Haploinsufficiency of the Ammonia Transporter Rhcg Predisposes to Chronic Acidosis

Background: Rhesus proteins transport NH3 and/or NH4+ in heterologous expression systems. Results: Heterozygous Rhcg mice develop delayed metabolic acidosis, whereas homozygous KO mice display severe metabolic acidosis. RhCG functions as an NH3 transporter on apical and basolateral membranes. Conclusion: RhCG is an NH3 but not NH4+ transporter. Significance: Loss or reduced expression of RhCG may underlie inherited or acquired forms of human acidosis. Ammonia secretion by the collecting duct (CD) is critical for acid-base homeostasis and, when defective, causes distal renal tubular acidosis (dRTA). The Rhesus protein RhCG mediates NH3 transport as evident from cell-free and cellular models as well as from Rhcg-null mice. Here, we investigated in a Rhcg mouse model the metabolic effects of Rhcg haploinsufficiency, the role of Rhcg in basolateral NH3 transport, and the mechanisms of adaptation to the lack of Rhcg. Both Rhcg+/+ and Rhcg+/− mice were able to handle an acute acid load, whereas Rhcg−/− mice developed severe metabolic acidosis with reduced ammonuria and high mortality. However, chronic acid loading revealed that Rhcg+/− mice did not fully recover, showing lower blood HCO3− concentration and more alkaline urine. Microperfusion studies demonstrated that transepithelial NH3 permeability was reduced by 80 and 40%, respectively, in CDs from Rhcg−/− and Rhcg+/− mice compared with controls. Basolateral membrane permeability to NH3 was reduced in CDs from Rhcg−/− mice consistent with basolateral Rhcg localization. Rhcg−/− responded to acid loading with normal expression of enzymes and transporters involved in proximal tubular ammoniagenesis but reduced abundance of the NKCC2 transporter responsible for medullary accumulation of ammonium. Consequently, tissue ammonium content was decreased. These data demonstrate a role for apical and basolateral Rhcg in transepithelial NH3 transport and uncover an incomplete dRTA phenotype in Rhcg+/− mice. Haploinsufficiency or reduced expression of RhCG may underlie human forms of (in)complete dRTA.

type in Rhcg ؉/؊ mice. Haploinsufficiency or reduced expression of RhCG may underlie human forms of (in)complete dRTA.

Ammonium (NH 4
ϩ ) is the main component of urinary acid excretion. Renal synthesis and excretion of NH 4 ϩ rise in response to an acid load, allowing kidneys to regenerate bicarbonate and increase net acid excretion (1,2). Impaired renal acid excretion characterizes type I distal renal tubular acidosis (dRTA) 3 with low urinary ammonium and inappropriately alkaline urinary pH (3).
Ammonium is formed in the proximal tubule from the metabolism of glutamine and added to the luminal fluid. It is reabsorbed into the medullary interstitium in the thick ascending limb creating a cortico-papillary NH 3 /NH 4 ϩ gradient (4,5). The final step of NH 3 /NH 4 ϩ excretion is achieved by the CD (6). The high tissue concentration of NH 3 /NH 4 ϩ and the pH gradient between interstitium and urine provide the driving forces for NH 4 ϩ excretion into urine. NH 4 ϩ secretion results from the trapping of NH 3 in the tubular lumen after being titrated by H ϩ ions stemming from active secretion by V-type H ϩ -ATPases (7).
The mechanisms mediating NH 3 /NH 4 ϩ transport across cell membranes into urine have only recently become uncovered. Members of the Rhesus protein family have been identified as pathways for NH 3 /NH 4 ϩ transport in yeast, plants, fish, and mammals (2,5,8). In the kidney CD, RhCG and RhBG are expressed (9,10). Mice lacking Rhbg show either no phenotype or only a very mild reduction in urinary ammonium excretion (11,12). In contrast, cell-specific or complete Rhcg deficiency causes a massive reduction in urinary ammonium excretion in three different mouse models (13)(14)(15). Microperfusion experiments using CDs from Rhcg-deficient mice demonstrated that RhCG is critical for the apical exit of NH 3 into urine (13). Thus far, a phenotype of dRTA has only been reported in Rhcg Ϫ/Ϫ mice, whereas potential changes caused by haploinsufficiency in Rhcg have not been investigated. This issue is relevant because heterozygous abnormalities in RhCG might be more frequent and may affect the renal capacity to cope with an acid load leading to incomplete dRTA. Furthermore, Rhcg ϩ/Ϫ mice may also serve as a model to examine the consequences of reduced RhCG expression that may occur during kidney disease. Finally, RhCG localization has been controversial for years. RhCG has been localized by some groups only at the apical side of CD cells, whereas others have found RhCG on both the apical and basolateral membranes (9,10,16,17). The functionality of basolateral RhCG protein remains unknown.
