The proximal tubule through an NBCe1-dependent mechanism regulates collecting duct phenotypic and remodeling responses to acidosis

Abstract

The renal response to acid-base disturbances involves phenotypic and remodeling changes in the collecting duct. This study examines whether the proximal tubule controls these responses. We examined mice with genetic deletion of proteins present only in the proximal tubule, either the A variant or both A and B variants of isoform 1 of the electrogenic Na⁺-bicarbonate cotransporter (NBCe1). Both knockout (KO) mice have spontaneous metabolic acidosis. We then determined the collecting duct phenotypic responses to this acidosis and the remodeling responses to exogenous acid loading. Despite the spontaneous acidosis in NBCe1-A KO mice, type A intercalated cells in the inner stripe of the outer medullary collecting duct (OMCD_is) exhibited decreased height and reduced expression of H⁺-ATPase, anion exchanger 1, Rhesus B glycoprotein, and Rhesus C glycoprotein. Combined kidney-specific NBCe1-A/B deletion induced similar changes. Ultrastructural imaging showed decreased apical plasma membrane and increased vesicular H⁺-ATPase in OMCD_is type A intercalated cell in NBCe1-A KO mice. Next, we examined the collecting duct remodeling response to acidosis. In wild-type mice, acid loading increased the proportion of type A intercalated cells in the connecting tubule (CNT) and OMCD_is, and it decreased the proportion of non-A, non-B intercalated cells in the connecting tubule, and type B intercalated cells in the cortical collecting duct (CCD). These changes were absent in NBCe1-A KO mice. We conclude that the collecting duct phenotypic and remodeling responses depend on proximal tubule-dependent signaling mechanisms blocked by constitutive deletion of proximal tubule NBCe1 proteins.

NEW & NOTEWORTHY This study shows that the proximal tubule regulates collecting duct phenotypic and remodeling responses to acidosis.

INTRODUCTION

The collecting duct has a critical role in acid-base homeostasis. There is a specialized population of collecting duct cells, intercalated cells, that have the primary role of collecting duct acid-base transport. Type A intercalated cells secrete H⁺, which promotes filtered bicarbonate reabsorption and titratable acid excretion, and type B intercalated cells secrete bicarbonate, which is essential for recovery from metabolic alkalosis. Type A intercalated cells also secrete ammonia¹, the predominant component of the renal net acid excretion response to acid-base disturbances (1–3). Importantly, H⁺ secretion, ${\text{HCO}}_3^{-}$ secretion, and ammonia secretion are under dynamic control, enabling tight control of acid-base homeostasis. Dysregulation of collecting duct acid-base transport, as occurs with type I (distal) renal tubular acidosis (RTA), leads to severe acid-base disturbances (4, 5).

Changes in collecting duct acid-base transport result from a combination of phenotypic and remodeling responses. The phenotypic responses of type A intercalated cells to metabolic acidosis include hypertrophy (6), increased apical plasma membrane H⁺-ATPase expression (7–9), increased anion exchanger 1 (AE1) expression (10, 11), and increased expression of the ammonia transporters, Rhesus B glycoprotein (Rhbg) and Rhesus C glycoprotein (Rhcg) (12–14). Type B intercalated cells respond with decreased apical pendrin expression (15–17). Acidosis also induces a remodeling response that involves changes in the proportion of the different intercalated cell types, such that the proportion of acid-secreting type A intercalated cells increases, whereas the proportion of bicarbonate-secreting type B intercalated cells decreases (18–20). Together, these phenotypic and remodeling responses enable adaptive changes in collecting duct acid-base transport that contribute to systemic acid-base homeostasis. However, the mechanisms through which acidosis alters these responses are incompletely understood.

The present study tested whether the collecting duct response to acidosis is innate to the collecting duct or requires involvement of the proximal tubule. Na⁺-bicarbonate cotransporter, electrogenic, isoform 1 (NBCe1, SLC4A4) protein is exclusively in the proximal tubule in the kidney (21, 22), where it plays a crucial role in filtered bicarbonate reabsorption (23, 24). Two splice variants of NBCe1, NBCe1-A and NBCe1-B, are present in the proximal tubule, and NBCe1-A is the predominant variant (24–27). Deletion of just NBCe1-A or both NBCe1-A/B induces severe metabolic acidosis (28, 29) and inhibits proximal tubule ammonia and citrate responses to this acidosis (28, 30). Thus, studying mice with NBCe1-A deletion allows one to determine whether proximal tubule-dependent signaling mechanisms are necessary for the collecting duct response to acidosis. To consider the possibility that the genome-editing procedure used to generate NBCe1-A knockout (KO) mice had off-target effects, we examined a second genetic model, combined NBCe1-A/B deletion, which we recently reported (25). Our results show that proximal tubule NBCe1 deletion inhibits intercalated cell phenotypic and remodeling responses to metabolic acidosis.

METHODS

Animals

Mice with NBCe1-A deletion have been previously described (28). We used offspring from breeding heterozygous males and females and homozygous NBCe1-A deletion offspring with age-matched wild-type (WT) offspring. All mice were genotyped using tail-clip DNA samples. Some, but not all, samples came from mice used in previous studies that did not examine specifically the collecting duct (28–30).

As a second model of NBCe1 deletion, we used tissue from mice with kidney-specific deletion of both renal expressed NBCe1 splice variants, NBCe1-A and NBCe1-B. We used mice with floxed NBCe1 alleles (31) and induced kidney-specific deletion using Cre recombinase controlled by a constitutively active Pax8 promoter (32). All breeding involved homozygous floxed, Cre-negative mice with homozygous floxed, and Cre-positive mice. Cre-negative offspring were used as control mice. All offspring were genotyped using tail-clip DNA samples. Some of the tissue used came from mice in a previous study that did not examine the collecting duct (33).

The Gainesville Veterans Affairs Medical Center and the University of Florida College of Medicine Institutional Animal Care and Use Committees approved all experiments. We used both male and female mice in this study.

Metabolic Cage Experiments

Mice were placed in metabolic cages for 2–3 days and fed a powdered normal diet (18% protein, Harlan Teklad) mixed 1:1 with water. Daily food consumption was measured, and daily 24-h urine samples were collected. Urine samples were collected in tubes containing water-equilibrated mineral oil. After collection, daily urine volume and pH were recorded, and urine samples were then stored at −20°C until analyzed further.

Acid Loading

Animals were acid-loaded using standard techniques (28, 34). Briefly, HCl (0.4 M) was added to powdered rodent chow for 7 days at a 1 mL/g chow ratio.

Antibodies

Table 1 shows the antibodies used, which have been previously described.

Table 1.

Antibodies used

Target	Supplier	Location	Cat. No.
AE1	Alpha Diagnostics	San Antonio, TX	AE11-A
H+-ATPase a4 subunit (immunohistochemistry)	Fiona Karet, PhD	Cambridge Institute for Medical Research, Cambridge, UK	NA
H+-ATPase (immunogold electron microscopy)	Dennis Stone, MD	University of Texas Southwestern, Dallas, TX	U-172
Pendrin	Susan Wall, MD	Emory University, Atlanta, GA	NA
Rhbg	Weiner Laboratory	University of Florida, Gainesville, FL	NA
Rhcg	Weiner Laboratory	University of Florida, Gainesville, FL	NA

AE1, anion exchanger 1; NA, not applicable; Rhbg, Rhesus B glycoprotein; Rhcg, Rhesus C glycoprotein.

