An example of a double-labelled preparation with anti-Hu and NSE is shown in Fig. 1.

Figure 1.

 Double labelling of colonic myenteric ganglion cells with antisera for Hu (A) and neuron specific enolase (NSE) (B). The great majority of nerve cells labelled by the antiserum to Hu were also immunoreactive for NSE (arrows). However, a significant number of very small cells were clearly labelled by Hu but were not distinguished adequately by NSE antisera (arrowheads). Calibration: 20 μm.

Double labelling of anti-Hu and S100

Many small, spindly cells were labelled with S100 antiserum. No labelling of any nerve cell body identified with anti-Hu antiserum was observed to also occur for S100 protein. An example of a double-labelled preparation with anti-Hu and S100 is shown in Fig. 2.

Figure 2.

 Double labelling of human colonic myenteric ganglia with antisera for Hu (A) and S100 protein (B). No nerve cell bodies labelled for Hu were labelled for S100 (marker of glial cells) – see examples of nerve cell bodies marked with arrows. Calibration: 50 μm.

NOS/ChAT/anti-Hu triple labelling

ChAT+/NOS− myenteric neurones 

Of myenteric plexus nerve cell bodies labelled with anti-Hu, a mean of 48% (SEM, 3%; n = 4) were ChAT immunoreactive but NOS negative. These neurons varied in appearance from small unipolar cells, through bigger irregularly shaped cells to infrequent very large cells with a smooth outline (Fig. 3B). These cells were found to have a mean cell area of 537.3 μm2 (SEM 24.5 μm2) with a mean major axis of 33.6 μm (SEM 7.1 μm), and a mean minor axis of 21.3 μm (SEM 6.1 μm). These cells however, varied widely in size, ranging from some of the smallest to some of the largest neurones in the ganglia, with suggestion of a polymodal distribution (see Fig. 4).

Figure 3.

 Triple labelling of a colonic myenteric ganglion with antisera for Hu (A), choline acetyltransferase (ChAT) (B) and nitric oxide synthase (NOS) (C). Many Anti-Hu labelled cells were immunoreactive for either ChAT (arrows) or for NOS (open arrows) and occasional cells were immunoreactive for both ChAT and NOS (large arrowhead) or for neither of the transmitter-related antigens (small arrowhead). Calibration: 50 μm.

Figure 4.

 Distributions of sizes of human colonic myenteric neurones defined by choline acetyltransferase (ChAT) and nitric oxide synthase (NOS) immunoreactivity. The soma-dendritic area of labelled cells was measured in μm2 from digital micrographs and the distribution of sizes plotted as percentage of cells with each combination of immunoreactivity. ChAT+/NOS− cells (n = 200) had the widest distribution of sizes with a clear tendency to show a polymodal distribution. ChAT−/NOS+ cells (n = 200) had a more restricted distribution. ChAT+/NOS+ cells (n = 40) were on average slightly larger, whereas ChAT−/NOS- cell bodies (n = 40) were consistently amongst the smallest in the ganglia, suggesting that they may belong to a single discrete class.

NOS+/ChAT− myenteric neurones 

Of myenteric plexus cell bodies labelled with anti-Hu, 43% (SEM, 2.5%; n = 4) contained NOS immunoreactivity without ChAT (see Fig. 3). NOS immunoreactive cells had a mean area of 519.3 μm2 (SEM, 15.9 μm2) with a mean major axis of 31.8 μm (SEM 4.2 μm), and a mean minor axis of 19.9 μm (SEM 3.4 μm). The distribution of sizes of NOS immunoreactive cells revealed less variability in neuronal size compared with ChAT immunoreactive neurons (see Fig. 4).

ChAT+/NOS+ myenteric neurones

A mean of 4% (SEM 0.5%; n = 4) of the total number of myenteric neurons counted with anti-Hu, labelled with NOS and ChAT. This group of neurons accounted for a mean of 7% (SEM, 0.9%; n = 4) of the ChAT immunoreactive neurons and 8% of all NOS−immunoreactive neurons. These neurons had a mean area of 678.5 μm2 (SEM, 45.4 μm2) with a mean major axis of 39.8 μm (SEM 8.9 μm), and a mean minor axis of 23.5 μm (SEM 6.4 μm). These cells appeared to comprise a discrete subset of medium sized neurons (Fig. 4) with multiple lamellar dendrites.

