Influence of proteolytic cleavage of ENaC’s γ subunit upon Na⁺ and K⁺ handling

Abstract

The epithelial Na⁺ channel (ENaC) γ subunit is essential for homeostasis of Na⁺, K⁺, and body fluid. Dual γ subunit cleavage before and after a short inhibitory tract allows dissociation of this tract, increasing channel open probability (P_O), in vitro. Cleavage proximal to the tract occurs at a furin recognition sequence (¹⁴³RKRR¹⁴⁶, in the mouse γ subunit). Loss of furin-mediated cleavage prevents in vitro activation of the channel by proteolysis at distal sites. We hypothesized that ¹⁴³RKRR¹⁴⁶ mutation to ¹⁴³QQQQ¹⁴⁶ (γ^Q4) in 129/Sv mice would reduce ENaC P_O, impair flow-stimulated flux of Na⁺ (J_Na) and K⁺ (J_K) in perfused collecting ducts, reduce colonic amiloride-sensitive short-circuit current (I_SC), and impair Na⁺, K⁺, and body fluid homeostasis. Immunoblot of γ^Q4/Q4 mouse kidney lysates confirmed loss of a band consistent in size with the furin-cleaved proteolytic fragment. However, γ^Q4/Q4 male mice on a low Na⁺ diet did not exhibit altered ENaC P_O or flow-induced J_Na, though flow-induced J_K modestly decreased. Colonic amiloride-sensitive I_SC in γ^Q4/Q4 mice was not altered. γ^Q4/Q4 males, but not females, exhibited mildly impaired fluid volume conservation when challenged with a low Na⁺ diet. Blood Na⁺ and K⁺ were unchanged on a regular, low Na⁺, or high K⁺ diet. These findings suggest that biochemical evidence of γ subunit cleavage should not be used in isolation to evaluate ENaC activity. Furthermore, factors independent of γ subunit cleavage modulate channel P_O and the influence of ENaC on Na⁺, K⁺, and fluid volume homeostasis in 129/Sv mice, in vivo.

NEW & NOTEWORTHY The epithelial Na⁺ channel (ENaC) is activated in vitro by post-translational proteolysis. In vivo, low Na⁺ or high K⁺ diets enhance ENaC proteolysis, and proteolysis is hypothesized to contribute to channel activation in these settings. Using a mouse expressing ENaC with disruption of a key proteolytic cleavage site, this study demonstrates that impaired proteolytic activation of ENaC’s γ subunit has little impact upon channel open probability or the ability of mice to adapt to low Na⁺ or high K⁺ diets.

INTRODUCTION

The epithelial Na⁺ channel (ENaC) mediates Na⁺ reabsorption in the aldosterone-sensitive distal nephron, respiratory tree, and distal colon (1–4). This channel is highly Na⁺-selective and is canonically comprised of three distinct transmembrane subunits (α, β, and γ), each of which is essential for fluid and electrolyte handling (1, 2, 5). Post-translational proteolytic cleavage of ENaC modulates its activity. This was first suggested in studies demonstrating that short-circuit Na⁺ currents in toad bladders were inhibited by treatment with a serine protease inhibitor (6). Experimental application of the serine protease, trypsin, was subsequently shown to stimulate ENaC activity in Xenopus oocytes or in renal epithelial cells in culture (7). Evidence that extracellular proteases can enhance distal nephron Na⁺ reabsorption in vivo came from the observation that experimentally applied proteases and protease inhibitors modulate ENaC activity in distal nephrons, as assessed by patch clamp, micropuncture, or microperfusion (8–10). In parallel, numerous studies show that ENaC subunits are processed by proteases in vivo and that this proteolytic processing is regulated by a variety of factors (8, 11, 12).

