Role of Receptor Protein Tyrosine Phosphatase γ in Sensing Extracellular CO₂ and HCO₃⁻

Abstract

Regulation of blood pH—critical for virtually every facet of life—requires that the renal proximal tubule (PT) adjust its rate of H⁺ secretion (nearly the same as the rate of HCO₃⁻ reabsorption, J_HCO3) in response to changes in blood [CO₂] and [HCO₃⁻]. Yet CO₂/HCO₃⁻ sensing mechanisms remain poorly characterized. Because receptor tyrosine kinase inhibitors render J_HCO3 in the PT insensitive to changes in CO₂ concentration, we hypothesized that the structural features of receptor protein tyrosine phosphatase-γ (RPTPγ) that are consistent with binding of extracellular CO₂ or HCO₃⁻ facilitate monitoring of blood CO₂/HCO₃⁻ concentrations. We now report that PTs express RPTPγ on blood-facing membranes. Moreover, RPTPγ deletion in mice eliminated the CO₂ and HCO₃⁻ sensitivities of J_HCO3 as well as the normal defense of blood pH during whole-body acidosis. Thus, RPTPγ appears to be a novel extracellular CO₂/HCO₃⁻ sensor critical for pH homeostasis.

The regulation of pH is critically important because nearly all biologic processes, both inside cells and in the extracellular fluid, are pH sensitive.^1–3 Intracellular pH (pH_i) depends on extracellular fluid pH (pH_o); the latter, in turn, depends on arterial pH (pH_a), which reflects the ratio of arterial carbon dioxide concentration to arterial bicarbonate concentration [CO₂]_a/[HCO₃⁻]_a. Pulmonary ventilation controls [CO₂]_a, and the kidneys—particularly the proximal tubules (PTs)—control [HCO₃⁻]_a by adjusting the rate of HCO₃⁻ transfer (J_HCO3) from the epithelial cell, across the basolateral (BL) membrane, into the blood. Curiously, J_HCO3 is acutely independent of pH at the BL surface (pH_BL) of PTs yet rises steeply with increases in [CO₂]_BL or decreases in [HCO₃⁻]_BL.⁴ Nevertheless, how PTs, or other cells, sense changes in [CO₂]_BL or [HCO₃⁻]_BL has remained elusive. However, two key insights on how this sensing occurs are that (1) drugs blocking the ErbB receptor protein tyrosine kinases eliminate PT CO₂ sensitivity⁵ and (2) the receptor protein tyrosine phosphatases RPTPγ (which are approximately ubiquitous) and RPTPζ (mainly in the brain) have extracellular ligand-binding domains that are 35%–40% identical to the catalytic domains of human carbonic anhydrases (CAs).⁶ Our goal was to determine whether RPTPγ, with no previously established function in ligand sensing, plays a role in CO₂/HCO₃⁻ sensing.

In the early 1950s, three groups demonstrated in dogs that respiratory acidosis—a rise in [CO₂] that produces a fall in pH and a rise in [HCO₃⁻]—causes a rapid increase in renal acid secretion.^7–9 Later investigators showed that alkalosis inhibits acid secretion at the level of renal tubules.^10,11 However, with classic acid-base disturbances at least two of the three chemical parameters ([CO₂], [HCO₃⁻], and pH) must change under equilibrium conditions, making it impossible to define the key underlying parameters. The invention of out-of-equilibrium (OOE) CO₂/HCO₃⁻ solutions¹² made it possible to alter each of the three parameters independently, leading to the discovery that PTs sense [CO₂]_BL and [HCO₃⁻]_BL per se.⁴

PCR data show that mouse kidney expresses two splice variants (designated A and B) of the Ptprg gene (Supplemental Figure 1). We developed a polyclonal antibody to a unique 17-residue sequence in the carbonic-anhydrase–like domain (CALD) of RPTPγ. Western blotting of mouse whole-kidney lysates shows reactivity at approximately 200 kDa (predicted, unglycosylated molecular mass: approximately 160 kDa) for wild-type (WT) and RPTPζ^−/− mice, but not for RPTPγ^−/− mice (Supplemental Figure 2). Thus, kidneys expressed RPTPγ protein, for which our antibody is specific. Figure 1 shows cryosections of kidney cortex from WT versus RPTPγ^−/− mice. For both, the sodium-potassium adenosine triphosphatase showed the expected reactivity at the PT BL membrane (Figure 1, A and C), including the extensive infoldings that nearly surround a specialized extracellular space into which PT cells deposit reabsorbed fluid before moving into the blood vessels. The RPTPγ antibody (Figure 1, B and D) reacted only with PTs from WT mice, predominantly staining portions of the BL membrane that face blood vessels, not the infoldings (Supplemental Figure 3 shows additional control experiments). Thus, RPTPγ in the PT is ideally situated to monitor blood—not reabsorbate—composition.