ChAT−/NOS− myenteric neurones 

A small number of myenteric nerve cell bodies in each preparation labelled with the Hu antibody lacked detectable immunoreactivity for either ChAT or NOS (Fig. 3). These comprised 4.9% (SEM 2.8%; n = 4) of all myenteric neurons. They were much smaller than the other cells with a mean cell area of 294.5 μm2 (SEM 25.9 μm2, see Fig. 4), with a mean major axis of 25.9 μm (SEM 4.6 μm), and a mean minor axis of 14.4 μm (SEM 2.9 μm).


Anti-Hu labelling

In this study we have shown that a commercially produced mouse monoclonal antibody, anti-Hu, which binds to neuronal specific Hu antigens, is a reliable, accurate and effective marker in labelling all myenteric neurons both in fresh fixed and cultured specimens of human colon. As a first step towards the full quantification of all classes of cells, we have characterized three major populations of neurones, defined by the presence of one or both functionally important markers, ChAT and NOS. In addition, the first evidence has been obtained for the existence of a population of enteric neurones that appears to lack both ChAT and NOS and which is likely therefore to be neither cholinergic nor nitrergic.

Anti-Hu antiserum labels neuronal somata, with minimal labelling of nerve fibres, and this enabled cell bodies to be easily distinguished and counted. This is of particular importance in human intestine, where the ganglia contain nerve cells in several layers, even when placed under maximal stretch. Improved accuracy in counting was demonstrated with the use of anti-Hu, by a greater number of nerve cell bodies, being counted, when compared with counting, using other general neuronal markers such as NSE. When compared with similar attempts at quantification of the proportions of cholinergic and nitrergic neurons using NSE12 or PGP 9.513 to label cell bodies, we showed a lower proportion of labelling (52% as compared with 63% or 65% for cholinergic neurons) as a result of increased identification of cell bodies. It is likely that in other studies utilizing NSE to count all neurons33,34 proportions of markers have also been overestimated.

Labelling of nerve cell bodies by anti-Hu was effective in both fresh fixed and cultured tissue, thus expanding its potential applications. Furthermore, it can be readily combined with double and triple labelling immunohistochemistry, thus giving added advantages over techniques requiring serial processing or sectioning.13,22,35 The small neuronal cells bodies identified by anti-Hu in our study were not glial cells as demonstrated by lack of staining for the S100 protein and also by the shape of the cells.36,37 Similarly Lin et al. found no cross reactivity for S100 protein and Hu in guinea pig myenteric ganglia,31 although Phillips et al. found a few cells (<1%), which were labelled with anti-Hu, yet which also demonstrated immunoreactivity for S100.22 Cuprolinic blue is an alternative histochemical method of labelling all enteric neurons, which has been shown to be successful in both small animals,22,24,25 and humans,26 and suggested as an ideal pan-neuronal marker. Recent work has shown that neuronal quantification with cuprolinic blue is comparable with quantification with anti-Hu,26, however, this contrasts with earlier work from Phillips et al., that demonstrated anti-Hu to label more neurons than cuprolinic blue.22

Proportions of markers of human myenteric neurones

Markers for cholinergic and nitrergic neurons were chosen for the initial investigation of proportions of markers, as previous work has revealed that these two populations account for the majority of myenteric neurons in a number of species,3,8,11,12,38 and because of their known importance in neurotransmission. Of relevance to the study of pathological tissue was the observation, that there was very little variation in proportions between the specimens,12,13 so comparison of small numbers of specimens will enable conclusions to be made, regarding changes in neuronal populations.

Cholinergic neurons belong to a number of different functional classes. In the guinea pig ileum, excitatory motor neurones to the longitudinal and circular muscle layers, ascending interneurons, several classes of descending interneurons, primary afferent neurones, some vasomotor and secretomotor neurones have all been shown to be immunoreactive for ChAT and thus are likely to be cholinergic.2,3,39 The polymodal distribution of cell area measurements of cholinergic neurons labelled in the present study reflects the existence of multiple different functional classes within this general immunohistochemical type. Retrograde labelling studies in human colon have shown cholinergic neurons project to the circular muscle, and in ascending and descending interneuronal pathways.10,11,40,41 Markers to discriminate these subclasses of cholinergic neurons will be needed to achieve a full quantitative analysis such as, has been achieved in the guinea pig

In our study, 47% of human colonic myenteric neurons were NOS immunoreactive, a proportion similar to the 43%, found by Porter et al.11 The reason for this small discrepancy may be that NOS was not found in the small neurones, not identified readily by NSE in Porter's study, but counted by us, with use of anti-Hu. Retrograde tracing studies in humans reveal NOS in motor neurones and descending interneurons,11,41 thus, the number of classes of nitrergic neurones, is less than the number of cholinergic classes, as is supported by the absence of a polymodal distribution of cell area measurements for NOS immunoreactive neurones.