Blood Analysis

Blood was obtained by cannulation of the abdominal aorta in mice anesthetized with isoflurane, drawn into a heparinized syringe, and immediately analyzed for Na⁺, K⁺, and bicarbonate using a Siemens Microanalytic Blood Gas Analyzer (RAPIDLab 348 Analyzer, Siemens, Munich, Germany). As previously discussed (29), while arterial sample pH and Pco₂ were measured and recorded, we do not formally report these values. Because the anesthesia and surgical procedure necessary for aortic blood collection are likely to alter respiration, which will rapidly alter both pH and Pco₂, and respiration was not explicitly assessed, we cannot conclude that their measurement accurately reflects in vivo conditions.

Urinary Analysis

Urine ammonia was measured using a modification of a commercially available kit (A7553, Pointe Scientific, Canton, MI) (35, 36). Urine pH was measured using a micro-pH electrode (ROSS semi-micro pH, Orion 8103BN, Thermo Scientific, Waltham, MA).

Protein Preparation

Animals were anesthetized with inhalant isoflurane, and kidneys were rinsed by in vivo cardiac perfusion with PBS (pH 7.4) containing Na-heparin (6,000 U/L) and lidocaine (120 mg/L). The right kidney was removed rapidly, and the inner stripe of the outer medulla (ISOM) was isolated on a cold stage under a dissecting microscope. Samples were snap-frozen in liquid nitrogen and stored frozen at −80°C until used. Tissues were homogenized in T-PER tissue protein extraction reagent (Pierce Biotechnology, Rockford, IL) using microtube pestles (USA Scientific, Ocala, FL), and protein was extracted according to the manufacturer’s recommendations. An aliquot was used for total protein quantification using a bicinchoninic acid (BCA) assay, and the remainder was stored at −80°C until use.

Immunoblot Procedures

Immunoblot analysis was performed as previously detailed (12, 37, 38). Results were visualized using enhanced chemiluminescence (SuperSignal West Pico Substrate, Pierce) and a Kodak Image Station 440CF digital imaging system. In selected experiments, blots were stripped. Band density was normalized such that the mean density in WT tissue was 100. The absence of saturation was confirmed by examining pixel intensity distribution in all immunoblots.

Tissue Preparation for Immunohistochemistry

Tissues were obtained for immunohistochemistry using standard techniques (12, 37, 38). Briefly, under isoflurane anesthesia, kidneys were preserved by in vivo cardiac perfusion with PBS (pH 7.4) containing Na-heparin (6,000 U/L) and lidocaine (120 mg/L) followed by periodate-lysine-2% paraformaldehyde (PLP), cut transversely into 2- to 3-mm-thick slices, and immersed for 24–30 h at 4°C in the same fixative. Samples were embedded in polyester wax made with polyethylene glycol 400 distearate (Polysciences, Warrington, PA) with 10% 1-hexadecanol, and 2-µm-thick sections were cut and mounted on gelatin-coated slides.

Tissue Preparation for Immunogold Electron Microscopy

Tissue was preserved for immunogold electron microscopy using a similar approach to immunohistochemistry with minor changes. Tissues were initially rinsed using Tyrode buffer made with 0.9% NaCl and 4% hydroxyethyl starch with an osmolality of 331 mosmol/kgH₂O and then fixed with the same solution with the addition of 4% paraformaldehyde (osmolarity ∼1,400 mosmol/kgH₂O). Perfusion was started at 145 mmHg and then gradually increased to 170 mmHg. After the kidneys were blanched, they were perfused for an additional 8–9 min.

Immunohistochemistry

Immunohistochemistry used standard immunoperoxidase procedures previously described (35, 39–43). Briefly, sections were dewaxed in ethanol, rehydrated, heated in Trilogy (Cell Marque, Rocklin, CA) at 88°C for 30 min and then at 96°C for 30 min, cooled for 30 min, and rinsed in PBS. Endogenous peroxidase activity was blocked by incubating sections in 3% H₂O₂ for 45 min. Sections were blocked for 15 min with serum-free protein block (Dako Cytomation) and then incubated overnight at 4°C with primary antibody. Tissue sections were washed in PBS and incubated for 30 min with polymer-linked, peroxidase-conjugated horse anti-rabbit IgG (ImmPRESS, Vector Laboratories, Burlingame, CA), washed again with PBS, and then exposed to diaminobenzidine for 5 min. Sections were washed with distilled water, dehydrated with graded ethanols, mounted, and observed by light microscopy. All comparisons of labeling were made using tissue sections from the same immunohistochemistry experiment. Each experiment included a tissue section exposed to immunolabeling procedures without primary antibody to ensure that the label was due to primary antibody.

Double-Immunolabel Procedures

We performed double-immunolabel procedures as previously described (41, 44–46). Briefly, tissue sections were labeled with the first primary antibody using Vector SG (Vector Laboratories, Burlingame, CA) as the chromogen to produce a blue label, as described earlier. After the Vector SG reaction, sections were washed in PBS and blocked using 3% H₂O₂ in methanol. The aforementioned procedure was repeated using the second primary antibody and the substitution of diaminobenzidine for Vector SG. This brown label was easily distinguishable from the blue label produced by Vector SG. Sections were washed with glass distilled water, dehydrated with xylene, mounted with Permount, and observed by light microscopy.

Quantitative Immunohistochemistry

Quantitative immunohistochemistry was performed as we have previously described and validated (35, 36, 42–44). Briefly, we used high-resolution digital micrographs of randomly selected fields of the renal cortex and ISOM without imaging enhancement techniques. For experiments quantifying cell height and both total and apical polarization of protein expression, we measured pixel intensity across a straight line drawn from the tubule lumen through an individual cell using ImageJ software (National Institutes of Health, Bethesda, MD). Mean background intensity, measured outside the cell, was subtracted from absolute pixel intensity to determine net pixel intensity. Total expression was the integrated net pixel intensity in the cell. Cell height was the distance in pixels between the apical and basolateral edges of the cells and converted to absolute height using calibrated determination of individual pixel size. Apical polarization was calculated as the proportion of total expression in the apical 25% of the cell. We carefully outlined intercalated cells using ImageJ software in experiments that only quantified single-cell expression. Background intensity was determined and subtracted from individual pixel intensity, and net pixel intensity was then integrated within the region to determine single-cell expression. In both approaches, a minimum of 15 individual cells from at least four photomicrographs from each kidney were analyzed.

Remodeling Response

We determined the proportion of the different intercalated cell types in the connecting tubule (CNT), cortical collecting duct (CCD), and inner stripe of the outer medullary collecting duct (OMCD_is) using standard double-immunolabel approaches. In the CNT, intercalated cell type was determined using antibodies to pendrin and AE1. AE1-positive cells were identified as type A intercalated cells, and pendrin-positive cells were identified as non-A, non-B intercalated cells. In the CCD and OMCD_is, we used double immunolabel with antibodies to H⁺-ATPase and AE1. Apical H⁺-ATPase and basolateral AE1 immunolabel identified type A intercalated cells. Basolateral H⁺-ATPase and the absence of AE1 immunolabel identified type B intercalated cells, which were present only in the CCD. We obtained high-resolution digital micrographs of randomly selected fields of the renal cortex and OMCD_is. A minimum of 15 intercalated cells in at least four different photomicrographs were examined in each kidney. The “auto contrast” feature in Adobe Photoshop was used to simplify the detection of low-intensity immunolabel. In preliminary studies, this enhanced visualization of low levels of immunolabel intensity but did not result in immunolabel not visible in the unenhanced micrographs being visible in the contrast-enhanced image. Sections were imaged using DIC optics, and we only counted cells in which the nucleus was visible. The proportion of different cell types was determined in each region in a kidney and used as a single data point for statistical analysis.