Here, we used a novel Rhcg mouse model to provide the first evidence that haploinsufficiency in Rhcg impairs the handling of a chronic acid load in Rhcg ϩ/Ϫ mice, which develop an incomplete metabolic acidosis. Microperfusion studies provide the functional basis of the defect and show that RhCG is absolutely required for apical and partially for basolateral NH 3 transport. Moreover, loss of Rhcg is associated with a profound down-regulation of NKCC2 and reduced medullary accumulation of ammonium impairing the gradient necessary for the final excretory step. These data provide new insights into the complex role of RhCG and suggest that congenital or acquired defects in RhCG protein expression may be associated with incomplete dRTA.

EXPERIMENTAL PROCEDURES
Animals-Rhcg ϩ/Ϫ mice were purchased from the Texas Institute of Genomic Medicine (Houston TX). Mice were generated by replacing exon 1 by a vector carrying a LacZ/Neo cassette (Fig. 1A). Mice were genotyped by PCR directly on a 3-l 25 mM NaOH tail digestion product. Genomic DNA was amplified using primer pairs specific for exon 1 forward (AGACCCCACAATGGAAAGCTATAA), wild type reverse (CAACCAGAACTCCCCAGTGTCAGA), and knock-out reverse (ATGGGCTGACCGCTTCCTCGTGCTTTAC). The products were separated by electrophoresis in 1% agarose gel (mutant product, 522 bp; wild type product, 376 bp). Mice were generated by mating Rhcg ϩ/Ϫ mice, and mice were bred in the Êntres Vivants Exempts d'Organismes Pathogènes Spécifiques Animal Facility. For acid-loading experiments, mice in metabolic cages were given 0.2 M HCl added to powdered standard food. All experiments were performed according to Swiss Animal Welfare laws and approved by the local veterinary authority (Veterinäramt Zürich).
In Vivo Experiments-All experiments were performed using age-and sex-matched Rhcg wild type (Rhcg ϩ/ϩ ), Rhcg knockout (Rhcg Ϫ/Ϫ ), and Rhcg heterozygote (Rhcg ϩ/Ϫ ) littermate mice (3-4 month-old) that were housed in metabolic cages (Techniplast, Switzerland). Mice were given deionized water ad libitum and were fed with a standard powdered laboratory chow (Kliba, Augst, Switzerland). Mice were allowed to adapt to metabolic cages for 3 days, and a first retro-orbital blood sample was taken for blood gas analysis under base line. Then two 24-h urine samples were collected under light mineral oil in the urine collector to determine daily urinary parameters. Mice were then allowed to recover for 2 weeks before giving an HClcontaining diet (0.2 M HCl added to powdered standard food) in normal cages. Food, water intake, and urine excretion were monitored following the same procedures as under base-line conditions. Urine collections were performed on the 1st and 2nd day of acid loading and then on the 6th and 7th day. Retroorbital blood samples were taken on the 2nd and 7th days of HCl diet.
Immunoblotting-Crude membrane proteins or cytosolic fractions were obtained from kidneys homogenized in 250 mM sucrose, 10 mM Tris-HCl, pH 7.5, and in the presence of protease inhibitors.
Immunostaining and Immunogold-Immunohistochemistry and immunogold staining were performed on frozen sections, and specificity was demonstrated on Rhcg Ϫ/Ϫ tissues. Mouse kidneys were fixed by perfusion retrograde through the aorta with 3% paraformaldehyde in 0.1 M sodium cacodylate buffer, pH 7.2. The tissue was either trimmed into small blocks, further fixed by immersion in 1% paraformaldehyde, infiltrated with 2.3 M sucrose for 30 min, and frozen in liquid nitrogen or prepared for routine paraffin embedding.
For electron microscopy, 70 -90-nm cryosections were obtained at Ϫ100°C with an FCS Reichert Ultracut S cryoultramicrotome as described previously (24), and 2-m paraffin sections were obtained with a Leica RM2165 microtome. For LM immunolabeling, the sections were incubated with a rabbit polyclonal antibody against RhCG (a kind gift from Dr. Yves Colin, INSERM, Paris, France (16,25)) at room temperature for 1 h after preincubation in PBS containing 0.05 M glycine and 1% bovine serum albumin. The sections were subsequently incubated with peroxidase-conjugated secondary antibody (Dako); the peroxidase was visualized with diaminobenzidine, and the sections were counter-stained with Maier's stain for 2 min. The sections were examined in a Leica DMR microscope equipped with a Leica DFC320 camera. Images were transferred by a Leica TFC Twain 6.1.0 program and processed using Adobe Photoshop 8.0. For EM immunolabeling, the sections were incubated with the primary antibody at 4°C overnight followed by incubation at room temperature for 1 h with 10-nm gold particles coupled to anti-rabbit IgG (BioCell, Cardiff, UK). The cryosections were embedded in methylcellulose containing 0.3% uranyl acetate and studied under a Philips CM100 electron microscope. For controls, sections were incubated with secondary antibodies alone or with nonspecific IgG.