We have shown that the proportion of cells to be immunoreactive for NOS appears to be higher in the myenteric plexus of the human colon than in most other preparations studied to date.11,35,42 It is possible that this reflects age related changes. The total number of neurons within the myenteric plexus decreases with age,43,44 however, it has been reported that the proportion of cells immunoreactive for NOS increases in frequency, in the gastrointestinal tract of aged patients, compared with adult controls.45 Phillips et al. when examining cell numbers in rats, found that neuronal loss in aged rats was specific to cholinergic neurons,46 and this may also occur in humans, resulting in a higher proportion of nitrergic neurons. This finding may also reflect functional adaptations of the human colon for accommodation, which are important in its role as a reservoir.47 A greater number of specimens from the right colon in this study may also be of significance, as there are increased numbers of NOS immunoreactive neurons in rat proximal colon compared with distal colon,48 which may possibly also occur in humans. As the majority of NOS−immunoreactive neurons are likely to function as inhibitory motor neurons, it is possible that the high proportion of NOS−immunoreactive motor neurons reflects a functional requirement for powerful inhibitory neuronal input to the muscle layers.

We have confirmed the presence of myenteric neurons that are reactive for both ChAT and NOS, as reported previously in both guinea pigs2,49 and humans.41 These are known to be descending interneurones from retrograde tracing studies41 and the current study reveals these to comprise 4% of all human colonic myenteric neurons. A number of recent studies by Brehmer et al. have characterized human myenteric neurons on the basis of morphology and neurochemical staining.42,50–52 These investigators have characterized small intestine Dogiel type II neurons on the basis of cell morphology, neurochemistry and projections, visible in wholemounts.51 Whilst the authors of this paper suggest these neurons are primary afferent neurons, this can only be proven conclusively, with further investigations, such as retrograde labelling from the mucosa, and electrophysiological experiments. The same group has also characterized neurons immunoreactive for enkelphalin,52 and NOS and VIP, using the same methods, in both small and large intestine,42 however, an exclusive chemical code for a particular functional class has not been previously demonstrated in human colon. Descending interneurons with immunoreactivity for both ChAT and NOS, are the first identifiable exclusive code denoting a class of myenteric neurons in human colon.

A class of myenteric neurons (as revealed by Hu immunoreactivity) totalling 5% of all neurons revealed neither ChAT nor NOS immunoreactivity. To the best of our knowledge, this is the first description of enteric neurones lacking both ChAT and NOS. Cells with this coding were typically very small and are likely to form a discrete population. On the basis of cell size these may be longitudinal muscle motor neurons.53 A variant peripheral form of ChAT, peripheral ChAT has been recently described in the myenteric plexus of a number of experimental animals.54–57 In guinea pigs and sheep,55,57 the common form of ChAT, central ChAT, labels more neurons, than peripheral ChAT, however, it is possible that a small number of cholinergic neurons were not labelled with central ChAT, yet are still cholinergic and may account for some of these neurons without ChAT and NOS imunoreactivity.

To date studies of neuropathology in human gastrointestinal disorders have largely relied upon semi-quantitative assessments of the numbers of nerve cells and fibres.58–60 More recent analyses of nerve cell numbers and proportions of markers have provided better quantification, and revealed changes not detected by prior methods.33,61,62 These studies have relied upon NSE and PGP 9.5 to assess all nerve cell bodies in enteric ganglia. The use of anti-Hu antiserum to readily label such nerve cell bodies, as demonstrated in our study, offers the ability to improve quantification further, thus enabling more subtle detection of changes in neuronal populations.


This work was supported by DK56986 of NIDDK of the National Institutes of Health of the USA and by a grant from the Flinders Medical Centre Research Foundation. EMAM and DF were supported by grants from the Royal Australian College of Surgeons and the Clinician's Special Purpose Fund of Flinders Medical Centre. SJHB is a senior research fellow of the National Health and Medical Research Council of Australia. We would like to thank Janine Edwards for valuable assistance with tissue processing.