Background Human colon may secrete substantial amounts of water secondary to chloride (Cl−) and/or potassium (K+) secretion in a variety of diarrhoeal diseases. Ion secretion occurs via Cl− and K+ channels, which are generally assumed to be co-located in the colonocyte apical membrane, although their exact cellular sites remain unclear.
Objective To investigate the location of apical Cl− (CFTR) and apical K+ (large conductance; BK) channels within human colonic epithelium.
Design Whole-cell patch clamp recordings were obtained from intact human colonic crypts. Specific blockers of K+ channels and CFTR identified different types of K+ channel and CFTR under resting conditions and after stimulating intracellular cAMP with forskolin. The BK channel β3-subunit was localised by immunostaining.
Results Two types of crypt cells were identified. One (73% of cells) had whole-cell currents dominated by intermediate conductance (IK) K+ channels under resting conditions, which developed large CFTR-mediated currents in response to increasing intracellular cAMP. The other (27% of cells) had resting currents dominated by BK channels inhibited by the BK channel blocker penitrem A, but insensitive to both forskolin and the IK channel blocker clotrimazole. Immunostaining showed co-localisation of the BK channel β3-subunit and the goblet cell marker, MUC2.
Conclusions In human colon, Cl− secretion originates from the dominant population of colonocytes expressing apical CFTR, whereas K+ secretion is derived from a smaller population of goblet cells expressing apical BK channels. These findings provide new insights into the pathophysiology of secretory diarrhoea and should be taken into account during the development of anti-diarrhoeal drugs.
- BASIC SCIENCES
- COLORECTAL PHYSIOLOGY
- INTESTINAL ION TRANSPORT
- ION CHANNELS
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
Significance of this study
What is already known about this subject?
Water secretion, driven by active secretion of Cl− and/or K+, occurs in human colon in a variety of diarrhoeal diseases.
The main conductive pathways across the apical membrane of colonic epithelial cells are CFTR and large conductance K+ (BK) channels, which mediate Cl− secretion and K+ secretion, respectively.
The current model of colonic ion secretion has both Cl− and K+ originating from colonocytes, the dominant cell type within the epithelium.
What are the new findings?
In human colon, Cl− secretion originates from colonocytes, which account for ∼75% of the crypt cells and express apical CFTR and basolateral intermediate conductance K+ (IK) channels.
By contrast, K+ secretion originates from goblet cells, which account for ∼25% of the crypt cells and express apical BK channels but not apical CFTR.
How might it impact on clinical practice in the foreseeable future?
Partitioning of Cl− and K+ secretion between distinct types of cell within human colon suggests that they are regulated independently, which may influence the development of new anti-diarrhoeal drugs.
The development of specific BK channel activators may provide a means of stimulating K+-driven fluid secretion in cystic fibrosis, where Cl−-driven fluid secretion is absent.
The principal function of the mammalian colon is to minimise losses of salt and water.1 ,2 Human colon absorbs over 95% of the 1.5–2 litres of fluid passing through the ileocaecal valve each day. Human colon also has a capacity for water secretion, which is driven by Cl− and/or K+ secretion, such as occurs in a variety of diarrhoeal diseases of colonic origin.3–6 Although these secretory fluxes may result in net water fluxes which are close to zero, they are associated with clinically significant electrolyte and water imbalance.3 The two main conductive pathways for electrogenic ion secretion across the apical membrane of colonic epithelial cells are Cl− channels (CFTR, which mediate Cl− secretion) and large conductance K+ (BK) channels, which mediate K+ secretion. In both cases, transgenic knockout of the respective gene results in the predicted specific loss of Cl− or K+ secretion.7 ,8 The BK channel in human colon is homologous with the BK channel present in mouse, rat and human brain,9–11 and rabbit kidney,12 and has been variously referred to as the maxi-K or Slo channel. Most mammalian BK channels are composed of α-subunits encoded by a single gene (KCNMA1),11 ,13 cell-specific function being determined by alternative gene splicing,11 ,14 the presence of regulatory β-subunits15 ,16 and the effects of protein phosphorylation/dephosphorylation.14 ,17 At least nine isoforms of KCNMA1 exist in human brain,11 but the identity of the KCNMA1 isoform(s) in human colonic epithelial cells is unclear. The human β-subunit has four isoforms, RT-PCR suggesting that the β1- and β3-subunits are the dominant isoforms in human colon.18 We have also shown that intermediate conductance (∼25 pS), Ca2+-sensitive and inwardly rectifying K+ channels, encoded by KCNN4, are the commonest type of K+ channel in the basolateral membrane of human colonic crypt cells.19 ,20 Their properties are similar to IKCa channels (variously termed KCNN4, KCa3.1 and SK4 by different authors) identified in a variety of human tissues and rat colon.21–23
The surface area of the colon is amplified by crypts radiating from the colonic lumen, consisting mainly of columnar epithelial cells (colonocytes) and goblet cells, which make up approximately 75% and 25% of the crypt cell mass, respectively.24 The generally held view is that Cl− and K+ are both secreted by colonocytes, whereas goblet cells are responsible for mucus secretion.25 This study began as an investigation into how different types of K+ channel contribute to the membrane conductance of human colonic crypt cells. We were surprised to find that cells expressed currents mediated by either CFTR or BK channels, which suggests that Cl− and K+ are secreted by two distinct cell types within the colonic epithelium.
