The electroneutral Na⁺:HCO₃⁻ cotransporter NBCn1 is a major pH_i regulator in murine duodenum

Abstract

Duodenocyte pH_i control and HCO₃⁻ secretion protects the proximal duodenum against damage by gastric acid. The molecular details of duodenocyte pH control are not well understood. A selective duodenal expression (within the upper GI tract) has been reported for the electroneutral Na⁺:HCO₃⁻ cotransporter NBCn1 (Slc4a7). We aimed to determine the role of NBCn1 and NBCe2 in duodenocyte intracellular pH regulation as well as basal and agonist-stimulated duodenal bicarbonate secretion (J_HCO3⁻), exploiting mouse models of genetic slc4a7 and slc4a5 disruption. Basal and forskolin (FSK)-stimulated J_HCO3⁻ was measured by single-pass perfusion in the duodenum of slc4a7^−/− and slc4a7^+/+ as well as slc4a5^−/− and slc4a5^+/+ mice in vivo, and by pH-stat titration in isolated duodenal mucosa in vitro. Duodenocyte HCO₃⁻ uptake rates were fluorometrically assessed after acidification of intact villi and of isolated duodenocytes. Slc4a7^−/− mice displayed significantly lower basal and FSK-stimulated duodenal HCO₃⁻ secretion than slc4a7^+/+ littermates in vivo. FSK-stimulated HCO₃⁻ secretion was significantly reduced in slc4a7^−/− isolated duodenal mucosa. Na⁺- and HCO₃⁻-dependent base uptake rates were significantly decreased in slc4a7^−/− compared with slc4a7^+/+ villus duodenocytes when measured in intact villi. Carbonic anhydrase (CA)-mediated CO₂ hydration played no apparent role as a HCO₃⁻ supply mechanism for basal or FSK-stimulated secretion in the slc4a7^+/+ duodenum, but was an important alternative HCO₃⁻ supply mechanism in the slc4a7^−/− duodenum. NBCe2 (Slc4a5) displayed markedly lower duodenal mRNA expression levels, and its disruption did not interfere with duodenal HCO₃⁻ secretion. The electroneutral Na⁺:HCO₃⁻ cotransporter NBCn1 (slc4a7) is a major duodenal HCO₃⁻ importer that supplies HCO₃⁻ during basal and FSK-stimulated HCO₃⁻ secretion.

Key points

• The mucosa of the proximal duodenum is regularly exposed to the strong acid of gastric juice, and acid damage causes peptic ulceration and bleeding.

• The epithelial cells of the duodenum protect themselves against acid-induced chemical damage by secreting bicarbonate into the lumen and by regulating their intracellular pH, but the molecular mechanisms of duodenal pH_i control are incompletely understood.

• We have found a high expression of the electroneutral sodium bicarbonate transporter Slc4a7 (NBCn1) in the basolateral membrane of duodenal villous enterocytes.

• Genetic knockout of Slc4a7 resulted in a strong defect in duodenocyte pH_i recovery from acid loads, and reduced bicarbonate secretory rates.

• This suggests that the electroneutral sodium bicarbonate cotransporter Slc4a7 is a major import mechanism for HCO₃⁻ from the blood into the duodenal epithelial cells. Defects in this transporter may severely endanger duodenal mucosal integrity

Introduction

The gastric contents that are emptied into the duodenum occasionally reach pH values as low as 1, and the duodenal mucosa is required to protect itself against this acid load in order to avoid chemical damage (Allen & Flemstrom, 2005; Kaunitz & Akiba, 2006). The mechanisms for duodenal protection have been discussed extensively. On the one hand, a large body of circumstantial evidence suggests that the epithelium protects itself by secreting HCO₃⁻ ions into the lumen and thus neutralizing H⁺ ions that diffuse through the mucus gel (reviewed in Allen & Flemstrom, 2005; Seidler & Sjöblom 2012). However, the duodenal villus tips in mice are not consistently covered by mucus gel, and Akiba and Kaunitz provided evidence in rodent duodenum in vivo that basolateral HCO₃⁻ uptake and intracellular neutralization of acid may play a major role in preventing epithelial damage in the villous region (Akiba et al. 2001).

As recently reviewed (Seidler et al. 2011; Seidler & Sjöblom, 2012), attempts to elucidate the pH_i regulatory mechanisms in duodenocytes, using isolated cell suspensions or short-term culture, provided functional evidence for the involvement of Na⁺/H⁺ exchangers (NHEs), Na⁺/HCO₃⁻ cotransporters (NBCs) and even Cl⁻/HCO₃⁻ exchangers in the pH_i recovery from an intracellular acid load, and the application of ion transport inhibitors allowed some further characterization of the involved NHE isoforms. However, Akiba et al. (2001) showed that in the rodent duodenum in vivo, a DIDS-sensitive mechanism is essential for pH_i recovery from intracellular acidification, suggesting the possible involvement of NBCs in duodenal pH_i recovery from acidification.

The first identified member of the Na⁺:HCO₃⁻ cotransporter was the electrogenic Na⁺:HCO₃⁻ cotransporter (NBC1, NBCe1) of the basolateral membrane in renal proximal tubules (Burnham et al. 1997; Romero et al. 1997). A splice variant of this gene with a different N-terminus and a different stochiometry (first named p(ancreatic)NBC1 (Abuladze et al. 1998) or d(uodenal) NBC1 (Jacob et al. 2000)) was found with a widespread distribution in gastrointestinal epithelia (Rossmann et al. 1999; Jacob et al. 2000; Jakab et al. 2011). Due to its high expression in the duodenum, this NBC isoform is generally regarded as an important duodenal HCO₃⁻ importer, but due to the early death of the NBCn1 knockout mice, this has never been functionally confirmed (Gawenis et al. 2007).

In 2000, an electroneutral Na⁺:HCO₃⁻ cotransporter, NBCn1, was cloned (Choi et al. 2000). The study of its tissue distribution showed high expression levels in the vasculature but low levels in gastrointestinal (GI) organs (Abuladze et al. 1998), thus not suggesting a high relevance for GI ion transport. However, Praetorius and colleagues demonstrated the expression of NBCn1, in addition to NBCe1 (which is also expressed in other parts of the small intestine) in the basolateral membrane of villous duodenocytes (Praetorius et al. 2001). NBCn1 was found to also be expressed in human duodenum (Damkier et al. 2007). The expression of this transporter in a gastrointestinal epithelium with a particularly high demand for intracellular pH_i regulation as well as for protective HCO₃⁻ secretions raises the question of its involvement in the regulation of duodenocyte pH as well as for HCO₃⁻ import during duodenal alkaline secretion.

