Colonic Anion Secretory Defects and Metabolic Acidosis in Mice Lacking the NBC1 \batchmode \documentclass[fleqn,10pt,legalpaper]{article} \usepackage{amssymb} \usepackage{amsfonts} \usepackage{amsmath} \pagestyle{empty} \begin{document} \(\mathrm{Na}^{+}{/}\mathrm{HCO}_{3}^{-}\) \end{document} Cotransporter*

The NBC1 \batchmode \documentclass[fleqn,10pt,legalpaper]{article} \usepackage{amssymb} \usepackage{amsfonts} \usepackage{amsmath} \pagestyle{empty} \begin{document} \(\mathrm{Na}^{+}{/}\mathrm{HCO}_{3}^{-}\) \end{document} cotransporter is expressed in many tissues, including kidney and intestinal epithelia. NBC1 mutations cause proximal renal tubular acidosis in humans, consistent with its role in \batchmode \documentclass[fleqn,10pt,legalpaper]{article} \usepackage{amssymb} \usepackage{amsfonts} \usepackage{amsmath} \pagestyle{empty} \begin{document} \(\mathrm{HCO}_{3}^{-}\) \end{document} absorption in the kidney. In intestinal and colonic epithelia, NBC1 localizes to basolateral membranes and is thought to function in anion secretion. To test the hypothesis that NBC1 plays a role in transepithelial \batchmode \documentclass[fleqn,10pt,legalpaper]{article} \usepackage{amssymb} \usepackage{amsfonts} \usepackage{amsmath} \pagestyle{empty} \begin{document} \(\mathrm{HCO}_{3}^{-}\) \end{document} secretion in the intestinal tract, null mutant (NBC1-/-) mice were prepared by targeted disruption of its gene (Slc4a4). NBC1-/- mice exhibited severe metabolic acidosis, growth retardation, reduced plasma Na+, hyperal-dosteronism, splenomegaly, abnormal dentition, intestinal obstructions, and death before weaning. Intracellular pH (pHi) was not altered in cAMP-stimulated epithelial cells of NBC1-/- cecum, but pHi regulation during sodium removal and readdition was impaired. Bioelectric measurements of NBC1-/- colons revealed increased amiloride-sensitive Na+ absorption. In Ringer solution containing both Cl- and \batchmode \documentclass[fleqn,10pt,legalpaper]{article} \usepackage{amssymb} \usepackage{amsfonts} \usepackage{amsmath} \pagestyle{empty} \begin{document} \(\mathrm{HCO}_{3}^{-}\) \end{document}, the magnitude of cAMP-stimulated anion secretion was normal in NBC1-/- distal colon but increased in proximal colon, with the increase largely supported by enhanced activity of the basolateral NKCC1 Na+-K+-2Cl- cotransporter. Anion substitution studies in which carbonic anhydrase was inhibited and transepithelial anion conductance was limited to \batchmode \documentclass[fleqn,10pt,legalpaper]{article} \usepackage{amssymb} \usepackage{amsfonts} \usepackage{amsmath} \pagestyle{empty} \begin{document} \(\mathrm{HCO}_{3}^{-}\) \end{document} revealed a sharp decrease in both cAMP-stimulated \batchmode \documentclass[fleqn,10pt,legalpaper]{article} \usepackage{amssymb} \usepackage{amsfonts} \usepackage{amsmath} \pagestyle{empty} \begin{document} \(\mathrm{HCO}_{3}^{-}\) \end{document} secretion and SITS-sensitive current in NBC1-/- proximal colon. These results are consistent with the known function of NBC1 in \batchmode \documentclass[fleqn,10pt,legalpaper]{article} \usepackage{amssymb} \usepackage{amsfonts} \usepackage{amsmath} \pagestyle{empty} \begin{document} \(\mathrm{HCO}_{3}^{-}\) \end{document} absorption in the kidney and demonstrate that NBC1 activity is a component of the basolateral mechanisms for \batchmode \documentclass[fleqn,10pt,legalpaper]{article} \usepackage{amssymb} \usepackage{amsfonts} \usepackage{amsmath} \pagestyle{empty} \begin{document} \(\mathrm{HCO}_{3}^{-}\) \end{document} uptake during cAMP-stimulated anion secretion in the proximal colon.

Human patients with proximal renal tubular acidosis resulting from mutations in NBC1 have been reported (12)(13)(14)(15)(16), thereby confirming a bicarbonate-absorptive role for NBC1 in kidney. The primary mutations were single amino acid substitutions (R298S, T485S, R510H, A799V, R881C, and S427L), which appeared to cause decreased function of the cotransporter rather than loss of function (12)(13)(14). One patient had an inactivating mutation in the unique N terminus of the kidney NBC1 variant (Q29X), but the pancreatic variant, which is expressed in many other tissues and at low levels in kidney (4), was intact (15). Only a single patient has been identified with a complete inactivating mutation, a nucleotide deletion that causes a frameshift at codon 721 (16). The pRTA resulting from NBC1 mutations clearly shows that this transporter is essential for renal HCO 3 Ϫ absorption; however, clinically significant intestinal disease has not been reported.
NBC1 has been localized to the basolateral membrane of epithelial cells lining both the small and large intestine (8,17,18). In the colon, its expression was greatest in crypt cells, consist-ent with a role in transepithelial anion secretion (19). Previous studies investigating Na ϩ /HCO 3 Ϫ cotransport activity in intestinal epithelia using relatively nonspecific inhibitors of HCO 3 Ϫ transport support the hypothesis that NBC1 is a component of the basolateral anion uptake mechanisms that facilitate transepithelial anion secretion and regulate pH i (18 -20); however, at least two additional Na ϩ /HCO 3 Ϫ cotransporters, the electroneutral NBCn1 and the electrogenic NBC4, are also expressed in intestinal epithelia (17,21). NBCn1 is present on the basolateral membrane of small intestinal enterocytes (17) and is co-expressed with NBC1 in colonic crypts (18), whereas the membrane localization of NBC4 has not been determined in intestinal epithelia (21). To understand the role of NBC1 in transepithelial anion secretion in the intestinal tract and also to assess its importance in kidney and other tissues, we developed a mouse carrying a targeted disruption of the Slc4a4 gene. Our data show that the loss of NBC1 in mice causes severe metabolic acidosis and impaired transepithelial HCO 3 Ϫ secretion in the colon. These results are consistent with the known function of NBC1 in bicarbonate absorption in the kidneys and demonstrate that it also serves a major secretory function in the intestinal tract.