Both the α and γ subunits of ENaC are subject to proteolytic processing. Each of these two subunits contains a short inhibitory tract that suppresses channel open probability (P_O) and can only be removed after two cleavage events occur: 1) at a site proximal (N-terminal) to the inhibitory tract and 2) at a site distal (C-terminal) to the tract (4, 13). Removal of the α subunit’s inhibitory tract moderately stimulates the channel in vitro, compared with noncleaved channels (14). Removal of the γ subunit’s inhibitory tract further increases channel activity, regardless of α subunit cleavage (as depicted in Fig. 1C) (15, 16). Evidence that removal of the subunit’s inhibitory tract (and not cleavage alone) is the essential event leading to activation of the channel is provided by studies where γ subunit cDNA was modified by deleting the inhibitory tract and mutating a remaining protease site. In this setting, no γ subunit cleavage occurred, but channel activity was markedly enhanced (15). Furthermore, synthetic peptides with sequences corresponding to inhibitory tracts reduced channel activity (4). Because the effects of γ subunit cleavage predominate over α subunit cleavage in vitro, we suggested that a loss of the γ subunit’s inhibitory tract has a dominant role in channel activation (15, 16). Removal of the γ subunit inhibitory tract occurs following furin-mediated cleavage at a site proximal to its inhibitory tract (mouse γ¹⁴⁰RKRR¹⁴³) in combination with cleavage at any one of several sites just distal to the subunit’s inhibitory tract. Extracellular proteases capable of performing this second cleavage event include prostasin [also known as channel activating protease 1 (CAP1)] (7, 17), TMPRSS2 (18), TMPRSS4 (CAP2) (19), neutrophil and pancreatic elastases (20), kallikrein (21), matriptase (CAP3) (22), cathepsins (23, 24), plasmin (25, 26), and urokinase-type plasminogen activator (27).

Figure 1.

Disruption of the mouse ENaC γ subunit furin cleavage site,¹⁴⁰RKRR¹⁴³ (WT), by replacement with ¹⁴⁰QQQQ¹⁴³ (Q4) impairs proteolytic activation of the channel. A: representative two-electrode voltage-clamp current recordings are shown (left) from Xenopus oocytes expressing mouse ENaC α and β subunits and a γ subunit that was either WT or Q4. ENaC was coexpressed with or without the serine protease prostasin. Current amplitudes are shown (right). Na⁺ currents were similar in WT and Q4 ENaC in the absence of prostasin. Coexpression of prostasin with WT, but not Q4, ENaC dramatically increased current amplitude (N = 15 oocytes per group, ****P < 0.0001 by t test). B: trypsin perfusion of WT ENaC expressed in oocytes enhanced amiloride-sensitive currents significantly more than trypsin perfusion of Q4 ENaC. Baseline currents were recorded as in A, and oocytes were perfused with 2 µg/mL trypsin for ∼1 min (until currents stabilized). Amiloride (10 µM) was then added to the bath solution to determine the amiloride-insensitive component of the whole cell current. The percentage increase in the amiloride-sensitive current in response to trypsin was calculated. WT ENaC activation (92 ± 41%, N = 9) was significantly greater than Q4 ENaC activation by trypsin (11 ± 11%, N = 9; ****P < 0.0001 by t test). Error bars represent standard deviation of the mean. C: schematic showing proteolytic activation of ENaC. For clarity, only α and γ subunits are depicted. Cleavage of the α subunit at furin recognition sequences (solid arrowheads) removed an inhibitory tract (not shown), resulting in a channel with intermediate open probability (P_O) (left). Removal of the γ subunit’s inhibitory tract (red hatched marks) requires cleavage at both a furin recognition sequence proximal to the tract and at a site distal to the tract (open arrowheads). Dual cleavage of the γ subunit resulted in loss of the γ subunit inhibitory tract, producing a highly active channel with a high P_O (middle). Disruption of the γ subunit’s furin recognition sequence (red X, indicating the Q4 substitution) abrogated activation of the channel by proteases that cleave distal to the γ subunit inhibitory tract (right). ENaC, epithelial Na⁺ channel; WT, wild type.