Therefore, the role of NBCn1 in duodenal pH_i regulation as well as basal and agonist-stimulated duodenal HCO₃⁻ secretion were studied in NBCn1-deficient (slc4a7^−/−) mice and their WT littermates in vivo and in vitro. We also studied the compensatory mechanisms for HCO₃⁻ uptake that are used by NBCn1-deficient (slc4a7^−/−) duodenocytes.

Methods

Chemicals and solutions

All reagents were purchased from Sigma-Aldrich (Deisenhofen, Germany) if not indicated otherwise. Forskolin (FSK) was diluted in DMSO (10⁻² m), stored at −20°C and added to an isotonic NaCl solution to a final concentration as indicated in the text on the same day of use.

Animals

Transcription of slc4a7 (gene for NBCn1) was interrupted by insertion of a gene trap-cassette in a CG-rich region upstream of exon1. Details of the mouse generation and general characterization are given elsewhere (Boedkjer et al. 2011). Transcription of slc4a5 (gene for NBCe2) was disrupted as described in the legend to Fig. 4. Mice were bred in the animal facilities of Hannover Medical School and Department of Biomedicine, Aarhus University, on a congenic C57Bl/6 background. The mice born from slc4a7^+/− breeding pairs had a non-Mendelian distribution of the genotypes at birth, with less than expected slc4a7^−/− pups by approximately one third. No difference in growth, vital parameters and blood acid–base status during the experimental procedure (Table 1) were observed in the slc4a7^−/− mice, and the slc4a5^−/− mice appeared grossly healthy up to the time that we used them to perform HCO₃⁻ secretion experiments. As the slc4a5^−/− mice did not display a HCO₃⁻ secretory defect in the duodenum and the expression of NBCe2 was markedly lower than that for NBCn1 and NBCe1 (Figs 1 and 2), a full characterization of this strain was not pursued. All studies were approved by the Hannover Medical School Committee on investigations involving animals and the ‘Regierungspräsidium’. Care was taken to gender-match littermates. The mice were housed under standard temperature and light conditions (12:12 h light–dark cycle; temperature, 21–22°C) and were allowed free access to food and water. Primers for genotyping slc4a7-deficient mice are found in Supplementary Table S1.

Figure 4. Basal and FSK-activated HCO₃⁻ secretion in the duodenum of anaesthetized slc4a5^−/− and slc4a5^+/+ mice.

A, time course of duodenal HCO₃⁻ secretion in anaesthetized slc4a5^−/− mice and slc4a5^+/+ littermates displayed no difference in at any time point. B, basal and C, FSK-stimulated Δ in % of basal were identical in slc4a5^−/− and slc4a5^+/+ duodenum (n= 7 for slc4a7^−/− and 11 for slc4a7^+/+ mice). The NBCe2 (Slc4a5) gene-targeted mouse model was prepared by deleting the 272-nucleotide exon 13 of Ensemble transcript variant Slc4a5-201 (GenBank Accession number NM_001166067.1). This removed codons 474–563 (beginning with the amino acid sequence ANGLDY and ending with CVAPDT) and created a frameshift that eliminated the potential for translation of the ∼500 amino acid membrane-spanning domains of the protein, which are essential for anion transport activity. Details of the gene-targeting strategy and genotyping protocol (E. M. Bradford, Yinhuai Chen and G. E. Shull, unpublished data) will be described elsewhere. The slightly higher in these experiments compared with those in Fig. 3 is probably due to the different mouse line, a different investigator, a more advanced anaesthesia equipment and more frequent blood gas analyses.

Table 1.

Blood gas analysis of the anaesthetized mice at the end of the experiments shown in Figs 3 and 7

Group	pH	PCO2(mmHg)	HCO3− (mm)	SBE	No. of mice
Slc4a7+/+ (FSK)	7.41 ± 0.04	35.4 ± 3.5	22.8 ± 3.2	−1.5 ± 3.4	7
Slc4a7−/− (FSK)	7.46 ± 0.02	31.3 ± 2.9	22.0 ± 1.2	−1.7 ± 0.9	3
Slc4a7+/+ (FSK+mtz/acz)	7.35 ± 0.02	51.2 ± 3.7	27.1 ± 0.8	1.9 ± 0.7	10
Slc4a7−/− (FSK+mtz/acz)	7.38 ± 0.02	45.6 ± 1.4	26.5 ± 1.1	2.0 ± 1.2	8

The last column indicates the number of mice for which analysis could be done. Not all mice used in the experiments could be analysed (frequent occupation of the blood gas analyser).

Figure 1. mRNA expression for NBCe1, NBCn1 and NBCe2 in the mucosa from different intestinal segments of C57B/6 mice, and duodenocyte basolateral NBCn1 immunostaining in WT and KO mice.

A, the bars show the mRNA expression levels for the three NBCs in different segments of the GI tract, in relation to ß-actin as a control gene. NBCn1 is expressed in similar levels to that of NBCe1 in the duodenum. NBCe2 mRNA expression levels are low. n= 6. B, strong immunohistochemical staining of the basolateral membrane of the villous duodenocytes was completely absent in the slc4a7^−/− duodenum. Representative staining for 5 examined pairs of mice. LU lumen, LP lamina propria.

Figure 2. NHE1–3, and NBCe1 and NBCe2 mRNA expression in slc4a7^−/− and slc4a7^+/+ duodenal mucosa.

A, mRNA expression levels of NHE1–3 and of NBCe1 and NBCe2 in slc4a7^+/+ and slc4a7^−/− duodenal mucosa. qPCR was performed with epithelial-specific villin as the control gene, and the respective expression level of the gene of interest was normalized to WT level. No significant change in expression level was found for the studied acid–base transporters in the absence of NBCn1 (n= 6). B, NBCe1 immunostaining in slc4a7^+/+ and slc4a7^−/− duodenal mucosa. The images are representative for 5 mice examined. No difference was observed in intensity or localization of NBCe1 in slc4a7^+/+ and slc4a7^−/− duodenal mucosa. LU lumen, LP lamina propria

Measurement of mRNA expression of NBCe1, NBCn1, NBCe2 and NHE1–3

mRNA expression levels in scraped duodenal mucosa was performed using a quantitative real-time PCR protocol exactly as described before (Broere et al. 2009). The primer pairs are given in Supplementary Table 1.

