|Home | About | Journals | Submit | Contact Us | Français|
Lubiprostone is a chloride channel activator in clinical use for the treatment of chronic constipation, but the mechanisms of action of the drug are poorly understood. The aim of this study was to determine whether lubiprostone exerts secretory effects in the intestine by membrane trafficking of ion transporters and associated machinery.
Immunolabeling and quantitative fluorescence intensity were used to examine lubiprostone-induced trafficking of the cystic fibrosis transmembrane conductance regulator (CFTR), sodium/potassium-coupled chloride co-transporter 1 (NKCC1), electrogenic sodium/bicarbonate co-transporter 1 (NBCe1), down-regulated in adenoma (DRA), putative anion transporter 1 (PAT1), sodium/proton exchanger 3 (NHE3), Ca2+ activated chloride channel 2 (ClC-2) serotonin and its transporter SERT, E prostanoid receptors EP4 and EP1, sodium/potassium ATPase (Na-K-ATPase) and protein kinase A (PKA). The effects of lubiprostone on mucus exocytosis in rat intestine and human rectosigmoid explants were also examined.
Lubiprostone induced contraction of villi and proximal colonic plicae and membrane trafficking of transporters that was more pronounced in villus/surface cells compared to the crypt. Membrane trafficking was determined by: (1) increased membrane labeling for CFTR, PAT1, NKCC1, and NBCe1 and decreased membrane labeling for NHE3, DRA and ClC-2; (2) increased serotonin, SERT, EP4, EP1 and PKA labeling in enterochromaffin cells; (3) increased SERT, EP4, EP1, PKA and Na-K-ATPase in enterocytes; (4) and increased mucus exocytosis in goblet cells.
These data suggest that lubiprostone can target serotonergic, EP4/PKA and EP1 signaling in surface/villus regions; stimulate membrane trafficking of CFTR/NBCe1/NKCC1 in villus epithelia and PAT1/NBCe1/NKCC1 in colonic surface epithelia; suppress NHE3/DRA trafficking and fluid absorption; enhance mucus-mobilization and mucosal contractility.
Lubiprostone is an FDA-approved prostaglandin E1 derivative in clinical use for the treatment of chronic constipation  and irritable bowel syndrome with constipation in adults . Enhancement of luminal liquidity is a major factor in its efficacy in patients as the majority report bowel movement within 24 hours . Lubiprostone also increases fasting gastric volume, delays gastric emptying, accelerates transit in the small intestine and all colonic segments , and has minor effects on colonic tone and compliance . However, the underlying mechanisms responsible for these observations are not fully established. According to the manufacturer, lubiprostone acts locally within the gastrointestinal (GI) tract . How therapeutic concentrations of the drug are achieved in the colon to stimulate bowel contractions is not clear.
Lubiprostone stimulates electrogenic Cl− secretion in human and rodent intestinal epithelia [6-9]. There is evidence that the E prostanoid receptor 4/cAMP/protein kinase A/cystic fibrosis transmembrane conductance regulator (EP4/cAMP/PKA/CFTR) pathway is involved in lubiprostone-induced Cl− and HCO3− secretion [6, 9-11]. But cAMP/CFTR mediated secretion  cannot fully explain lubiprostone's mild diarrhea-inducing profile since it has been shown to be beneficial in CF patients where CFTR is dysfunctional [1-2, 12]. Of note, misoprostol, a prostaglandin E1 derivative ulcer-treating drug, promotes HCO3− and mucus secretion, induces CFTR-mediated fluid secretion in jejunum of healthy subjects, but also reduces Na+/H+ exchanger 3 (NHE3) mediated fluid absorption in CF patients in a CFTR-independent manner . Since E1 prostaglandins regulate Na-K-ATPase and salt absorption through both cAMP and Ca2+ pathways , and cAMP and Ca2+ can inhibit (NHE3) mediated fluid absorption, a similar cAMP/Ca2+ co-activation by lubiprostone cannot be excluded. Furthermore, it has been suggested that lubiprostone could activate an alternative Cl− exit pathway, such as the voltage-gated ClC-2 channel [7-8, 15]. However, direct evidence linking lubiprostone to apical ClC-2 activity in the intestine has not been provided [6, 9]. Lubiprostone was reported to activate ClC-2 without affecting CFTR in human colon T84 cells . However, two subsequent studies conducted in the same cell model reported the opposite: that lubiprostone activates CFTR without affecting ClC-2 [6, 9]. Furthermore, studies in kidney cells suggested that lubiprostone could activate ClC-2 at low (< 100nM) but CFTR at higher concentration. But these conclusions have similarly been challenged [9,15]. Thus, there remains a lack of consensus regarding lubiprostone's mechanism of action.
