Inhibition of Na+/H+ exchanger isoform 3 improves gut fluidity and alkalinity in cystic fibrosis transmembrane conductance regulator‐deficient and F508del mutant mice

Constipation and intestinal obstructive episodes are major health problems in cystic fibrosis (CF) patients. Three FDA‐approved drugs against constipation‐prone irritable bowel syndrome were tested for their ability to increase luminal fluidity and alkalinity in cystic fibrosis transmembrane conductance regulator (CFTR) null (cftr−/−) and F508del mutant (F508delmut/mut) murine intestine.


| INTRODUCTION
With markedly improved clinical outcomes for lung disease and therefore much longer life expectancy in cystic fibrosis (CF) patients, CFrelated intestinal disease becomes an increasing burden for many patients. Malnutrition in CF patients has been attributed to pancreatic insufficiency, low duodenal and upper jejunal pH and has indeed been amenable to pancreatic enzyme substitution and gastric acid blockade.
In contrast, the incidence of distal intestinal obstruction and chronic obstipation increases as patients reach adulthood and is particularly high in CF patients with severe phenotypes that have received a lung transplant (Abraham & Taylor, 2017;Houwen et al., 2010;Mavilia, 2019). A recent database search confirmed CF-related intestinal disease to be a "very poorly studied area compared to respiratory complications in CF" (Green, Gilchrist, & Carroll, 2018, p. 2) and demonstrated that no consensus guidelines exist for its treatment (Green, Gilchrist, & Carroll, 2018).
The cystic fibrosis transmembrane conductance regulator (CFTR) channel is strongly expressed in all segments of the mammalian intestine, with crypt predominance (Ameen, Alexis, & Salas, 2000;Strong, Boehm, & Collins, 1994). CFTR activation, followed by fluid and HCO 3 − secretion will result in a "flushing" of the crypts and is necessary for proper mucus granule expulsion (Liu, Walker, Ootani, Strubberg, & Clarke, 2015), mucin release from the membrane (Schutte et al., 2014), and mucus expansion and hydration (Yang, Garcia, & Quinton, 2013). In CF patients and CF animal models, dilated and mucus filled cryptal lumina and signs of mucosal inflammation have been observed, which are considered to be major drivers of mucus hypersecretion and intestinal obstruction (Munck, 2014). The restoration of the flushing effect in the cryptal base will only be possible with new treatment strategies that target the basic defect (Burgener & Moss, 2018) or with an activation of crypt-expressed "alternative" anion channels (Li, Salomon, Sheppard, Mall, & Galietta, 2017). Whether these recently introduced therapies that target the basic defect have any effect on intestinal secretory function in patients with CF-related intestinal obstructive disease is unknown.
CFTR-deficient mice do not show major histopathological alterations in their lungs and pancreas, which is possibly related to the expression of alternative anion channels in those organs (Clarke et al., 1994). In contrast, they are strongly affected in the intestine and the majority of them die of intestinal obstructions during the weaning period, if not on lifelong treatment with osmotic laxatives (Clarke, Gawenis, Franklin, & Harline, 1996). Interestingly, they also show a marked survival benefit if the major intestinal sodium absorptive transporter, the NHE3 (Slc9a3) Na/H exchanger, is knocked out (Bradford, Sartor, Gawenis, Clarke, & Shull, 2009). This suggests that an inhibition of fluid absorption may also improve intestinal obstructive episodes in the CF gut. However, it has also been shown that NHE3 expression and/or function is down-regulated in CFTRdeficient small intestine (Gawenis et al., 2004). In the isolated small intestinal mucosa of CFTR-deficient mice, cAMP analogues did not inhibit 22 Na + absorption  and this was explained by a lack of CFTR-mediated cell shrinkage, with shrinkage being the NHE3-inhibiting function (Gawenis et al., 2003). However, we had previously observed an electroneutral cAMP-dependent increase in luminal alkalinization in CFTR-deficient colonic epithelium that was inhibited by the NHE3 inhibitor S1611 .
We therefore considered it important to test the ability of currently available drugs, with proven or suspected NHE3 inhibitory activity, to increase luminal fluidity and alkalinity in the small and large intestine of mice that express either no CFTR protein (cftr −/− ) or express the most frequent mutant (F508del mut/mut ) CFTR protein.
NHE3 is inhibited by a variety of agonists that result in an increase in enterocyte cAMP, cGMP or Ca 2+ levels (Zachos, Tse, & Donowitz, 2005). Linaclotide binds to the guanylate cyclase C (GC-C) in the apical membrane of enterocytes and results in cGMP production, which stimulates enterocyte CFTR-mediated anion secretion and presumably inhibits NHE3-mediated salt absorption (Steinbrecher, 2014). Lubiprostone is a PGE 1 analogue and chloride channel activator and does not elicit a short circuit response in the intestine of CFTR null mice (Ambizas & Ginzburg, 2007;Bijvelds, Bot, Escher, & de Jonge, 2009, Figures S3 and S4). Tenapanor is an intestinal-selective NHE3 inhibitor (Spencer et al., 2014). All three agents result in low systemic exposure and have been approved for What is already known • Constipation, inflammation and intestinal obstructive episodes are major health problems in cystic fibrosis patients.