RNA Extraction and Reverse Transcription-Snap-frozen kidneys (five kidneys for each condition) were homogenized in RLT-Buffer (Qiagen, Basel, Switzerland) supplemented with ␤-mercaptoethanol to a final concentration of 1%. Total RNA was extracted from 200-l aliquots of each homogenized sample using the RNeasy mini kit (Qiagen, Basel, Switzerland) according to the manufacturer's instructions. Quality and concentration of the isolated RNA preparations were analyzed on the ND-1000 spectrophotometer (Nano-Drop Technologies). Total RNA samples were stored at Ϫ80°C. Each RNA sample was diluted to 100 ng/l, and 3 l was used as a template for reverse transcription using the TaqMan reverse transcription kit (Applied Biosystems, Foster City, CA). For reverse transcription, 300 ng of RNA template were diluted in a 40-l reaction mix that contained (final concentrations) RT buffer (1ϫ), MgCl 2 (5.5 mM), random hexamers (2.5 M), RNase inhibitor (0.4 unites/l), the multiscribe reverse transcriptase enzyme (1.25 units/l), dNTP mix (500 M each), and RNase-free water.
Real Time Quantitative PCR-Quantitative real time quantitative RT-PCR was performed on the ABI PRISM 7700 sequence detection system (Applied Biosystems, Foster City, CA). Primers were chosen using the BLAST tool of Ensemble to result in amplicons no longer than 150 bp spanning intronexon boundaries to exclude genomic DNA contamination. The specificity of all primers was first tested on mRNA derived from kidney and always resulted in a single product of the expected size (data not shown). Probes were labeled with the reporter dye 6-carboxyfluorescein at the 5Ј end and the quencher dye 5-(and 6-)-carboxytetramethylrhodamine at the 3Ј end (Microsynth, Balgach, Switzerland). The primers were designed to target the deleted sequence in Rhcg Ϫ/Ϫ animals (accession number NM_019799) with 5Ј-ATGCAGGGATGGTTCCATTA-3Ј (238 -258 bp) as the left primer located within exon 1 and 5Ј-TGGAAGAAGGTCATAATGAGCAG-3Ј (373-392 bp) as the right primer located within exon 2, and 5Ј-TTAC-TATCGCTACCCGAGCTTCCAG-3Ј as the probe (292-317 bp). Real time PCRs were performed using TaqMan Universal PCR master mix (Applied Biosystems, Foster City, CA). Briefly, 3 l of cDNA, 0.8 l of each primer (25 M), 0.4 l of labeled probe (5 M), 5 l of RNase-free water, and 10 l of TaqMan Universal PCR master mix reached 20 l of final reaction volume. Reaction conditions were denaturation at 95°C for 10 min followed by 40 cycles of denaturation at 95°C for 15 s and annealing/elongation at 60°C for 60 s with auto ramp time. All reactions were run in triplicate. For analyzing the data, the threshold was set to 0.2 as this value had been determined to be in the linear range of the amplification curves for all mRNAs in all experimental runs. The expression of the genes of interest was calculated in relation to hypoxanthine-guanine phosphoribosyltransferase (accession number NM_013556, forward primer, 5Ј-TTATCAGACTGAAGAGCTACTGTAAG-ATC-3Ј (442-471), reverse primer, 5Ј-TTACCAGTGTCAA-TTATATCTTCAACAATC-3Ј (539 -568), and probe, 5Ј-TGAGAGATCATCTCCACCAATAACTTTTATGTCCC-3Ј (481-515)). Relative expression ratios were calculated as R ϭ 2 ( Ct(HPRT) Ϫ Ct(test gene) , where Ct represents the cycle number at the threshold 0.02; and HPRT is hypoxanthine-guanine phosphoribosyltransferase.
Measurement of Renal Ammonia Content-The renal tissue ammonia content was measured by an enzymatic technique (ammonia assay kit, Sigma) as described previously (26). Mice were anesthetized, and the kidneys were removed and immediately frozen in liquid nitrogen. They were then sliced frozen to yield a column of tissue, which extended from the cortex to the tip of the papilla. Sections were cut along the corticomedullary axis to yield three slices as follows: cortex, outer medulla, and inner medulla. Two kidneys from the same animal were pooled for each sample. Tissue slices were then homogenized in 300 l of ice-cold 7% trichloroacetic acid, and the solution was centrifuged. The supernatant was drawn off, and the pH of a 250-l sample was adjusted to near neutral by the addition of 12 l of 10 mM Na 2 HPO 4 in 9 N NaOH. A 200-l sample of buffered supernatant was then analyzed for ammonium. The pellet was resuspended in 1 N NaOH, shaken overnight, and analyzed for total protein using the Bio-Rad protein assay.
Microperfusion Studies on Isolated Tubules-Mice were anesthetized with 50 mg/kg pentobarbital sodium or xylazine/ ketamine intraperitoneally. Both kidneys were cooled in situ with control bath solution (see below) for 1 min and then removed and cut into thin coronal slices for tubule dissection. CCDs were dissected from the cortex at 10°C in the control solution. In vitro microperfusion of single CCD segments, intracellular pH, and transepithelial NH 3 permeability measurements were performed as described previously (13).