Figure 1.

PTs express RPTPγ on the portion of the basalateral membrane that faces blood vessels. (A–D) Immunofluorescent staining of cortical sections of kidneys from WT or RPTPγ^−/− mice, showing predominantly PTs. The sodium-potassium adenosine triphosphatase (Na⁺-K⁺ ATPase) antibody (green, BL membrane marker) stains all regions of the BL membrane in PTs from WT and RPTPγ^−/− mice. Our RPTPγ antibody (red) establishes that RPTPγ protein was present only in the subset of the PT BL membrane facing the blood vessels in WT PTs. Images are representative of data obtained from three mice per group. Scale bar =20 μm.

With the RPTPγ^−/− mouse,¹³ it is possible to explore the role of RPTPγ in acid-base homeostasis. Figure 2 summarizes experiments on isolated, perfused PTs from WT versus RPTPγ^−/− mice. The filled circles in Figure 2A show, for WT tubules, the dependence of J_HCO3 on [CO₂]_BL, as we used OOE technology to fix [HCO₃⁻]_BL and pH_BL. As noted previously for PTs from rabbits⁴ and mice,¹⁴ lowering [CO₂]_BL to <5% (the physiologic value) decreased J_HCO3, whereas raising [CO₂]_BL increased J_HCO3—responses appropriate for stabilizing pH_a during “respiratory”/[CO₂]_a acid-base disturbances. The open circles, representing RPTPγ^−/− tubules, reveal a depressed J_HCO3 at 5% CO₂ and, more strikingly, a complete insensitivity to changes in [CO₂]_BL. Figure 2B shows that for both WT and RPTPγ^−/− tubules, [CO₂]_BL changes did not significantly affect fluid-volume reabsorption (J_V). The WT data further confirm previous results from rabbits⁴ and mice¹⁴ by implying that PTs compensate for J_NaHCO3 changes (Figure 2A) with reciprocal J_NaCl changes; in the intact animal, this compensation would stabilize blood volume and BP.

Figure 2.

Isolated PTs from RPTPγ knockout mice do not respond to changes in either [CO₂]_BL or [HCO₃⁻]_BL. (A) Dependence of J_HCO3, in WT versus RPTPγ^−/− PTs, on BL [CO₂], using OOE solutions to vary [CO₂]_BL at a fixed [HCO₃⁻]_BL of 22 mM and a fixed pH_BL of 7.40 (37°C). (B) Dependence of J_V on [CO₂]_BL in the same tubules as in panel A. (C) Dependence of J_HCO3, in WT versus RPTPγ^−/− PTs, on [HCO₃⁻]_BL, using OOE solutions to vary [HCO₃⁻]_BL at a fixed [CO₂]_BL of 5% and a fixed pH_BL of 7.40. (D) Dependence of J_V on [HCO₃⁻]_BL in the tubules from panel C.

Figure 2, C and D, is analogous to Figure 2, A and B, except here we varied [HCO₃⁻]_BL at a fixed [CO₂]_BL and pH_BL. Figure 2C shows that, for WT tubules, J_HCO3 varied reciprocally with [HCO₃⁻]_BL—the appropriate compensation for “metabolic”/[HCO₃⁻]_a disturbances, extending previous work on rabbit PTs.⁴ However, RPTPγ^−/− tubules were completely insensitive to changes in [HCO₃⁻]_BL. Previous work with rabbit⁴ PTs showed that J_V is insensitive to changes in [HCO₃⁻]_BL, again implying that J_NaHCO3 and J_NaCl vary reciprocally. Figure 2D confirms this result for PTs from WT mice.