Second messenger dependent trafficking of transporters from subcellular endosomes to the plasma membrane is an important physiological mechanism underlying intestinal anion secretion and can be demonstrated in native tissues using indirect morphological approaches . We hypothesized that lubiprostone's secretory effects [7, 10] could involve trafficking of intestinal ion transporters including CFTR, the basolateral Na+/K+/2Cl− co-transporter 1 (NKCC1), the electrogenic basolateral Na+/HCO3− co-transporter 1 (NBCe1), and the apical Cl−/HCO3− exchangers PAT1 (a.k.a. Putative Anion Exchanger 1; SLC26A6) and/or DRA (a.k.a. Down-Regulated in Adenoma; SLC26A3). On the other hand, lubiprostone–induced inhibition of fluid absorption could be explained by trafficking of apical NHE3. Because of the lack of consensus on the role of ClC-2 in mediating lubiprostone's action, we examined the distribution of ClC-2 using three independent antibodies, and tested lubiprostone's influence on CFTR and ClC-2 membrane trafficking. We also assessed its effects on E prostanoid receptors 4 and 1 (EP4 and EP1) because we identified both receptors in enterochromaffin (EC) cells in the intestine. Similar to naturally occurring prostaglandins lubiprostone exerts pro-secretory effects in small and large intestine through binding to prostanoid receptors on epithelial cells to stimulate intracellular adenyl cyclase and cAMP . Finally, lubiprostone's effect on mucus exocytosis was examined because the amount and composition of intestinal mucus can greatly influence both liquidity and transit of luminal contents, and nothing is known about lubiprostone's effects on mucus release. The results of this study using human and rat intestine suggest that lubiprostone has multiple targets; stimulating CFTR-dependent as well as CFTR- and ClC-2-independent anion secretion, suppressing fluid absorption, enhancing mucus secretion and promoting local mucosal contractility.
Rectosigmoid biopsy specimens (provided by Dr. Anthony Bauer; University of Pittsburgh, PA) were obtained from healthy subjects after informed consent. The study was approved by the Institutional Review Board (IRB) of the University of Pittsburgh. Colonic biopsies (1.5 mm × 3 mm) were maintained in Dulbecco's modified Eagle medium and left untreated, or treated ex vivo with normal saline (pH 7.4) or 2.5 μg/ml or 7.5 μg/ml lubiprostone (mol. w. 390.46 g) in saline (Takeda Pharmaceutical North America, Deerfield, IL, USA) for 30 minutes at 37 °C in 5% CO2/90% air atmosphere incubator. Tissues were fixed in 2% PFA for 1 hour.
The Institutional Animal Care and Use Committee of Yale University School of Medicine approved the study. Male Sprague-Dawley rats (200-250 g, Charles River Laboratories) were fasted overnight but allowed access to drinking water, anesthetized with Inactin (120 mg/kg i.p.), and euthanized (Inactin; 200 mg/kg, i.p.) at the end of the experiment.
Ligated jejunal, ileal and proximal colonic loops (~3 cm length) were instilled with 0.1 ml normal saline (pH 7.4, 37 °C) or 0.25 μg (0.64 nMol) lubiprostone in 0.1 ml saline (6.4 nMol/ml; i.e. 2.5 μg/ml) for 30 min.
Intestinal segments were isolated, fixed and processed for immunofluorescence and histochemical labeling. See Supplementary Methods for a detailed description and full list of primary antibodies.