Intracellular pH Measurement-The isolated tubule was transferred to the bath chamber on the stage of an inverted microscope (Axiovert 200, Carl Zeiss, Germany) in the control solution containing (in mM) 138 NaCl, 1.5 CaCl 2 , 1.2 MgSO 4 , 2 K 2 HPO 4 , 10 HEPES, 5.5 glucose, 5 alanine, pH 7.40, and then was mounted on concentric pipettes and perfused in vitro with Na ϩ -free, ammonium-free solution, where N-methyl-D-glutamine ϩ (NMDG ϩ ) replaced Na ϩ . All solutions were equilibrated with 100% O 2 passed through a 3 N KOH CO 2 trap. Once the solutions were gassed and the pH checked, they were placed in a reservoir and were continuously bubbled with 100% O 2 . The average tubule length exposed to bath fluid was limited to 300 -350 m to prevent motion of the tubule. CCDs or OMCDs were loaded with 5 M of the fluorescent probe 2Ј,7Јbis(2-carboxyl)-5-(and-6)-carboxyfluorescein (BCECF Invitrogen) for ϳ20 min at 37°C in the control bath solution. The loading solution was then washed out by initiation of bath flow, and the tubule was equilibrated with dye-free control bath solution for 5 min. Bath solution was delivered at a rate of 20 ml/min and warmed to 37°C by a water jacket immediately upstream to the chamber. After this temperature equilibration in control solution, tubules were first transiently acidified by peritubular Na ϩ removal (sodium-free, ammonium-free solution) (10 min duration), replaced by NMDG ϩ to avoid exit of NH 4 ϩ by basolateral Na ϩ -coupled transport. This maneuver was done in the luminal absence of Na ϩ . During the fluorescence recording, perfusion solution was delivered to the perfusion pipette via a chamber under an inert gas (N 2 ) pressure (around 1 bar) connected through a manual six-way valve. With this system, opening of the valve instantaneously activates flow of solutions. The majority of the fluid delivery to the pipette exits the rear of the pipette system through a drain port at 4 ml/min. This method results in a smooth and complete exchange of the luminal or the peritubular solution in less than 3-4 s (27). After the fluorescence signal stabilization, luminal medium was instantly (at the rate of 4 ml/min in the draining) replaced by a Na ϩ -free solution containing 20 mM NH 4 Cl (and 118 mM NMDG-Cl) that elicited a rapid intracellular alkalinization, followed by a sharp acidification. The rate of intracellular alkalinization has been associated with the entry of NH 3 , whereas the subsequent phase of intracellular acidification in the continued presence of extracellular NH 4 Cl reflects mostly NH 4 ϩ entry (28). Fluorescence monitoring and calibration were performed with a video imaging system (Visitron Systems, Germany) as described previously (11,13). For peritubular ammonium pulse, peritubular solution was changed by a 6 mM NH 4 Cl solution, pH 7.40. in the presence of 1 mM furosemide and 2.5 mM ouabain in the bath.
Intrinsic Buffering Capacity Determination-The intrinsic buffering capacity (␤i) of CCD cells was determined, as reported previously (29), using a 40 mM NH 4 Cl basolateral pulse to acidify the cells. To exclude HCO 3 Ϫ /CO 2 as a buffering component and block Na ϩ -dependent pH i regulatory mechanisms, Na ϩ -free, HEPES-buffered solutions were used in the perfusate; the bath contained 1 mM amiloride (to inhibit Na ϩ /H ϩ exchangers), and bath and perfusate also contained 100 M Sch28080 (to block H ϩ /K ϩ -ATPases) and 200 nM concanamycin A (to block H ϩ -ATPases). Addition of 40 mM NH 4 Cl to the bath induced an increase following by a decrease in pH i . The pK a of ammonia (9.03) was used to calculate the intracellular NH 4 ϩ concentration when cell acidification plateaued. ␤i was calculated as the ratio of the change in intracel-lular NH 4 ϩ concentration to the change in pH i . Therefore, we determined the correlation between intracellular pH and ␤ i , which was Ϫ32.7-22.3⅐pH in wild types, Ϫ2.3-8.2⅐pH in heterozygotes, and 1,4 -5.0⅐ pH i in Rhcg Ϫ/Ϫ mice. We measured cellular buffering power (␤ i ) in CCDs from all three genotypes of mice.