Because the knockout of RPTPγ eliminates the ability of PTs to respond to acute changes in [HCO₃⁻]_BL, we hypothesized that RPTPγ^−/− mice also have a reduced ability to defend pH_a during chronic metabolic acidosis (MAc). Figure 3 summarizes experiments in which we cannulated the carotid arteries of WT or RPTPγ^−/− mice at day −7 and, after recovery, sequentially sampled arterial blood from conscious mice. On day 0, RPTPγ^−/− mice had a mean pH_a that was not significantly different from that of WT mice in this (Figure 3A) or a previous¹⁵ study. Thus, even though PTs from RPTPγ^−/− mice show a modest decrease in J_HCO3 at 5% CO₂/22 mM HCO₃⁻ (Figure 2, A and C), RPTPγ^−/− mice maintained a normal pH_a, perhaps because of PT ammoniagenesis and modest compensation by nephron segments downstream of the PT.

Figure 3.

RPTPγ knockout mice cannot appropriately defend against chronic metabolic acidosis. Blood was sampled at days 0, 2, and 7. (A) pH_a. (B) P_CO2. (C) Standard bicarbonate concentration (SBC). (D) Oxygen saturation of hemoglobin (SO₂, %). Values are the mean±SEM (n=8–16). *P<0.05 versus genotype-matched control (i.e., WT+MAc versus WT+Ctrl, or RPTPγ^−/−+MAc versus RPTPγ^−/−+Ctrl). ^†P<0.05 versus water-matched WT (RPTPγ^−/−+Ctrl versus WT+Ctrl, or RPTPγ^−/−+MAc versus WT+MAc). Single diamond denotes a statistically significance change from day 0; two diamonds denote significant change from day 2 (within the respective treatment group, P<0.05). Please refer to the Supplemental Material for additional details on time-dependent statistical differences.

Immediately following day 0 arterial sampling, we either continued to provide normal drinking water (control) or switched to water containing 1% ammonium chloride (wt/vol) to elicit MAc.¹⁶ On control water, both WT and RPTPγ^−/− mice maintained a stable pH_a over the 7 days (Figure 3A). However, a slight increase in the WT value on day 2 led to a difference that reached statistical significance, suggesting a modest defect in net acid excretion in RPTPγ^−/− mice. Compared with day 0, on day 2 mice challenged with MAc exhibited substantial pH_a declines, greater in RPTPγ^−/− than WT. Compared with day 2, on day 7, pH_a for WT+MAc mice recovered, whereas RPTPγ^−/−+MAc mice continued to show substantial acidosis. As expected, the calculated standard [HCO₃⁻] (Figure 3B) exhibited a pattern similar to that of pH_a, whereas P_CO2 (Figure 3C) and hemoglobin-oxygen saturation (Figure 3D) underwent less remarkable changes. (Supplemental Table 1 provides all arterial blood gas data, including full statistical analyses, and Supplemental Tables 2 and 3 provide blood chemistry and hematology data). Thus, RPTPγ^−/− mice cannot appropriately defend pH_a or standard [HCO₃⁻] when challenged by MAc.

To further characterize the RPTPγ^−/− phenotype at MAc day 7, we used quantitative PCR to assess a panel of gene targets related to ammoniagenesis and PT hydrogen (H⁺) secretion. Regarding ammoniagenesis, increased PT metabolism of plasma glutamine during MAc would increase ammonium excretion and thereby compensate for the increased acid load.¹⁷ As expected, we observed a trend for MAc to upregulate Snat3, Glud1, and Pck1 in WT tissue (Figure 4). The knockout tended to downregulate ammoniagenesis genes with control mice but produced a far more robust response to MAc. Interestingly, MAc did not upregulate Aqp8, and the knockout markedly reduced Aqp8. Regarding H⁺ secretion, MAc in WT tissue downregulated Slc9a1 and Slc9a3 and had no significant effect on the other components. MAc in RPTPγ^−/− tissue tended to downregulate all genes involved in PT H⁺ secretion.

Figure 4.

