Dr Paul Kelly, Adult and Paediatric Gastroenterology Research Group, Institute of Cell and Molecular Science, Barts and The London School of Medicine, Turner Street, London E1 2AD, UK. E-mail: firstname.lastname@example.org
Paneth cells are important contributors to the intestinal antimicrobial barrier through synthesis and release of antimicrobial peptides and proteins. Animal studies indicate that Paneth cell numbers, location and granule morphology are altered by infection and zinc status. We examined human tissue to determine whether Paneth cell numbers, distribution or granule morphology are altered in infective, inflammatory and nutritional disorders. Archival sections from infective disorders (giardiasis, cryptosporidiosis, HIV, helminth infection) were compared with active inflammatory conditions (coeliac, Crohn's and graft-versus-host diseases) and histologically normal tissues. A subset of tissues was studied by electron microscopy and TUNEL staining for apoptosis. Human defensin-5 (HD5) peptide and mRNA was analysed by immunohistochemistry, in situ hybridization and quantitative reverse transcription polymerase chain reaction. Sections from a tropical population cohort study were then analysed to determine the relationship of granule depletion to infection, nutritional status and plasma zinc concentration. In HIV-related cryptosporidiosis, but not other disorders, Paneth cells were reduced in number and markedly depleted of granules. Paneth cell granule depletion was associated with reduced HD5 immunoreactivity, but this was not due to apoptosis and there was no reduction in mRNA transcripts. In the tropical population studied, depletion of granules was associated with reduced body mass index, reduced plasma zinc levels and HIV infection. Paneth cell granules in human small intestine may be depleted in response to infective and nutritional stress. We postulate that this is one mechanism through which zinc status influences host susceptibility to intestinal infection.
The role of Paneth cells in small intestinal physiology and pathology was uncertain for many years. Believed at first to play a digestive or phagocytic role, it is now clear that they contribute to innate mucosal defence through the expression of a powerful array of antimicrobial and immunoregulatory molecules including lysozyme, α defensins, secretory phospholipase-A2, α1 antitrypsin and tumour necrosis factor [1,2]. Paneth cells, located at the base of the crypts of Lieberkühn, release a high concentration of antimicrobial molecules adjacent to the epithelial stem cells, offering protection against colonization to these cells and to the crypt [3–5].
There are few examples of changes in Paneth cells in human small intestinal disease, and whether any changes are primary or secondary is unclear. Paneth cells expressing lysozyme are almost absent in necrotizing enterocolitis  and their intracellular granules are abnormal in the congenital zinc deficiency disorder acrodermatitis enteropathica. In the latter, the secretory granules are empty in some but not all Paneth cells , and in some cells the granules are small and pleiomorphic . A reduction in Paneth cell numbers has been observed in coeliac disease, and it has been suggested that this reduction identifies patients unresponsive to gluten withdrawal, but this remains equivocal [9,10]. In small intestinal Crohn's disease, Paneth cell depletion occurs in the most heavily inflamed areas, but occasional specimens show increased numbers and Paneth cell degranulation has been observed in cystic fibrosis .
In rats, zinc chelation with diphenylthiocarbazone leads to Paneth cell apoptosis rather than just granule changes , indicating that the precise interaction between zinc and Paneth cell biology may be species-dependent. In primates with organic mercury poisoning, Paneth cell degranulation and inclusions of the rough endoplasmic reticulum and Golgi apparatus are observed .
In addition to changes in cell number and granule morphology, two types of metaplasia of Paneth cells have been described. First, within the small intestine Paneth cells appear outside the base of the crypt, as for example in murine infection with the nematode Trichinella spiralis, when numerous Paneth cells can be found much higher up the crypt and on the sides of the villi . Secondly, Paneth cells can be observed outside the small intestine in ulcerative colitis (colon), colorectal carcinoma (colon) or intestinal metaplasia (stomach and oesophagus).