NH 3 Permeability Measurement-The basic approach used to determine NH 3 permeability involved construction of a transepithelial gradient of NH 3 and measurement of the resulting NH 3 flux from the basolateral to the luminal side as described previously (11,13,30). The isolated CCD was transferred to the bath chamber on the stage of an inverted microscope (Axiovision A1, Zeiss, Germany) and mounted on concentric glass pipettes for microperfusion. Bath solution was delivered at a rate of 20 ml/min and warmed to 37°C by a water jacket immediately upstream of the chamber. The perfusion rate was adjusted by hydrostatic pressure to ϳ10 nl/min. The tubules were equilibrated for 20 -30 min at 37°C before the beginning of collections. To construct a transepithelial NH 3 gradient, the perfusion (lumen) solution contained (in mM) 139 NaCl, 1 NH 4 Cl, 2.5 K 2 HPO 4 , 2 NaHCO 3 , pH 6.4; the bath solution contained (in mM) 117 NaCl, 1 NH 4 Cl, 2.5 K 2 HPO 4 , 23 NaHCO 3 , pH 7.4, and in addition, both solutions contained (in mM) 5.5 glucose, 2 CaCl 2 , 1.2 MgSO 4 and 10 HEPES. The osmolarity of the solution was 295 Ϯ 5 mosmol/kg H 2 O. All solutions were equilibrated with 95% O 2 , 5% CO 2 . Once the solutions were gassed and the pH checked, they were placed in a reservoir and continuously bubbled with 95% O 2 , 5% CO 2 . The actual pH of the solutions was monitored several times during the experiments, and the pH of solutions was checked at the end of the experiment to ensure that changes did not occur. Carbonic anhydrase (Sigma) was added to the perfusate solution (1 mg/10 ml of solution). The purpose of carbonic anhydrase was to prevent any pH disequilibrium that might arise from proton secretion or NH 3 transport. Total ammonium concentration was measured in 10 -12-nl samples of peritubular, perfused, and collected fluids using an NH 3 diagnostic kit (Sigma) and the flow-through microfluorometer Nanoflow apparatus (World Precision Instruments, UK) (31).
Calculations of Transepithelial NH 3 Permeability-Assuming an absence of osmotic or hydrostatic pressure gradients across the epithelium and therefore an absence of net fluid transport, the passive transepithelial transport of total ammonium (Am) may be described by Equation 1, where P NH3 is diffusive permeability of NH 3 (cm/s), A S is tubule luminal surface area (cm 2 ), and ⌬C NH3 is the transepithelial concentration difference for NH 3 (mM). To calculate the permeability to NH 3 , Equation 1 is rearranged as Equation 2, The net rate of transport J Am is calculated as shown in Equation 3, The pK a equals 9.03 at physiological pH and temperature.
Heterozygous Mice Develop Metabolic Acidosis While on Long Term Acid Load-We first assessed acid-base status under basal conditions and during an acid load in Rhcg litter-mates. Throughout the study, food intake was similar in all three genotypes. At base line, no difference in acid-base and electrolyte levels was observed (Tables 1 and 2).
The effects of both acute (2 days) and chronic (7 days) HCl load were tested (Tables 1-3 and Fig. 2). On the 2nd day of the HCl load, blood pH and HCO 3 Ϫ concentration were decreased in all genotypes, as compared with base line (Table 1, and Fig. 2,  D and E). Blood pH and HCO 3 Ϫ concentrations were significantly lower in Rhcg Ϫ/Ϫ mice but similar in Rhcg ϩ/Ϫ and Rhcg ϩ/ϩ mice. Urinary ammonium excretion rate increased significantly on the 1st day of acid loading in Rhcg ϩ/ϩ and Rhcg ϩ/Ϫ mice but much less in Rhcg Ϫ/Ϫ mice (Tables 2 and 3 and Fig. 2A). Urinary pH decreased in Rhcg ϩ/ϩ and Rhcg ϩ/Ϫ mice as compared with base-line values ( Table 2 and Fig. 2B) but to a lesser extent in Rhcg Ϫ/Ϫ mice. Urinary titratable acid excretion was unaltered in all three genotypes during the acute acid load. Long term HCl loading resulted in the death of most Rhcg Ϫ/Ϫ mice, which poorly excreted ammonium and exhibited a very severe metabolic acidosis. These animals showed a higher loss of body weight after the acute HCl load presumably due to dehydration (Table 3).
In contrast to the Rhcg ϩ/ϩ mice, which adapted and nearly normalized their blood pH and HCO 3 Ϫ concentration, Rhcg ϩ/Ϫ remained acidotic at day 7 of the HCl load, even though both genotypes maintained a high NH 4 ϩ excretion. At the end of the chronic acid load, Rhcg ϩ/Ϫ mice showed less increase in their titratable acid excretion and had a more alkaline urine pH (Tables 1 and 2 and Fig. 2, B and C). Thus, both Rhcg Ϫ/Ϫ and Rhcg ϩ/Ϫ exhibited renal acid handling defects.
Absence of RhCG Abolishes Medullary Ammonium Accumulation-To assess the cortico-papillary gradient of NH 3 /NH 4 ϩ in kidneys from Rhcg mice, we measured ammonium content in the cortex and outer and inner medulla after 4 days of HCl treatment. There was no difference between Rhcg ϩ/ϩ and Rhcg ϩ/Ϫ mice. However, the inner medulla ammonium content was strongly reduced to 39% in Rhcg Ϫ/Ϫ mice (Fig. 3). Thus, the absence of Rhcg impairs the ability to concentrate ammonium in the interstitium of the inner medulla.