Immunolabeled sections were examined on a Nikon Eclipse E800 epifluorescent microscope equipped with a Hamamatsu Orca R2-C10600 digital camera. The acquisition parameters were standardized in relation to the highest intensity regions to avoid oversaturation of pixel intensity. Digital images (8 bits/channel; 1344 × 1024 pixels) were taken at 40× magnification and same exposure time. The images were digitally converted into a grey scale image for analysis. Densitometry analysis was performed as described before . The Record Measurements tool of Adobe Photoshop CS4 Extended provided the pixel intensities (mean gray value) of the selected areas. For background correction, pixel intensity values from the lamina propria were subtracted. For apical membrane labeling, ~2.0 μm thick ribbons were traced centered on apical fluorescence. For lateral membrane labeling, ~1.5 μm thick ribbons were centered on lateral membrane fluorescence. For intracellular apical pole labeling in enterocytes, ~3.0 μm thick ribbons were traced. Fluorescence intensity (FI) levels were measured on digital images as before : (1) FI levels for CFTR, ClC-2, DRA, PAT1, EP1, EP4, NBCe1, NKCC1, NHE3 and SERT over the intracellular apical pole, apical membrane, or lateral membrane, and over entire cells for 5-HT (2) FI levels in short rows of mucosal cells were measured for Na-K-ATPase and PKA. Data from 4 to 12 selected areas were averaged in each image, 6 images were analyzed for each measurement group in one animal, and data were collected from three animals (n=3). The number of mucus-filled and empty goblet cells were counted from at least six 10× images of Alcian Blue/PAS stained tissue per animal. All values were presented as means ± SEM. Statistical significance between two individual measurement groups was determined by unpaired t-test. Differences among groups were determined using oneway ANOVA and the Tukey's post hoc method of multiple comparisons. The level of significance was set at P< .05.
The mucosal layers in human and rodent colon are organized into ridges called plicae circulares which increase the surface area by ~ 3-fold. Small intestinal villi further increase surface area by ~10-fold. All lubiprostone-treated proximal colonic plicae (consisting of mucosal sheets and lamina propria) were contracted (see Figs. 1A, 4C-D; 5A-B). Area measurements revealed that the mucosa shrunk from 100% to 86.5%. The lamina propria shrunk to 44%, less than half of controls, and was the main contributing factor to plica shrinkage. These size reductions resulted in 22% increased gut lumen area (see graphs in Fig. 1A). Many lubiprostone treated villi in jejunum (Fig. 6A and D) and ileum were also contracted, exhibiting a “Christmas-tree” shape. Contracted and non-contracted villi exhibited similar anion transporter distribution profile. Contractions were not observed in untreated and saline-treated intestine.
Since lubiprostone-induced intestinal contractions have been shown to be EP1 and EP4-mediated , we investigated the drug's effect on the distribution of these receptors. In rat proximal colonic mucosa, both EP4 (Fig. 1A-C) and EP1 receptor-positive cells were detected by immunolabeling (Fig. 1D-E). Labeled cells included colonic enterocytes (also called colonocytes) and 5-HT-positive and 5-HT-negative enterochromaffin (EC) cells (Fig. 1C-E). In untreated and saline treated conditions, EP4 and EP1 labeling were predominantly intracellular (Fig. 1B, 1D-E) in both surface enterocytes and 5-HT-labeled EC cells. EP4 labeling was prominent in surface cells and low in crypt cells, whereas EP1 labeling was relatively even in surface and crypt cells (see graphs of Fig. 1F). Consistent with our findings, both EP4 and EP1 were detected in ileal EC cells in Mastomys . EP4-positive cells were also detected in the lamina propria (Fig. 1B).
Lubiprostone increased EP4 labeling in proximal colonic surface mucosa (Fig. 1A-B). EP4 labeling shifted from intracellular stores to the apical membrane of enterocytes and cell membranes of EC cells (Fig. 1B). Membrane-localized EP4 increased ~ 3.3-fold in surface EC cells and ~ 3.6-fold in surface enterocytes, and intracellular EP4 decreased in parallel (Fig. 1F). In 5-HT/EP4 double-labeled tissues, EP4 was recruited to the cell membrane of serotonergic (Fig. 1C) and non-serotonergic surface EC cells (not shown). In contrast, in crypts, EP4 localization and labeling intensity did not change significantly (Fig. 1F). After lubiprostone treatment, EP1 labeling remained intracellular. EP1 labeling intensity increased ~ 1.7-fold in surface EC cells and ~ 1.4-fold in surface enterocytes (Fig. 1F). However, EP1 levels were not changed in crypt cells.