Immunohistochemistry

Mice were fixed via the left ventricle with 3% paraformaldehyde in phosphate-buffered salt solution (PBS, in mm: 167 Na⁺, 2.8 H₂PO₄⁻, 7.2 HPO₄²⁻; pH 7.4). The duodeni were removed, dehydrated, embedded in paraffin wax, and 2 μm sections were cut using a rotary microtome (Leica, Wetzlar, Germany). The sections were dewaxed and stepwise rehydrated. Epitopes were retrieved by boiling the sections in 10 mm Tris buffer (pH 9) with 0.5 mm EGTA, and then quenched with 50 mm NH₄Cl. Unspecific binding was blocked by 1% BSA in PBS with 0.2% gelatin and 0.05% saponin. The sections were incubated overnight at 4°C with primary antibody, diluted in 0.1% BSA in PBS with 0.3% Triton X-100 added. The anti-NBCn1 antibody was previously described and validated (Damkier et al. 2006). Horseradish peroxidase-conjugated goat anti-rabbit secondary antibodies (Dako, Hamburg, Germany) were used for visualization with 3,3′-diaminobenzidine tetrahydrochloride dissolved in PBS with 0.1% H₂O₂ for 5 min in PBS. Mayer's haematoxylin was used for counterstaining, and the sections were dehydrated in graded alcohol and xylene and mounted in hydrophobic Eukitt mounting medium (Kindler, Freiburg, Germany). Microscopy was performed on a Leica DMRE bright-field microscope equipped with a Leica DM300 digital camera.

In vivo experimental protocols

The mice were anaesthetized by spontaneous inhalation of isoflurane (Abbott Germany, Wiesbaden) and the experiments performed as described (Singh et al. 2009), with the following modifications: a catheter was placed in the left carotid artery for continuous infusion of a solution containing (in mm) 130 Na⁺, 5.4 K⁺, 0.9 Ca²⁺, 1 Mg²⁺, 55 Cl⁻, 55 CO₃²⁻, 27 acetate, gassed with 5% CO₂–95% O₂ at a rate of 0.35 ml h⁻¹, to correct systemic acid–base and fluid balance. Blood gas analysis was performed and the values are given in Table 1. The earlier in vivo experiments (Figs 3 and 7) were performed with the Univentor 400 anaesthesia unit (with oxygen mixed by a separate valve to room air by hand); the later ones (Fig. 4) with the univentor 1250 unit (where oxygen is added more precisely by the system and the air is dehumidified). The role of slc4a7 in the regulation of duodenal luminal alkalinization rate was assessed before and after intrasegmental perfusion of forskolin (FSK, 100 μm) as described (Singh et al. 2008). Mild respiratory alkalosis was sometimes seen; it was due to hyperventilation during anaesthesia and was not statistically different in NBCn1-deficient and WT littermates. As systemic acetazolamide causes mild acidosis, a higher rate of sodium carbonate (75 mm CO₃²⁻, 35 mm Cl⁻) was infused to keep blood pH_i above pH 7.3 (Table 1)

Figure 3. Basal and FSK-activated HCO₃⁻ secretion in the duodenum of anaesthetized slc4a7^−/− and slc4a7^+/+ mice.

A, time course of duodenal HCO₃⁻ secretion in anaesthetized slc4a7^−/− mice and slc4a7^+/+ littermates displayed significantly lower at all time points. There was a significant reduction of both basal (Fig. 3B) and FSK-stimulated HCO₃⁻ secretion in the slc4a7^−/− duodenum, but the HCO₃⁻ secretory response in per cent above basal (C) was not significantly different (n= 7 for slc4a7^−/− and 11 for slc4a7^+/+ mice). *P < 0.05, **P < 0.01.

Figure 7. Effect of CA inhibition on basal and FSK-stimulated HCO₃⁻ secretory rate in the duodenum of slc4a7^−/− and slc4a7^+/+ mice.

A, systemic (10 mg kg⁻¹ i.a. acetazolamide) and luminal (1 mm methazolamide) application of CA inhibitors did not significantly inhibit basal or FSK-stimulated duodenal HCO₃⁻ secretion in slc4a7^+/+ mice, but caused a significant reduction of the duodenal HCO₃⁻ secretory response to FSK in the slc4a7^−/− mice. B, peak duodenal HCO₃⁻ secretory rate in slc4a7^+/+ and slc4a7^−/− mice in % above basal. C, the % increase in HCO₃⁻ secretory rate after FSK was significantly decreased in slc4a7^−/− compared with slc4a7^+/+ mice (n= 4).

Ussing chamber experiments

Isolated duodenal mucosa of NBCn1-deficient mice and WT littermates were placed in Ussing chambers and secretory studies performed according to identical protocols as described before (Tuo et al. 2006), except that open circuit conditions were used, potential difference (PD) and electrical resistance (R_t) were continuously recorded, and the short circuit current (I_sc) calculated as described in Xiao et al. (2012).

Fluorometric pH_i measurements and determination of base uptake rates into slc4a7^−/− and slc4a7^+/+ duodenocytes within intact villi

For measuring pH_i in isolated duodenal villi, a piece of duodenal tissue was transferred onto the cooled stage of a dissecting microscope, and individual villi were detached from the intestine and attached to a glass coverslip pre-coated with Cell-Tak adhesive (BD Biosciences, Bedford, MA, USA). Isolated individual villi were loaded for approximately 10 min with 16 μm 2′,7′-bis(2-carboxyethyl)-5(6)-carboxy-fluorescein acetoxymethyl ester (BCECF-AM) (Invitrogen, Darmstadt, Germany) in Ringer solution with 500 μm DTT. After being loaded, the villi were fixed on the coverslip by covering them with a clear filter, and perfused with pre-warmed (37°C) O₂-gassed (A) or O₂–CO₂-gassed (B) buffer as needed per experimental protocol (for buffer composition see Supplementary methods). Fluorometric pH_i measurements were performed as previously described (Broere et al. 2009), with modifications described in Chen et al. (2010). Calibration was performed as described for pancreatic ducts (Hegyi et al. 2004). Intrinsic buffering capacity (βi) of duodenal villi was determined according to Roos & Boron (1981) as described previously (Bachmann et al. 2003). A detailed description of the method is given in Chen et al. (2010), except that the buffer capacity was determined both by acidifying the villous enterocytes with the ammonium prepulse technique and then adding small concentrations of NH₄Cl and recording the rapid increase in pH_i for each increase of [NH₄Cl], and also by alkalinizing the cells by 80 mm NH₄Cl, followed by a stepwise decrease in [NH₄Cl], and recording of the change in pH_i, in the absence of Na⁺ (for stepwise increase in pH_i, or Na⁺ and Cl⁻ for stepwise decrease in pH_i). One such pH_i trace is depicted in Supplementary Fig. S1B, while the intrinsic as well as the total β (intrinsic and CO₂/HCO₃⁻-mediated β) is shown in Supplementary Fig. S1C. The rates of pH_i change measured in the experiments were converted to proton flux using the equation =ΔpH/Δt×β_total.