Preparation of Targeting Construct and Generation of
Mutant Mice-PCR generated fragments of the mouse Slc4a4 gene and the pMJKO vector (22) were used for the targeting construct. A 3.2-kb genomic fragment starting in intron 7 and ending in exon 9 was inserted into a cloning site 5Ј to the promoter of the neomycin resistance gene, and a 2.3-kb fragment corresponding to sequences in intron 9 was inserted between the 3Ј end of the neomycin resistance gene and the herpes simplex virus thymidine kinase gene. Targeting of ES cells (cells derived from 129S6/SvEv Tac (Taconic, Germantown, NY) mice) was performed as previously described (22). ES cells that were positive for homologous recombination were identified by Southern blot analysis of EcoRV-digested DNA using a 5Ј probe corresponding to genomic sequences from intron 6. Targeted ES cells were used to generate chimeric mice that were bred with wild-type Black Swiss females (Taconic). All studies were performed using mice of the mixed 129S6/SvEv and Black Swiss background with wild-type age-matched littermates serving as controls. At ϳ5 days of age, offspring were genotyped by PCR analysis of tail DNA using the following primers: a forward primer from the deleted region of intron 9 (5Ј-TCACAAACC-TTTCAGCAAAAGAGTGC-3Ј) that identifies only the wildtype allele; a reverse primer from intron 9 (5Ј-CAAAGAGCA-ACAGTCAGACAGC-3Ј) that identifies both wild-type and mutant alleles; and a primer from the neomycin resistance gene (5Ј-GACAATAGCAGGCATGCTGG-3Ј) that identifies only the mutant allele. Amplification using DNA from a tail biopsy and all three primers in the same reaction yields 269-and 241-bp products for the wild-type and mutant alleles, respectively.
Northern Blot Analysis-Total RNA (30 g) isolated from tissues of 14-day-old mice using Tri Reagent (Molecular Research Center, Inc., Cincinnati, OH) was denatured with glyoxal, fractionated by electrophoresis in 1% agarose, transferred to a nylon membrane, and hybridized with a 32 P-labeled NBC1 cDNA probe.
Histology-Tissue samples (brain, eye, kidney, spleen, pancreas, small intestine, cecum, proximal colon, distal colon, lung, and heart) were fixed in 10% neutral buffered formalin, dehydrated, and embedded in paraffin. Samples were stained with hematoxylin and eosin and examined for histological changes by light microscopy.
Analysis of Blood and Blood Counts-At 14 -15 days of age, NBC1 ϩ/ϩ , NBC1 ϩ/Ϫ , and NBC1 Ϫ/Ϫ pups were sacrificed by decapitation, and trunk blood was collected into heparinized tubes for analysis on a blood gas analyzer (Chiron Diagnostics model 248, Norwood, MA). Plasma Na ϩ and K ϩ concentrations were determined using flame photometry (model 480; Corning Glass). For peripheral blood counts, blood smears were obtained from the tail vein and allowed to dry. Samples were then stained for 2 min with Giemsa stain, washed in distilled water, and coverslipped using CytoSeal. Smears were analyzed for the percentage of nucleated red blood cells, myelocytes (includes myelocytes, neutrophils, eosinophils, and basophils), lymphocytes, and mononuclear white blood cells.
Analysis of Serum Aldosterone-Mice were euthanized with CO 2 , and blood was drawn by cardiac puncture. Serum was separated from whole blood via centrifugation and was stored at Ϫ80°C. Serum from two mice was pooled together and diluted 1:4 in phosphate-buffered saline; a total of 14 mice were used for each genotype. Aldosterone concentrations were determined using a commercially available radioimmunoassay kit (Diagnostic Products, Los Angeles, CA) according to the manufacturer's directions. Samples were counted using a ␥ counter (Packard Instrument Co.).
BCECF Microfluorimetry-The method used for imaging intact intestinal epithelium of the cecum was based on a previously described technique for imaging epithelial cells in intact duodenum (23). Following asphyxiation in 100% CO 2 and bilateral pneumothorax, the cecum of 14 -18-day-old animals was removed and placed immediately in an oxygenated, ice-cold Ringer solution. The ceca were opened along the mesenteric border, and the serosa and muscularis externa were removed from the underlying mucosa as previously described (24). The muscle-stripped preparations were mounted basolateral (serosal) side up on a horizontal bilateral perfusion chamber (24). The luminal (mucosal) and serosal surfaces were independently bathed with Ringer solution containing 1 M indomethacin to minimize the effect of endogenous prostaglandins and 10 M forskolin (to stimulate anion secretion and basolateral anion uptake) (25,26). In addition, the serosal bathing medium contained 0.1 M tetrodotoxin to minimize neural tone and 1 M EIPA to inhibit activity of the basolateral Na ϩ /H ϩ exchanger. Ussing Chamber Analysis-Following asphyxiation of 14 -18-day-old animals in 100% CO 2 and bilateral thoracotomy, proximal, and distal colon samples were removed by a midline incision and placed in oxygenated, ice-cold Ringer solution (with 1 M indomethacin). Tissues were opened along the mesenteric border and the muscle layers underlying the mucosa of the proximal colon segments removed by sharp dissection. Distal colon segments were used with the muscle layer intact. Colons were mounted in standard Ussing chambers (0.1-cm 2 exposed surface area), and the mucosal and serosal surfaces were independently bathed in 4 ml of Krebs-bicarbonate Ringer solution (37°C, gassed with 95% O 2 and 5% CO 2 ) as described previously (27,28). For Cl Ϫ substitution experiments, gluconate was substituted on an equimolar basis. To minimize the variations in neural tone and the generation of endogenous prostanoids, tetrodotoxin (0.1 M serosal) and indomethacin (1 M, mucosal and serosal) were added to the bathing solutions (25,26). Forskolin (10 M) and 3-isobutyl-1methylxanthine (100 M) were added to the mucosal and serosal baths of amiloride-pretreated (10 M) tissues to stimulate intracellular cAMP prior to the sequential addition of bumetanide (100 M) and SITS (1 mM) to the serosal side. For all experiments, the final concentration of dimethyl sulfoxide in the bath Ringer solution was maintained at or below 0.1%.
Transepithelial short circuit current (I sc ; reported as A/cm 2 tissue surface area) was measured using an automatic voltage clamp (VCC-600; Physiologic Instruments, San Diego, CA) as previously described (29). Transepithelial conductance (G t , reported in mS/cm 2 tissue surface area) was determined at 5-min intervals during the experiment by measuring the magnitude of the current deflection from application of a 5-mV pulse across each tissue and applying Ohm's law. All experiments were performed under short circuited conditions with the serosal bath serving as ground.