Numerous physiologic and pathophysiologic stimuli enhance proteolytic processing of the γ subunit. Body fluid volume depletion occurring in response to dietary Na⁺ deprivation enhances γ subunit cleavage in kidneys from rats and mice (11, 28–35). Increased dietary K⁺ increases the proteolytically processed fraction of the γ subunit in kidneys (36, 37). Aldosterone signaling in response to fluid depletion or hyperkalemia likely contributes to these findings, as aldosterone infusion in rats and mice stimulates γ subunit cleavage (11, 31, 34, 38), and primary aldosteronism in humans is associated with increased urinary protease expression that resolves following adrenalectomy (12). Proteolysis of the γ subunit is also increased in kidneys from animals with experimentally induced proteinuria or humans with proteinuric kidney disease (25, 39–42). This phenomenon may occur as a consequence of leakage of plasma proteases into the tubule through damaged glomeruli. This suggestion is supported by the observation that proteases capable of activating ENaC are found in the urine of individuals with nephrotic syndrome (41), preeclampsia (43), and diabetic nephropathy (44, 45). These observations have led us and others to hypothesize that γ subunit cleavage contributes to Na⁺ retention in the setting of body fluid depletion (9), and proteinuric kidney disease (4, 46–48).

Given the profound effects of ENaC γ subunit proteolysis noted in in vitro studies, we hypothesized that γ subunit proteolytic cleavage with the release of its inhibitory tract will impact extracellular volume and Na⁺ and K⁺ homeostasis. To test this hypothesis, we generated a mouse with genetic resistance to proteolytic removal of the γ subunit inhibitory tract. This was performed by replacing the γ subunit furin cleavage site (¹⁴⁰RKRR¹⁴³, or WT) with residues conferring resistance to furin cleavage (¹⁴⁰QQQQ¹⁴³, or Q4). This mutation successfully prevented activation of mouse ENaC by prostasin in Xenopus oocytes. We confirmed the viability of homozygous Q4 (γ^Q4/Q4) mice and determined biochemically that this mutation prevented proteolytic processing at the furin cleavage site of the γ subunit. We examined the P_O of ENaC using on-cell patch clamp of principal cells in isolated connecting tubule/collecting ducts (CNT/CCDs) and measured transepithelial Na⁺ and K⁺ fluxes in microperfused tubules. We measured whole blood electrolytes including K⁺ and tCO₂, and plasma aldosterone in mice on low Na⁺ and high K⁺ diets. We examined amiloride-sensitive currents in the distal colon. We assessed whole body fluid volume in response to low Na⁺ diet using live animal quantitative magnetic resonance. Surprisingly, the phenotypic differences between the WT and Q4 mice were modest, suggesting that other factors have important roles in regulating ENaC activity.

MATERIALS AND METHODS

Two-Electrode Voltage-Clamp Electrophysiology

cDNA encoding the mouse γ subunit was mutated to replace ¹⁴⁰RKRR¹⁴³ with ¹⁴⁰QQQQ¹⁴³ using the QuickChange II XL mutagenesis kit (Agilent Technologies). Plasmids with cDNA encoding mouse ENaC α, β, and γ [previously generated in the Kleyman laboratory (49)], or encoding mouse prostasin [obtained from Open Biosystems, Inc. and then subcloned into pcDNA3.1(+)], were transcribed using the mMessage mMachine transcription kit (Life Technologies). Resulting cRNA encoding each of the subunits was injected into stage V–VI Xenopus oocytes. Oocytes were incubated in Barth’s saline for 24–48 h to allow expression of the channel. Whole cell currents were recorded using standard two-electrode voltage-clamp techniques as previously described (50, 51). In experiments with trypsin perfusion, Na⁺ currents were measured before, and after, perfusion with 2 µg/mL trypsin.