The aim of this study was to test the hypothesis that Paneth cell numbers and morphology are altered in infectious disease of the human small intestine. We went on to assess whether the observed granule depletion is a consequence of apoptosis or of changes in transcription of granule peptides or proteins, using the human defensin HD5, a peptide stored exclusively in Paneth cell granules, as an example. In view of the ultrastructural similarity to zinc deficiency, we used samples from a community study in Zambian adults to determine the relationship of granule depletion to plasma zinc concentration and other potential determinants, including HIV, which is highly prevalent in this population.
MATERIALS AND METHODS
In order to determine the distribution of Paneth cells along the human small intestine and to assess differences in disease states, haematoxylin and eosin (H&E)-stained sections of normal and diseased tissue (largely endoscopic biopsies) were obtained from the archives of the Department of Histopathology or from tissue obtained during the course of studies of intestinal infection in Zambia [15,16]. Normal sections were obtained from duodenum, jejunum and ileum; sections representing disease states were obtained from duodenum except in the case of Crohn's disease, for which tissue from jejunum and ileum was also used. The histological appearance of all sections was confirmed as compatible with the final diagnosis by either or both of the expert pathologists involved in the study (R. F., P. D.). Approval was obtained from the Research Ethics Committee of the East London Health Authority.
Sections of coeliac disease showed subtotal or partial villous atrophy, and from Crohn's disease were from actively inflamed tissue. Sections from graft-versus-host disease (GVHD) showed minor changes only. Sections from patients with intestinal infections were taken from biopsies at the site where infection was demonstrable on the day of biopsy. In ascariasis and the majority of hookworm infections, sections were used only when worms had actually been recognized in the jejunum during enteroscopy immediately prior to biopsy. In two of the hookworm infections, and in the cases of cryptosporidiosis, parasites were seen in histological sections or in stool samples taken just prior to biopsy. In giardiasis, trophozoites were seen in the sections. Sections from asymptomatic HIV infection were taken from patients undergoing endoscopy for abdominal complaints but whose endoscopy was normal. Sections from AIDS-related cryptosporidiosis were obtained from previously published studies of this disease . Sections with tropical enteropathy were taken from asymptomatic participants in the Zambian studies, among whom subclinical infection with low-grade pathogens would be expected. All sections had been fixed in neutral-buffered saline (Merck, Poole, UK).
In order to elucidate the pathology behind the apparently degranulated Paneth cell, duodenal and jejunal biopsies, and plasma samples for zinc estimation were also obtained from recent studies of tropical enteropathy and HIV-related intestinal disorders in Zambian adults . Tissue was used for histology, inmmunohistochemistry, TUNEL staining and quantification of HD5 mRNA (see below). Ethical committee approval for these studies was obtained from the Research Ethics Committees of the University of Zambia and the London School of Hygiene and Tropical Medicine.
Paneth cell counts and granule morphology
Sections were evaluated by counting Paneth cells and by evaluating granules. For cell counts, crypts (between six and 24 in number) were evaluated only if aligned along the longitudinal axis such that the lumen of the crypt could be seen along its length. The results were expressed as Paneth cells per crypt, but each crypt was divided into three zones (Fig. 1) to take into account distribution along the crypt–villus axis. Each section was also given a qualitative assessment of Paneth cell granule morphology using a three-point scale (normal, mildly depleted granules, severely depleted granules). These scores were given by one observer (P. K.), but the validity of the scores was assessed by one of the expert pathologists (R. F.) who assessed the sections from patients with cryptosporidiosis. There was 100% concordance between the two observers.
To determine whether Paneth cell counts differ along the small intestine, counts were performed in sections of normal mucosa taken from duodenum, jejunum or ileum. As no significant differences were found (see below), sections from any region of the small intestine were compared in different disease states.