Immunoblot Analysis-Mice were euthanized, and the colons were removed, cleaned in ice-cold 1ϫ phosphate-buffered saline, flash-frozen in liquid nitrogen, and stored at Ϫ70°C until further processing. Frozen tissues were pulverized in liquid nitrogen using a tissue grinder, and the tissue powder was suspended in prechilled 1ϫ homogenization buffer (10 mM NaCl, 20 mM Pipes (pH 7.0), 5 mM EDTA, 0.5% Nonidet P-40, 2 mM dithiothreitol plus protease inhibitors and phosphatase inhibitors). The samples were then homogenized using a Polytron 3000 homogenizer, and the proteins were allowed to solubilize over ice for 2 h. Protein concentration was estimated by the Bradford method.
The presence of NKCC1 protein and phosphorylated NKCC1 was determined by immunoblotting after separation of lysate proteins by electrophoresis on a discontinuous, 7% reducing SDS-polyacrylamide gel. The antibodies (described in Refs. 30 and 31) used were T4 for NKCC1 (contributed to the Developmental Studies Hybridoma Bank, University of Iowa (Iowa City, IA) by C. Lytle and B. Forbush), R5 for NKCC1 that was phosphorylated on two regulatory threonine residues (a gift from Dr. Biff Forbush, Yale University), and antiactin (Sigma A4700) for the loading control.
Statistics-A two-tailed unpaired Student's t test assuming equal variances was used to compare data from two treatment groups. A one-way analysis of variance with a post hoc Tukey's t test was used for comparisons among more than two treatment groups. A probability value of p Ͻ 0.05 was considered statistically significant. All values are reported as the mean Ϯ S.E.
Materials-All reagents were obtained from Sigma. Tetrodotoxin was dissolved at a stock concentration of 100 M in 0.2% acetic acid. Forskolin and indomethacin were dissolved at stock concentrations of 10 mM in dimethyl sulfoxide. 3-Isobutyl-1methylxanthine and amiloride were dissolved at stock concentrations of 10 mM in sterile water. EIPA was dissolved at a stock concentration of 1 mM in dimethyl sulfoxide. Acetazolamide and SITS were dissolved at stock concentrations of 10 and 100 mM in the perfusion Ringer solution for the experiment in which they were used (i.e. Krebs bicarbonate Ringer or Cl Ϫ -free Ringer solution). Bumetanide was dissolved at a stock concentration of 100 mM in ethanol.

Generation of NBC1 Null Mutant Mice-
The targeting procedure replaced sequences that include part of exon 9 (including the 3Ј splice site), which is present in the mRNAs for both NBC1 variants, and part of intron 9 with the neomycin resistance gene (Fig.  1A). Chimeric male mice generated using the targeted ES cells were bred to wild-type females to produce NBC1 ϩ/Ϫ mice, and breeding of heterozygous mutant mice resulted in live offspring of all three genotypes (Fig. 1, B and C). Northern blot analysis of kidney, small intestine, and colon demonstrated that the NBC1 mRNA was ablated in NBC1 Ϫ/Ϫ mice (Fig. 1D).
Gross Phenotype-Genotype frequencies of pups obtained from heterozygous matings exhibited a normal 1:2:1 Mendelian ratio (25.9% wild type, 48.6% heterozygous, and 25.5% null mutant among more than 1000 pups), with no alteration in the percentage of male or female null mutant mice. At birth, ϳ25% of NBC1 Ϫ/Ϫ mice were notably smaller than their wild-type and heterozygous littermates. As they developed during the postnatal period, all of the knockouts were emaciated, exhibited severe growth retardation (Fig. 2, A and B), and their teeth had a chalky white appearance and were easily chipped. NBC1 null mutants had poor survival, with death beginning about 5 days after birth and no survival beyond 24 days (Fig. 2C).
Of the knock-out pups that survived to at least 20 days of age, ϳ80% exhibited mild to severe intestinal impactions in the terminal ileum, cecum, and colon; however, the intestinal tracts of mice that died at much earlier ages were not obstructed. Regardless of whether they had intestinal impactions, 12-21-day-old NBC1-null mice had small, corkscrew ceca, a phenotype that has been observed in mice lacking either the CFTR Cl Ϫ channel (32,33) or the NKCC1 Na ϩ -K ϩ -2Cl Ϫ cotransporter (34), both of which have defects in intestinal anion secretion (34,35).
A broad range of tissues from 14 -18-day-old wild-type and NBC1 Ϫ/Ϫ mice were analyzed by light microscopy. Despite the intestinal anion secretory defect described below and the ocular defects and mental retardation observed in some patients with NBC1 deficiency (12)(13)(14)(15)(16), all segments of the intestinal tract, eye, and brain appeared normal. Although NBC1 is expressed in renal proximal tubule epithelial cells, the proximal tubule appeared histologically normal. As described below, spleen and blood abnormalities were observed, although these appear to be secondary to the metabolic acidosis. Analysis of bone was not performed; however, when brains were isolated for histological analysis, the skulls were observed to be very thin and virtually transparent, which is likely to be due to the severe acidosis (see "Discussion").
Analysis of Blood pH, Gasses, Ions, and Aldosterone Levels-Because the most prominent phenotype in humans with NBC1 mutations is pRTA, resulting from a reduction in NBC1 activity in the kidney, we analyzed blood from 14 -17-day-old mice of all three genotypes ( Table 1). The hematocrit was significantly lower in NBC1 Ϫ/Ϫ mice. Blood pH and HCO 3 Ϫ concentrations were sharply reduced in null mutants, and because there was no diarrheal state, which is the only other condition that could lead to significant loss of HCO 3 Ϫ , this systemic acidosis is clearly of renal origin. Blood HCO 3 Ϫ concentrations were also significantly reduced in heterozygous mutants, indicating that NBC1 haploinsufficiency causes a mild pRTA. Plasma Na ϩ concentrations were mildly reduced in NBC1 Ϫ/Ϫ mice, whereas there was no significant difference in K ϩ concentrations (Table 1). Serum aldosterone levels were sharply increased in NBC1 Ϫ/Ϫ mice (Fig. 3), as would be expected in response to a proximal tubule absorptive defect and resulting hypovolemia.