To determine whether lubiprostone specifically affects the EC cells, we examined the drug's effect on 5-HT levels in EC cells by measuring fluorescence intensities of 5-HT and its transporter (SERT) labeling. These experiments also examined whether lubiprostone could potentially modulate intestinal serotonin, as several other constipation drugs . Lubiprostone significantly increased 5-HT labeling in proximal colonic surface EC cells (Fig. 2A), and jejunal/ileal villus EC cells (not shown). In proximal colon, lubiprostone increased 5-HT FI ~ 1.36-fold in surface EC cells, but only ~ 1.1-fold in crypt EC cells (Fig. 2C). Lubiprostone also increased SERT labeling in proximal colon (Fig. 2B), jejunum and ileum (not shown). In proximal colon SERT FI increased ~ 2-fold in the apical membrane of surface enterocytes, ~ 1.6-fold in the cell membrane of SERT/5-HT double-labeled surface EC cells, and ~ 1.2-fold in crypt EC cells (Fig. 2D). In summary, both 5-HT and SERT labeling were mainly increased in surface/villus regions by the drug, suggesting that lubiprostone may selectively target these epithelial regions.
Could lubiprostone selectively target surface/villus epithelia at the level of ion transporters as well? In the next series of experiments we examined how lubiprostone influences the trafficking of ion transporters involved in fluid secretion/absorption, by examining changes in the subcellular distribution of transporters. In rat proximal colon, lubiprostone decreased basolateral NKCC1 in lower crypt enterocytes and goblet cells to 78%, but increased it in upper crypt enterocytes and goblet cells ~ 1.4 fold, and in surface enterocytes (~ 2.2-fold) and goblet cells (~ 1.3-fold) (Fig. 3A, 3D, 4A-B). In lower crypt cells NBCe1 levels remained low, but lubiprostone increased basolateral NBCe1 in upper crypt enterocytes ~ 1.5-fold and surface enterocytes ~ 2.3-fold (Fig. 5A-B). In summary, in the rat proximal colon, the largest lubiprostone induced increases in both NKCC1 and NBCe1 membrane labeling occurred in surface cells.
In rat proximal colon, lubiprostone decreased apical CFTR in lower crypt cells to 71%, but increased it in upper crypt cells ~ 1.45-fold (Fig. 4A-B, 4E). CFTR was not detected in surface epithelia (Fig. 4A-B). In parallel, lubiprostone increased apical PAT1 ~ 1.8-fold (Fig. 3A-B), but reduced apical DRA to 37% (Fig. 3C) and NHE3 to 61% (Fig. 5B) in surface enterocytes.
Lubiprostone induced similar ion transporter redistribution in rat jejunum (Figs. 6--7)7) and ileum (Fig. 8). In jejunum, lubiprostone (1) reduced apical membrane NHE3 of villus enterocytes to 65% (Figs. 6C, ,7A);7A); (2) increased apical membrane CFTR of crypt enterocytes ~ 1.4-fold and villus enterocytes ~ 3-fold (Figs. 6A, ,7B)7B) and also increased apical PAT1 in villus enterocytes (not shown); (3) increased lateral membrane NBCe1 of villus enterocytes ~ 1.5-fold (Figs. 6D-E, ,7C);7C); (4) increased enterocyte lateral membrane NKCC1 in crypt ~ 1.4 fold and villus ~ 2.3-fold (Figs. 6A-B, ,7D);7D); and (5) increased goblet cell lateral membrane NKCC1 in crypt by ~ 1.4-fold and villus ~ 1.3-fold (Fig. 7D). Both apical NHE3 reduction and basolateral NBCe1 increase were particularly robust in the upper villus (Fig. 6C-D). Intracellular NBCe1 partially co-localized with the early endosome marker EEA1 in saline- but not lubiprostone-treated jejunum (Fig. 6E), consistent with membrane traffic. In all tissues and for each ion transporter, intracellular FI changes were reciprocal to the respective membrane FI changes. The decrease in membrane label for NHE3 and DRA and the increased membrane label for CFTR, PAT1, NBCe1, and NKCC1 that were induced by drug were all most prominent in surface/villus regions, further supporting selectivity for lubiprostone targeting to these epithelial regions.