What does this study adds
• This study investigates the ability of three FDA-approved drugs to increase gut fluidity and alkalinity.
• Experiments were carried out in mice expressing no CFTR protein or the F508del mutant protein.

What is the clinical significance
• These drugs may improve gut fluidity and reduce obstructive episodes in CF patients.
the treatment of constipation-prone irritable bowel (IBS-C) syndrome.
Despite an interest to apply these agents in the fight against  (Ratcliff et al., 1993), as well as with mice that were homozygous for the F508del mutation (F508del mut/mut ) (FVB/N(B6)129P2-CFTR tm1Eu ) (French et al., 1996). The strains had been bred for >10 generations on a congenic FVB/N background at Hannover Medical School. For each experimental animal, an age-and sex-matched littermate was raised under identical breeding conditions. Special breeding conditions were used for cftr −/− and F508del mut/mut mice, as well as the respective controls, as previously described (Xiao, Juric, et al., 2012): A pilot experiment was performed to study the effect of chronic application of the diet/drinking solution on the intestinal basal and NHE3-mediated fluid absorption, but no significant difference was seen between WT mice with and without the diet/drinking solution ( Figure S1).

| Histochemistry
Colonic tissues were harvested and fixed in 4% paraformaldehyde (PFA); 3 μm paraffin-embedded sections were stained for the different mucins with Alcian blue/periodic acid-Schiff stain (AB/PAS) previously described (Xiao, Juric, et al., 2012). For the determination of the  After each experiment and prior to killing, a sample of arterial blood was taken and used for blood gas analysis in a radiometer blood-gas analyser (Radiometer, Kopenhagen) and haematocrit measurement in a haematocrit centrifuge (Hematokrit 210, Hettich, Kirchlengern, Germany). The haematocrit was measured in a series of mice before and after the experiment, and subsequently in each mouse after the end of the experiment, before killing.

| Ussing chamber experiments
Because of considerable discrepancies in the literature regarding the mechanism of action and the dependency of the anion secretory effect of lubiprostone on CFTR expression (Mizumori, Akiba et al., 2009;Bijvelds, Bot et al., 2009;Musch, Wang et al., 2013;Cuppoletti, Chakrabarti, et al., 2014), we performed experiments in muscle-stripped and chemically denervated jejunal, proximal and distal colonic mucosa to study the concentration-response curve, the sidedness of its action, and its effect in cftr -/and F508del mut/mut mucosa (suppl. Figures 3,4). The experiments were performed as recently described (Liu, Tan et al., 2020). Muscle-stripped jejunum, proximal, and distal colonic mucosa was mounted in Ussing chambers with an exposed area of 0.126 cm 2 . To ensure a purely epithelial action, the muscle layers harbouring the mesenteric plexus neurons were stripped and 10 -6 M tetrodotoxin was added to the serosal side to chemically denervate the tissue. The serosal solution contained (in mM) 108 NaCl, 25 NaHCO 3 , 3 KCl, 1.3 MgSO 4 , 2 CaCl 2 , 2.25 KH 2 PO 4 , 8.9 glucose and 10 sodium pyruvate, and was gassed with 95%O2/5%CO2 (pH 7.4). Tetrodotoxin (1μM) and indomethacin (3μM) were added serosally. The mucosal solution contained (in mM) 146 NaCl, 5 KCl, 2 CaCl 2 , and was gassed with 100% O 2 . After basal parameters were measured for 30 minutes, the respective drugs, as indicated in the figure legends, were added to the mucosal or serosal solution. Bicarbonate secretion (J HCO3-) was continuously titrated for the subsequent minutes of the experiment. Transepithelial short circuit current was calculated as μEq.cm −2 tissue surface area. Measurements of J HCO3 were continuously recorded and averaged for 5-minutes periods and expressed as μmolÁcm −2 Áh −1 . The peak electrical response was averaged over 60 s and taken as Δshort circuit current response.