Analysis of Spleen and Peripheral Blood Cells-
The abdominal organs of NBC1 Ϫ/Ϫ mice were pale, which could be due in part to the observed anemia, and appeared to be poorly perfused when compared with wild-type mice. The most apparent anatomical abnormality in NBC1 Ϫ/Ϫ mice FIGURE 1. Generation of NBC1 null mutant mice. A, gene targeting strategy showing a map of the wild-type gene, a diagram of the targeting construct containing the neomycin resistance gene (NEO) and the thymidine kinase gene (TK), and the predicted structure of the targeted allele and location of the diagnostic probe. B, Southern blot analysis of tail DNA from offspring of a heterozygous mating. DNA was digested with EcoRV and hybridized with a 5Ј outside probe, which identifies a 6.8-kb fragment in the wild-type allele and a 6.3-kb fragment in the mutant allele. C, PCR genotyping of tail biopsies from NBC1 ϩ/ϩ , NBC1 ϩ/Ϫ , and NBC1 Ϫ/Ϫ mice. D, Northern blot analysis of total RNA (30 g/lane) from small intestine, colon, and kidney of 15-day-old NBC1 ϩ/ϩ , NBC1 ϩ/Ϫ , and NBC1 Ϫ/Ϫ mice hybridized with an NBC1 cDNA probe. FIGURE 2. Gross phenotype of NBC1 null mutant mice. A, compared with gender-matched NBC1 ϩ/ϩ littermates, 15-day-old NBC1 Ϫ/Ϫ mice exhibited growth retardation and emaciation. B, body weights of NBC1 Ϫ/Ϫ mice on day 13 were significantly reduced compared with NBC1 ϩ/ϩ mice (n ϭ 12 ϩ/ϩ, 15 Ϫ/Ϫ; p Ͻ 3 ϫ 10 Ϫ7 ); null mutants exhibited little growth beyond that age, with few exceeding 7 g. C, survival curve for NBC1 ϩ/ϩ and NBC1 Ϫ/Ϫ mice. The percentage of surviving mice was plotted against age in days, with day 0 being the day of birth. NBC1 Ϫ/Ϫ mice exhibited a sharp increase in mortality starting around 8 days of age with few mice surviving to 21 days of age (n ϭ 153 for each genotype).
was severe splenomegaly (0.46 Ϯ 0.03 and 1.30 Ϯ 0.39% of body weight for NBC1 ϩ/ϩ and NBC1 Ϫ/Ϫ spleen, respectively; n ϭ 9 -12; p Ͻ 0.009). Unlike wild-type spleen, in which a clear organization of the white pulp and red pulp was apparent ( Fig. 4A), follicular organization of the NBC1 Ϫ/Ϫ spleen was severely disrupted (Fig. 4B). Both white pulp and red pulp were increased in NBC1 Ϫ/Ϫ spleen, but areas of lymphocytes predominated (Fig. 4, compare C and D), and mitoses were frequent. A count of megakaryocytes at ϫ40 revealed fewer per field in null mutants (1.05 Ϯ 0.40 and 2.73 Ϯ 0.31 in wild type and null mutant, respectively, p Ͻ 0.01, n ϭ 5 for each genotype), and the megakaryocytes appeared smaller in size in NBC1 Ϫ/Ϫ spleens (Fig. 4, compare C and D). The spleen defect is probably secondary to the severe acidosis, since NBC1 was not detected in wild-type spleen by Northern blot analysis (data not shown). Counts of peripheral blood cells (Table 2) revealed a statistically significant decrease in the relative numbers of lymphocytes and a relative increase in myelocytes. There was a 7-fold increase in the percentage of nucleated red blood cells.
Absence of NBC1 Activity in NBC1 Ϫ/Ϫ Cecum-The effect of NBC1 deletion on the regulation of intracellular pH during cAMP stimulation was assessed in the cecal surface epithelium using the pH-sensitive dye BCECF-AM. Cecum was used, because the surface epithelial cells are easily visualized and, unlike colonic surface epithelium, retain dye sufficiently for pH i measurements. To isolate pH i changes due to activity of a baso-lateral NBC, 1 M EIPA was included in the serosal bath solution to inhibit regulation of pH i by the basolateral NHE1 Na ϩ /H ϩ exchanger. Following the addition of 10 M forskolin to stimulate intracellular cAMP levels and increase NBC1 activity (19), the effect of serosal Na ϩ removal and readdition on pH i was assessed in NBC1 ϩ/ϩ and NBC1 Ϫ/Ϫ surface epithelium bathed in physiological Ringer solution (containing both Cl Ϫ and HCO 3 Ϫ ). As shown in Fig. 5, the base-line pH i before Na ϩ removal and readdition was not significantly different between NBC1 ϩ/ϩ and NBC1 Ϫ/Ϫ epithelia. However, as shown by the pH i trace in Fig. 6A, NBC1 ϩ/ϩ cells exhibited rapid acidification following removal of Na ϩ (indicative of reversal of Na ϩ / HCO 3 Ϫ cotransport) and alkalinization upon the readdition of Na ϩ to the serosal solution. In contrast, NBC1 Ϫ/Ϫ cecal epithelial cells demonstrated marked attenuation of these pH i FIGURE 3. Serum aldosterone levels in NBC1 ؉/؉ , NBC1 ؉/؊ , and NBC1 ؊/؊ mice. Serum was collected from blood of euthanized mice, and aldosterone levels were determined by radioimmunoassay. Serum aldosterone was significantly elevated in NBC1 Ϫ/Ϫ mice when compared with both NBC1 ϩ/Ϫ and NBC1 ϩ/ϩ mice. ϩ , p Ͻ 0.001 by analysis of variance; n ϭ 7 sets of samples for each genotype, with each sample containing pooled serum from two mice.

Intestinal Secretory Function of NBC1
responses during Na ϩ removal and readdition (Fig. 6B). The cumulative data (Fig. 6, C and D) show that the rate of acidification when serosal Na ϩ was removed and the rate of pH i recovery when Na ϩ was restored were significantly reduced in NBC1 Ϫ/Ϫ cells, consistent with the loss of a major basolateral Na ϩ /HCO 3 Ϫ cotransport system.
Enhanced Na ϩ Absorption in NBC1 Ϫ/Ϫ Colons-Bioelectric measurements using Ussing chambers revealed increased amiloride-sensitive I sc , a measure of Na ϩ absorption via the amiloride-sensitive epithelial Na ϩ channel in both proximal and distal colons of NBC1 null mice as compared with wildtype controls (Fig. 7, A and B). This finding is not surprising given the elevation of plasma aldosterone, which is known to induce ENaC expression and activity in the colon (36).
Anion Secretory Capacity in the Colon of NBC1 Ϫ/Ϫ and Wildtype Mice-In small intestine and colon, bulk transepithelial anion secretion occurs predominantly in the crypts, and expression of both the apical membrane anion channel (CFTR) and NBC1 are higher in crypts than in villi or surface epithelium (18,37). Because of the small size of NBC1 null mutants (4 -7 g), it was not possible to analyze small intestinal tissue in Ussing chambers, and only colonic tissue was used to analyze the anion secretory functions of NBC1. Tissues were pretreated with amiloride to block ENaC activity.