Immunogold Labeling

The immunogold labeling procedure was performed as we have previously described (40, 47). We exposed ultrathin tissue sections to primary antibody and then to goat anti-rabbit IgG secondary antibody conjugated to 10-nm colloidal gold particles (British BioCell, Ted Pella, Redding, CA). Unless noted otherwise, all steps were done by floating the grids on solution droplets at room temperature. Sections were exposed to 0.1 M NH₄Cl for 1 h, rinsed with rinse buffer (0.01 M Tris·HCl and 0.15 M NaCl, in distilled water, pH 7.2), treated with BSA buffer (1% BSA, 0.01 M Tris·HCl, 0.5 M NaCl, and 4 mM NaN₃, in distilled water, pH 7.2) for 30 min, and then incubated in a humidified chamber overnight at 4°C with H⁺-ATPase primary antibody diluted in filtered BSA buffer. Sections were washed with rinse buffer, treated with Carbowax buffer (0.02% Carbowax PEG 20 M, 0.01 M Tris·HCl, 0.15 M NaCl, and 4 mM NaN₃, in distilled water, pH 7.2) for 30 min, and then exposed to secondary antibody diluted in Carbowax buffer for 1 h. Sections were washed with rinse buffer, washed with PBS, postfixed with 1.6% glutaraldehyde in PBS, and washed with PBS and then distilled water. Silver enhancement of the gold particles was done using Aurion EM SE (Electron Microscopy Sciences, Hatfield, PA) for 45 min, and the grids were then washed with distilled water, dried overnight, and counterstained with saturated uranyl acetate. Each group of sections subjected to the immunogold procedure included a control section exposed to BSA buffer in place of the primary antibody. Ultrathin sections were examined using a Hitachi 7600 transmission electron microscope (Hitachi High-Technologies America, Pleasanton, CA) equipped with a Macrofire monochrome progressive scan charge-coupled device camera (Optronics, Goleta, CA) and AMT image capture software (Advanced Microscopy Techniques, Danvers, MA).

2-Oxoglutarate Quantification

Urinary 2-oxoglutarate was measured in 24-h urine samples obtained in metabolic cages before and after a 7-day acid-loading protocol. Proton one-dimensional nuclear magnetic resonance (NMR) spectra were collected using a 14.1-T NMR magnet equipped with a CP TXI CryoProbe and an Avance II Console (Bruker Biospin, Billerca, MA). Each NMR sample consisted of 100 μL [50% (vol/vol), centrifuged and filtered] urine, 80 μL [40% (vol/vol)] deuterated 20× PBS, and 20 μL [10% (vol/vol)] of internal standard (5 mM DSS-D6 and 0.2% NaN₃ in D₂O, Chenomx, Alberta, ON, Canada). The pH was controlled at 7.2. A one-dimensional NOESY pulse sequence (tnnoesy.c) was used to acquire one-dimensional proton spectra, with a 3-s relaxation delay (d1), 100-ms mixing time, 64 scans (nt), 12-ppm spectral width, and 4-s acquisition time (48). All experiments were acquired at 25°C. Spectra were processed and analyzed using MestReNova 11.0.0 (Mestrelab Research, Santiago de Compostela, Spain). They were zero-filled to 131,072 points with an exponential line broadening of 0.5 Hz and Whittaker Smoother baseline correction. Chemical shifts were calibrated with respect to the DSS singlet signal at 0 ppm. 2-Oxoglutarate peaks were fitted to a mixed Gaussian/Lorentzian line shape and compared with the DSS standard to determine concentration.

Statistical Analysis

Results are presented as means ± SE; n refers to the number of animals studied. When performing quantitative immunohistochemistry, all measurements of a specific cell type in a mouse kidney were averaged and used as a single measurement for statistical analysis. ANOVA was performed using general linear modeling to assess genotype-dependent effects; sex was always included as an independent variable. When sex was determined to have a statistically significant effect, subgroup analyses comparing male and female mice were determined. Analyses were performed using SPSS, JASP, and Microsoft Excel software. P < 0.05 was considered statistically significant.

RESULTS

NBCe1-A Deletion Induces Metabolic Acidosis

Table 2 shows blood chemistries in the NBCe1-A KO and WT mice used in this study. Plasma bicarbonate was significantly lower in KO mice, confirming metabolic acidosis, although plasma Na⁺ and K⁺ concentrations did not differ significantly between WT and KO mice. Therefore, the NBCe1-A KO mice used in this study have severe metabolic acidosis resulting from genetic modifications affecting only the proximal tubule.

Table 2.

Blood and urine chemistries of NBCe1-A WT and KO mice used in this study

	WT [n = 8 (4 Males and 4 Females)]	KO [n = 8 (4 Males and 4 Females)]	P Value
Na+, mM	151 ± 1	149 ± 1	NS
K+, mM	3.8 ± 0.1	4.0 ± 0.4	NS
${\text{HCO}}_3^{-}$, mM	22.7 ± 1.2	12.0 ± 0.8	<0.001
Urine pH	6.32 ± 0.07	5.47 ± 0.14	<0.001
Urine ammonia, µmol/day	67 ± 14	168 ± 46	NS

These mice were a subset of mice previously reported (28). KO, knockout; NBCe1-A, Na⁺-bicarbonate cotransporter, electrogenic, isoform 1, splice variant A; NS, not significant; WT, wild type.

Effect of NBCe1-A Deletion on H⁺-ATPase and AE1 Expression

Type A intercalated cells are characterized by apical H⁺-ATPase and basolateral AE1 immunolabel, which normally exhibit increased expression in response to metabolic acidosis (1, 2). Both H⁺-ATPase and AE1 immunolabel intensity were lower in type A intercalated cells of the OMCD_is from NBCe1-A KO mice compared with WT mice (Fig. 1). Similarly, AE1 abundance was lower in ISOM lysates from NBCe1-A KO mice compared with WT mice (Fig. 2). Thus, the expected response of crucial intercalated cell acid-secretory proteins to metabolic acidosis in NBCe1-A KO mice is absent or blunted.

Figure 1.

Effect of proximal tubule Na⁺-bicarbonate cotransporter, electrogenic, isoform 1, splice variant A (NBCe1-A) knockout (KO) on inner stripe of the outer medullary collecting duct (OMCD_is) intercalated cells. The left column shows immunolabel expression for key proteins involved in collecting duct acid and ammonia secretion in wild-type (WT) mice, and the right column shows findings in NBCe1-A KO mice. The top row shows anion exchanger 1 (AE1) immunolabel and demonstrates that NBCe1-A KO decreases AE1 immunolabel intensity dramatically. The middle top row shows H⁺-ATPase immunolabel and demonstrates that NBCe1-A KO decreases apical H⁺-ATPase immunolabel intensity. The middle bottom row shows Rhesus B glycoprotein (Rhbg) immunolabel, demonstrating that NBCe1-A KO decreases Rhbg immunolabel intensity. The bottom row shows Rhesus C glycoprotein (Rhcg) immunolabel; NBCe1-A KO decreased Rhcg immunolabel intensity. Arrows indicate type A intercalated cells. Micrographs are representative of findings in eight mice (4 males and 4 females) in each genotype.

Figure 2.