Statistics

Data are presented as means ± SEM. The Mann–Whitney rank-sum test, Student's t test or, if appropriate, the ANOVA for multiple comparisons were used for statistics, and values of P < 0.05 were considered significant. *P < 0.05, **P < 0.01 and ***P < 0.001.

Results

NBCe1, NBCe2 and NHE1–3 expression is not altered in slc4a7^−/− duodenal mucosa

Quantitative PCR analysis of mRNA expression levels for NBCe1, NBCn1 and NBCe2 in relationship to β-actin expression in the different segments of C57B/6 mouse intestine (Fig. 1A) revealed that NBCn1 expression is particularly high in the duodenum. The basolateral localization of NBCe1 and NBCn1 and their villus-predominant expression was determined by immunohistochemistry; the results were identical to previously published data (Praetorius et al. 2001) and are therefore not shown. The specificity of the NBCn1 antibody is evident from the lack of basolateral staining in the NBCn1 KO mouse (Fig. 1B). NBCe2 expression was low, and since basal and FSK-stimulated duodenal HCO₃⁻ secretion was not decreased in NBCe2 KO mice (Fig. 4A–C), further attempts to localize this transporter were not pursued. We then studied whether major changes occur in the expression of other base importers in the absence of duodenal NBCn1 expression. The mRNA expression levels for NBCe1, NBCe2 and NHE1–3 were determined in scraped duodenal mucosa, and found to be not significantly altered in the slc4a7^−/− duodenal mucosa compared with WT (Fig. 2A). The localization of NBCe1 was also studied by immunohistochemistry in slc4a7^−/− and WT mucosa. Its distribution and intensity of staining was not significantly different in five pairs of KO and WT mice (Fig. 2B).

Basal and FSK-activated HCO₃⁻ secretion is reduced in the duodenum of anaesthetized slc4a7^−/− mice

In the duodenum of anaesthetized mice, basal secretory rates were significantly lower in the slc4a7^−/− than in slc4a7^+/+ mice (Fig. 3A and B). Perfusion of the lumen with FSK (10⁻⁴ m) elicited a HCO₃⁻ secretory response that was significantly lower in absolute HCO₃⁻ secretory rate (Fig. 3A) but not in per cent of basal rate in slc4a7^−/− vs. slc4a7^+/+ duodenum (Fig. 3C).

No significant difference in the basal or FSK-stimulated J_HCO3⁻ was found in the duodenum of slc4a5^−/− compared with slc4a5^+/+ anaesthetized mice, suggesting a minor role of this bicarbonate transporter in duodenal bicarbonate secretion (Fig. 4A–C).

Basal and FSK-stimulated HCO₃⁻ secretory rates are reduced in the isolated duodenal mucosa of slc4a7^−/− mice

In muscle-stripped duodenal mucosa from slc4a7^−/− and slc4a7^+/+ mice, basal HCO₃⁻ secretory rates were not significantly different (Fig. 5A), but FSK-stimulated HCO₃⁻ secretory response was significantly diminished (Fig. 5A and B). Figure 5C and D show the calculated short-circuit currents (from the continuously recorded R_t and PD) and electrical resistance (R_t) of slc4a7^−/− and WT mucosa, with no difference in the results.

Figure 5. Basal and FSK-stimulated HCO₃⁻ secretory rates in the isolated duodenal mucosa from slc4a7^+/+ and slc4a7^−/− mice.

Time course of HCO₃⁻ secretion before and after stimulation with FSK (10⁻⁵ m in serosal and luminal bath) in isolated slc4a7^+/+ and slc4a7^−/− duodenal mucosa (A). A significantly reduced HCO₃⁻ secretory response after FSK application was seen in the slc4a7^−/− duodenum (B). The FSK-induced I_sc response (C), as well as R_t (D), was not significantly different in slc4a7^+/+ and slc4a7^−/− mice (n= 8).

Steady-state pH_i and Na⁺-dependent pH_i recovery in enterocytes of intact villous tips

We next assessed the effect of NBCn1 ablation on the ability of the villus enterocytes to regulate pH_i. Figure 6A shows a BCECF-loaded villus with the regions of interest (ROIs). Steady-state pH_i of isolated C57B/6 duodenal villi was found to be higher in the absence than in the presence of CO₂/HCO₃⁻, indicating that in the absence of transepithelial gradients for pH_i and HCO₃⁻, the effect of HCO₃⁻ efflux from the duodenocyte may be larger than that of HCO₃⁻ import (Fig. 6B). Steady-state pH_i in CO₂/HCO₃⁻ was slightly higher in slc4a7^−/− compared with slc4a7^+/+ enterocytes within intact villi (Fig. 6C).

Figure 6. Steady-state pH_i and Na⁺-dependent pH_i recovery in enterocytes of intact villus tips.

A, BCECF-loaded microdissected duodenal villi. B, steady-state pH_i in C57B/6 duodenocytes in the absence and presence of CO₂/HCO₃⁻. C, steady-state pH_i in villous duodenocytes of slc4a7^−/− mice in CO₂/HCO₃⁻-containing buffer was slightly higher than that of slc4a7^+/+ mice. D, acid-activated base uptake rates were markedly lower in slc4a7^−/− compared with slc4a7^+/+ duodenocytes after inhibition of NHE1–3 (by 50 μm of Hoe 642 and 20 μm S1611) in a HCO₃⁻-containing buffer. E, acid-activated, Na⁺-dependent base uptake rates in slc4a7^+/+ enterocytes in CO₂/HCO₃⁻ buffer with sequential inhibition of NHE1, NHE2 and NHE3 (by 1 μm Hoe 642 for NHE1, 50 μm Hoe 642 for NHE1 and NHE2, and 20 μm S1611 for NHE3. F, the effect of DIDS on acid-activated base uptake in slc4a7^+/+ and slc4a7^−/− villi was studied to assess whether the absence of NBCn1 leads to activation of other NBCs. Indeed, 200 mm DIDS caused a significant inhibition of Na⁺-dependent base uptake rates in slc4a7^−/− but not in WT duodenocytes (n= 3–5).