Generation of Mouse Models

All animal experiments were approved by the Institutional Animal Care and Use Committee at the University of Pittsburgh or at the Icahn School of Medicine at Mt. Sinai, where appropriate. Animals were housed in AAALAC-accredited facilities at respective institutions. Mice in the 129/Sv background (129S2/SvPasCrl; Charles River Laboratories) were genetically modified using CRISPR-Cas9 technology, as previously reported (52). Exon 3 of the Scnn1g locus on mouse chromosome 7 was modified using a Cas9 sgRNA targeting the sequence CCTCGGAAACGCCGGGAAGCAGG, changing amino acids ¹⁴⁰RKRR¹⁴³ in the ENaC γ subunit to ¹⁴⁰QQQQ¹⁴³ and to add a BseY1 restriction endonuclease target sequence ( CCCAGC) to facilitate genotyping of genetically modified animals. Genotyping was performed by polymerase chain reaction-based amplification of the surrounding region of Scnn1g with primers ER61 ( GACTGTGGGACTACCAGCTC) and ER67 ( AAGGGACTGGTCAGGAGACA), resulting in a 498 nucleotide product. Restriction endonuclease digestion with BseY1 resulted in 213 and 285 nucleotide DNA bands in ¹⁴⁰QQQQ¹⁴³, but not littermate control, mice. Mice were housed in facilities with a 12-h/12-h light/dark cycle. Standard mouse chow for breeding and growth was Prolab IsoPro RMH 3000 (0.23% Na^+, 0.94% K; LabDiet). Low Na⁺ diet was Teklad TD.90228 (0.01%–0.02% Na⁺, 0.8% K⁺; Envigo). High K⁺ diet was Teklad TD.09075 (0.3% Na⁺, 5.2% K⁺ as KCl). All mice were provided with reverse osmosis-purified drinking water, ad libitum. Mice were sacrificed under general anesthesia provided by inhaled isoflurane.

Mouse Body Composition Measurements

Body composition of mice was measured in live, unanesthetized 12-wk-old mice by quantitative magnetic resonance using a 100H Body Composition Analyzer (EchoMRI) as previously described (53–55). Baseline body weight and body composition were measured daily for 3 days. On the third day, mice were transitioned to a low Na⁺ diet. Body weight and body composition were measured for 10 days on the low Na⁺ diet. All data shown are normalized to each animal’s average baseline body weight or composition.

Mouse Blood Metabolite Measurement

Blood was collected by intracardiac aspiration at the time of sacrifice from mice 10 to 16 wk of age. Electrolytes were measured immediately in whole blood using an iSTAT analyzer (Abbott). Measurement of plasma aldosterone was accomplished by separating plasma from red blood cells in heparinized plasma-separator tubes, freezing in liquid nitrogen, and storing at –80°C until analysis. Plasma was then thawed on ice, and aldosterone levels were measured using an aldosterone ELISA kit (Enzo).

ENaC γ Subunit Immunoblots

Immunoblots were prepared from kidneys collected at the time of sacrifice from 12-wk-old animals provided with a low Na⁺ diet for 10 days. Kidneys were snap-frozen in liquid nitrogen and stored at –80°C until use. Once thawed, whole kidneys were homogenized in CelLytic MT lysis buffer (Sigma #C3228) at 1:20 weight/volume using a glass homogenizer. Lysates were then centrifuged at 12,000g for 20 min to remove insoluble material. After determining sample protein concentration by BCA assay, 20 µg of protein from each sample was subjected to PNGase F treatment (NEB #P0704) to remove N-linked oligosaccharides prior to SDS-PAGE. This technique allows for greater separation of distinct γ-ENaC cleavage products detectable by immunoblot (35). Proteins were resolved via SDS-PAGE using 4%–15% Tris-glycine gels (Bio-Rad) and subsequently transferred to PVDF membranes. After transferring, membranes were blocked in 5% milk dissolved in Tris-buffered saline with 0.1% Tween-20 (TBST) for 2 h at room temperature, followed by overnight incubation in blocking buffer containing a primary antibody to the C-terminal region of γ-ENaC (StressMarq #SPC-405; 1:1,000 dilution). On the following day, membranes were washed five times in TBST before incubating in blocking buffer containing HRP-conjugated secondary antibody for 1 h at room temperature. After washing again five times with TBST, chemiluminescent substrate (Bio-Rad Clarity, #1705060) was applied and membranes were imaged with a Chemidoc multipurpose imager. Densitometric quantitation of γ-ENaC cleavage product abundance relative to total subunit abundance was determined using ImageJ software.