Effect of Na⁺-bicarbonate cotransporter, electrogenic, isoform 1, splice variant A (NBCe1-A) knockout (KO) on anion exchanger 1 (AE1) and Rhesus B glycoprotein (Rhbg) in the inner stripe of the outer medulla (ISOM). A: quantification of ISOM AE1 expression using immunoblot analysis. B: quantification of ISOM Rhbg expression using immunoblot analysis. Despite associated metabolic ancidosis, NBCe1-A KO decreased AE1 and Rhbg expression in the ISOM. NBCe1-A KO caused abnormal expression of critical inner stripe of the outer medullary collecting duct acid-base proteins. Immunoblot analysis results are representative of findings in eight mice (4 males and 4 females) of each genotype. WT, wild type.

Effect on Rhesus Glycoproteins

The kidney helps correct metabolic acidosis by increasing ammonia excretion, which occurs, in part, by upregulating Rhbg and Rhcg expression (12, 13, 37). However, Rhbg and Rhcg immunolabel intensity in type A intercalated cells in the OMCD_is was less in NBCe1-A KO mice than in nonacidotic WT mice (Fig. 1). Moreover, by immunoblot, Rhbg abundance was significantly less in KO mice than in WT mice (Fig. 2). Thus, the expected response to metabolic acidosis with increased expression of proteins critical to collecting duct ammonia secretion is blunted or absent in NBCe1-A KO mice.

Intercalated Cell Structural Response

Another phenotypic collecting duct response to metabolic acidosis involves type A intercalated cell hypertrophy and increased apical plasma membrane H⁺-ATPase. However, despite the associated metabolic acidosis, type A intercalated cell height was significantly less in KO mice than in WT mice [WT: 13.0 ± 0.4 µm and KO: 11.1 ± 0.2 µm, n = 7 (4 males and 3 females) and 7 (3 males and 4 females), respectively, P < 0.001; Fig. 3A]. Similarly, total H⁺-ATPase abundance was less in the OMCD_is of NBCe1 A KO mice than in WT mice (Fig. 3B). These responses were not limited to the OMCD_is; quantitative immunohistochemistry showed decreased AE1 immunolabel in CCD type A intercalated cells (Fig. 3C). Thus, type A intercalated cell size and H⁺-ATPase and AE1 abundance are reduced in NBCe1-A KO, despite the associated metabolic acidosis that should produce the opposite response.

Figure 3.

Na⁺-bicarbonate cotransporter, electrogenic, isoform 1, splice variant A (NBCe1-A) deletion decreases type A intercalated cell height and H⁺-ATPase and cortical collecting duct (CCD) anion exchanger 1 (AE1). A: height of type A intercalated cells in the inner stripe of the outer medullary collecting duct (OMCD_is). Despite the associated acidosis, which typically inreases type A intercalated cell height, cell height in NBCe1-A knockout (KO) mice was significantly less than in wild-type (WT) mice. n = 7 WT mice (4 males and 3 females) and n = 7 KO mice (3 males and 4 females). B: H⁺-ATPase immunolabel expression in OMCD_is type A intercalated cells. Type A intercalated cell H⁺-ATPase expression was significantly less in NBCe1-A KO mice than in WT mice. Expression normalized so that mean expression in WT mice equals 1.00. n = 7 WT mice (4 males and 3 females) and n = 7 KO mice (3 males and 4 females). C: CCD type A intercalated cell AE1 immunolabel expression. AE1 expression was significantly less in KO mice than in WT mice. Expression normalized so that mean expression in WT mice equals 1.00. n = 8 mice (4 males and 4 females) for each genotype.

H⁺-ATPase Subcellular Distribution

We used immunogold electron microscopy to examine further changes in type A intercalated cell H⁺-ATPase subcellular distribution. We observed that KO mice exhibited a qualitative change in H⁺-ATPase subcellular distribution, with a relative decrease in apical plasma membrane H⁺-ATPase expression and a relative increase in subapical vesicular cytoplasmic H⁺-ATPase expression compared with WT mice (Fig. 4). In addition, apical plasma membrane microprojections, an ultrastructural finding of type A intercalated cell activation, were qualitatively decreased in NBCe1-A KO mice. These responses are the opposite of those typically seen during metabolic acidosis (49).

Figure 4.

Na⁺-bicarbonate cotransporter, electrogenic, isoform 1, splice variant A (NBCe1-A) deletion decreases inner stripe of the outer medullary collecting duct (OMCD_is) type A intercalated cell apical plasma membrane H⁺-ATPase expression. High-power, representative immunogold electron microscopy micrographs of apical plasma membrane and subapical region of OMCD_is type A intercalated cells using antibodies to H⁺-ATPase in wild-type (WT) and NBCe1-A knockout (KO) mice are shown. KO mice had decreased apical plasma membrane H⁺-ATPase compared with WT mice. In addition, apical microprojections were reduced in KO mice. Arrows show H⁺-ATPase in the apical plasma membrane; arrowheads show H⁺-ATPase in subapical vesicles. Results are representative of findings in two male and two female kidneys for each genotype.

Pendrin-Positive Intercalated Cell Regulation

Pendrin-positive intercalated cells secrete bicarbonate through a mechanism dependent on apical pendrin, which helps correct metabolic alkalosis. The normal kidney responds to metabolic acidosis with decreased apical pendrin expression (15–17). If this response is innate to the collecting duct, pendrin abundance would be lower in KO mice than in WT mice due to the metabolic acidosis observed in the former. However, immunoblot analysis of the renal cortex showed that pendrin expression did not differ significantly between NBCe1-A KO and WT mice (Fig. 5A).

Figure 5.

Effect of Na⁺-bicarbonate cotransporter, electrogenic, isoform 1, splice variant A (NBCe1-A) knockout (KO) on pendrin expression. A: immunoblot analysis of pendrin expression in the renal cortex did not differ significantly between wild-type (WT) and KO mice. Expression was greater in females than in males (P < 0.001 by ANOVA), with no significant interaction of genotype with sex [P = not significant (NS)]. B: double-immunolabel immunohistochemistry (pendrin, blue; anion exchanger 1, brown) was used to characterize pendrin expression in connecting tubule (CNT) non-A, non-B and cortical collecting duct (CCD) type B intercalated cells in WT and NBCe1-A KO mice. Characteristic apical pendrin was seen in non-A, non-B intercalated cells (arrows) in the CNT and type B intercalated cells (arrows) in the CCD. C and D: single-cell pendrin expression in the CNT. Non-A, non-B per cell pendrin expression and relative apical pendrin expression was determined using quantitative immunohistochemistry. Relative apical pendrin expression was the proportion of total cellular pendrin in the apical 25% of the cell. Despite the associated metabolic acidosis in KO mice, neither single-cell pendrin nor percent apical pendrin were significantly altered in KO mice. n = 8 mice (4 males and 4 females) for each genotype. E and F: single-cell CCD type B intercalated cell pendrin expression and relative apical pendrin expression. Despite the associated metabolic acidosis in KO mice, neither single-cell pendrin nor percent apical pendrin were significantly altered in CCD type B intercalated cells in KO mice. Pendrin expression in C and E was normalized such that mean WT expression in each equals 1.00. n = 8 mice (4 males and 4 females) for each genotype.