We then measured the pH_i recovery rate after an ammonium prepulsing in the presence of both CO₂/HCO₃⁻ and of inhibitors for the Na⁺/H⁺ exchanger isoforms NHE1–3. The acid-activated, Na⁺-dependent base uptake rate was reduced to almost 50% in slc4a7^−/−-deficient duodenal enterocytes compared with that in slc4a7^+/+ enterocytes, demonstrating that NBCn1 is a major base uptake mechanism in villous duodenocytes in the presence of CO₂/HCO₃⁻ (Fig. 6D).

To assess the importance of NBCn1 for duodenal villus enterocytes in relation to that of other duodenal base uptake mechanisms, we compared the effect of NBCn1 ablation (Fig. 6D) with that of NHE1 inhibition (with 1 μm Hoe 642), NHE2 (in addition to NHE1) inhibition (with 50 μm Hoe 642), NHE3 inhibition (by 20 μm S1611), and NBCe1/e2 inhibition (with 200 μm H₂DIDS), in WT mice (Fig. 6E and F). The results demonstrated that NHE3 and NHE1 are the second most active pH_i recovery mechanisms in duodenal villi, with no contribution of base loaders that are sensitive to low concentrations of H₂DIDS (200 μm), presumably NBCe1 (Praetorius et al. 2001) and maybe to some extent the H₂DIDS-sensitive NBCe2 (Virkki et al. 2002) in slc4a7^+/+ duodenocytes. Interestingly, there was a significant contribution of a H₂DIDS-sensitive base loader to Na⁺ and HCO₃⁻-dependent pH_i recovery from an acid load in slc4a7^−/− duodenocytes (Fig. 6F), suggesting functional upregulation of NBCe1/e2 under these circumstances. The results demonstrate that NBCn1 is a major, but not the only base uptake mechanism, in villous duodenocytes.

Effect of CA inhibition on basal and FSK-stimulated HCO₃⁻ secretory rate in the duodenum of slc4a7^−/− and slc4a7 ^+/+ mice

As FSK enhanced HCO₃⁻ secretion even in the slc4a7^−/− duodeni, alternative HCO₃⁻ supply mechanisms must exist for the duodenocyte during FSK-stimulated secretion. The potential alternative mechanisms are HCO₃⁻ uptake by another NBC isoform or CA-facilitated CO₂ hydration and basolateral proton export by Na⁺/H⁺ exchangers (Jacob et al. 2000). In order to distinguish between these possibilities, we repeated the experiments in the presence of luminal and systemic CA inhibitors (which will result in virtually complete CA inhibition; Ilies et al. 2000) (Fig. 7). This manoeuvre did not significantly decrease basal or FSK-stimulated HCO₃⁻ secretion in the WT duodenum (compare the WT values in Fig. 7A and B with those in Fig. 3A and B), but significantly reduced FSK-stimulated HCO₃⁻ secretion in the slc4a7^−/− duodenum by approximately 50% (compare Fig. 7C with Fig. 3C). This indicates that CO₂ hydration has no rate-limiting effect in FSK-stimulated duodenal HCO₃⁻ secretion in vivo, but becomes an important pathway for HCO₃⁻ supply when NBCn1 is not active.

No significant inhibition of basal FSK-induced HCO₃⁻ secretory rate in isolated slc4a7^−/− or slc4a7^+/+ duodenal mucosa by moderate concentrations of DIDS

As the residual HCO₃⁻ secretory response suggested the existence of alternative pathways for basolateral HCO₃⁻ uptake, we studied the effect of 200 μm H₂DIDS in the serosal perfusate (Fig. 8). This concentration of H₂DIDS supposedly inhibits NBCe1 (and NBCe2), while NBCn1 is relatively insensitive (Burnham et al. 1997; Choi et al. 2000). H₂DIDS (200 μm) had a dual effect on the I_sc response: the addition of H₂DIDS caused a small increase in I_sc negativity, and subsequent FSK resulted in a rapid increase to peak values which were not significantly different from the peak values in the respective genotype in the absence of H₂DIDS (compare Fig. 8C with Fig. 5C), but no sustained plateau phase was observed (as seen in Fig. 5C). H₂DIDS, on the other hand, did not significantly inhibit basal or FSK-stimulated HCO₃⁻ secretory rates (Fig. 8A and B) in WT and slc4a7^−/− mucosa when compared with the HCO₃⁻ secretory responses shown in Fig. 5. This indicates that DIDS-sensitive base importers may not play a significant role as HCO₃⁻ uptake pathways for FSK-stimulated HCO₃⁻ secretion in isolated duodenal mucosa. While the addition of 200 μm H₂DIDS did not significantly change the time course of tissue resistance (R_t) when compared with that in the absence of H₂DIDS (compare Fig. 5D with Fig. 8D), higher concentrations of H₂DIDS resulted in a decrease in R_t, in particular after FSK addition, which limits interpretation (data not shown).

Figure 8. Basal and FSK-stimulated HCO₃⁻ secretory rates in the isolated duodenal mucosa of slc4a7^+/+ and slc4a7^−/− mice in the presence of 200 μm H₂DIDS.

A, in order to study the contribution of DIDS-sensitive HCO₃⁻ transporters to the HCO₃⁻ secretory response, 200 μm H₂DIDS was added to the serosal bath 5 min prior to FSK stimulation, and the per cent of stimulation was calculated for slc4a7^+/+ and slc4a7^−/− duodenum. B, when compared with the results in Fig. 4, 200 μm DIDS did not cause a significant reduction in the FSK-induced HCO₃⁻ secretory response in slc4a7^+/+ and slc4a7 ^−/− duodenal mucosa. However, there was a strong effect of DIDS on the I_sc (C): I_sc became slightly more negative after DIDS application, and the subsequent I_sc response to FSK became stronger but declined rapidly, as compared with the stable increase in Fig. 5. R_t was not significantly different in DIDS-treated slc4a7^+/+ and slc4a7^−/− mice (D). n= 9

Discussion

The duodenal epithelium is exposed to high concentrations of gastric acid under physiological conditions. Unlike gastric epithelial cells in the fundus region (Wallace, 1989; Boron et al. 1994), the duodenal epithelial lining does not possess intrinsic plasma membrane acid resistance (Nylander et al. 1994; Akiba et al. 2001; Slöblom et al. 2009). Consequently, the duodenum needs ‘extracellular’ as well as ‘intracellular’ protective mechanisms against acid damage.