Materials and methods
Isolation of colonic crypts
Sigmoid colonic biopsy specimens were obtained with written consent from patients undergoing routine colonoscopy for altered bowel habit and who were subsequently found to be free of mucosal disease. The study was approved by the Leeds Health Authority ethics committee. In all experiments, intact crypts were isolated by Ca2+ chelation,20 and used within 8 h. Attempts to separate isolated crypts into their individual and identifiable cellular components resulted in ‘rounded’ cells lacking polarity and it was impossible to differentiate between colonocytes and goblet cells. Each patch clamp experiment was therefore done on an intact crypt without any prior knowledge about the nature of the particular cell being studied.
Patch clamp recording
Whole-cell currents were measured using the perforated patch technique, having added amphotericin B (240 µg/ml) to the pipette solution, which contained 119 mM K+ gluconate, 6 mM KCl, 20 mM NaCl, 1 mM MgCl2 and 10 mM HEPES, titrated to pH 7.4 with KOH. The bath (high-Na+) solution ordinarily contained 140 mM NaCl, 4.5 mM KCl, 1.2 mM CaCl2, 1.2 mM MgCl2, 10 mM HEPES and 5 mM glucose titrated to pH 7.4 with NaOH. When a high-K+ bath solution was used, NaCl was replaced with equimolar KCl and titrated to pH 7.4 with KOH. Patch clamp recordings were made from basolateral membrane of cells in the mid-third of the crypts in the cell-attached and perforated whole-cell configurations. Patch pipettes were fabricated from borosilicate glass microhaematocrit tubes by using a two-stage pipette puller (Narashige, Tokyo, Japan; model PP-83) and had tip resistances of 2–5 MΩ. Experiments were carried out at room temperature (20–22°C) to maintain cellular viability. Single channel and whole-cell currents were amplified and low-pass filtered at 2 kHz (Axon Instruments, Union City, California, USA; Axopatch 200B). Currents were displayed on a computer monitor and saved to a hard drive via a Digidata 1320A interface (Axon Instruments). Data acquisition and analysis were performed using pCLAMP V.9.0 software (Axon Instruments). Whole-cell conductance was estimated from the slope of the linear portion of the relationship between the whole-cell current (nA) and the command voltage (mV), using linear regression analysis. The reversal potential of the cell was determined from the x-axis intercept of the current–voltage plot.
Isolated crypts were fixed in ice-cold 70% ethanol. The primary monoclonal antibody to the BK channel β3-subunit was raised in mice and used at 1:100 dilution (cat. 75–098; NeuroMab, Davis, California, USA). The primary polyclonal antibody to MUC2 was raised in rabbits and used at a dilution of 1:500 (cat. H-300; Santa Cruz Biotechnology, Santa Cruz, California, USA). Binding of the BK channel β3-subunit antibody was detected with biotinylated horse anti-mouse antibody at a dilution of 1:250 (cat. BA2001; Vector Laboratories, Peterborough, UK), followed by the application of streptavidin AlexaFluor 555 conjugate at a dilution of 1:500 (cat. S32355; Life Technologies Ltd, Paisley, UK). Binding of the MUC2 antibody was detected with AlexaFluor 488 donkey anti-rabbit at a dilution of 1:500 (cat. A21206; Life Technologies Ltd, Paisley, UK), Nuclei were stained with 4’,6-diamidino-2-phenylindole (DAPI) and sections mounted in Vectashield (Vector Laboratories, Peterborough, UK). Confocal images were obtained using a Zeiss LSM 510 META Axiovert 200M (Carl Zeiss, Welwyn Garden City, UK).
Data are presented as mean± SEM and analysed by paired or unpaired t test as appropriate.