In human colonic biopsy explants, treatment with lubiprostone for thirty minutes increased apical CFTR ~ 1.45-fold in colonic crypts (Fig. 9A-B, and E). CFTR was not detected in surface cells (not shown). Lubiprostone also increased basolateral NKCC1 in crypt enterocytes ~1.5 fold, surface enterocytes ~2.4-fold, crypt goblet cells~1.5 fold and surface goblet cells ~1.6-fold (Fig. 9C-D, and F). NHE3 levels were low in untreated and lubiprostone-treated explants (data not shown). Membrane recruitment of CFTR and NKCC1 was accompanied by concomitant decreased intracellular labeling FI (Fig. 9E-F).
We examined the intestinal localization of ClC-2 using three independent antibodies and tested lubiprostone's influence on ClC-2 membrane trafficking, because of the lack of consensus on these issues (see Introduction). Three ClC-2 antibodies (ACL-002, H-90, and YY9) produced similar labeling patterns in human tissues. In rat tissues, only the ClC-2 (ACL-002) antiserum produced labeling patterns similar to that in human tissues. In both rat proximal colon (Fig. 5C) and human colon tissues (Fig. 10) ClC-2 labeling was prominent in surface enterocytes, but not detected in crypt cells; goblet cell labeling was near background level. In untreated (not shown) and saline treated rat proximal colon (Fig. 5C) and untreated human colon tissue (Fig. 10) ClC-2 label was partially localized to the basolateral membrane and intracellular compartments of surface enterocytes. In saline treated proximal colon (graph in Fig. 5F) normalized lateral membrane ClC-2 (1.0) was ~ 2.5-fold higher than intracellular ClC-2 (0.41). In untreated human colon surface cells (Fig. 10A-C), lateral membrane ClC-2 (1.0) was ~2 to 2.5-fold higher (depending on the ClC-2 antibody used) than the corresponding intracellular ClC-2: 0.39 – ClC-2 (ACL-002), 0.52 – ClC-2 (H-90), and 0.49 – ClC-2 (YY9) (see graphs, Figure 10). Basolateral ClC-2 labeling in rat jejunal and ileal villus epithelia was weak (not shown). In summary, in agreement with earlier studies , ClC-2 labeling was partially basolateral and partially intracellular; strong in colonic surface and weak in small intestinal epithelia.
In contrast to untreated tissues, ClC-2 label was predominantly intracellular in surface epithelium in lubiprostone treated proximal rat colon (Fig. 5C) and human colon explants (Fig. 10A-C). In lubiprostone treated rat proximal colon (see graph in Fig. 5F) intracellular ClC-2 (0.75) was ~2-fold higher than lateral membrane ClC-2 (0.36). In lubiprostone-treated human colon (graphs in Fig. 10A-C), intracellular ClC-2 (0.80 – ClC-2 (ACL-002), 0.75 – ClC-2 (H-90), and 0.72 – ClC-2 (YY9) was ~ 3.4 to 2.4-fold higher, compared to corresponding lateral membrane ClC-2 FI: 0.24 – ClC-2 (ACL-002), 0.28 – ClC-2 (H-90), and 0.30 – ClC-2 (YY9).
The observation that lubiprostone induced recruitment of NKCC1 in goblet cells suggested the drug's involvement in stimulating activity of these mucus-producing cells. Thus, we tested lubiprostone's potential influence on mucus exocytosis. To assess the presence of mucus in goblet cells, tissue sections were labeled with a Mucin 2 antibody, and/or stained with Alcian blue and periodic acid-Schiff base (AB/PAS) or Alcian blue and Toluidin blue (AB/TB). In rat proximal colon, lubiprostone reduced the ratio of mucus-filled goblet cells from 78 % to 28% (Fig.4C-D, F). Lubiprostone also induced prominent mucus exocytosis in rat jejunum (Fig. 6F), ileum (Fig. 8D) and human colon explant (Fig. 11).