It was observed decades ago that disturbances of systemic acid–base balance, with low blood HCO₃⁻ concentrations, severely endanger the integrity of the duodenal mucosa during gastric acid exposure (Kivilaakso et al. 1981). Both a higher epithelial HCO₃⁻ secretory rate as well as interstitial and intracellular HCO₃⁻ ions were protective factors (see Allen & Flemstrom, 2005; Kaunitz & Akiba, 2006; Seidler & Sjöblom 2012, for review). This has spurred a strong interest in HCO₃⁻ secretion as well as HCO₃⁻ uptake mechanisms of duodenocytes (see Allen & Flemstrom, 2005; Kaunitz & Akiba, 2006; Seidler et al. 2011; Seidler & Sjöblom, 2012, for review). The HCO₃⁻ supply to the duodenocyte was first thought to depend on CA-mediated CO₂ hydration (Muallem et al. 1994; Knutson et al. 1995). However, recent work from this laboratory revealed that CA-augmented intracellular CO₂ hydration was essential for ‘sensing’ the luminal acidity and for responding to it by augmented HCO₃⁻ secretion, but was not necessary for the HCO₃⁻ secretory response per se (Sjöblom et al. 2009).

The involvement of a Na⁺:HCO₃⁻ cotransport system located in the basolateral membrane of the duodenocyte was first discussed in the 1980s during studies of isolated amphibian duodenum in Ussing chambers, but the idea was rejected at the time, as experimental evidence for a firm coupling of Na⁺ to HCO₃⁻ transport and SITS sensitivity could not be generated (Simson et al. 1981a,b,c). Later, the use of stilbene derivatives and the removal/readdition of CO₂/HCO₃⁻ to isolated duodenocyte basolateral membranes, duodenal cell suspensions and isolated duodenal mucosa provided functional evidence for the likely involvement of Na⁺/H⁺ exchangers as well as Na⁺:HCO₃⁻ cotransporters both in duodenocyte pH_i control (Isenberg et al. 1993; Ainsworth et al. 1998; Praetorius et al. 2000, 2001) and for base uptake during HCO₃⁻ secretion into the lumen (Goddard et al. 1998; Jacob et al. 2000).

As pointed out in the Introduction, the molecular nature of the Na⁺HCO₃⁻ cotransporter(s) that play a role in duodenal pH_i control and HCO₃⁻ uptake destined for secretion is under debate. Based on expression studies, NBCe1 (Slca4) and NBCn1 (Slc4a7) are two equally interesting candidates (Fig. 1). NBCe2 (Slc4a5), also reported with intestinal expression (Pushkin et al. 2000), was found with lower expression levels (Fig. 1).

Since no pharmacological inhibitors exist for NBCn1, an NBCn1-deficient mouse was created by the insertion of a gene trap into the putative NBCn1 promoter sequence, which resulted in a complete loss of expression of this protein in the duodenum. In the first set of experiments, we searched for a potential role of NBCn1 as a HCO₃⁻ supply pathway for duodenal HCO₃⁻ secretion. We found that basal as well as FSK-induced, duodenal HCO₃⁻ secretion in anaesthetized slc4a7^−/− mice was significantly reduced compared with WT mice (Fig. 3). This was not explained by changes in systemic acid–base parameters (Table 1). NBCe2 deletion, on the other hand, did not result in significant changes in basal or FSK-stimulated duodenal HCO₃⁻ secretion (Fig. 4), and since the NBCe2 expression within the murein GI tract was highest in the upper small intestine, yet low compared with NBCe1 and NBCn1 (Fig. 1A), the study of this mouse genotype was not pursued further. The Cl⁻-dependent isoforms Slc4a8 and Slc4a19 had previously been ruled out as major candidates because of low duodenal expression (Damkier et al. 2007), the lack of Cl⁻ dependence of duodenocyte pH_i recovery from an acid load (Ainsworth et al. 1998), and because the bilateral removal of Cl⁻ from the serosal as well as the luminal bath did not reduce FSK-induced Δ in rabbit duodenum (Spiegel et al. 2003).

The relationship between vascular tone, blood flow and duodenal HCO₃⁻ secretion is probably highly complex (Jönson et al. 1990, 1991). Since NBCn1 is expressed in the vascular system and has recently been shown to play a role in agonist-mediated vasodilatation (Broedtkjer et al. 2011), we next assessed whether the effect of slc4a7 ablation or duodenal HCO₃⁻ secretory rate was a direct effect due to the lack of this transporter in the isolated epithelium. In the isolated, muscle-stripped and chemically denervated duodenal epithelium, we found a strongly significant reduction in the FSK-elicited HCO₃⁻ secretory response compared with the WT mucosa (Fig. 5). Thus, NBCn1 is crucial for maximal HCO₃⁻ secretory rates in the duodenocytes themselves.

To study the role of NBCn1 in duodenocyte pH_i control, fluorometric pH_i measurements were performed in optically isolated duodenocytes in manually dissected intact villi. Steady-state pH_i in WT villous duodenocytes was found to be lower in the presence than in the absence of CO₂/HCO₃⁻ (Fig. 6B) and the absence of NBCn1 did not result in a lower steady-state pH_i (Fig. 6C). This indicates that at an external pH of 7.4, the overall action of all HCO₃⁻ transporters is that of lowering pH_i. This can be due both to electrogenic export of HCO₃⁻ via CFTR (Simpson et al. 2005) and/or NBCe1 (Praetorius et al. 2001), or the electroneutral export via Cl⁻/HCO₃⁻ exchange, or all of them. However, NBCn1 was found to be a major pH_i regulator in villous duodenocytes after an intracellular acid load (Fig. 6D).