Evidence for two cell types from whole-cell recordings
We measured whole-cell currents in the mid-third region of freshly isolated human colonic crypts using the perforated patch technique and found two distinct populations of cells based upon their electrophysiological characteristics. Under basal conditions, the dominant cell type (73% of cells, 58/79 recordings) showed an inwardly rectifying conductance which was K+-selective and which we have termed IRK cells (figure 1A,C). The other minority cell type (27% of cells, 21/79 recordings) exhibited an outwardly rectifying conductance which was also K+-selective and which we have termed ORK cells (figure 1B,D). As shown in figure 2A,C, these two cell types showed different sensitivities to the K+ channel blockers clotrimazole (CLT 10 µM; an inhibitor of IK channels),26 and 100 nM penitrem A (a highly selective inhibitor of BK channels).27 CLT inhibited whole-cell currents in IRK cells (figure 2A), the whole-cell conductance in these cells decreasing by 62±7% (n=7; p<0.01, figure 2B). This was accompanied by a positive shift in Erev of 21±5 mV, consistent with K+ channel inhibition. We have previously shown that inwardly rectifying K+ (IK) channels in human colonic crypt cells are inhibited by CLT,28 and also by 100 nM TRAM-34, a potent and selective IK channel inhibitor.29 By contrast, CLT had no effect on whole-cell currents (figure 2A) or whole-cell conductance (figure 2B) in ORK cells (n=3). However, the selective BK channel inhibitor penitrem A (100 nM) inhibited whole-cell currents (figure 2C) and the whole-cell conductance in ORK cells by 68.5±5.7% (n=8; p<0.01, figure 2D). This was not accompanied by a significant shift in Erev (Δ +3±3 mV), reflecting the fact that BK channels are inactive at the resting membrane potential. Furthermore, penitrem A had no significant effect on IRK cells (figure 2C,D). Thus, ORK cells contain BK channels, which are located in the apical membrane of human colonic crypt cells and are responsible for K+ secretion.18 On the other hand, IRK cells express IK channels, which are the main type of K+ channel found in the basolateral membranes of human colonic crypts.30
Evidence for two cell types from single channel recordings
The molecular identities of the two K+ channel species were further evaluated by single channel recordings. Cell-attached recordings from basolateral membrane patches revealed IK channels with a unitary conductance of ∼30 pS and an inwardly rectifying current–voltage relationship similar to that observed with the IRK whole-cell currents (figure 3A). The occurrence of IK channels in 75% of basolateral membrane patches in human colonic crypts,30 may now be at least partially explained by the selective cellular distribution of IK channels. The conductance of the ORK cells in the negative voltage range was low. However, owing to the large amplitude of the BK channel currents, it was possible to evaluate single BK channel activity in whole-cell recordings from ORK cells (figure 3B), which indicated a typically high unitary conductance of 215 pS.31 Such single channel activity can also be seen in figure 1B. BK channels with a conductance of this magnitude are found mainly in the apical membrane of surface cells in human colon,18 but not in the basolateral membrane of human colonic crypt cells, in which 138 pS K+ channels have previously been identified in 25% of cell-attached patches.30 In our study, 138 pS K+ channels were seen in 19/123 (15%) of cell-attached basolateral membrane patches which also exhibited IK channel activity (ie, in IRK cells), but not during whole-cell recordings from ORK cells, in which only 215 pS (BK) K+ channels could be identified.
Evidence for two cell types from channel activation
A clinically relevant feature of small intestinal and colonic epithelia is their capacity for cAMP-induced Cl− secretion via apical CFTR.7 ,32 Increasing cytosolic cAMP with 10 µM forskolin elicited a marked rise in whole-cell conductance in IRK cells, which was accompanied by membrane depolarisation (Δ +26±6 mV; n=11) (figure 4A,C,D). The cAMP-activated whole-cell conductance in IRK cells was insensitive to 200 µM 4,4-diisothiocyanatostilbene-2,2-disulfonic acid (DIDS; n=4; p>0.05), but was inhibited 86±18% by 200 µM 5-nitro-2-(3-phenylpropylamino)benzoic acid (NPPB; n=4; p<0.05, figure 4B), which has a similar effect on whole-cell Cl− currents in pancreatic duct cells.33 Its biophysical characteristics and marked sensitivities to cAMP and NPPB identifies this conductance within IRK cells as CFTR. By contrast, ORK cells were insensitive to forskolin (figure 4A,C,D).
Evidence for two cell types from immunostaining
We have previously identified BK channel β3-subunit mRNA in human colonic crypt cells by RT-PCR.18 To determine whether BK channel protein was expressed in colonocytes or goblet cells, we stained with antibodies to both the BK channel β3-subunit and the goblet cell marker, MUC2.34 BK channel β3-subunit and MUC2 proteins were co-localised in cells within the human colonic crypt (figure 5). MUC2-containing goblet cells had a distinctive flower head-like appearance and were clearly delineated from the other cells in the crypts. Thus, BK channels and MUC2 are expressed in goblet cells, whereas IK channels are present in CFTR-containing cells, which we take to be colonocytes.