Connecting Tubule/Collecting Duct Single-Channel Recording

Patch clamp studies were performed on manually dissected tubules from kidneys harvested from 14- to 17-wk-old mice provided with a low Na⁺ diet for 6–7 days. The mid-section of the mouse kidney was sliced into 1 mm sections with a sharp razor, from which the renal cortex was removed and cut into small blocks. The blocks were then treated with 1 mg/mL collagenase type II in Leibovitz’s L-15 Medium (ThermoFisher Scientific Inc.) at 37°C for 30–45 min. The cortex blocks were washed with cold L-15 media and dissected immediately or stored temporarily on ice. Connecting tubules/cortical collecting ducts (CNTs/CCDs) were isolated manually with dissecting needle and tweezers and transferred to a cover glass coated with poly-lysine (Sigma-Aldrich, Inc., St. Louis, MO) for patch clamp studies, as previously described (56). CNTs/CCDs were opened with sharp pipettes under an inverted microscope, and principal cells were identified by morphology. Heat-polished patch pipettes were filled with a pipette solution containing 140 mM LiCl, 2 mM MgCl₂, and 10 mM HEPES at pH 7.40 with a tip resistance of 4–7 MΩ. Cell-attached patch clamp experiments were conducted at room temperature using a PC-ONE Patch Clamp Amplifier (Dagan Corporation, Minneapolis, MN), DigiData 1440 A, and Clampex 10.4 software (Molecular Devices, San Jose, CA). Patches were clamped at multiple voltages to obtain current-voltage (IV) relationships, and long recordings (>5 min) were performed at –60 mV to determine open probability. Data were sampled at 10 kHz and filtered at 1 kHz. Single-channel recording data were analyzed using Clampfit 10.4. Single-channel activity (NP_O) was obtained using the single-channel searching function of Clampfit 10.4 (Molecular Devices), and channel number was determined by visual inspection of the whole recordings. Unitary currents were estimated by cursor measurements and represented by the average of three measurements at a clamping voltage. The slope conductance was estimated from the linear fit of unitary currents and clamping voltages in the range of –20 to –100 mV.

Colonic Epithelium Electrophysiology

Mice were placed on a low Na⁺ diet for 14 days and sacrificed at an age of 10–16 wk for harvesting of colons. Mucosal/submucosal colon preparations were prepared by opening the colon longitudinally along the mesenteric line, pinning the tissue mucosal side down in a dissecting dish, and peeling away the muscularis layers using fine forceps. Tissues were then mounted on 0.3 cm² sliders for use in an Ussing-style recording chamber (Physiologic Instruments) and bathed in Ringer’s solution containing, in mM: 140 Na⁺, 119.8 Cl^–, 25 HCO₃^–, 5.2 K⁺, 1.2 Ca²⁺, 1.2 Mg²⁺, 2.4 HPO₄^2–, 0.4 H₂PO₄^–, and 10 glucose. The solution was warmed to 37°C and maintained at pH 7.41 by gassing with 5% CO₂ balanced with oxygen. Tetrodotoxin (0.5 μM) was added to the serosal bath to inhibit neurogenic secretion. After allowing approximately 20 min for equilibration, short-circuit current (I_SC) was measured using a multichannel voltage clamp/amplifier (VCC MC6; Physiologic Instruments) controlled by computer-operated software (pClamp 10; Molecular Devices). After currents stabilized, ENaC-dependent Na⁺ absorption was measured as the change in I_SC after addition of 100 μM amiloride to the apical bath compartment.