Despite previous studies indicating that NBC1 is important in HCO 3 Ϫ uptake to support transepithelial anion secretion (18,19), NBC1 Ϫ/Ϫ proximal colons exhibited no deficit but had a FIGURE 5. pH i of surface epithelial cells from cAMP-stimulated NBC1 ؉/؉ and NBC1 ؊/؊ cecum. NBC1 Ϫ/Ϫ and wild-type ceca were mounted in a horizontal Ussing chamber after removal of the underlying serosa and muscularis externa, and pH i of the surface epithelial cells was measured using BCECF-AM microfluorimetry. The tissues were treated with 1 M EIPA to inhibit the NHE1 Na ϩ /H ϩ exchanger and then stimulated with cAMP (10 M forskolin). Under these conditions, there was no significant difference in the pH i value between NBC1 ϩ/ϩ and NBC1 Ϫ/Ϫ cecal surface epithelial cells. n ϭ ceca from three mice of each genotype. FIGURE 6. Rate of change in pH i of NBC1 ؉/؉ and NBC1 ؊/؊ cecal epithelial cells in response to removal or readdition of Na ؉ . NBC1 Ϫ/Ϫ and wild-type cecal preparations (n ϭ 3) from the experiment shown in Fig. 5 were subjected to removal and then readdition of Na ϩ on the basolateral side of the tissue. As indicated, both apical (Ap) and basolateral (BL) surfaces were bathed in Ringer solution containing CO 2 and HCO 3 Ϫ ; Na ϩ was manipulated only in the basolateral solution and was present in the apical solution throughout the experiment. A, during inhibition of NHE1, removal and then readdition of Na ϩ resulted in cellular acidification and then recovery of pH i in NBC1 ϩ/ϩ ceca, consistent with basolateral Na ϩ /HCO 3 Ϫ cotransport activity. B, pH i of NBC1 Ϫ/Ϫ cecal surface cells exhibited little change in response to removal or readdition of Na ϩ . C, the rate of cellular acidification following removal of Na ϩ was significantly reduced in NBC1 Ϫ/Ϫ cecum (*, p Ͻ 0.01). D, the rate of pH i recovery following the readdition of Na ϩ was significantly reduced in NBC1 Ϫ/Ϫ cecum (*, p Ͻ 0.001). significantly increased maximal I sc response to cAMP stimulation when experiments were performed in physiologic Ringer solution (containing both Cl Ϫ and HCO 3 Ϫ ) (Fig. 8A). The increase in I sc response was supported by increased activity of the NKCC1 Na ϩ -K ϩ -2Cl Ϫ cotransporter, as indicated by a greater bumetanide-sensitive I sc (Fig. 8A). There were no significant differences between wild-type and NBC1 Ϫ/Ϫ tissues in either the SITS-sensitive I sc or the residual cAMP-stimulated I sc remaining after sequential treatment with bumetanide and SITS. In distal colon, there were no differences between wildtype and NBC1 Ϫ/Ϫ tissues in the I sc response to cAMP stimulation or subsequent treatment with bumetanide (Fig. 8B), indicating that increased activity of NKCC1 in response to NBC1 deficiency is limited to the proximal segment of the colon.
Northern blot analysis revealed no significant change in NKCC1 mRNA levels (data not shown), indicating that tran-scriptional up-regulation of NKCC1 is not involved in the observed increase in NKCC1 activity in NBC1 Ϫ/Ϫ proximal colon. Immunoblot analysis of NKCC1 protein (Fig. 9A) also showed no significant differences between the two genotypes, indicating that increased NKCC1 protein was not involved. Because phosphorylation of NKCC1 is a major mechanism for acute up-regulation of NKCC1 activity, immunoblot analysis was performed using an antibody (31) that recognizes two of the phosphothreonine residues involved in activation of NKCC1 (Fig. 9B). This experiment also revealed no significant differences between the two genotypes.
To analyze the role of NBC1 in HCO 3 Ϫ secretion, we removed Cl Ϫ from the bathing medium, thereby eliminating activity of the Cl Ϫ -dependent basolateral Na ϩ -K ϩ -2Cl Ϫ cotransporter. This manipulation left HCO 3 Ϫ , which could be taken up across the basolateral membrane and/or generated inside the cell by the activity of carbonic anhydrase as the only CFTRpermeant anion. Under these conditions, we observed no difference in the I sc response to cAMP between NBC1 Ϫ/Ϫ and wild-type tissues (Fig. 10A). Furthermore, cAMP-stimulated NBC1 Ϫ/Ϫ and wild-type proximal colons exhibited essentially the same reduction in I sc response following treatment with SITS (Ϫ36.0 Ϯ 9.2 and Ϫ32.3 Ϯ 9.5 A/cm 2 in wild-type and NBC1 Ϫ/Ϫ tissues, respectively). The presence of a SITS-sensitive current in NBC1 Ϫ/Ϫ proximal colon, at levels equivalent to those observed in wild-type tissue, suggests the presence of a non-NBC1, SITS-sensitive HCO 3 Ϫ transporter in NBC1 Ϫ/Ϫ proximal colon that mediates ϳ20% of the anion current under these experimental conditions. However, the experiments shown in Fig. 10A did not reveal a contribution by NBC1, which has been reported to be sensitive to inhibition by SITS (20).