Because there are two distinct pendrin-positive cells in the cortex, CNT non-A, non-B intercalated cells and CCD type B intercalated cells, we used quantitative immunohistochemistry to assess cell-specific pendrin expression in each cell population (Fig. 5B). In CNT non-A, non-B intercalated cells, per cell pendrin expression did not differ significantly between WT and KO mice [WT: 1.00 ± 0.04 and KO: 0.94 ± 0.05, P = not significant (NS), n = 8 (4 males and 4 females) in each group; Fig. 5C]. Moreover, apical pendrin polarization (WT: 43 ± 1% and KO: 40 ± 2%, P = NS, n = 8 (4 males and 4 females) in each group] did not differ significantly between NBCe1-A KO and WT mice (Fig. 5D).

Similar findings were present in CCD type B intercalated cells. Neither pendrin immunolabel intensity per cell [WT: 1.00 ± 0.04 and KO: 0.94 ± 0.05, P = NS, n = 8 (4 males and 4 females) in each group] nor apical pendrin polarization (WT: 43 ± 1% and KO: 40 ± 2%, P = NS, n = 8 (4 males and 4 females in each group)] differed significantly between NBCe1-A KO and WT mice (Fig. 5, E and F). Thus, the expected acidosis-induced decrease in pendrin expression is absent in spontaneously acidotic NBCe1-A KO mice.

Is the Effect of NBCe1-A Deletion an Off-Target Effect of the Gene-Editing Procedure?

We next asked if the observed NBCe1-A KO collecting duct phenotype resulted from off-target effects of the gene-editing procedure rather than the absence of proximal tubule NBCe1-A. To address this question, we studied NBCe1 KO generated using a different gene-editing approach than the TALEN techniques used for NBCe1-A mice (28). We used mice with kidney-specific deletion of both NBCe1 splice variants, NBCe1-A and NBCe1-B, that were generated using Cre loxP technology and were recently described (33). Since NBCe1-A and NBCe1-B are found in the kidney at detectable levels only in the proximal tubule (25, 28, 33), we term these “PT-NBCe1-KO” mice. PT-NBCe1-KO mice exhibited severe metabolic acidosis, although serum Na⁺ or K⁺ concentrations were unchanged (Table 3), consistent with our original description of these mice (33).

Table 3.

Blood and urine chemistries of proximal tubule NBCe1-A/B WT and KO mice

	WT [n = 8 (4 Males and 4 Females)]	KO [n = 8 (4 Males and 4 Females)]	P Value
Na+, mM	155 ± 1	152 ± 1	NS
K+, mM	3.9 ± 0.1	3.7 ± 0.2	NS
${\text{HCO}}_3^{-}$, mM	18.1 ± 0.7 (3 males and 3 females)	8.8 ± 0.6	<0.001
Urine pH	6.12 ± 0.04	5.50 ± 0.11	<0.001
Urine ammonia, µmol/day	124 ± 16	155 ± 38	NS

These mice were a subset of mice previously reported (33). KO, knockout; NBCe1-A, Na⁺-bicarbonate cotransporter; electrogenic, isoform 1, splice variant A; NS, not significant; WT, wild type.

Immunohistochemistry showed that AE1, H⁺-ATPase, Rhbg, and Rhcg immunolabel intensities were lower in type A intercalated cells of the OMCD_is from the PT-NBCe1-KO mice (Fig. 6). Moreover, immunoblot analysis of the ISOM showed decreased expression of both AE1 and Rhbg (Fig. 7). These findings parallel those observed with NBCe1-A KO (see Figs. 1 and 2, respectively) and thereby further demonstrate the critical role of the proximal tubule in regulating intercalated cell acid- and ammonia-secretory phenotype.

Figure 6.

Effect of kidney-specific deletion of Na⁺-bicarbonate cotransporter, electrogenic, isoform 1, splice variant A (NBCe1-A) splice variants NBCe1-A and NBCe1-B (PT-NBCe1-KO) on inner stripe of the outer medullary collecting duct (OMCD_is) anion exchanger 1 (AE1), H⁺-ATPase, Rhesus B glycoprotein (Rhbg), and Rhesus C glycoprotein (Rhcg) expression. The left column shows findings in wild-type (WT) mice, and the right column shows findings in PT-NBCe1-KO mice. The top row shows AE1 immunohistochemistry and demonstrates that PT-NBCe1-KO decreases AE1 immunolabel intensity dramatically. The middle top row shows H⁺-ATPase immunohistochemistry and demonstrates decreased immunolabel intensity in the PT-NBCe1-KO mouse kidney. The middle bottom row shows representative Rhbg immunohistochemistry and demonstrates decreased Rhbg immunolabel intensity in the PT-NBCe1-KO kidney. The bottom shows Rhcg immunohistochemistry in the WT and PT-NBCe1-KO kidney. Rhcg immunolabel intensity was less in the PT-NBCe1-KO kidney than in the WT kidney. Arrows indicate type A intercalated cells. All micrographs are representative of findings in n = 8 mice (4 males and 4 females) of each genotype. KO, knockout.

Figure 7.

Effect of kidney-specific deletion of Na⁺-bicarbonate cotransporter, electrogenic, isoform 1 (NBCe1-A) splice variants NBCe1-A and NBCe1-B (PT-NBCe1-KO) on anion exchanger 1 (AE1) and Rhesus B glycoprotein (Rhbg) expression in the inner stripe of the outer medulla (ISOM). A: immunoblot for AE1 in the ISOM of wild-type (WT) and PT-NBCe1-KO kidneys. NBCe1 deletion significantly decreased AE1 expression despite the associated metabolic acidosis. B: immunoblot analysis for Rhbg. PT-NBCe1-KO decreased Rhbg expression significantly, also, despite the associated metabolic acidosis. n = 8 mice (4 males and 4 females) for each genotype. These results parallel those for NBCe1-A KO, suggesting that the observed effects are not a nonspecific effect of the gene-editing procedurdes.

Intercalated Cell Remodeling Response to Acidosis

Because the response of the collecting duct to acidosis appeared to involve changes in the proportion of the different intercalated cell types, we asked if the proximal tubule modulates this remodeling response. We separately assessed changes in the intercalated cell type distribution in the CNT, CCD, and OMCD_is.

In the CNT of WT mice, acid loading increased the proportion of type A intercalated cells [basal: 9.1 ± 0.4% and acid: 11.0 ± 0.3%, P < 0.001, n = 8 (4 males and 4 females) and 12 (6 males and 6 females), respectively] and decreased the number of non-A, non-B intercalated cells [basal: 24.9 ± 0.4% and acid: 22.7 ± 0.3%, P < 0.001, n = 8 (4 males and 4 females) and 12 (6 males and 6 females), respectively]. NBCe1-A KO altered this CNT response significantly (P < 0.001 by ANOVA). In NBCe1-A KO mice, the intercalated cell response differed significantly (P < 0.001 by ANOVA). Specifically, acid loading did not increase and instead slightly decreased the proportion of type A intercalated cells [basal: 20.8 ± 1.3% and acid: 16.9 ± 0.9%, P < 0.02, n = 8 (4 males and 4 females) and 12 (6 males and 6 females), respectively], and it completely blocked any change in the proportion of non-A, non-B intercalated cells (basal: 20.0 ± 1.0% and acid: 21.8 ± 1.0%, P = NS, n = 8 and 12, respectively; Fig. 8, A and B).

Figure 8.