RhCG Is Located at the Apical and Basolateral Sides of Cells
Along the Distal Nephron-RhCG has been localized to most cells of the CD, including type A intercalated cells as well as principal cells (9,17). However, the subcellular localization of RhCG has remained controversial because some groups reported both apical and basolateral staining (10, 17), whereas   others detected only apical staining for RhCG (9) based on different immunohistochemical methods. We confirmed a strong labeling of both apical and basolateral poles of CD cells in mouse kidneys (Fig. 4 and Table 4). This staining was absent from Rhcg Ϫ/Ϫ kidneys demonstrating its specificity. In the kidney cortex, RhCG was localized in the distal convoluted tubule, connecting tubule, and CCD (Fig. 4A). In the distal convoluted tubule cells, RhCG was mainly present at the apical side (Fig.  4A). In the connecting tubule, CCD and OMCD, both intercalated and principal cells, were stained. Principal cells and some intercalated cells exhibited RhCG staining at both the apical and basolateral sides (Fig. 4, A-C). However, in some cells RhCG was strictly localized at the apical pole (see arrowheads in Fig. 4, A, B, and D). These cells have been previously identified as non-A/non-B type intercalated cells (33,34). Finally, in the inner medulla, only strong apical and faint basolateral staining were found in intercalated cells ( Fig. 4 and Table 4). Immunogold electron microscopy demonstrated RhCG associated with the apical membrane as well as the basolateral interdigitations of segment specific and intercalated cells (Fig. 4, E and F).
No specific immunogold labeling was found in kidneys from Rhcg Ϫ/Ϫ mice (data not shown). Thus, RhCG protein is expressed on both apical and basolateral membranes in mouse segment-specific principal and intercalated CD cells. Total and Apical Membrane Permeabilities for NH 3 Are Reduced in CDs from Rhcg ϩ/Ϫ Mice-We assessed total transepithelial permeability for NH 3 in in vitro microperfused cortical CDs from Rhcg mice after a 2-day HCl diet, a condition causing a strong difference in urinary ammonium excretion between Rhcg ϩ/ϩ and Rhcg Ϫ/Ϫ mice. Imposing a bath-to-lumen NH 3 gradient in the nominal absence of an NH 4 ϩ gradient generated a measurable NH 3 secretory flux, which was significantly lower in CCDs from Rhcg ϩ/Ϫ and Rhcg Ϫ/Ϫ mice versus Rhcg ϩ/ϩ mice. These differences were due to a decrease in transepithelial permeability to NH 3 by 54 and 83%, respectively (Table 5 and Fig. 5). Thus, one Rhcg allele was not sufficient to sustain normal transepithelial permeability to NH 3 in the mouse collecting duct.
Next, we tested whether the apical permeability to NH 3 was affected in CDs from Rhcg ϩ/Ϫ mice. Therefore, we measured the effects of an inwardly directed gradient on pH i on CCDs isolated from Rhcg ϩ/ϩ and Rhcg ϩ/Ϫ mice (13). Fig. 6A depicts the typical time course of pH i changes when NH 3 /NH 4 ϩ was added to the lumen tubule. The initial rate of cellular alkalinization is proportional to the rate of apical NH 3 entry as described previously (13,35,36). To directly compare transport rates, we measured intracellular buffering power and calculated the amount of H ϩ used to titrate NH 3 transported across the membrane. Fig. 6B depicts the calculated rate of NH 3 transported into CDs from Rhcg ϩ/ϩ and Rhcg ϩ/Ϫ mice, which was drastically reduced in Rhcg ϩ/Ϫ tissue.
Basolateral RhCG Accounts for Peritubular Membrane Permeability to NH 3 -Because RhCG is expressed at both sides of at least a subset of CD cells (Fig. 4 and Table 4), we next tested the effect of Rhcg disruption on NH 3 transport across the basolateral membranes of CCD cells. Several transport pathways for NH 4 ϩ have been proposed at the basolateral side of CD cells, including the Na ϩ -K ϩ -2Cl Ϫ cotransporter NKCC1 and the Na/K-ATPase where NH 4 ϩ would substitute for K ϩ (37, 38) . Therefore, we performed experiments in the nominal absence of sodium to block the activity of both transport pathways.
Peritubular NH 3 /NH 4 ϩ prepulses were performed on cortical CDs from Rhcg ϩ/ϩ , Rhcg ϩ/Ϫ , and Rhcg Ϫ/Ϫ mice submitted to HCl loading for 2 days. As summarized in Fig. 7, when NH 3 /NH 4 ϩ (6 mM) was applied to the basolateral side, the calculated rate of NH 3 transport into CD cells was unchanged in Rhcg ϩ/Ϫ tissue but strongly reduced in Rhcg Ϫ/Ϫ tissue. Thus, RhCG sustains both apical and basolateral transport of NH3 in CD cells, in agreement Kidneys from all three genotypes were dissected after 4 days of HCl load. Cortex (C), outer medulla (OM), and inner medulla (IM) were separated, and tissue ammonium content was measured. In wild type and heterozygous mice, a gradient in renal tissue ammonia content from cortex to inner medulla was observed. In Rhcg Ϫ/Ϫ mice, the ammonia content increased from cortex to outer medulla, but ammonium content was lower in the inner medulla than in mice from the other two genotypes. (n ϭ 5-8 mice/genotype), * statistically different from Rhcg ϩ/ϩ mice (p Ͻ 0.05). NS, not significant. Incomplete dRTA in Rhcg-targeted Mice FEBRUARY 22, 2013 • VOLUME 288 • NUMBER 8 with structural data and reconstituted RhCG suggesting that the protein forms a NH 3 -permeable channel (39,40).