In contrast to the above findings, inhibition of carbonic anhydrase activity by acetazolamide, which would restrict the source of HCO 3 Ϫ to that transported into the cell from the serosal solution, greatly attenuated the cAMP-stimulated I sc response in the NBC1 Ϫ/Ϫ proximal colon (Fig. 10B). The I sc response to cAMP in NBC1 ϩ/ϩ proximal colons was equivalent to that observed in the FIGURE 8. Transepithelial Cl ؊ and HCO 3 ؊ secretion in NBC1 ؊/؊ proximal and distal colon. Tissues were mounted in Ussing chambers, pretreated with amiloride to inhibit ENaC, and I sc measurements were performed during the sequential addition of pharmacologic agents (n ϭ 4 of each genotype). A, change in I sc in wild-type and NBC1 Ϫ/Ϫ proximal colon in response to increased cAMP (forskolin/3-isobutyl-1-methylxanthine), bumetanide (an inhibitor of NKCC1), and SITS (a relatively nonspecific inhibitor of NBC1 and other HCO 3 Ϫ transporters). Bumetanide-sensitive and SITS-sensitive components of the cAMP-stimulated I sc are indicated by negative currents. The residual cAMP-stimulated I sc remaining after treatment with bumetanide and SITS is also shown. NBC1 Ϫ/Ϫ proximal colon exhibited an increased I sc response to cAMP (*, p Ͻ 0.01) and greater inhibition of the I sc response by bumetanide (*, p Ͻ 0.001) when compared with wild-type tissue. B, change in I sc in wild-type and NBC1 Ϫ/Ϫ distal colon in response to increased cAMP and bumetanide (SITS was not used because wild-type distal colon exhibits no response to SITS). absence of carbonic anhydrase activity. However, the cAMP-stimulated I sc in acetazolamide-treated wild-type proximal colon exhibited greatly increased sensitivity to inhibition by SITS. The magnitude of this change was approximately equal to the difference in the I sc response to cAMP between wild-type and NBC1 Ϫ/Ϫ proximal colons under these conditions, suggesting that NBC1 activity can be increased during carbonic anhydrase inhibition to maintain normal levels of HCO 3 Ϫ secretion. The fact that the SITSsensitive current in the NBC1 Ϫ/Ϫ proximal colons was sharply reduced or absent during carbonic anhydrase inhibition (Fig. 10,   compare A and B) also suggests that the putative non-NBC1, SITSsensitive HCO 3 Ϫ transporter is dependent on carbonic anhydrase activity. Finally, the residual current detected in both NBC1 ϩ/ϩ and NBC1 Ϫ/Ϫ proximal colon indicated the presence of a non-NBC1, SITS-insensitive HCO 3 Ϫ transporter that contributes to transepithelial, electrogenic HCO 3 Ϫ secretion (Fig. 10B).

DISCUSSION
Although many of the NBC1 Ϫ/Ϫ mice were small at birth when compared with their wild-type or heterozygous littermates, the normal Mendelian ratio of null mutants suggests that the loss of NBC1 does not seriously impair embryogenesis. Those that survived beyond day 1 were severely growth-retarded and exhibited increasing mortality rates at around day 5, and few survived beyond 20 days of age. Nevertheless, analysis of blood pH and HCO 3 Ϫ concentrations allowed confirmation of the metabolic acidosis that would be predicted on the basis of the well established HCO 3 Ϫ -absorptive function of NBC1 in the kidney. More importantly, because intestinal anion secretion is well established before weaning, we were able to assess the role of NBC1 in colonic anion secretion.
Metabolic Acidosis and Mild Hyponatremia-Severe metabolic acidosis, with plasma HCO 3 Ϫ concentrations of 4.0 -7.6 mM and pH values of 6.80 -6.93, was the phenotypic trait most likely to be responsible for the high morbidity and mortality among NBC1 null mutants. Metabolic and respiratory acidosis and decreased plasma HCO 3 Ϫ concentrations in neonates have been associated with blood hyperviscosity and poor perfusion of organs (brain, kidney, and digestive tract) (38). Given (i) the expression of NBC1 in the renal proximal tubule, including that of the mouse (39), where it mediates HCO 3 Ϫ reabsorption across the basolateral membrane (1-3); (ii) the absence of a diarrheal state, which can also cause HCO 3 Ϫ wasting; and (iii) the occurrence of pRTA in human patients with mutations in NBC1, it seems likely that the acidosis in the NBC1 mouse model is a proximal renal tubular acidosis. Interestingly, NBC1 heterozygous mouse mutants exhibited a small but significant decrease in blood HCO 3 Ϫ , suggesting that NBC1 haploinsufficiency in mice causes a mild metabolic acidosis. This observation is consistent with reports that some of the NBC1 mutations associated with pRTA in humans may cause decreased ion transport rather than complete loss of activity (12,14).
A small reduction in plasma Na ϩ and an increase in serum aldosterone, with no increase in plasma K ϩ , were also observed, consistent with a role for NBC1 in Na ϩ -fluid volume homeostasis. The mild hyponatremia was surprising, because reductions in plasma Na ϩ have not been reported in humans with NBC1 deficiency, and mice lacking either the NHE3 Na ϩ /H ϩ exchanger (40), the NKCC2 Na ϩ -K ϩ -2Cl Ϫ cotransporter (41), or the thiazidesensitive NaCl cotransporter (42), which function as the major transporters for uptake of Na ϩ across the apical membranes of the proximal tubule, thick ascending limb, and distal convoluted tubule, respectively, exhibit no reductions in plasma Na ϩ .
Dental Abnormalities-There are at least two reports of pRTA patients with confirmed NBC1 mutations that also had dental abnormalities (13,16) and an earlier report described two patients with inherited pRTA of unknown causes that had similar dental abnormalities (43). The occurrence of a tooth Ϫ currents were analyzed in Ussing chambers using Cl Ϫ -free solutions in which gluconate was substituted for Cl Ϫ . Tissues (n ϭ 4 of each genotype) were pretreated with amiloride to inhibit ENaC, stimulated with cAMP, and then treated with SITS on the serosal side of the epithelium. A, HCO 3 Ϫ -mediated cAMP-stimulated I sc responses, SITS-sensitive component of the I sc , and residual I sc in NBC1 ϩ/ϩ and NBC1 Ϫ/Ϫ proximal colon did not differ significantly when HCO 3 Ϫ was available from both enzymatic generation via carbonic anhydrase and uptake across the basolateral membrane. B, restriction of anion secretion to HCO 3 Ϫ taken up across the basolateral membrane (Cl Ϫ -free Ringer solution and presence of acetazolamide to inhibit carbonic anhydrase activity) resulted in reductions in both the I sc response to cAMP (*, p Ͻ 0.02) and the SITS-sensitive component of the I sc (*, p Ͻ 0.001) in NBC1 Ϫ/Ϫ proximal colon relative to wild-type. In NBC1 ϩ/ϩ colons, this manipulation did not cause a reduction in total cAMP-stimulated I sc , as occurred in NBC1 Ϫ/Ϫ colons (compare A and B), but did lead to a substantial increase in the SITS-sensitive component of the current. The failure of a similar switch to a SITS-sensitive current in NBC1 Ϫ/Ϫ proximal colon is consistent with the absence of the SITS-sensitive NBC1. defect in NBC1 Ϫ/Ϫ mice that results in incisors that are prone to enamel chipping suggests that NBC1 mutations contribute to this defect in humans. Whether this trait is due to the loss of NBC1 activity in tissues involved in tooth development or is secondary to systemic effects of the loss of NBC1 in kidney has not been determined. However, mice lacking the CFTR Cl Ϫ channel have a similar tooth defect that involves altered mineral content and abnormal formation of enamel (44,45). CFTR mRNA expression has been shown to occur in the apical tooth bud (45), and the extracellular pH of incisor enamel was more acidic in CFTR-deficient mice (46). A model has been proposed in which the CFTR and an electrogenic Na ϩ /HCO 3 Ϫ cotransporter regulate enamel pH via transepithelial HCO 3 Ϫ fluxes (46). The results of the current study demonstrate a role for NBC1 in transepithelial HCO 3 Ϫ transport in the intestine and support the hypothesis (46) that this isoform might serve a similar function during the development of tooth enamel.