Connecting tubule (CNT), cortical collecting duct (CCD), and inner stripe of the outer medullary collecting duct (OMCD_is) intercalated cell type responses to acid loading. A and B: CNT response. CNT intercalated cell types were counted in mice on either a basal diet or acid-loading diet for 1 wk. In wild-type (WT) mice, acid diet increased type A intercalated cell numbers and decreased non-A, non-B intercalated cell numbers. Na⁺-bicarbonate cotransporter, electrogenic, isoform 1, splice variant A (NBCe1-A) knockout (KO) blocked the effect of acid loading on both type A intercalated cells and non-A, non-B intercalated cells (P < 0.001 and P < 0.005 by ANOVA). n = 8 WT and KO mice (4 males and 4 females) on the basal diet and n = 12 WT and KO mice (6 males and 6 females) on the acid diet. C and D: CCD response. Intercalated cell type was defined using double-immunolabel, H⁺-ATPase and AE1, immunohistochemistry. In WT mice, acid loading decreased the number of type B intercalated cells, and NBCe1-A KO blocked this response (P < 0.001 by ANOVA). n = 8 WT mice (4 males and 4 females) and 7 KO mice (3 males and 4 females) on the basal diet and n = 12 WT mice (6 males and 6 females) and 10 KO mice (4 males and 6 females) on the acid diet. E: OMCD_is response. In WT mice, acid loading increased the number of OMCD_is type A intercalated cells. NBCe1-A KO blocked this response (P < 0.001 by ANOVA). n = 7 WT mice (4 males and 3 females) and 7 KO mice (3 males and 4 females) on the basal diet and n = 11 WT mice (5 males and 6 females) and 12 KO mice (6 males and 6 females) on the acid diet. NS, not significant.

A similar dependence of the pendrin-positive cell remodeling response to exogenous acidosis was observed in the CCD. In WT mice, acid loading decreased the proportion of type B intercalated cells significantly (basal: 22.3 ± 0.8% and acid: 13.9 ± 0.5%, P < 0.001, n = 8 and 12, respectively). However, this response was absent in NBCe1-A KO mice (basal: 16.6 ± 0.5% and acid: 15.2 ± 0.7%, P = NS, n = 7 and 10, respectively; Fig. 8D).

This effect was also present in the outer medulla. In WT mice, acid loading increased the proportion of OMCD_is type A intercalated cells [basal: 40.2 ± 1.0% and acid: 44.3 ± 0.8%, P < 0.006, n = 7 (4 males and 3 females) and 11 (5 males and 6 females), respectively; Fig. 8E]. In NBCe1-A KO mice, the proportion of type A intercalated cells did not increase and instead decreased slightly with acid loading [basal: 45.6 ± 0.8% and acid: 39.4 ± 1.0%, P < 0.001 by ANOVA, n = 7 (3 males and 4 females) and 12 (6 males and 6 females), respectively]. Thus, the OMCD_is, CCD, and CNT intercalated cell remodeling responses to acid loading are dependent on proximal tubule NBCe1-A expression.

Effects of Sex

Sex alters numerous components of acid-base homeostasis, including effects on renal structure, protein expression, and ammonia metabolism (50–53). In the present study, sex significantly affected several parameters of intercalated cell phenotype. These effects are shown in Table 4. However, none of these effects abrogated the effect of either NBCe1-A or PT-NBCe1-KO on collecting duct phenotypic or remodeling responses.

Table 4.

Effect of sex

Model	Measure	Male Vs. Female		WT Vs. KO
		WT	KO	Male	Female
NBCe1-A KO	AE1 expression	Male, 1.29 ± 0.17 (n = 4) Female, 0.71 ± 0.11 (n = 4) Male > Female, P < 0.02	Male, 0.47 ± 0.07 (n = 4) Female, 0.49 ± 0.13 (n = 4) NS	WT > KO, P < 0.005	NS
	Rhbg expression	Male, 1.26 ± 0.09 (n = 4) Female, 0.74 ± 0.09 (n = 4) Male > Female, P < 0.005	Male, 0.51 ± 0.11 (n = 4) Female, 0.53 ± 0.05 (n = 4) NS	WT > KO, P < 0.001	WT > KO, P < 0.05
	Pendrin expression	Male, 0.71 ± 0.03 (n = 4) Female, 1.29 ± 0.12 (n = 4) Male < Female, P < 0.001	Male, 0.70 ± 0.08 (n = 4) Female, 1.06 ± 0.07 (n = 4) Male < Female, P < 0.01	NS	NS
	OMCDis A cell height	Male, 12.3 ± 0.3 µm (n = 4) Female, 13.9 ± 0.2 µm (n = 3) Male < Female, P < 0.005	Male, 11.3 ± 0.4 µm (n = 3) Female, 11.0 ± 0.1 µm (n = 4) NS	WT > KO, P < 0.05	WT > KO, P < 0.001
	CCD A cell AE1 expression	Male, 1.15 ± 0.05 (n = 4) Female, 0.85 ± 0.04 (n = 4) Male > Female, P < 0.02	Male, 0.73 ± 0.07 Female, 0.74 ± 0.07 NS	WT > KO, P < 0.05	NS
PT-NBCe1-KO	Rhbg expression	Male, 1.02 ± 0.08 (n = 4) Female, 0.98 ± 0.05 (n = 4) NS	Male, 0.43 ± 0.04 (n = 4) Female, 0.73 ± 0.10 (n = 4) Male < Female, P < 0.05	WT > KO, P < 0.001	WT > KO, P < 0.05

Subgroup analysis of the effect of sex was only performed when ANOVA indicated statistically significant effects of sex alone or on the interaction of sex with genotype. Protein expression normalized to mean in wild-type (WT) mice is equal to 1.00. Results in the WT vs. knockout (KO) column do not repeat data, which are present in the male vs. female column, and only show the statistical analysis. AE1, anion exchanger 1; CCD, collecting duct; NBCe1-A, Na⁺-bicarbonate cotransporter; electrogenic, isoform 1, splice variant A; NS, not significant; OMCD_is, inner stripe of the outer medullary collecting duct; PT-NBCe1-KO, kidney-specific NBCe1 KO; Rhbg, Rhesus B glycoprotein; Rhcg, rhesus C glycoprotein.

Effects of NBCe1-a KO on 2-Oxoglutarate

2-Oxoglutarate (α-ketoglutarate) stimulates intercalated cell ${\text{HCO}}_3^{-}$ secretion (54), and its excretion is altered in acid-base disturbances (55). To assess the possibility of 2-oxoglutarate functioning as a proximal tubule-dependent signaling molecule, we measured its urinary excretion in NBCe1-A KO mice. We found that urinary 2-oxoglutarate did not differ significantly between WT and NBCe1-A KO mice on the basal diet (WT: 3.25 ± 0.41 µmol/day and KO: 3.06 ± 0.61 µmol/day, P = NS, n = 6 in each group). Acid loading decreased 2-oxoglutarate excretion by 92.9 ± 1.4% in WT mice, to 0.21 ± 0.03 µmol/day (n = 6), and by 95.7 ± 2.6% in NBCe1-A KO mice, to 0.08 ± 0.02 µmol/day (n = 6). The proportionate decrease in 2-oxoglutarate excretion did not differ between WT and KO mice (P = NS). Thus, 2-oxoglutarate excretion cannot explain the NBCe1A-dependent effects of the proximal tubule on intercalated cell responses to acidosis.