Two distinct populations of cells were identified in intact human colonic crypts on the basis of their whole-cell current characteristics: inwardly rectifying IK channels were present only in cells containing cAMP-activated CFTR (IRK cells), whereas outwardly rectifying BK channels were present in cells which did not contain cAMP-activated CFTR (ORK cells). BK channel β3-subunit protein always co-localised with the goblet cell marker MUC2 (figure 5), indicating that BK channels which mediate colonic K+ secretion are located in goblet cells. Furthermore, IK channel immunoreactivity has been identified in colonocytes but not mucus cells in rat colon.35 In addition, whole-cell currents inhibited by the IK channel blocker clotrimazole were present in 73% of cells in the mid-crypt region, whereas whole-cell currents inhibited by the BK channel blocker penitrem A were present in 27% of cells. Although possibly coincidental, this electrophysiological partitioning was almost identical to the 75%:25% histological partitioning of colonocytes and goblet cells in human colonic crypts.24 In addition, it should be noted that BK channel β3-subunit protein only co-localised to MUC2-containing goblet cells. We did not perform quantitative confocal imaging, but close inspection of the DAPI image (figure 5) shows nuclear staining of ∼140 cells, while the merged image shows co-localisation of MUC2 antibody and BK channel β3-subunit antibody in ∼30 cells, equivalent to 21% of the total.
Thus, our results strongly suggest that Cl− and K+ secretion originates from cell types which are different and distinct. We propose that Cl− originates from colonocytes expressing apical CFTR, whereas K+ originates from goblet cells expressing apical BK channels (figure 6). Our findings differ from the generally held view that both Cl− and K+ are secreted by a single cell type—namely, colonocytes.2 However, they are consistent with the results of recent studies in rat distal colon using the Ussing chamber technique, which showed that cAMP-stimulated K+ secretion is mediated via apical BK channels, whereas cAMP-stimulated Cl− secretion requires the activity of basolateral IK (KCNN4b) channels and is independent of the concurrent stimulation of apical BK channels.36 Furthermore, our electrophysiological data showing that IK channels, which are known to dominate the basolateral membrane, are expressed together with CFTR in colonocytes but not goblet cells, are consistent with several pieces of evidence linking IK channel activity to Cl− secretion. For example, we have previously demonstrated that carbachol stimulates basolateral IK channels in human colonic crypt cells,19 and carbachol-induced Cl− secretion has been shown to depend on the simultaneous activation of basolateral Ca2+-sensitive IK channels and apical CFTR.37 Moreover, intestinal Cl− secretion is abolished by the IK channel inhibitors CLT and charybdotoxin,26 and IK channel knockout mice have a significantly lower stool water content and a complete absence of Ca2+-mediated Cl− secretion compared with wild-type animals.38
In addition, we have demonstrated marked downregulation of basolateral IK channels in patients with active ulcerative colitis, where the main transport defect is a failure of Na+ absorption (and hence Cl− and water absorption), rather than increased Cl− secretion.39 We have also previously reported molecular and electrophysiological evidence for the existence of low conductance (∼7 pS) basolateral K+ channels (KCNE3/KCNQ1) in human colonic crypt cells, which have been implicated in Cl− secretion and are activated by cAMP and inhibited by chromanol 293B.40 ,41 Although we did not study these channels in this study, they may coexist with basolateral IK channels in IRK cells, even though they may not be present in ORK cells. In the case of goblet cells, basolateral anion channels may provide the driving force for K+ secretion, the most likely candidate being the ClC-2 Cl− channel. This inward rectifying Cl− channel is present in the basolateral membrane of surface cells in mouse colon and in human Caco-2 cells,42 ,43 as well as in surface cells of guinea pig distal colon.44 There may well be significant interspecies variability in ClC-2 channel expression and in the sensitivities of the antibodies used, since immunostaining in human colon detected ClC-2 protein mainly in the cytosolic supranuclear region and at the apical pole of surface colonocytes and in cells in the upper two-thirds of colonic crypts.45 None of these studies of ClC-2 Cl− channel localisation differentiated between separate types of colonic epithelial cells, as we did in this study. The pattern of ClC-2 immunostaining in human colon does not therefore preclude the presence of sufficient basolateral ClC-2 channels to support K+ secretion by goblet cells.