Compensatory Adaptations to the Loss of Rhcg-We finally examined whether Rhcg deletion affects mechanisms involved in renal ammoniagenesis and ammonium excretion. Surprisingly, immunoblots of whole kidney extracts from mice subjected to a 2-day HCl load revealed no differences in the expression of PEPCK expression (Fig. 8A). In contrast, the expression of the PDG, mediating the initial step of ammoniagenesis was decreased in Rhcg ϩ/Ϫ kidneys as compared with both Rhcg ϩ/ϩ and Rhcg Ϫ/Ϫ kidneys (Fig. 8B). The abundance of the Na ϩ /H ϩ exchanger NHE3, involved in NH 4 ϩ secretion into the proximal tubule lumen, was unchanged (Fig. 8C). Moreover, the expression of the Na ϩ -K ϩ -2Cl Ϫ cotransporter NKCC2, the main apical NH 4 ϩ transporter in the thick ascending limb, was highly down-regulated in kidneys from Rhcg ϩ/Ϫ and Rhcg Ϫ/Ϫ mice (Fig. 8D).

DISCUSSION
Renal ammonium excretion is critical for acid-base homeostasis and is achieved by a complex process involving various nephron segments (2,5,6). A critical role for the Rhesus protein RhCG was demonstrated in tissue-specific and complete knock-out mouse models (13)(14)(15). However, metabolic effects of Rhcg haploinsufficiency, the role of RhCG in basolateral NH 3 transport in CD cells, and the response of other pathways critical for renal ammoniagenesis and ammonium transport have remained unknown. Based on a novel Rhcg mouse model, we show that Rhcg haploinsufficiency causes incomplete metabolic acidosis in mice. Next, RhCG contributes to peritubular NH 3 uptake giving functional evidence to the basolateral localization of RhCG in CD cells.
Rhcg Ϫ/Ϫ Mice Develop a Severe Incomplete Renal Tubular Acidosis-Rhcg Ϫ/Ϫ develop incomplete dRTA as evident from the inability to respond to an acute oral acid load. They showed severe hyperchloremic metabolic acidosis and did not increase urinary ammonium excretion resulting in low net acid excretion. Moreover, these animals had a drastic reduction in their blood HCO 3 Ϫ concentration and pH. Rhcg Ϫ/Ϫ mice developed severe dehydration as indicated by high blood hemoglobin and weight loss (Tables 2 and 4). These phenotypes are more pronounced than in mice with partial deletion of Rhcg (14,15).
Rhcg Haploinsufficiency Leads to Metabolic Acidosis-Similarly, Rhcg ϩ/Ϫ mice develop also an incomplete dRTA. However, these mice initially responded to the acid load by increasing urinary ammonium excretion to the same extent as   Rhcg ϩ/ϩ mice. Later, they did not fully adapt to chronic acid loading and remained acidotic after 7 days of HCl loading, being unable to correct blood HCO 3 Ϫ concentration and excreting more alkaline urine, whereas total net acid excretion was comparable with Rhcg ϩ/ϩ animals. The incomplete chronic metabolic acidosis in Rhcg ϩ/Ϫ mice is at least in part due to the inability to maximally acidify urine and to excrete adequate amounts of acid in the form of titratable acidity. The metabolic phenotype of Rhcg ϩ/Ϫ mice was further supported by functional studies on microperfused CCDs. Transepithelial permeability to NH 3 was reduced in CCDs from Rhcg Ϫ/Ϫ and Rhcg ϩ/Ϫ mice. The reduction in NH 3 permeability in CCDs from Rhcg ϩ/Ϫ mice was less than in Rhcg Ϫ/Ϫ mice, suggesting that a 50% reduction in net NH 3 permeability in CCDs is still sufficient to support high ammonium excretion but is not enough to correct metabolic acidosis. Of note, medullary accumulation of NH 3 /NH 4 ϩ appeared to be intact. Together, these results repre-sent the first evidence that loss of a single Rhcg allele can lead to chronic hyperchloremic metabolic acidosis. Renal Adaptation to Rhcg Invalidation-We further examined compensatory mechanisms in Rhcg ϩ/Ϫ and Rhcg Ϫ/Ϫ mice. The generation of the cortico-papillary gradient of ammonium is required for the subsequent uptake of NH 3 and NH 4 ϩ by intercalated cells and the secretion of NH 3 into urine. Surprisingly, Rhcg Ϫ/Ϫ mice had low tissue ammonium content in the inner medulla after 4 days of acid loading. This result suggests that the absence of RhCG affects the ability of the medulla to generate or maintain a high interstitium ammonium content. Because the key enzymes of proximal tubular ammoniagenesis, PEPCK and PDG, were normally expressed in Rhcg Ϫ/Ϫ kidney tissue, mice most likely form adequate amounts of ammonium. In Rhcg Ϫ/Ϫ mice, this ammonium might be either shunted back into systemic circulation or not be absorbed at the level of the thick ascending limb.