Isolation and In Vitro Microperfusion of CCDs for Measurement of Net Transepithelial Na⁺ and K⁺ Transport

Tubular Na⁺ and K⁺ fluxes (J_Na and J_K, respectively) were measured in isolated, perfused collecting ducts, as previously described (57–59). Mice between 8 and 10 wk old were fed a low Na⁺ diet for 6–7 days before euthanization, at which time kidneys were harvested, sectioned coronally, and placed in chilled Ringer’s solution (containing, in mM: 145 NaCl, 2.5 K₂HPO₄, 2.0 CaCl₂, 1.2 MgSO₄, 4.0 Na⁺ lactate, 1.0 Na⁺ citrate, 6.0 l-alanine, and 5.5 glucose. pH 7. 4). Single collecting ducts (0.3–0.4 mm in length) were manually microdissected and transferred to a temperature and O₂/CO₂-controlled specimen chamber, mounted on concentric glass pipettes, and perfused and bathed at 37°C with Burg’s solution containing, in mM: 120 NaCl, 25 NaHCO₃, 2.5 K₂HPO₄, 2.0 CaCl₂, 1.2 MgSO₄, 4.0 Na⁺ lactate, 1.0 Na₃ citrate, 6.0 l-alanine, and 5.5 d-glucose; pH 7.4. During 45 min of equilibrium and thereafter, the bath solution was suffused with 95% O₂ and 5% CO₂ and continuously replaced at a rate of 10 mL/h using a syringe pump (Razel Scientific). After equilibration, three to four samples of tubular fluid per experimental condition were collected under water saturated light mineral oil by timed filling of a calibrated ∼7 nL volumetric constriction pipette. Each tubule was perfused at low (∼1.0 ± 0.1 nL·min⁻¹·mm⁻¹) and high (∼5.0 ± 0.2 nL·min⁻¹·mm⁻¹) flow rates. The flow rate was varied by adjusting the height of the perfusate reservoir. At the end of each experiment, ouabain (200 mM) was added to the bath to inhibit all active transport, and three additional samples of tubular fluid were obtained for analysis to determine the composition of the solution actually delivered to the lumen of the tubule. Na⁺ and K⁺ concentrations of perfusate and collected tubular fluid were determined by helium glow photometry. The rates of net ion transport (J_X, in pmol·min⁻¹·mm⁻¹) were calculated using previously described standard flux equations (60). As transport measurements were performed in the absence of transepithelial osmotic gradients, water transport was assumed to be zero. Calculated ion fluxes were averaged to obtain a mean rate of ion transport for the CCD at each flow rate.

Statistics

All statistics were performed using GraphPad Prism v. 9.5.0. Outlier testing was performed using the ROUT method (Q = 1%). Pairwise comparisons were performed using a two-sided Student’s t test with α < 0.05 considered significant. For time series analysis (body composition as a function of time and genotype after transition to a low Na⁺ diet), significance was determined by mixed-effects analysis.

RESULTS

To confirm that disruption of the furin cleavage site (¹⁴⁰RKRR¹⁴³, WT) in the mouse ENaC γ subunit impairs proteolytic activation of the channel, we mutated the furin cleavage site (to ¹⁴⁰QQQQ¹⁴³, Q4) in the cDNA encoding the γ subunit and examined the influence of the extracellular protease, prostasin, on ENaC currents in Xenopus oocytes. RNA made from the cDNAs of mouse α, β, and γ subunits was injected into oocytes with or without RNA encoding mouse prostasin (Fig. 1A). Measured currents were confirmed to be ENaC currents by perfusion of the oocyte with the ENaC blocker, amiloride, at a concentration of 10 µM. Amiloride-sensitive current magnitudes were significantly greater in oocytes coexpressing prostasin with WT ENaC than in oocytes expressing ENaC alone. In contrast, currents in oocytes with Q4 ENaC and coexpressing prostasin were no different than in oocytes expressing the Q4 subunit with no prostasin. To examine the impact of the Q4 substitution on activation by another protease, WT and Q4 cDNAs were expressed in oocytes, and resulting currents were measured before and after perfusion with 2 µg/mL trypsin (Fig. 1B). Wild-type ENaC activity was significantly increased by trypsin perfusion, whereas little or no change in current was seen with Q4 ENaC. These data confirmed that substitution of ¹⁴⁰RKRR¹⁴³ with ¹⁴⁰QQQQ¹⁴³ rendered the mouse ENaC channel insensitive to activation by an extracellular serine protease (depicted in Fig. 1C).