Previous studies have tended to focus on the distribution of BK channels along the surface cell–crypt cell axis, or between the apical and basolateral membrane domains. Immunostaining has identified BK channel protein predominantly in surface and upper crypt cells of human and rabbit colon,18 ,46 but within crypt cells in mouse colon.8 Apical membrane staining was seen in all of these species, with additional basolateral membrane staining in the rabbit.46 Patch clamp studies also showed BK channels localised to the apical membrane of surface cells in human and rat colon,18 ,47 whereas BK channels with a unitary conductance of >200 pS were absent from basolateral membrane patches from human colonic crypts.30 It should be noted that while the immunostaining of BK channels was most marked in surface and upper crypt cells in human colon, close inspection of the immunostaining along the surface–crypt axis shows that cells in the lower parts of the crypt are not entirely devoid of BK channel protein,18 which probably explains the presence of BK channels in ORK cells in the mid-third of crypts in this study.
We also found that the pattern of immunostaining of BK channel protein along the surface–crypt axis was similar in human ascending and sigmoid colon,18 and it seems reasonable to speculate that the differentiation between ORK and IRK cells in crypts from the sigmoid applies throughout the colon. Although apical BK channels from rat colon were activated by cAMP,48 we found in this study that the outwardly rectifying whole-cell K+ currents in ORK cells from human colonic crypts (which reflect BK channel activity) were unchanged by cAMP (figure 4A). Since human and rat colonic BK channels have a similar unitary conductance under symmetrical high-K+ conditions (∼210 pS), the species difference in BK channel sensitivity to cAMP may reflect differences between BK channel α-subunit splice variants and/or β-subunits, which have both been shown to dramatically alter the regulation of BK channels by cAMP and Ca2+.15 ,49 ,50 The basolateral 138 pS K+ channels previously identified in human colonic crypt cells,30 and again in this study, are likely to be significantly different from apical BK channels. Although their function is unclear, they are found in basolateral membrane patches containing IK channels (ie, in Cl−-secreting colonocytes),51 but apparently not in K+-secreting goblet cells.
To our knowledge, this study is the first to demonstrate the presence of the BK channel β3-subunit protein within human colonic epithelium and the first to identify BK channels within colonic goblet cells but not within Cl−-secreting colonocytes. It is becoming clear that apical BK channels play an important role in colonic K+ secretion, as shown by the correlation between BK channel expression and colonic K+ secretion in patients with active ulcerative colitis and end-stage renal disease,18 ,52 as well as loss of K+ secretion in BK channel knockout animals.8 The notion that CFTR and BK channels are expressed in different cell types within colonic epithelia is consistent with the earlier view that colonic Cl− and K+ secretion are regulated independently by secretagogues,53 and show differing sensitivity to bumetanide.54 Interestingly, optical and electrophysiological studies demonstrated that cAMP-activated Cl− secretion and mucin secretion probably originate from distinct cell types in isolated rabbit colonic crypts.55 Similarly, there is evidence that isolated overactivity of either CFTR or BK channels may play an important role in certain clinical situations. For example, stimulation of CFTR during acute cholera results in Cl−-driven secretory diarrhoea.56 On the other hand, overexpression of colonic apical BK channels (and by inference, enhanced colonic K+ secretion) has been described in a patient with severe diarrhoea, massive stool K+ losses and profound hypokalaemia.6 This raises the possibility that pharmacological activation of colonic BK channels may prove useful in the treatment of idiopathic constipation.57 Indeed, the presence of apical BK channels in mucus-secreting epithelial cells may have wider clinical implications. If goblet cells present in abundance in bronchi and small intestine also express apical BK channels, stimulation of K+ secretion using BK channel activators could be explored as a new method of increasing fluid production in the respiratory and alimentary tracts of patients with cystic fibrosis.
In summary, our results suggest that Cl− secretion originates from colonocytes, whereas colonic K+ secretion appears to be mediated by goblet cells, and these secretory processes are likely to be regulated independently. These findings may therefore have a bearing on strategies for developing new anti-diarrhoeal drugs, and also point to BK channels as a potential target for eliciting K+-driven epithelial fluid secretion in cystic fibrosis.
The authors thank Jim Deuchars and Ian Edwards for their guidance on immunofluorescence.
Contributors JL, MH and GIS conceived the experiments, analysed the data and were involved in writing the paper. JL, AL, SK and MH carried out the experiments and acquired and analysed the data. All authors approved the submitted version of the paper.
Funding Yorkshire Cancer Research (project grant L309) and the Leeds Teaching Hospitals NHS Trust Charitable Foundation.
Competing interests None.
Patient consent Obtained.
Ethics approval Leeds Health Authority ethics committee.
Provenance and peer review Not commissioned; externally peer reviewed.