A series of experiments was performed to delineate the other Na⁺-dependent base import mechanisms that are active in duodenal villous enterocytes in the presence of CO₂/HCO₃⁻. Interestingly, the second most active base uptake mechanisms under the experimental conditions were NHE1 and NHE3 (Fig. 6E). This is, to our knowledge, the first investigation of the relative contribution of the different NHE isoforms expressed in the duodenocytes in comparison with that of NBC in the presence of CO₂/HCO₃⁻. The potential contribution of the NBCe1 and NBCe2 was assessed by exploiting their sensitivity to low concentrations of the stilbene DIDS (Boyarsky et al. 1988; Abuladze et al. 1998; Choi et al. 1999; Virkki et al. 2002). No contribution of DIDS-sensitive NBCs to pH_i recovery from an acid load was observed in WT duodenocytes, despite the obvious cellular and membrane co-expression of NBCe1 with NBCn1 (Figs 1 and 2). Since the membrane potential has been reported to become more negative during an ammonium prepulse as soon as the extracellular NH₄⁺ is removed in a variety of cell types (Wu & Vaughan-Jones, 1994; Bonnet & Wiemann, 1999), the electrogenicity of NBCe1- and NBCe2-mediated HCO₃⁻ transport may make them relatively uneconomical and/or slow for cellular base uptake in comparison with NBCn1. This concept is consistent with the finding of Praetorius et al. (2001) of a strong increase in pH_i in isolated duodenocytes upon addition of DIDS, which was discussed at the time as a possible indication of HCO₃⁻ exit via DIDS-sensitive transporters including NBCes. In the absence of NBCn1, however, the DIDS-sensitive base uptake mechanisms did play a significant role in recovery from acid pH_i (Fig. 6F).

As FSK was still able to elicit HCO₃⁻ secretion in slc4a7^−/− duodenal mucosa, additional HCO₃⁻ uptake mechanisms are operative during FSK-stimulated HCO₃⁻ secretion. The apically located NHE3, which was found to be a major pH_i regulator after intracellular acidification in duodenocytes, is ruled out as a base loader for HCO₃⁻ secretion because its action has been shown to acidify the lumen (Clarke et al. 2001; Furukawa et al. 2004; Singh et al. 2010). Alternative options are CO₂ diffusion into the enterocyte, intracellular carbonic anhydrase (CA)-facilitated CO₂ hydration, and export of the generated protons via a basolateral NHE (i.e. NHE1), and electrogenic Na⁺(HCO₃⁻)_n cotransport via NBCe1/NBCe2 (Jacob et al. 2000). As the application of systemic DIDS to an anaesthetized mouse will cause multiple confounding effects, we studied the first option in vivo and the second one in vitro. Duodenal CA inhibition was achieved by luminal and systemic application of non-selective CA inhibitors to slc4a7^−/− and slc4a7^+/+ anaesthetized mice prior to stimulation of HCO₃⁻ secretion (Fig. 7). CA inhibition did not significantly affect basal or FSK-stimulated HCO₃⁻ secretory rates in WT mice, but reduced both parameters by approximately 50% in slc4a7^−/− mice, indicating that this HCO₃⁻ supply pathway becomes important in the absence of the major duodenal HCO₃⁻ importer, NBCn1. This agrees with previously published data from rabbit duodenum on the reciprocal substitution of NBC and NHE/CA for the provision of HCO₃⁻ during stimulated HCO₃⁻ secretion (Jacob et al. 2000) and suggests that when NBCn1 activity is absent, CA-generated HCO₃⁻ with H⁺ extrusion via NHE1 compensates partially for the HCO₃⁻ uptake defect (Fig. 9).

Figure 9. Model of the localization and function of villous duodenocyte acid–base transporters.

A, apical and basolateral ion transporters that result in movement of HCO₃⁻ across the membrane, either directly or via proton movement, with demonstrated localization in villous duodenocytes (Jacob et al. 2000, 2002; Praetorius et al. 2000; Wang et al. 2002; Simpson et al. 2005; Walker et al. 2008; Jakab et al. 2011; this paper). B, hypothetical model depicting the major route(s) of bicarbonate entry into and exit from the villous duodenocyte in the resting state (left panel) and during FSK stimulation (right panel). Resting state is defined as: nutrient-free interdigestive phase, no acid discharge from the stomach, no massive HCO₃⁻ output from the pancreas, no neurohumoral stimulation of epithelial secretion. The villous duodenocyte is in an absorptive state. The major HCO₃⁻ exit pathway is probably Slc26a3 (DRA), with minor HCO₃⁻ exit via CFTR. HCO₃⁻ will be generated from CO₂ hydration, with the CO₂ being recycled from the lumen, as well as diffusing from the vasculature/interstitium. HCO₃⁻ will also enter the cell, possibly via NBCe1 > NBCn1 (Na⁺ as well as Cl⁻ are absorbed from the lumen, and a conductive basolateral Cl⁻ exit may lower membrane potential, only constitutive basolateral K⁺ channels active). After FSK stimulation, apical CFTR is phosphoryated and may become more HCO₃⁻ permeable secondary to WNK-SPAK/OSR1 activation (Park et al. 2010). The large HCO₃⁻ efflux via CFTR and the FSK-mediated inhibition of NHE3 will both inhibit HCO₃⁻ efflux via Slc26a3, but stimulate HCO₃⁻ uptake via NBCn1 >NBCe1 (this paper). FSK also activates basolateral K⁺ channels, which may result in less favourable conditions for NBCe1- than NBCn1-mediated HCO₃⁻ uptake.

The potential contribution of NBCe1/e2 for HCO₃⁻ supply during FSK-activated HCO₃⁻ secretion was tested by using the reported DIDS sensitivity of the electrogenic NBCs. However, 200 μm DIDS in the serosal perfusate did not significantly inhibit FSK-stimulated HCO₃⁻ secretion in WT or in NBCn1-deficient mucosa, and 500 μm DIDS caused an increase in epithelial permeability, but still did not inhibit the FSK-induced HCO₃⁻ response (data not shown). Therefore, we could not find evidence for a role of NBCe1/e2 for HCO₃⁻ secretion under the conditions of our experiments. Given the fact that the true sensitivity of the murine pancreatic/intestinal NBCe1 is not known, and DIDS targets many anion transporters but also other proteins, this conclusion is considered provisional and studies in NBCe1-deficient epithelia, possibly generated as intestinal epithelium-specific knockout, are necessary for a final assessment. However, the electrogenicity, the high resting state pH_i in duodenal enterocytes, and the fact that FSK activates basolateral K⁺ channels and thus decreases the electrochemical driving force for NBCe1-mediated HCO₃⁻ uptake, may mask a role for electrogenic NBC in FSK-stimulated duodenal HCO₃⁻ secretion (Fig. 9). Obviously, this may be different for other NBCe1/e2 expressing epithelia like pancreatic ducts or colonic mucosa.