Next we attempted to gather further evidence for the phenomenon that lubiprostone may preferentially target ion transporter trafficking in surface cells. We examined Na-K-ATPase labeling, since throughout the gut, fluid and electrolyte transport is driven primarily by Na-K-ATPase across the basolateral membrane. We also studied PKA labeling, since this signaling pathway plays a major role in lubiprostone-induced Cl− and HCO3− secretion (see Introduction). In untreated and saline treated rat proximal colon, Na-K-ATPase (Fig. 12A and D) and PKA labeling (Fig. 12B and E) were both most intense in upper crypts, where NKCC1 and NBCe1 levels were highest. Lubiprostone decreased Na-K-ATPase labeling in lower crypt cells to 61%, but increased it in upper crypt cells ~ 1.2-fold and ~ 1.7-fold in surface cells (Fig. 12A and D). Lubiprostone decreased PKA labeling in lower crypt cells to 55%, in upper crypt cells to 57% (Fig. 12B and E), but increased it ~ 1.6-fold in surface cells (Fig. 12A and E). Lubiprostone also increased PKA labeling ~ 1.6-fold in 5-HT-labeled surface EC cells, without significantly changing PKA levels in crypt EC cells (Fig. 12B inset, and F). In summary, lubiprostone (in parallel with the labeling increase for NKCC1 and NBCe1) also induced the largest increase in Na-K-ATPase and PKA labeling intensity in surface cells. We examined p-catenin levels to further confirm this effect, since PKA activity is known to regulate β-catenin levels at cell borders. Indeed, lubiprostone selectively increased β-catenin labeling in surface cells (Fig. 12C).
Overall, the results suggest that lubiprostone preferentially targets small intestinal villus and colonic surface epithelia, to activate fluid secretion machinery (Fig. 13). Specifically lubiprostone selectively increases NKCC1, NBCe1, PKA and Na-K-ATPase labeling intensities in villus/surface cells, and appears to signal EP4/PKA, EP1, and serotonergic pathways in these regions. Although previous studies demonstrated lubiprostone activation of EP4 receptors, the current study extend those observations by identifying EP4 and EP1 receptors in intestinal 5-HT-containing EC cells and effects of lubiprostone on these cells, particularly in villus/surface regions . These observations suggest that at least part of the drug's action could perhaps be due to 5HT4 action in the colon. EP4 and EP1 have also been shown to coexist in serotonergic ileal EC cells . Thus, lubiprostone may also exert some action by modulating serotonin levels . We also found that lubiprostone possesses multiple cellular targets and transporters in the intestine. The data suggest that lubiprostone can suppress apical NHE3/DRA- and basolateral ClC-2 mediated fluid absorption; and stimulate (1) CFTR-dependent anion secretion in crypt and villus epithelia, (2) PAT1-dependent, but CFTR- and ClC-2-independent anion secretion in colonic surface epithelia, (3) mucus exocytosis from goblet cells, (4) transformation of condensed mucus to mobile expanded mucus via enhanced alkaline secretions, and (5) mixing/movement of luminal content via mucosal contractions. As a luminally active agent, lubiprostone may transiently enhance a combination of events in intestinal segments of limited length. This profile may contribute to its constipation-alleviating effects, without disturbing nutrient absorption.
The therapeutic dose of lubiprostone for clinical use is 24 μg (61.4 nMol). About 100-fold less lubiprostone, 0.25 μg (0.64 nMol) was injected into ~3-cm long ligated segments of the rat intestine (considering the two-magnitude difference between human and rat body weight). However, in human patients, lubiprostone is partially metabolized in the stomach, and then distributed to at least 100-fold longer intestinal segments with 104 to 105-fold larger mucosal surface areas, compared to a ~3 cm ligated loop. Therefore, lubiprostone effects described here could represent effects induced by the drug at 102-103 fold higher local concentration compared to that of human intestine. The potential clinical implications of our findings will require further investigation, because one cannot reliably compare local lubiprostone concentrations in the intestine of patients versus those affecting ligated rat intestinal loops and human colon explants.
In untreated and saline treated tissues, CFTR (part apical, part intracellular) was localized to enterocytes of small intestinal crypts and villi and colonic crypts, as shown before . CFTR was not detected in rat and human colon surface enterocytes, confirming earlier reports . As shown before in rat , NKCC1 present in enterocytes and goblet cells was largely basolateral in crypt cells, and mainly intracellular along small intestinal villus and colonic surface epithelia. The current study extended these findings to the human colon. NBCe1 (part basolateral, part intracellular) was present in small intestinal villus enterocytes and colonic crypt and surface enterocytes. NHE3 (mainly apical membrane and also intracellular) was localized to small intestinal villus and colonic surface enterocytes, as described .