Expression of NKCC2 was decreased in both Rhcg ϩ/Ϫ and Rhcg Ϫ/Ϫ kidney tissues. NKCC2 is crucial for thick ascending limb NH 4 ϩ absorption and concentration in the medulla. Thus, our results suggest that mechanisms contributing to create a high medullary ammonium concentration are altered in Rhcgdeficient mice. Other mechanisms such as increased removal of ammonium from the interstitium with venous blood might contribute also to the low medullary ammonium content. Metabolic acidosis alters concentrations of various vasoactive substances in the kidney, including higher levels of endothelin and prostaglandins and lower concentrations of NO (41,42), which may alter medullary blood flow and the ability of the kidneys to maintain the cortico-papillary ammonium gradient.
RhCG Mediates NH 3 Transport Also at the Basolateral Side of CD Cells-Subcellular RhCG localization has remained controversial for many years (9,10,17). Here, we confirmed a strong labeling of both apical and basolateral poles of CD cells in mouse kidneys. These data were corroborated by our functional study on in vitro microperfused CCDs. We measured a 60% reduction in the NH 3 permeability at the basolateral side of FIGURE 6. RhCG is required for apical NH 3 permeability of cortical collecting duct cells. Cortical collecting ducts were isolated from kidneys of Rhcg ϩ/ϩ , Rhcg ϩ/Ϫ , and Rhcg Ϫ/Ϫ mice after 2 days of HCl loading, and intracellular pH was monitored with BCECF. 20 mM NH 4 Cl was applied with the luminal perfusate. A, pH i recording from CCD exposed to a luminal NH 4 Cl pulse in Rhcg ϩ/ϩ and Rhcg ϩ/Ϫ CCD. Exposure to NH 4 Cl caused a rapid alkalinization corresponding to NH 3 entry into cells. The initial slopes (⌬pH i /⌬t) of the alkalinization phase were measured, and the amount of NH 3 titrated into NH 4 ϩ was finally calculated based on intracellular buffering power. B, bar graph summarizing the amount of titrated NH 3 during luminal NH 4 Cl pulses (n ϭ 8 -12 tubules/genotype). *, p Ͻ 0.05; NS, not significant. Cortical collecting ducts were isolated from kidneys of Rhcg ϩ/ϩ , Rhcg ϩ/Ϫ , and Rhcg Ϫ/Ϫ mice after 2 days of HCl loading, and intracellular pH was monitored with BCECF. After an equilibrium phase, 6 mM NH 4 Cl was applied to the bath. Exposure to NH 4 Cl caused a rapid alkalinization corresponding to NH 3 entry into cells. The initial slopes (⌬pH i /⌬t) of the alkalinization phase were measured, and the amount of NH 3 titrated into NH 4 ϩ was finally calculated based on intracellular buffering power. The bar graph summarizes the amount of titrated NH 3 (alkalinization phase) during basolateral NH 4 Cl pulses (n ϭ 5-7 tubules/genotype). *, p Ͻ 0.05; NS, not significant. FEBRUARY 22, 2013 • VOLUME 288 • NUMBER 8

JOURNAL OF BIOLOGICAL CHEMISTRY 5527
Rhcg Ϫ/Ϫ CCD cells. This is the first evidence that Rhcg is also functional at the basolateral side of CDs cells and participates in basolateral NH 3 uptake. Previously, it has been assumed that basolateral uptake occurs mostly if not exclusively in the form of NH 4 ϩ and that NH 3 plays no major role. Several pathways for NH 4 ϩ uptake have been delineated from pharmacological and functional experiments that demonstrated a role for the Na ϩ -K ϩ -2Cl Ϫ cotransporter NKCC1, the Na ϩ /K ϩ -ATPase, and possibly potassium channels where NH 4 ϩ always would replace potassium (37,38,43). We performed our experiments in the absence of sodium to reduce NKCC1 and Na ϩ /K ϩ -ATPase activity, respectively. Thus, our experiments do not allow conclusions regarding the relative importance and contribution of these different pathways to total NH 3 /NH 4 ϩ uptake across the basolateral membrane, but they modify the current model of ammonium transport across the basolateral membrane. However it appears from the data on Rhcg ϩ/Ϫ CCD that apical NH 3 flux critically depends on full RhCG expression, whereas basolateral NH 3 flux is sustained with reduced RhCG expression suggesting that transport of the NH 3 form is most important at the apical side. Finally, our findings could also explain the complete lack or only mild phenotype observed in Rhbg knock-out mice (11,44), where RhCG may have compensated for the lack of RhBG.
In summary, reduced expression or complete loss of expression of RhCG affects the ability of the kidneys to excrete acidic urine and appropriately regulate acid/base homeostasis. We provide the first functional evidence for basolateral RhCG activity and show that RhCG is expressed and functional on apical and basolateral membranes of most cells lining the renal collecting duct. Thus, Rhcg haploinsufficiency or Rhcg deletion may contribute to orphan forms of inherited distal renal tubular acidosis in humans as well as to acquired forms of dRTA.