Why do we see a marked difference in basal HCO₃⁻ secretory rates between NBCn1 WT and KO mice in vivo but not in vitro? We know that the ‘basal’ i.e. non-experimentally stimulated HCO₃⁻ secretion, in vivo is under a ‘secretory tone’ because it can be inhibited by a variety of manoeuvres, i.e. the application of β-adrenergic blockers (Singh et al. 2009) or cyclooxygenase inhibitors (Flemstrom et al. 1982). In the experiments in isolated duodenal mucosa, endogenous prostaglandin production was suppressed by indomethacin and any stimulation by neurotransmission blocked by chemical denervation. Thus, the spontaneous secretory tone is presumed to be very low, and indeed basal HCO₃⁻ secretion of isolated duodenal mucosa is only about one fifth to one tenth of that seen in vivo, when calculated in cm². Since NBCn1 is not needed to maintain pH_i in ambient CO₂/HCO₃⁻ and pH (Fig. 6C), we assume that this transporter becomes important when cellular HCO₃⁻ is lost, either by export into the lumen or during cellular acidification. Thus, its importance is seen in vivo without additional external stimulation, but in isolated duodenal mucosa only when HCO₃⁻ secretion is experimentally stimulated.

The conceptual differences between measuring duodenal HCO₃⁻ secretion in vivo and in the isolated mucosa also need to be discussed. The isolated small intestinal murine mucosa in the Ussing chamber is well oxygenated due to its extreme thinness. The epithelium remains in a net salt absorptive mode for hours, suggesting an intact energy supply and tight junctional integrity (Clarke & Harline, 1996; Seidler et al. 2008; Walker et al. 2008). However, for certain substances, there may nevertheless be diffusion from the serosal side to the villous tips. While the problem with agonist diffusion in isolated chambered duodenum may be solved by adding the agonist to both compartments, the same problem may exist with HCO₃⁻, and this cannot be solved experimentally, since the lumen needs to be unbuffered in order to be able to titrate HCO₃⁻ secretion. In addition, CO₂ from the lumen has been shown to enhance duodenal HCO₃⁻ secretion (Furukawa et al. 2005), but it also enhances apical Na⁺/H⁺ exchange and thus HCO₃⁻ absorption from the lumen. (Hubel, 1973; Turnberg et al. 1970; Xia et al. Xia, W, Yu,Q, Seidler U unpublished observations). FSK is known to inhibit luminal Na⁺/H⁺ exchange and may thus mimick a ‘HCO₃⁻ secretory response’. This mechanism is probably completely missing or greatly diminished in the Ussing chamber, because the large reservoir is continuously gassed and any CO₂ diffusion from the serosal bath will be eliminated. It is therefore unclear what the HCO₃⁻ concentration will be at the basolateral membrane of upper villi duodenocytes under those circumstances. Thus, working with isolated epithelium to study small intestinal HCO₃⁻ secretion may result in a relative overestimation of what occurs in the crypt/lower villous region and underestimation of the processes in the upper villous region.

Measurements of duodenal bicarbonate secretion in vivo also have confounding factors, most prominently due to the complexity of signalling in the whole organism. Firstly, the application of agonists may change HCO₃⁻ secretory rates not only by changing the ion transport rates at the epithelial level, but also by changing HCO₃⁻ transport to the epithelium, via changes in vascular and systemic acid–base parameters. Secondly, anaesthesia may compromise the respiratory pattern of the mice, resulting in hyper- or hypoventilation, and this compromises acid–base balance and changes duodenal HCO₃⁻ secretion. We therefore record the ventilation frequency, adjust the isoflurane concentration accordingly, infuse base if necessary, and take blood gas measurements. In the luminally perfused duodenum in vivo, we assume that CO₂ diffusion into the rather narrow lumen is rapid, ambient probably close to identical in blood and lumen, and luminal CO₂ available for HCO₃⁻ supply to the villous enterocytes. We also know that the in vivo‘basal’ HCO₃⁻ secretory rate is actually under a ‘secretory tone’, as shown by further inhibition by ß-adrenergic antagonists (Singh et al. 2009). Therefore, what we call ‘basal HCO₃⁻ secretory rate’in vivo may actually to some extent be activated, FSK-induced Δ; on the other hand, it may in part result from an inhibition of apical Na⁺/H⁺ exchange. Significant differences in ‘basal HCO₃⁻ secretion’ between KO and WT in vivo but not in vitro may be due to the in vivo‘HCO₃⁻ secretory tone’, whereas a significant percental increase in Δ in vitro but not in vivo may be due to a larger interference of FSK with apical proton secretion in vivo than in vitro. To ascertain that NBCn1 is a physiologically important base uptake mechanism during duodenal HCO₃⁻ secretion, we have therefore measured basal and FSK-induced duodenal HCO₃⁻ secretion in slc4a7^−/− mice and WT littermates both in vivo and in isolated mucosa in vitro. The combination of both methods seems most suitable to fully describe the HCO₃⁻ secretory and pH_i regulatory defect in slc4a7^−/− murine duodenum.

In summary, this study demonstrates a prominent role for the villus-expressed electroneutral Na⁺:HCO₃⁻ cotransporter derived from the Slc4a7 gene as a major pH_i regulatory mechanism in murine villous duodenocytes. Furthermore, NBCn1 is essential to maintain basal and stimulated HCO₃⁻ secretory rates. Alternative HCO₃⁻ uptake mechanisms, such as CA-mediated CO₂ hydration or the equally expressed electrogenic NBCe1, are functionally up-regulated in its absence, but cannot maintain normal HCO₃⁻ secretory rates. However, more insight into the regulation of NBCn1 and NBCe1 under different physiological conditions is needed to fully understand the reason for the equally strong expression of these two HCO₃⁻ transporters in the villous duodenocyte basolateral membrane.

Glossary

CA

carbonic anhydrase

FSK

forskolin

Author contributions

All authors designed, performed and analysed experiments, and U.S. and J.P wrote the manuscript. All authors approved the final version.

Author's present address

M. Chen: The College of Medicine, Shaoxing University, Shaoxing, China.

W. Zheng: Department of Gynecology and Obstetrics, Hannover Medical School, Hannover, German.

F. Xiao: Department of Gastroenterology, Tongji Medical Colledge, Wuhan, China.

J. Wang: Department of Surgery, Jinling Hospital, Nanjing University School of Medicine, Nanjing, China.

Supplementary material

Supplementary Table S1