Immunohistochemistry was used to define expression of HD5 in biopsies with or without Paneth cell granule depletion (scored as above). Sections 4 µm thick were dewaxed, rehydrated and endogenous peroxidase blocked in 0·3% hydrogen peroxide in methanol for 5 min. The primary antibody was a monoclonal HD5 antibody generated against HD5 propeptide which will be described elsewhere (B. Shen, in prep.). Incubation with the primary anti-HD5 antibody [1 : 15 000 dilution in 1·5% horse serum in phosphate buffered saline (PBS)] was at 4°C overnight, and with the biotinylated horse antimouse secondary antibody (1 : 200 dilution; Vectastain Elite ABC, Vector Laboratories, Burlinghame, USA) was at room temperature for 20 min. Colour was developed using diaminobenzidine tetrahydrochloride (Vector DAB substrate kit). Sections were counterstained with 0·024% Light Green (Merck, Poole, UK) in 0·012% glacial acetic acid. The positive control used in each run was a section of normal human jejunum cut from a single block.
Paneth cell ultrastructure and TUNEL staining
Glutaraldehyde-fixed biopsies were processed for electron microscopy as described previously . In order to determine whether the abnormality of granules was due to Paneth cell apoptosis, unstained formalin-fixed sections were stained by the TUNEL technique using a modified protocol of the TdT in situ kit (R&D Systems) as described previously .
HD5 mRNA distribution was defined using in situ hybridization as described previously [18,19]. We have described recently a reverse transcription polymerase chain reaction (RT-PCR) assay for HD5 mRNA in which a standard synthetic RNA is co-reverse transcribed and co-amplified with target mRNA . Different sizes of the amplicons from synthetic and target (native) RNA allows them to be distinguished by agarose gel electrophoresis (Fig. 2). Sequencing of the PCR product confirmed identity with HD5 .
Determinants of Paneth cell granule depletion
As granule depletion was observed in patients with HIV-related cryptosporidiosis, who have advanced AIDS complicated by malnutrition, and in view of the Paneth cell abnormalities seen in zinc depletion, it was decided to explore further the association between granule depletion, HIV, reduced body mass index  and plasma zinc concentration. Sections were examined from a series of biopsies from a longitudinal study of intestinal infection in adults . Zinc concentrations in plasma were assayed in blood which had been specifically collected into zinc-free lithium–heparin collection tubes using zinc-free needles and syringes. Zinc was assayed using an in vitro colorimetric method (Wako Zinc Test, Alpha Laboratories, Eastleigh, Hampshire, UK). The coefficient of variation of the assay was 3·7%.
Paneth cell counts in zone 1 were found to be distributed normally using the Shapiro–Wilk test, and comparisons between means was performed using analysis of variance. As counts in zones 2 and 3 were skewed, with the majority being zero, these zones were analysed using the Kruskal–Wallis test. In all cases, statistical comparisons were made across all groups and post hoc analysis was not performed. Analysis of the association of HIV and low plasma zinc concentration with Paneth cell granule depletion used Fisher's exact test, and both Mantel–Haenszel analysis and logistic regression were used to identify the dominant factor.
Paneth cell were reduced in AIDS-related cryptosporidiosis
Sections from different regions of small intestine showed little variation in the number of Paneth cells per crypt. Mean counts in zone 1 were 2·7 in nine sections of normal duodenum, 3·3 in five sections of normal jejunum and 2·6 in seven sections of normal ileum (P = 0·48). As no significant difference was found in regional distribution, results of counts in tissue from disease states were pooled irrespective of region of origin, and tissue from 96 patients was evaluated together. The great majority of Paneth cells also showed consistent granule morphology, with each cell full of plump eosinophilic granules (Fig. 3).
Paneth cell counts in the crypt base (zone 1) were comparable in sections from normal and diseased small intestine, but reduced in sections from patients with HIV-related cryptosporidiosis (Table 1). Outside this zone, but in the deep part of the crypt (zone 2), numbers were increased in coeliac and Crohn's disease. Few Paneth cells were seen in zone 3 in any disorder.
Table 1. Paneth cell counts in health and disease
Zones 1 + 2
P-values (anova or Kruskal–Wallis test, see Methods) refer to the statistical probability that differences across diagnosis groups have arisen by chance. n refers to the number of patients from whom biopsies were taken; in sections from each patient at least six crypts were counted, but in most cases 8–12 crypts were counted. GVHD, graft-versus-host disease. Tissue from patients with HIV infection were divided into two groups, those with cryptosporidiosis and those with no evidence of intestinal disease undergoing routine endoscopy for other reasons .