The recruitment of NKCC1 to the basolateral membrane of small and large intestinal epithelial cells by lubiprostone is consistent with NKCC1-mediated Cl− secretion [6-7, 10]. Increased apical CFTR in rat and human intestine is consistent with previous observations that lubiprostone induces apical recruitment of CFTR in cultured human intestinal T84 cells , and with the findings that CFTRinh-172 suppressed lubiprostone-induced Cl− transport, confirming its dependence on CFTR [6, 9-10]. The lubiprostone-induced NHE3 internalization is also consistent with the drug's anti-absorptive properties, and the concomitant DRA internalization supports the notion that NHE3 and DRA are functionally coupled . Acting upon EP4/PKA-containing enterocytes, lubiprostone could activate the well-established inhibitory effect of cAMP/PKA signaling on salt and water absorption. Detection of ClC-2-type currents as the dominant anion conductance of enterocyte basolateral membrane  suggests that enterocytes extrude Cl− through basolateral ClC-2 during fluid absorption. The lubiprostone-induced removal of ClC-2 from the basolateral membrane suggests that lubiprostone suppresses, rather than stimulates ClC-2 activity. This effect appears to be another anti-absorptive property of the drug. The basolateral localization of ClC-2 at steady state condition and its lubiprostone-induced internalization are at odds with prevailing hypotheses: that (1) ClC-2 is an apical channel and that (2) lubiprostone's pro-secretory actions would involve the activation of ClC-2 [8, 15].
Lubiprostone's “dual action”, recruiting apical CFTR, and basolateral NKCC1/NBCe1 to the membrane, while internalizing apical NHE3 in small intestinal villus enterocytes is probably an EP4/PKA/cAMP mediated effect [6, 10]. The upper villus and colonic surface exhibit the most robust NBCe1 and NHE3 responses (see Figs. 5 and 6C-D) and both regions are also distinctly enriched in EP4 protein (present study) and mRNA . These data suggest that the upper villus and the colonic surface epithelia may be a major target site for the drug.
The unstimulated colonic surface epithelium absorbs salt and fluid from the lumen: in this condition, apical NHE3 and basolateral ClC-2 activities promote lumen-to-serosa fluid flow (i.e. fluid absorption), while intracellular NKCC1 and NBCe1 gate serosa-to-lumen fluid flow (i.e. fluid secretion). The lubiprostone-induced membrane trafficking changes suggest a switch to fluid secretion: basolateral membrane recruitment of NKCC1 and NBCe1 allow serosa-to-lumen Cl−/HCO3− flow, while internalized NHE3 and ClC-2 limit lumen-to-serosa Na+/Cl− flow. As our results indicate, HCO3− can exit to the lumen through apical PAT1 (SLC26A6) anion exchanger, which does not require CFTR . However, the apical channel for Cl− exit is uncertain in the absence of apical CFTR and ClC-2 in colonic surface epithelia: here lubiprostone may activate apical Cl−channels not identified in this study including calcium activated chloride channels .
Lubiprostone possesses EP1 and EP4 agonist properties [6, 10, 26]. Our findings indicate that intestinal villus/surface enterocytes and EC cells possess both receptors, and our data suggest that the drug acts upon these receptors. EP1 activation increases intracellular Ca2+, whereas EP4 activation increases cytoplasmic cAMP . cAMP/Ca2+ co-activation elicits synergistic, but time-constrained biphasic secretory response: cAMP (forskolin) activation alone sustains NKCC1 activity over 4 hours, but forskolin/acetylcholine (Ca2+ agonist) co-stimulation promotes NKCC1 internalization and fluid secretion declines within 30-60 min in human colon ex vivo . In our lubiprostone-treated human colon explants, NKCC1 reverted to an intracellular localization after 60 min (data not shown) suggesting co-stimulation of cAMP/Ca2+ pathways. Similar to this scenario, it is possible that the lubiprostone-induced EP1 internalization (accompanied with a EP1 labeling intensity increase) observed after 30 min is a secondary event that follows a rapid EP1 activation. Furthermore, in rat jejunum, lubiprostone elicited a transporter trafficking profile and cell shrinkage similar to that induced by carbachol, but the lubiprostone-induced changes were less robust . The physiological findings that both forskolin [6, 29] and carbachol  enhanced lubiprostone effects also implicate cAMP/Ca2+ co-activation as a mechanism of action for lubiprostone. Lubiprostone enhanced both carbachol  and forskolin effects , and its secretory response appeared biphasic with a peak and decline within 35 min  or 40 min . The time-constrained biphasic secretory response characteristic of cAMP/Ca2+ (EP4/EP1) co-activation could explain lubiprostone's mild diarrhea-inducing effect.