| Statistics
The study design included an a priori power analysis to determine the sample size for each experiment/subgroup. We used data from prior/ similar experiments to determine the effect sizes and SDs. These were used in a power analysis with the software G*Power (Faul, Erdfelder, Lang, & Buchner, 2007). To determine the necessary sample sizes, we assumed a type I error of α = 5% and a statistical power of at least 80%. The Ko/WT and mut/WT pairs were allocated to the experimental groups by a different person than the experimenter and this person also re-genotyped the mice after killing and allocated the genotype to the mouse number. A fully blinded experiment is not possible, because the experienced experimenter has a high chance of correctly guessing the genotype based on the different phenotypes of Ko and WT mice.
Sample sizes were calculated as biological replicates and were allo-  (Table 1). During perfusion with unbuffered saline, the basal fluid absorptive rates were significantly higher in the jejunum and proximal colon but not in the distal colon in both cftr −/− and F508del mut/mut compared to WT mice. The HCO 3 − output rates were significantly lower in all studied segment of the cftr −/− and F508del mut/mut than in the respective WT mice. The basal fluid absorptive rates and HCO 3 − output rates were not significantly different between F508del mut/mut and cftr −/− intestine.
T A B L E 1 Basal fluid absorptive and HCO 3 − output rates in cftr −/− , F508del mut/mut and the respective WT jejunum, proximal and mid-distal  all experiments at the given time point, whereas the dots in the right panels compare the rate before drug addition with the rate of maximal response to drug, for each mouse (shown as a dot), n = 8 (jejunum), n = 5 (proximal and mid-distal colon). *P < 0.05. NS, no significance 3.3 | Linaclotide reduces fluid absorptive rates and increases HCO 3 − output in cftr −/− and F508del mut/mut jejunum but not in the colon Linaclotide, 10 −7 M, applied luminally had been shown to elicit a robust short circuit current response in murine jejunal (Liu et al., 2020) and rat proximal colonic (Ahsan et al., 2017) isolated mucosa. In this study, it also stimulated a short circuit current response in isolated jejunal, proximal and distal colonic mucosa that was absent in cftr −/− , minimal in F508del mut/mut and robust in WT jejunal, proximal colonic and lower in mid-distal colonic mucosa ( Figure S3). This concentration is considered to be comparable to that as well as F508del mut/mut jejunum reached $25% of that seen in the respective WT jejunum (Figure 4a, FA, right panels). Surprisingly, the same effect was not seen in the colon, where a complete inhibition of fluid absorption and an increase in HCO 3 − output was observed in WT, but no significant change in both cftr −/− and F508del mut/mut mice was found, even after increasing the linaclotide concentration to 5 × 10 −7 M (Figures 3 and 4b,c,e,f).  (Figures 5d and 6d). The proximal colon was not tested due to shortage of mice.
3.5 | Tenapanor results in reduction of fluid absorptive rates and increase in HCO 3 − output in WT, cftr −/− and F508del mut/mut jejunum and mid-distal colon 3.7 | Difference in basal fluid absorptive rate between WT and ko/Mut, as well as the response to tenapanor and linaclotide, is dependent on the luminal pH and luminal CO 2 At first glance, the higher fluid absorptive rate in ko/mut jejunum (Table 1) might be explained by a lack of CFTR-mediated secretion.
However, we had not observed this difference in previous experiments in which buffered solutions had been perfused (Liu et al., 2020). During luminal perfusion of the jejunum with unbuffered solution the pH decreased in WT mice and significantly more so in ko/mut mice (Table 2). We wondered whether the higher fluid absorp- We also considered it necessary to perform an experiment with a CO 2 /HCO 3 − buffered perfusate, because luminal CO 2 has been shown to stimulate fluid absorption presumably by stimulating an apical Na + /H + exchange process (Turnberg, Fordtran, Carter, & Rector, 1970) We therefore perfused the jejunum of cftr −/− and WT mice with a HEPES-buffered pH of 7.4 (Figure 10d), or with a