RPTPγ gene deletion downregulates ammoniagenesis genes at baseline and exaggerates the response to MAc. The diagram summarizes proteins potentially involved in the adaptive response to 7 days of MAc. Relative gene expression was assessed in kidney tissue by quantitative PCR. Data are presented as fold change of gene expression between the groups indicated. ^§P<0.05 based on relative expression values (n=4–5). αKG, α-ketoglutarate; Gln, glutamine; Glu, glutamate; Mal, malate; Na⁺, sodium; NH₃, ammonia; NH₄⁺, ammonium; OA, oxaloacetate; PEP, phosphoenolpyruvate; TCA, tricarboxylic acid cycle. Columns 2 and 3 of the table present the gene and protein abbreviations, respectively.

Established acid-base sensors¹⁸ include soluble adenylyl cyclase¹⁹ and soluble guanylyl cyclase,^20,21 which sense intracellular [HCO₃⁻]; pH_o-sensitive ion channels^22–25; pH_o-sensitive G-protein–coupled receptors^26–29; and the tyrosine kinase Pyk2, which responds to decreased pH_i.³⁰ However, none can account for the ability of the PT to sense [CO₂]_o and [HCO₃⁻]_o.4 Our work suggests a new class of acid-base sensor, RPTPγ. The CALD domains of RPTPγ⁶ and RPTPζ³¹—homologous to the noncanonical CAs (VIII, X, and XI)—lack the full complement of His residues believed essential for enzymatic activity. Indeed, it is not clear how a CALD could sense CO₂ versus HCO₃⁻ if it were interconverting them. The most straightforward hypothesis is that CO₂ and HCO₃⁻ compete for binding to the CALD of RPTPγ and that CO₂–CALD or HCO₃⁻–CALD promotes dimerization and thus inactivation of this tyrosine phosphatase.³² Recent PT work shows that acid-base disturbances lead to specific changes in tyrosine phosphorylation of ErbB1/2.³³

Previous work provided limited insight into specific signaling roles for RPTPγ, which has been implicated as a tumor-suppressor gene.⁶ The initial characterization of RPTPγ^−/− mice revealed minor behavioral changes,¹³ with more recent work suggesting that RPTPγ knockdown has an antidepressive effect.³⁴ Other studies show that a portion of the CALD surface—distant from regions homologous to the CO₂/HCO₃⁻ binding sites of canonical CAs—interacts with contactins,^35,36 implying a role in cell–cell interactions. The present work suggests a specific signaling role to RPTPγ and reveals robust PT and whole-animal phenotypes consistent with the molecular data. Finally, our work raises the question of whether RPTPζ and the noncanonical CAs (VIII, X, and XI) also serve as CO₂/HCO₃⁻ sensors.

Concise Methods

Immunoblotting and Immunostaining

We generated a novel rabbit polyclonal antibody specific for the CALD of the RPTPγ protein to assess expression by Western blotting and subcellular localization by immunofluorescent staining in mouse renal tissue.

RPTPγ^−/− Mice

We used an established RPTPγ^−/− mouse line,¹³ backcrossed with our WT line.

Perfusion of Isolated Mouse PTs

As described previously,¹⁴ the luminal perfusate always contained 5% CO₂/22 mM HCO₃⁻ plus ¹⁴C-methoxyinulin as a volume marker. We used OOE technology to control [CO₂] and [HCO₃⁻] in the BL solution surrounding the PTs.^4,12,37 We collected the fluid that had passed through the PT lumen and analyzed it for ¹⁴C-methoxyinulin by scintillation counting and for total CO₂ using an enzyme-linked microfluorometry approach and computed J_V and J_HCO3.^14,38

Serial Assessment of Arterial Blood in Mice Challenged with MAc

Mice were implanted with indwelling catheters in the left carotid artery to permit serial sampling. Following 7 days of recovery from surgery, mice were dosed via drinking water with 1% ammonium chloride for 7 days to induce MAc.

Transcription Profile of Targeted Genes

Gene expression was assessed using total RNA isolated from mouse kidneys harvested following the 7-day MAc study. Quantitative PCR was performed at Case Western University’s Gene Expression & Genotyping facility using validated TaqMan assays from a commercial vendor.

Please refer to the Supplemental Material for detailed information on methods.

Disclosures

The research described in this article was completed while Margaret Chandler was an employee at Case Western Reserve University. The opinions expressed in this article are the authors' own and do not reflect the view of the National Institutes of Health, the US Department of Health and Human Services, or the US Government.