During the collection of brains for histological analyses, we noticed an apparent deficiency in mineralization of the skull bones. Whether NBC1 plays a direct role in the ion transport processes involved in bone formation is unclear, however, because metabolic acidosis itself can inhibit bone formation (47) and induce bone resorption (48). Thus, regardless of whether NBC1 is directly involved in bone formation, the acidosis resulting from the renal defect would be expected to impair bone formation.
Absence of Pancreatic Histopathology-Despite evidence that NBC1 is required for pancreatic duct HCO 3 Ϫ secretion (2), pancreatic abnormalities have not been reported in humans with NBC1 mutations, and no overt histological evidence of pancreatic duct obstruction was observed in NBC1 Ϫ/Ϫ mice (data not shown). As noted above, however, most of the human patients that have been identified so far carry mutations that reduce rather than eliminate NBC1 activity, which may not be sufficient to cause a serious perturbation of duct function. In our mouse studies, histological analysis of the pancreas was performed only on young animals (14 days of age). Thus, it is possible that pathological changes resulting in overt pancreatic duct obstruction may be age-dependent. Also, it should be noted that CFTR deficiency, which causes severe pancreatic insufficiency in humans, causes only minimal histological changes in mouse pancreas (32). Therefore, the mouse may be less susceptible than humans to pancreatic duct pathology.
Splenomegaly and Increased Nucleated Red Blood Cells-NBC1 Ϫ/Ϫ mice exhibited severe splenomegaly, an expansion of the red pulp, and alterations in peripheral blood components, including an increase in circulating nucleated red cells, which are not normally seen at significant numbers in wild-type animals. This is unlikely to be a primary effect of NBC1 deficiency, because Northern blot analysis showed that NBC1 is not expressed at detectable levels in spleen (data not shown). An increase in the number of circulating nucleated red cells is known to occur in infants suffering from fetal acidemia (49,50), and an expansion of the red pulp occurs in spleens of human patients with chronically increased rates of erythrocyte breakdown (51), which would be expected to increase the numbers of immature red blood cells in the circulation. The anemia observed in NBC1 Ϫ/Ϫ mice is also consistent with a loss of mature erythrocytes. Thus, the increased size and histological alterations of the spleen and the changes in peripheral blood may be secondary to the severe acidosis in NBC1 Ϫ/Ϫ mice. As far as we are aware, splenomegaly, alterations in blood cells, and anemia have not been reported in human patients with NBC1 mutations, which could be due to reduced severity of the acidosis in humans.
Role of NBC1 in Anion Secretion in the Intestinal Tract-On the basis of studies using relatively nonspecific inhibitors, a role in anion secretion in the small intestine and colon has been proposed for NBC1 (18,19). Direct evidence of this role for NBC1 in native intestine is lacking, however, and intestinal defects in humans with NBC1 mutations have not been reported.
In the intestinal tract, NBC1 is expressed on basolateral membranes, where it is thought to play a major role in anion uptake to support transepithelial HCO 3 Ϫ secretion (18,19) and also to support transepithelial Cl Ϫ secretion by operating in a coupled system with the AE2 Cl Ϫ /HCO 3 Ϫ exchanger (52). To obtain a direct measurement of NBC1 activity in intestinal tract tissue, we measured the serosal Na ϩ -dependent changes in pH i during cAMP stimulation of cecal surface epithelium. When basolateral Na ϩ /H ϩ exchange was inhibited, the rates of cellular acidification and pH i recovery during removal or readdition of Na ϩ from the serosal side of the tissue were significantly reduced in NBC1 Ϫ/Ϫ cecum as compared with control tissue. These findings are consistent with major activity of NBC1 as a basolateral Na ϩ /HCO 3 Ϫ cotransporter in wild-type large intestine. The impactions in the NBC1 Ϫ/Ϫ small intestine, cecum, and colon initially suggested the occurrence of a severe secretory defect similar to that observed in mice lacking the CFTR Cl Ϫ channel (32). However, the data in Fig. 8 showed that maximum cAMP-stimulated anion secretory capacity of the isolated NBC1 Ϫ/Ϫ proximal colon, rather than being reduced, was significantly increased, whereas the secretory capacity of NBC1 Ϫ/Ϫ distal colon was essentially unchanged when compared with these colonic segments in the wild-type animal. Thus, these data indicate that the loss of NBC1 in the intestinal tract does not, by itself, cause a massive deficit in anion secretory capacity.
However, loss of NBC1 caused major alterations in the basolateral transport mechanisms that support transepithelial anion secretion. When analyzed in the presence of both Cl Ϫ and HCO 3 Ϫ , the bumetanide-sensitive component of the cAMPstimulated current was sharply increased (Fig. 8A), indicating that much of the increased anion secretory capacity observed in isolated tissues was due to increased activity of NKCC1. The mechanism underlying the increased NKCC1-supported secretion in the NBC1 Ϫ/Ϫ colon is unclear but does not involve increased NKCC1 protein levels. Western blot analysis using an antibody (31) that detects phosphorylation of two threonine residues (corresponding to amino acids 212 and 217 of human NKCC1) that serve as regulatory phosphorylation sites detected no differences between the levels of phosphorylated NKCC1 in wild-type and NBC1-deficient colons. However, it is known that NKCC1 activity can be rapidly altered by phosphorylation events, with the involvement of at least 5 threonine residues located in the N-terminal cytoplasmic domain (53,54), including the two mentioned above. The apparent compensation by NKCC1 observed in isolated tissues correlates well with studies suggesting that the combined activities of the AE2 Cl Ϫ / HCO 3 Ϫ exchanger and a Na ϩ /HCO 3 Ϫ cotransporter that was presumed to be NBC1 provide an alternative basolateral anion uptake mechanism in NKCC1-deficient intestine (52) and with studies showing that up-regulation of NKCC1-supported secretion occurs in the proximal colon of AE2 null mutant mice (55).