| DISCUSSION
This study explores the ability of three recently FDA-approved drugs for treatment of constipation-prone irritable bowel syndrome, to improve gut fluidity and alkalinity in the cftr −/− and F508del mut/mut murine intestine and identifies the mechanisms of action of these drugs in the CF gut. All three drugs were able to decrease fluid absorption in the CF intestine but with significant differences regarding segmental efficacy or mode of action.
In order to get FDA-approval for the treatment of a benign disease such as constipation-prone irritable bowel syndrome, strict safety criteria have to be fulfilled. Safety is also essential for CF patients, because treatment should be started early and given  (Colledge et al., 1995;Ratcliff et al., 1992;Ratcliff et al., 1993). It was also shown that chronic PEGbased oral laxative application reduces the incidence of intestinal obstructions and prolongs survival in cftr −/− mice (Clarke, Gawenis, Franklin, & Harline, 1996). More recent studies demonstrated that this beneficial effect was due to the prevention of early pathological alter- and F508del mut/mut mice compared to WT littermates (Figure 1a,b) and NHE3 mRNA expression was not down-regulated (Figure 1c).
Linaclotide binds to the heat-stable Escherichia coli toxin/guanylin receptor GC-C in the apical membrane of intestinal epithelial cell (Bryant et al., 2010). Upon binding, cGMP production is stimulated.
This activates the cGMP-dependent kinase II (PKG2), resulting in the activation of the CFTR channel (Ahsan et al., 2017;Pfeifer et al., 1996) and the inhibition of the Na + /H + exchanger NHE3 (Cha et al., 2005). In one study, 50 μgÁkg −1 linaclotide had been applied to  -buffer than HEPES-buffer perfusate at the same pH value of 7.4. Both the speed and magnitude of inhibition of fluid absorptive rate with tenapanor is higher in CO 2 /HCO 3 − buffer. However, the response to linaclotide is similar, that is, completely inhibited by tenapanor in cftr −/− and F508del mut/mut jejunum. N = 5 for each group. *P < 0.05. NS, no significance short circuit current response was therefore tested in isolated jejunal, proximal and distal colonic mucosa of the three genotypes in vitro ( Figure S3) and subsequently on jejunal, proximal and distal colonic fluid movements and HCO 3 − output in vivo. A reduction of jejunal fluid absorption and an increase in bicarbonate secretion were found in cftr −/− and F508del jejunum that was approx. 25% that in WT mice.
Surprisingly, linaclotide did not significantly inhibit fluid absorption in the cftr −/− and F508del mut/mut proximal and mid-distal colon, even in at the higher concentration of 5 × 10 −7 M, which elicited a strong secretory response in WT mice. Possible reasons are a thicker and less permeable mucus layer and an altered GC-C expression in cftr −/− and F508del colon. A significant reduction or even complete loss of the biological effect of luminally applied toxins by the presence of the adherent mucus gel was demonstrated in rat duodenum (Flemstrom et al., 1999). Ikpa et al. recently demonstrated that the expression of Gucy2c mRNA in the murine colon is strongest in the base of the crypts, as compared to a villous expression in the small intestine (Ikpa et al., 2016). It is thus possible that linaclotide does not reach its receptor because of the viscous mucus in the CF colon. In a recent study in human volunteers, a dose of 870 μg did not result in a biological effect in the colon, unless a PEG-containing oral laxative solution was used to cleanse the bowel prior to linaclotide ingestion (Weinberg et al., 2017). Since it is unclear whether an exclusive action in the small bowel is sufficient to relieve CF-associated constipation, a galenic formulation of linaclotide that ensures colonic release seems necessary before clinical trials are started in chronically constipated CF patients. This would indeed be desirable, because linaclotide is generally well-tolerated, with diarrhoea being the most frequent side effect, which would likely not be an issue in CF patients (Black et al., 2018).