Compared to normal controls and most pathological specimens, many of the Paneth cells from patients with AIDS-related cryptosporidiosis appeared to be depleted qualitatively of granules (Fig. 3). This depletion was confirmed on immunohistochemistry using the anti-HD5 antibody (Fig. 4).
Granule depletion was not due to apoptosis
Using electron microscopy, Paneth cell granules in sections exhibiting depletion were seen to be pleiomorphic, smaller than normal and reduced in number (Fig. 5). Furthermore, some Paneth cells were observed to exhibit dilatation of the endoplasmic reticulum and Golgi apparatus, but none of the nuclei showed apoptotic changes. To determine if the empty Paneth cell appearance could be due to apoptosis, TUNEL staining was performed on sections showing granule depletion from the Zambian study. No TUNEL-positive Paneth cells were seen in experiments although appropriate controls were positive.
Granule depletion was not associated with reduction in HD5 mRNA
To determine whether granule depletion was associated with changes in transcription of granule constituents, HD5 mRNA was analysed in biopsies with and without granule depletion from the Zambian study. Riboprobe hybridization to mRNA was as strong in sections with granule depletion (assessed by H&E and by immunohistochemistry), as in sections in which granule morphology and immunoreactivity was normal (Fig. 4). In sections of any description, riboprobe hybridization to cells outside zone 1 was very infrequent (data not shown).
Tissue content of human defensin mRNA measured by quantitative RT-PCR was not reduced in biopsies from patients with depleted granules compared to those without (Table 2).
Table 2. Human HD5 mRNA in relation to Paneth cell granule depletion
HD5 mRNA (log transcripts/µg total RNA)
Values given are median (interquartile range), and expressed as log transcripts/µg total RNA. There was no statistically significant difference using the Kruskal–Wallis test.
Granule depletion was related to low zinc levels, low body mass index and HIV
As AIDS-related cryptosporidiosis is a complex disorder characterized by intestinal infection and advanced nutritional impairments, we analysed these interactions in participants undergoing endoscopy in the Zambian cohort. Plasma zinc concentrations (median, interquartile range) in samples from these same individuals were 16·7, 14·2–18·1 µmol/l in HIV seropositive adults and 17·6, 12·4–18·8 µmol/l in HIV seronegative adults (P = 0·97). Using a cut-off of 12·7 µmol/l to define abnormality (reference lower limit, University of Southampton), 3/12 (25%) of the HIV seropositive adults had reduced levels and 5/34 (15%) of the HIV seronegative adults (P = 0·41). Paneth cell granule depletion was associated with low zinc concentration and HIV infection (Table 3). Granule depletion was also associated with reduced body mass index (BMI): median, interquartile range BMI was 22, 19–26 kg/m2 in those with normal granules, 19, 17·4–21 in those with some granule depletion and 17·3, 15·9–20·7 in those with severe granule depletion (P = 0·01 by non-parametric test for trend). Multivariate analysis using both Mantel–Haenszel stratification and logistic regression indicated that granule depletion was associated with low plasma zinc concentration rather than with HIV infection or reduced body mass index.
Table 3. Paneth cell granule depletion in relation to HIV and in relation to plasma zinc concentration
The severity of granule depletion in relation separately to two potential determinants: HIV infection (total number analysed = 102) and plasma zinc (n = 46). In multivariate analysis, the association between low plasma zinc concentration and granule depletion remained after taking HIV infection into account. Stratification by zinc abolished the association between granule depletion and HIV. P-values derived using Fisher's exact test.