DISCUSSION

The results of this study provide important information regarding the collecting duct response to metabolic acidosis. Genetic deletion of proteins present only in the proximal tubule led to severe metabolic acidosis. Despite this metabolic acidosis, there was decreased type A intercalated cell height, decreased expression of essential proteins involved in collecting duct intercalated cell-mediated acid secretion, a change in the subcellular distribution of H⁺-ATPase, with a shift in its subcellular distribution from the apical plasma membrane to subapical vesicles, and decreased expression of essential proteins required for ammonia secretion, Rhbg and Rhcg. There also was a failure of the expected response of bicarbonate-secreting type B intercalated and non-A, non-B intercalated cells. Finally, the collecting duct remodeling response to an exogenous acid load was absent in NBCe1-A KO mice. Thus, abrogation of the proximal tubule acid-base response leads to abnormal collecting duct phenotypic responses and blocks the remodeling responses to acidosis; this suggests that the proximal tubule controls these responses.

Over the past several decades, numerous studies have shown significant effects of systemic acidosis on collecting duct intercalated cells. These responses include hypertrophy, changes in the subcellular distribution of H⁺-ATPase, increased basolateral AE1 expression, and increased expression of the ammonia transporter family members, Rhbg and Rhcg (1–3). Each response contributes to an increased ability to secrete H⁺ and ammonia. Pendrin-positive intercalated cells, which secrete bicarbonate, typically respond to metabolic acidosis with decreased pendrin expression and apical pendrin polarization (56).

The first significant finding in this study is that genetic changes affecting only the proximal tubule, NBCe1-A KO or combined NBCe1-A/B KO, lead to severe metabolic acidosis but do not lead to the expected intercalated cell acidosis-induced phenotypic responses. Instead, there was decreased type A intercalated cell size, decreased expression of critical proteins involved in type A intercalated cell H⁺ secretion, H⁺-ATPase and AE1, reduced expression of essential proteins required for intercalated cell ammonia secretion, Rhbg and Rhcg, and decreased apical plasma membrane localization of H⁺-ATPase. Essentially identical findings were observed in a second proximal tubule-dependent metabolic acidosis model, PT-NBCe1-KO. This effect was not limited to type A intercalated cells; pendrin-positive intercalated cells exhibited pendrin expression unchanged by metabolic acidosis. Thus, collecting duct intercalated cell phenotypic responses to acidosis depend on the proximal tubule.

The second finding is that the intercalated cell remodeling response to acidosis depends on the proximal tubule. Similar to findings in some (2, 19, 20, 57), but not all (6), previous studies, exogenous acid loading altered the intercalated cell distribution in the CNT, CCD, and OMCD_is. However, this response was completely absent in mice with NBCe1-A deletion. Since these proteins are present only in the proximal tubule in the kidney, these findings indicate that the proximal tubule also controls the collecting duct remodeling response to acid loading.

The finding that NBCe1-A and NBCe1-B are critical for the collecting duct response to acidosis extends previous findings of the role of these proteins in other pathways critical to acid-base homeostasis. Their deletion is known to cause proximal RTA through their central role in the basolateral bicarbonate transport necessary for the reabsorption of filtered bicarbonate (58, 59). Studies in rodent models have shown they are critical to the proximal tubule ammoniagenic response to acidosis (28, 33, 60). Third, they are critical for proper citrate metabolism by regulating proximal tubule NaDC1 expression (30). However, this role does not extrapolate necessarily to other organic anions, as NBCe1-A deletion did not alter 2-oxoglutarate excretion. Thus, NBCe1-A and NBCe1-B contribute to multiple aspects of acid-base regulation that involve both the proximal tubule and collecting duct. Because we used congenital NBCe1 splice variant deletion from the proximal tubule to alter proximal tubule acid-base responsiveness, we cannot exclude the possibility that different effects might be seen in mice with inducible deletion.

NBCe1-A and NBCe1-B are likely to mediate these multiple effects on components of acid-base regulation through their effects on intracellular pH. Basolateral Na⁺-coupled electrogenic bicarbonate transport, i.e., NBCe1 activity, is the primary mechanism through which extracellular bicarbonate alters proximal tubule intracellular pH (61, 62). Because NBCe1-A and NBCe1-B are bicarbonate extruding transporters, their deletion is likely to cause baseline intracellular alkalinization and inhibit the intracellular pH response to extracellular acidosis. This intracellular pH change may then regulate proximal tubule ammoniagenesis, proximal tubule citrate transport and metabolism, and proximal tubule regulation of the collecting duct phenotypic and remodeling responses to acidosis. However, because these NBCe1 variants are electrogenic and transport Na⁺, we cannot exclude indirect effects related to changes in membrane voltage or intracellular Na⁺.

Ammonia is likely to contribute to the mechanism through which NBCe1-A and NBCe1-B and the proximal tubule regulate collecting duct responses to acidosis. Ammonia is generated predominantly in the proximal tubule and is concentrated in the interstitium, and acidosis stimulates both of these processes (63–65). Thus, acidosis increases collecting duct exposure to ammonia. We and others have shown that ammonia stimulates collecting duct H⁺ secretion in the CCD (66–69), outer medullary collecting duct (70), and inner medullary collecting duct (71). The observation that ammonia causes renal epithelial cell hypertrophy in the proximal tubule (72–75) suggests that proximal tubule-derived ammonia may also cause collecting duct type A intercalated cell hypertrophy. Finally, both genetic models of altered proximal tubule function examined in the present study, NBCe1-A KO and PT-NBCe1-KO, have been shown to inhibit proximal tubule ammoniagenesis (28, 33). These considerations suggest that altered signaling involving ammonia likely contributes to the proximal tubule’s regulation of the collecting duct.

Ammonia-dependent regulation of the collecting duct can explain the intercalated cell response to dyskalemic conditions. In a genetic model of hyperkalemia-induced metabolic acidosis, there is inhibition of both proximal tubule ammoniagenesis and the collecting duct acid-secretory phenotype (76). Because hyperkalemia in this model was associated with metabolic acidosis, the collecting duct response is the opposite of the expected acid-base response and instead parallels changes in ammonia metabolism. Similarly, hypokalemia, which increases proximal tubule ammoniagenesis (29, 77, 78), leads to an increased number of intercalated cells in some studies (79–82) and in stimulation of the intercalated cell acid-secretory phenotype (6, 39, 40, 82–86). Because hypokalemia is typically associated with metabolic alkalosis, rather than metabolic acidosis, the collecting duct responses cannot be ascribed to the systemic acid-base disorder. Thus, in dyskalemic conditions, the collecting duct intercalated cell response parallels the proximal tubule ammonia response instead of the systemic acid-base status.

Another possible mechanism involves urinary extracellular vesicles. These are nanosized particles released from renal epithelial cells, containing proteins, RNA, and microRNA, and have the potential to deliver signals to downstream renal epithelial cells (87). Their contents are altered in response to experimental stimuli, including high-K⁺ diets (88) and mineralocorticoids (89). Thus, urinary extracellular vesicles are a second consideration to explain the effect of the proximal tubule on the collecting duct acid-base phenotypic and remodeling responses to acidosis.

It is also possible that the proximal tubule-dependent signaling mechanisms may involve intermediary signaling involving one or more intervening cell populations.