Lubiprostone has been shown to stimulate HCO3− secretion in duodenum . Recruitment of NBCe1 to the basolateral membrane of small and large intestinal enterocytes observed in this study suggests that a lubiprostone-induced HCO3− secretion may also occur in jejunum, ileum and colon. This effect is most likely CFTR-dependent, since CFTRinh-172 has been shown before to decrease lubiprostone-induced HCO3− secretion . However, colonic surface cells, where CFTR is not detectable, but NBCe1 is upregulated by lubiprostone, may be a source of CFTR-independent HCO3− secretion. The mucus exocytosis data and the increased basolateral NKCC1 in goblet cells indicate that lubiprostone can enhance intestinal mucus release, similar to the airway . After the removal of mucus “plug”, colonic goblet cells may secrete HCO3− through uptake by the Ca2+ activated basolateral bestrophin-2 channel . Polyanionic mucins, stored in granules with Ca2+ and protons screening the negative charges, expand 1000-fold when exocytosed . HCO3− plays a key role unmasking the negative charges , promoting access of proteases to mucins; thus producing low-viscosity mucus. Lubiprostone's ability to load the lumen with low-viscosity mucus could contribute to its therapeutic effect . In human subjects lubiprostone accelerates intestinal transit , but the same research group  could not replicate the effect in subjects who performed overnight colon cleansing. This mucus-depleting procedure could accelerate lubiprostone's degradation in proximal intestine, and prevent the drug from reaching distal intestinal segments.
Constipation and distal intestinal obstruction syndrome (DIOS) due to viscous accumulation of mucus are common in Cystic Fibrosis patients and are relieved by lubiprostone . Two CFTR-independent mechanisms may play a role: (1) an inhibitory effect on NHE3-mediated fluid absorption ; and (2) lubiprostone's potential to alkalinize mucus independent of CFTR could directly target the “mucoviscidosis” in CF patients. Indeed, recent observations suggested that lubiprostone could promote mucus expansion in CF mice . The drug could simultaneously stimulate alkaline secretions  and mucus release that may be beneficial for treating duodenal ulcer disease, under conditions where the alkaline response to acid is impaired.
Lubiprostone has been shown to induce EP1-mediated contraction in stomach longitudinal muscles, and EP4-mediated inhibition of contractions in colon circular muscles . EP1-mediated Ca2+ mobilization contracts smooth muscle cells, whereas EP4-mediated cAMP increase induces relaxation. The lubiprostone-induced contraction of small intestinal villi and proximal colonic ridges observed in this study resembles the piston-like villus motility, that is experimentally inducible e.g. by increasing capillary fluid filtration .
In summary, by selectively targeting serotonergic, EP4/PKA and EP1 signaling in surface/villus regions, lubiprostone may (1) preferentially stimulate CFTR/NBCe1/NKCC1-dependent anion secretion by villus epithelia and PAT1/NBCe1/NKCC1-dependent, but CFTR/ClC-2-independent anion secretion by colonic surface epithelia (2) suppress NHE3/DRA- mediated fluid absorption (3) augment mucus-mobilization and (4) promote bowel movement via enhanced mucosal contractility.
We thank Drs. Fred Gorelick, Marie Egan, Carol Soroka, and Ms. Nadia Hoekstra for reviewing the manuscript.
Grant Support: This study was supported by Takeda-Sucampo Pharmaceuticals and by NIH R01 DK 077065 grants to N.A. and P30 DK 34989 grant to the Yale Liver Center and Digestive Diseases Research Core at Yale University.
Competing interests: None to declare.