Paneth cells are specialized epithelial cells of the small intestinal crypts of Lieberkühn, which synthesize antimicrobial peptides and proteins, store them in large cytoplasmic granules and release them into the crypt lumen [21,22]. We observed that Paneth cell numbers varied little in different disease states of the small intestine. However, Paneth cell granule depletion was observed in association with infective and nutritional stress. This makes it difficult to enumerate Paneth cells (which are recognizable by their characteristic granules) and the results of our electron microscopy suggest that the apparently reduced cell numbers in AIDS-related cryptosporidiosis are due to the reduced granule content of those terminally differentiated Paneth cells which are present. This effect has been noted previously in coeliac disease . Our data suggest that Paneth cell granule depletion may be induced by conditions associated with low plasma zinc concentrations. These reduced plasma zinc concentrations could be due to reduced intake, impaired absorption by compounds such as phytate, or altered transport by albumin in plasma due to infectious or inflammatory processes.
Paneth cell degranulation takes place physiologically following cholinergic stimulation of the gut and of isolated crypts [4,24]. Atropine leads to reduced secretion of secretory granules and their accumulation in cytoplasm, a similar state to the engorged Paneth cells seen in germ-free rats . The granule depletion we observed in some patients is unlikely to be physiological, as patients are fasted routinely prior to undergoing endoscopy and intestinal biopsy. It seems more likely that the granule depletion we observed is a pathological process which would indicate either a reduced rate of synthesis of granule constituents, an impairment of protein packing into granules, or an increased rate of release due to some signal indicating higher demand for antimicrobial defence molecules. However, our quantitative measurements of mRNA and the results of in situ hybridization suggest that gene transcription of HD5 is not reduced. Interestingly, neither was there any compensatory increase in mRNA, as might be predicted if the depletion were simply due to increased demand.
We did not observe metaplasia of the Paneth cell population along the crypt–villus axis as has been observed in mice infected with Trichinella spiralis and other helminths . We found some evidence of this kind of metaplasia, but not in the protozoal or helminth infections analysed. The technique employed here, counting cells in longitudinal sections, would probably not be as sensitive as counting serial sections through microdissected crypts, but allowed us to evaluate archival material, and we assume that cells counted in longitudinal sections will be proportional to cells present. We did observe an increased cell count in the region of the crypt which we labelled ‘zone 2’ in coeliac disease, Crohn's disease, and graft-versus-host disease, all autoreactive enteropathies associated with Th1 activation. The degree of Paneth cell expansion was, however, very much less marked than that seen in the murine helminth models which would be associated with Th2 activation.
The development of the Paneth cell lineage in mice is not dependent on the intestinal microflora , but introduction of a novel luminal flora to germ-free rats does induce Paneth cell degranulation . It does not, however, alter α-defensin propeptide processing . Stimulation of isolated murine crypts with bacteria or lipopolysaccharide leads to Paneth cell degranulation and release of α-defensins , and these defensins seem to play a significant role in host defence. Mice which do not express matrilysin (the α-defensin processing enzyme in mice) were more susceptible to salmonellosis , and there is evidence that salmonellosis may inhibit Paneth cell antimicrobial activity . Transgenic mice expressing human defensin 5 were completely protected from Salmonella typhimurium infection .
Zinc is an important micronutrient with a role in maintaining the integrity of mucosal defence against intestinal infection . Zinc supplementation consistently reduces morbidity in diarrhoeal disease in children [34–36]. Zinc-deficient animals have reduced intestinal α-defensin mRNA , Paneth cell granule abnormalities and invasion of Paneth cells by intestinal bacteria . Zinc chelation leads to Paneth cell apoptosis in rats . We postulate that undernutrition, and specifically that poor zinc status, adversely impacts on the ability of Paneth cells to package and retain antimicrobial peptides and proteins in intracellular granules and to respond appropriately to the stress induced by intestinal infection. This may be one important pathway through which mucosal defence in the host is dependent on zinc as a micronutrient.
We are indebted to a team of nurses and researchers for their help in the conduct of the population study: Rosemary Banda, Vera Yambayamba, Stayner Mwanamakondo, Rose Soko, Max Katubulushi, Coillard Kaunga and Samson Mbewe, and to Angela Hounslow for technical assistance with the zinc assays. We are grateful to The Wellcome Trust for financial support.