Several mechanisms known to regulate collecting duct function are less likely to explain the findings in the present study. 2-Oxoglutarate stimulates intercalated cell ${\text{HCO}}_3^{-}$ secretion (54). However, NBCe1-A deletion did not alter basal or acidosis-stimulated 2-oxoglutarate excretion. Succinate, which is reabsorbed in the proximal tubule through an NaDC-1-dependent mechanism, can activate cellular processes through GPR91 (also referred to as SUCNR1) activation (90). However, GPR91 protein has been reported only in the proximal tubule, distal convoluted tubule, and juxtaglomerular apparatus (90), and transcriptome analysis found it only in the proximal tubule and macula densa in a second report (91) and only in the proximal tubule in a third report (92). This spatial expression pattern makes it unlikely that succinate-dependent GPR91 signaling explains the results in this study. Soluble adenylyl cyclase, via sensing of luminal bicarbonate, can stimulate intercalated cell apical H⁺-ATPase (93). The lack of direct measurement of luminal bicarbonate delivery to the collecting duct makes conclusions regarding the role of soluble adenylyl cyclase in the present study difficult. However, if luminal bicarbonate delivery is increased due to decreased proximal tubule bicarbonate reabsorption, then soluble adenylyl cyclase would be expected to result in increased apical H⁺-ATPase accumulation rather than the decrease observed in the present study. Although chronic metabolic acidosis can lead to more rapid progression of chronic kidney disease, the glomerular filtration rate, estimated as urea clearance, does not differ between NBCe1-A KO and WT mice (28). Aldosterone stimulates intercalated cell function (94, 95) but is unlikely to explain the present findings because NBCe1-A KO increases aldosterone (29), which would have the opposite effect on intercalated cells as that observed in the present study. Another possibility is whether NBCe1 protein is expressed in collecting duct at levels that could regulate collecting duct acid-base responses. However, multiple studies have not shown detectable NBCe1 protein in the collecting duct (22, 27–29, 33, 96–99). Although one transcriptome study examining isolated renal segments reported low levels of NBCe1 mRNA expression in the collecting duct, levels <1% of that observed in the PST1 (92), another study reporting single-cell transcript expression did not find detectable expression in either collecting duct principal cells or intercalated cells (91). NBCe1 is also expressed in the thyroid gland, where its expression may have been altered in Pax8-Cre, floxed NBCe1 PT-NBCe1-KO mice. However, the identical findings in NBCe1-A KO mice suggest that possible changes in thyroid hormone production in PT-NBCe1-KO mice are unlikely to mediate the observed findings.

NBCe1-A KO and PT-NBCe1-KO led to more acidic urine pH despite inhibiting the collecting duct acid-secretory phenotype. Urine pH predominantly reflects competing effects of collecting duct H⁺ secretion, which leads to a more acidic pH, and collecting duct NH₃ secretion, which leads to a more alkaline pH. H⁺ secretion depends primarily on H⁺-ATPase and AE1 expression, both of which were decreased in the models studied in this report. NBCe1-A KO and PT-NBCe1-KO also decrease proximal tubule ammoniagenesis (28, 33), which will decrease ammonia for collecting duct NH₃ secretion. The present study also showed decreased expression of the collecting duct NH₃ transporters. Thus, decreased collecting duct NH₃ secretion, resulting from the combination of decreased ammoniagenesis and decreased Rhbg and Rhcg expression, may exceed changes in H⁺ secretion, resulting in more acidic urine. This contrasts with distal RTA in humans, where the defect is limited to H⁺ secretion and where there is an alkaline, not acidic, urine pH (59, 100).

Understanding the effect of sex on biological responses to experimental stimuli is essential. In the kidney, sex affects basal and acidosis-stimulated ammonia metabolism and renal structure (34, 35). Many of these differences are dependent on testosterone (43) and involve the androgen receptor (101). In the present study, sex significantly affected several parameters of intercalated cell phenotype, as shown in Table 4. Most of these effects were no longer present in mice with NBCe1-A deletion, which suggests the sex differences in WT mice may be partly due to the sexual dimorphism in proximal tubule ammonia metabolism (35, 51). We cannot explain why Rhbg expression was greater in males than in females in WT mice for the NBCe1-A KO experiments but not in WT (Cre-negative) mice for the PT-NBCe1-KO experiments.

Several other proteins are critical for the collecting duct response to acidosis. These include SDF1 (19, 102), hensin (103), GDF15 (104), Pyk2 (105), and GPR4 (106), all of which are expressed in the collecting duct. Understanding the interaction of these gene products with the proximal tubule-dependent pathway identified in the present project with these will be necessary for future studies.

Proximal tubule RTA resulting from NBCe1 gene deletion has multiple effects on renal acid-base homeostasis. First, because Na⁺-coupled bicarbonate transport is the primary mechanism of proximal tubule basolateral bicarbonate transport (107), it is highly likely that NBCe1 deletion decreases proximal tubule bicarbonate reabsorption, although this has not been explicitly tested. Second, NBCe1 deletion impairs proximal tubule net ammoniagenesis leading to impaired urinary ammonia excretion (28, 29, 33, 60). Third, it decreases proximal tubule NaDC1 expression leading to increased urinary citrate excretion (30, 108). Finally, as shown in the present study, it leads to impaired collecting duct phenotypic and remodeling responses to acidosis.

The effects of distal RTA on proximal tubule function are less clear. Distal RTA is associated with decreased urinary citrate excretion (109), which is the expected response to the associated acidosis. We are unaware of studies examining genetic models of distal RTA that have directly examined proximal tubule bicarbonate transport or ammonia generation. However, our observations that inhibition of collecting duct Rhbg and Rhcg expression leads to adaptive changes in proximal tubule ammoniagenic enzyme expression (12, 37, 41). Thus, while proximal tubule RTA induced by NBCe1 deletion leads to maladaptive collecting duct responses, distal RTA leads to changes in proximal tubule citrate and ammonia metabolism that are expected to maintain acid-base homeostasis.

Another implication of the current study is that one cannot differentiate the different types of RTA based on collecting duct H⁺-ATPase or AE1 expression. Type I (distal) RTA typically results from genetic or acquired loss of H⁺-ATPase or AE1 (100, 110, 111). In type IV (hyperkalemic) RTA, the associated hyperkalemia decreases H⁺-ATPase apical expression (76). Finally, the present study shows that NBCe1 deletion, which causes type II (proximal) RTA (23, 24), decreases both H⁺-ATPase and AE1 expression. Thus, changes in the expression of these critical proteins cannot differentiate the varying forms of RTA.

Perspectives and Significance

In summary, the present study shows critical new aspects of collecting duct acid-base transport. Genetic deletion affecting only the proximal tubule, involving deletion of either NBCe1-A or both NBCe1-A and NBCe1-B, led to severe metabolic acidosis. Despite this metabolic acidosis, there was inhibition of type A intercalated cell phenotypic acid- and ammonia-secretory characteristics, abrogation of the non-A, non-B, and type B intercalated cell responses to acidosis, and blockade of the collecting duct remodeling responses to exogenous acid loading. Thus, the intercalated cell phenotypic and remodeling responses critical to acid-base homeostasis depend on signaling derived from the proximal tubule.

GRANTS

This work was supported by funds from the National Institute of Diabetes and Digestive and Kidney Diseases Grants R01DK107798 (to I.D.W. and J.W.V.), K08DK120873 (to A.N.H.), and R01DK110375 (to S.M.W.).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

J.W.V. and I.D.W. conceived and designed research; J.W.V. and H.-W.L. performed experiments; J.W.V., H.-W.L., A.N.H., and I.D.W. analyzed data; J.W.V., H.-W.L., S.M.W., A.N.H., and I.D.W. interpreted results of experiments; H.-W.L. and I.D.W. prepared figures; I.D.W. drafted manuscript; H.-W.L., A.N.H., and I.D.W. edited and revised manuscript; J.W.V., H.-W.L., S.M.W., A.N.H., and I.D.W. approved final version of manuscript.