Next, we examined how the loss of proteolytic activation of ENaC influences in vivo electrolyte handling. Using a CRISPR-Cas9 strategy in mice, we modified Scnn1g, encoding ENaC’s γ subunit, to reproduce the γ subunit ¹⁴⁰QQQQ¹⁴³ mutation that abrogated proteolytic activation of ENaC by prostasin in Xenopus oocytes (Fig. 2A) Mice were bred into the 129/Sv background. Mice lacking expression of the ENaC γ subunit exhibit perinatal mortality (1). However, there was no evidence of perinatal mortality in γ^Q4/Q4 mutant mice, as crosses of heterozygous parent mice resulted in litters with γ^WT/WT, γ^WT/Q4, and γ^Q4/Q4 pups in ratios consistent with Mendelian inheritance (Table 1). Health and behavior of γ^Q4/Q4 mice were subjectively normal. Weights of adult γ^Q4/Q4 mutant mice resembled γ^WT/WT littermate controls (Table 2). Homozygous male γ^Q4/Q4 mice were fertile.

Figure 2.

γ^Q4/Q4 mice demonstrate impaired γ subunit proteolytic processing. A: DNA and amino acid sequences within the mouse Scnn1g locus and the ENaC γ subunit. Bold lowercase nucleosides represent those modified by CRISPR/Cas9. Underlined nucleosides indicate a silent BseY1 restriction endonuclease sequence introduced to identify genetically modified mice. Red amino acids represent the wild type or disrupted γ subunit furin recognition sequence. B: schematic of the anticipated effects of the Q4 substitution on the γ subunit cleavage fragments as detected by immunoblot using a C-terminal anti-γ subunit antibody. Full-length γ subunit following PNGase F treatment migrates at 72 kDa. Cleavage of the γ subunit at the furin recognition sequence (filled arrowhead) results in a 57-kDa C-terminal fragment. Cleavage distal to the inhibitory tract (open arrowhead) results in a 52-kDa cleavage fragment. Mutation of the furin recognition sequence with Q4 (red X) would be predicted to eliminate the 57-kDa cleavage fragment. C: immunoblot of kidney lysates from γ^Q4/Q4 and γ^WT/WT control mice. Immunoblots of lysates run before (–) or after (+) treatment with PGNase F, used to remove N-linked oligosaccharides from the subunit, improving resolution of the proteolytic cleavage fragments. “>” notes a nonspecific band, predominantly in the kidney from male mice, of unclear significance. D: quantification of 57 kDa band intensity from both males and females, indicative of the furin-cleaved γ subunit, normalized to total subunit abundance, revealed a significant reduction of the 57 kDa band in γ^Q4/Q4 mice (N = 8 γ^WT/WT and 8 γ^Q4/Q4 mice including 4 females and 4 males in per group. ****P < 0.0001 as determined by unpaired t test). Quantification included the nonspecific band that migrated close to the 57-kDa C-terminal furin cleavage fragment. E: quantification of 52 kDa band intensity, normalized to total subunit abundance, revealed a significant increase in cleavage product abundance in kidneys from γ^Q4/Q4 compared with γ^WT/WT mice (N = 8 mice per group. ***P < 0.001 by unpaired t test). Error bars represent standard deviation of the mean.