Lubiprostone is a PGE 1 analogue that has already been tested in CF patients in a small published pilot trial, with benefit in some patients (O'Brien, Anderson, & Stowe, 2011). In our study, 10 −6 M lubiprostone significantly inhibited fluid absorptive rates and increased HCO 3 − output in both the small and large intestine of cftr −/− and F508del mut/mut , reaching $50% of the effect observed in the jejunum and colon of WT mice. This is in line with previous findings that showed an inhibition of ileal fluid absorption in CF patients by luminally applied PGE 1 analogue misoprostol (Coates et al., 2004).
While this result at first glance makes lubiprostone a very promising candidate for the treatment of CF-associated constipation and obstructive episodes, the concentrations that have been used in isolated mucosa and in animal studies, including ours, or in the study by Lubiprostone has originally been characterized as a stimulant of ClC-2-mediated intestinal fluid secretion (Ambizas & Ginzburg, 2007), but this concept was later questioned. Reviews by Akiba and Kaunitz (2012) and Jin and Blikslager (2015) discuss the controversial experimental results (Jin & Blikslager, 2015). Meanwhile, lubiprostone has been reported to target multiple cellular structures, including muscle (Chan & Mashimo, 2013), pacemaker cells of Cajal (Jiao et al., 2014), goblet cells (De Lisle, 2012;Jakab, Collaco, & Ameen, 2012) and possibly nerves (Bassil et al., 2008). This broad range of target structures may explain side effects of lubiprostone such as nausea, which may limit its applicability in higher concentrations for increasing gut fluidity and alkalinity in the CF gut.
Tenapanor, a selective NHE3-inhibitor whose action is limited to the intestine because of very low systemic absorption (Spencer et al., 2014), received FDA approval in September 2019 for the treatment of constipation-prone irritable bowel syndrome (Markham, 2019). Dose-finding experiments, using the identical method used in the present study, in anaesthetized C57B/6 mice had shown that 10 −5 M tenapanor inhibited jejunal fluid absorption reaching a plateau that was even somewhat lower than the fluid absorptive rate in NHE3-deficient mice, which may have undergone a degree of compensation (Xia et al., 2014). Therefore, we can assume full NHE3 inhibition with this concentration. Tenapanor inhibited fluid absorptive rate and increased HCO 3 − output both in the jejunum and colon to a similar percentage in cftr −/− , F508del mut/mut and WT mice, although the time course of inhibition was slower in the former. This difference in the onset of action was abolished when a CO 2 /HCO 3 −buffered perfusate was applied, which stimulates NHE3 activity (Turnberg, Fordtran, Carter, & Rector, 1970). Dose-ranging studies with tenapanor have been performed in healthy volunteers (Rosenbaum, Yan, & Jacobs, 2018) and doses of 50 mg twice daily were chosen for phase III trials for constipation-prone irritable bowel syndrome (Chey, Lembo, & Rosenbaum, 2020). One capsule achieves the concentration of 10 −5 M, as used in our perfusates, in $5 L of gastrojejunal fluid. Our chosen concentration is therefore likely to be in the range of the currently prescribed therapeutic drug concentration.
An issue of concern for the CF patient population is a potential interference of tenapanor with the absorption of nutrients that are taken up via a proton-coupled transport system such as the dipeptide transporter PEPT1 (SLC15A1) or the proton-coupled amino acid transporter PAT1 (SLC36A1) or the proton-coupled folate and heme transporter PCFT (SLC46A1), which depend on the action of NHE3 to restore intracellular pH i -homeostatis (Thwaites & Anderson, 2007).
This has been studied for PEPT1 but with a concentration of 2 × 15 mg, which is low compared to the recommended dose and found to not cause a significant decrease in the maximal serum concentration of the PEPT1 substrate cefadroxil (Johansson et al., 2017).
Nevertheless, nutrient/micronutrient absorption should be closely monitored in a future tenapanor clinical trial in CF patients with severe constipation and/or prior DIOS episodes.