Our data suggest that the impactions in the lower intestinal tract of NBC1 Ϫ/Ϫ mice do not result from deficiencies in total anion secretory capacity but rather from an imbalance between absorption and secretion in vivo, with absorption predominant. As expected in response to increased serum aldosterone, which probably occurs as a secondary effect of hypovolemia, the amiloride-sensitive I sc (reflecting activity of the epithelial Na ϩ channel) was increased in both the proximal and distal colon of NBC1 Ϫ/Ϫ mice. Increased ENaC activity is associated with increased absorption of Na ϩ from the lumen of the colon but does not, by itself, lead to impactions (56). We were unable to analyze transport activity in the small intestine; however, systemic acidosis in rats has been shown to increase activity of the NHE3 Na ϩ /H ϩ exchanger (57), the dominant mechanism of Na ϩ absorption in the small intestine (28,40). Thus, the severe acidosis in NBC1 Ϫ/Ϫ mice may lead to an increase in absorption of Na ϩ , with accompanying fluid, in both small and large intestine, thereby contributing to dehydration of the intestinal contents.
Despite the increase in absorption, one might anticipate that the enhanced secretory capacity, as observed in the proximal colon, would allow a balance between absorption and secretion and maintenance of appropriate fluidity of the luminal contents. However, it should be noted that the in vitro measurements of anion secretion were obtained using solutions containing 110 mM Cl Ϫ and 25 mM HCO 3 Ϫ or 25 mM HCO 3 Ϫ alone. Under these in vitro conditions, increased activities of other basolateral membrane transporters and/or carbonic anhydrase appear to compensate for the loss of NBC1 activity to maintain anion secretion (and HCO 3 Ϫ secretion in the absence of Cl Ϫ ) at or above wild-type levels. However, the in vivo anion secretory activity is not known, and it is apparent that the increase in secretory capacity observed in vitro does not overcome the significant increases in absorption in vivo. Given the reduced availability of plasma HCO 3 Ϫ for uptake across the basolateral membrane (ϳ5 mM in NBC1 Ϫ/Ϫ mice as compared with ϳ22 mM in wild-type mice), the capacity for secretion of anions, and especially for HCO 3 Ϫ , may be severely limited in the NBC1 Ϫ/Ϫ intestine under in vivo conditions. Studies of the NBC1 Ϫ/Ϫ colon also provided insight into the complex interactions that exist between basolateral membrane HCO 3 Ϫ uptake proteins and carbonic anydrases in the support of transepithelial HCO 3 Ϫ secretion. In experiments in which HCO 3 Ϫ was the only CFTR-permeant anion, NBC1 Ϫ/Ϫ proximal colon exhibited normal I sc responses to cAMP, and there was no significant reduction in the SITS-sensitive I sc when compared with that of wild-type tissue (Fig. 10A). Thus, NBC1 activity was not apparent under these conditions. However, when HCO 3 Ϫ availability was limited by inhibition of carbonic anhydrase with acetazolamide (Fig. 10B), NBC1 Ϫ/Ϫ proximal colon exhibited a distinct reduction in cAMP-stimulated I sc . In wild-type proximal colon, the activity of a SITS-sensitive transporter apparently maintained the I sc response at levels equiva-lent to those observed in tissue that had not been treated with acetazolamide. Furthermore, the SITS-sensitive current, which accounted for ϳ65% of the I sc , was not observed in NBC1 Ϫ/Ϫ proximal colon during carbonic anhydrase inhibition, and the difference between the cAMP-stimulated I sc in NBC1 ϩ/ϩ and NBC1 Ϫ/Ϫ proximal colon was equivalent to the SITS-sensitive I sc in wild-type tissue. These findings are consistent with previous reports that NBC1 is sensitive to SITS (20) and demonstrate that a switch to NBC1-supported transepithelial HCO 3 Ϫ secretion can occur in wild-type proximal colon when intracellular generation of HCO 3 Ϫ by carbonic anhydrase activity is insufficient. In addition to demonstrating a role for NBC1 in anion secretion, the data obtained in the presence of HCO 3 Ϫ alone provided evidence of a non-NBC1, SITS-sensitive HCO 3 Ϫ transporter (the SITS-sensitive activity in NBC1 Ϫ/Ϫ tissue in Fig. 10A) and a SITS-insensitive HCO 3 Ϫ transporter (the residual activity in Fig. 10A). Two candidates for these activities are NBC4 and NBCn1, both of which are expressed in colon (19,21); however, there is insufficient information about their inhibitor profiles, cell type distribution, and membrane expression patterns in colon to resolve their contributions to the cAMP-stimulated I sc observed in Fig. 10. NBCn1 is coexpressed with NBC1 on basolateral membranes of villus cells of the duodenum (17), and NBCn1 mRNA levels in colon were 30% of the levels of NBC1 mRNA (19). NBCn1 is relatively insensitive to the stilbene disulfonate DIDS (3), so it may correspond to the SITS-insensitive transporter. The reduction in SITS-sensitive activity that occurred in NBC1 null proximal colon in response to acetazolamide (Ϫ32.3 Ϯ 9.5 A/cm 2 in Fig. 10A versus Ϫ8.0 Ϯ 2.0 A/cm 2 in Fig. 10B) suggests that the putative non-NBC1, SITS-sensitive HCO 3 Ϫ transporter requires carbonic anhydrase for maximum activity. This may reflect a direct interaction between carbonic anhydrase and the putative Na ϩ /HCO 3 Ϫ cotransporter, as several HCO 3 Ϫ transporters, including NBC1 and several Cl Ϫ /HCO 3 Ϫ exchangers, have been shown to bind carbonic anhydrase to form a bicarbonate transport metabolon (6,58,59).
Conclusion-Our results confirm that NBC1 is required for maintenance of systemic pH homeostasis in mice, consistent with its established role in HCO 3 Ϫ reabsorption in the kidney. More importantly, they provide direct evidence that NBC1 contributes to anion secretion in the gut. In the proximal colon, NBC1 appears to be an important component of a complex system for HCO 3 Ϫ uptake and generation to support electrogenic secretion of HCO 3 Ϫ across the apical membrane in response to cAMP. Other sources of HCO 3 Ϫ for electrogenic secretion include carbonic anhydrase-catalyzed hydration of CO 2 , an additional SITS-sensitive HCO 3 Ϫ transporter, and possibly a SITS-insensitive HCO 3 Ϫ transporter. In the absence of NBC1, other basolateral ion transporters and carbonic anhydrase appear to compensate to maintain Cl Ϫ and HCO 3 Ϫ secretory capacity in the NBC1 Ϫ/Ϫ proximal colon in vitro, with NKCC1 playing a particularly prominent role. Nevertheless, it appears that increased absorption resulting from deficits in systemic Na ϩ -fluid volume and acid-base homeostasis and possibly reduced secretion due to the very low plasma HCO 3 Ϫ resulting from severe metabolic acidosis leads to desiccation of the luminal contents and impactions in the terminal ileum, cecum, and colon in vivo.