In order to evaluate the mechanism of the inhibitory action of 10 −7 M linaclotide and 10 −6 M lubiprostone in the CF gut, we tested the effect of either drug after tenapanor preincubation. In contrast to linaclotide, lubiprostone resulted in a further decrease in fluid absorption in F508del mut/mut mice, albeit significantly less than without tenapanor preincubation ( Figure 10). This does not reflect merely a further inhibition of NHE activity, because this would be associated with an increase in HCO 3 − output. As mentioned above, lubiprostone shares with other prostaglandins the broad range of target structures.
In some laboratories, the F508del mutant mouse models showed better survival rate than that described for CFTR null mice, which was related to low levels of functional CFTR protein in the brush border membrane of intestinal epithelium and appeared to be dependent on the genetic background (van Doorninck et al., 1995;Wilke et al., 2011). This is the mouse strain and background that we also use and we also observe a residual short circuit current response to linaclotide and lubiprostone in isolated small and large intestinal mucosa of and F508del mut/mut mice, which is not seen in cftr −/− mucosa ( Figure S3). It is therefore interesting that the changes in fluid absorptive and HCO 3 − output rates did not differ in a major way in cftr −/− and F508del mut/mut intestinal segments, suggesting that the short circuit current response in isolated mucosa in vitro may not be a good readout of fluid secretory response in vivo.
As a potential predictor of treatment success of the tested drugs in CF-associated constipation and obstructive episodes, our study has several limitations:-(1) The method assesses the direct effect on fluid movements and HCO 3 − output in defined intestinal segments but ignores the indirect effects that links fluid secretion to intestinal motility via the enteric nervous system; (2) While the expression of CFTR and NHE3 has been studied in both murine and human intestine and is comparably distributed in the intestinal segments under investigation, differences in receptor or transporter affinity for the studied drugs may exist between mice and humans and (3), drug-related side effects are not recognized. Nevertheless, they yield valuable information about the drugs' effects on intestinal fluid and acid/base transport in vivo.
In summary, this study assessed fluid absorptive and HCO 3 − output rates in anaesthetized and tightly acid/base and BP-controlled cftr −/− , F508del mut/mut and WT mice in response to three recently FDA-approved drugs developed for the treatment of constipationprone irritable bowel syndrome. Each of these substances has the ability to significantly inhibit the fluid absorptive rate and to increase the HCO 3 − output in the jejunum of cftr −/− as well as F508del mut/mut mice, acting exclusively (linaclotide, tenapanor) or predominantly (lubiprostone) via the inhibition of the sodium/hydrogen exchanger NHE3. Tenapanor and lubiprostone, but not linaclotide, also inhibited fluid absorption and increased HCO 3 − output in the colon of cftr −/− and F508del mut/mut mice. Future studies are needed to test the longterm action of these drugs on the prevention of obstructive episodes and the reversibility of inflammatory alterations and dysbiosis in CF animal models and in patients.

ACKNOWLEDGEMENTS
This study was funded by the Cystic Fibrosis Trust SRC011, the DFG SE460/19-1 and SE460/21-1 and the Volkswagen Foundation grant ZN1953. The mouse care by Anna Knoll-Schluch and Silke Thiele is gratefully acknowledged.

AUTHOR CONTRIBUTIONS
Q.T., G.dS., X.T., X.R. and U.S. performed experiments, acquired and analysed data. G.dS., D.R. and U.S. planned and supervised the mouse breeding. S.R.T. helped with statistical calculations. Q.T. and U.S. drafted the figures and wrote the manuscript.

CONFLICT OF INTEREST
The authors declare no conflicts of interest.

RIGOUR
This Declaration acknowledges that this paper adheres to the principles for transparent reporting and scientific rigour of preclinical research as stated in the BJP guidelines for Design and Analysis, Animal Experimentation and as recommended by funding agencies, publishers and other organizations engaged with supporting research.

DATA AVAILABILITY STATEMENT
Data are available on request from the authors.