Table 1.

Genotypes of pups from crosses of heterozygous (γ^WT/Q4) mice

	γWT/WT	γWT/Q4	γQ4/Q4
Number of pups	87	183	89
% Pups per litter	24 ± 15%	53 ± 22%	24 ± 16%

Errors represent standard deviation of the mean. N = 59 litters.

Table 2.

Adult mouse weights

	γWT/WT	γQ4/Q4
Males	25.6 ± 2.7 g (11)	25.9 ± 2.4 g (11)
Females	23.0 ± 2.4 g (20)	21.7 ± 2.3 g (14)

Errors represent standard deviation of the mean. Weights of mice ages 10–16 wk. P = not significant for within-sex comparisons by Student’s t test.

To confirm that the γ subunit ¹⁴⁰RKRR¹⁴³ to ¹⁴⁰QQQQ¹⁴³ mutation reduced production of the γ subunit furin cleavage product, we placed mice on a low (<0.02%) Na⁺ diet for 10 days and harvested kidneys. Dietary Na⁺ depletion enhances proteolytic processing of the γ subunit in an aldosterone-dependent fashion (11, 31). Immunoblots of kidney lysates from γ^WT/WT mice probed with a C-terminal γ subunit antibody exhibited three predominant bands following PNGase F treatment, which separated the γ-ENaC cleavage products (Fig. 2, B and C). The full-length γ subunit migrated with an apparent molecular weight of 72 kDa. A 52-kDa band is consistent with cleavage distal to the channel’s inhibitory tract. A 57-kDa band is consistent with the C-terminal γ subunit furin cleavage product. In lysates from γ^WT/WT kidneys, the furin cleavage product represented 31 ± 5% of the total subunit abundance (N = 8). In lysates from γ^Q4/Q4 kidneys, the furin cleavage product abundance was reduced to 7 ± 3% of total subunit abundance (N = 8, P < 0.0001) (Fig. 2D). The extent of furin cleavage of γ^Q4/Q4 is likely lower, as we observed a nonspecific band (noted with “>”) in males migrating at a slightly higher molecular weight than the γ furin cleavage product following PNGase F treatment, and we included this band in measurements of cleavage product densitometry. This band was present in the absence and presence of PNGase F treatment in the male γ^Q4/Q4 lanes. In contrast, distal cleavage product abundance rose from 33 ± 8% (N = 8) in γ^WT/WT kidneys to 57 ± 14% of total subunit abundance in γ^Q4/Q4 kidneys (N = 8, P < 0.001) (Fig. 2E).

We examined blood electrolytes in γ^WT/WT and γ^Q4/Q4 mice. In mice on a standard (0.23% Na⁺, 0.94% K⁺) diet, we observed no differences in blood Na⁺, K⁺, Cl^–, total carbon dioxide (tCO₂), blood urea nitrogen (BUN), hemoglobin (Hb), or ionized calcium (iCa²⁺) (Table 3). As mice with large reductions in expression of specific ENaC subunits exhibit increased sensitivity to dietary Na⁺-restriction (61), we examined plasma electrolytes and aldosterone levels in response to a low Na⁺ (0.01%–0.02% Na⁺, 0.8% K⁺) diet for 10 days. No differences in plasma electrolytes or aldosterone levels were observed. ENaC activity is largely required for connecting tubule and collecting duct K⁺ secretion, although a component of ENaC independent K⁺ secretion has been observed in rodents on a high K⁺ diet (62). We examined plasma electrolytes in mice on a high K⁺ diet (0.3% Na⁺, 5.2% K⁺ as KCl) for 7 days. With the exception of a higher plasma Cl^– concentration in male γ^Q4/Q4 mice, no differences in plasma electrolytes were observed on this diet.

Table 3.

Mouse blood electrolytes

	Male	Male	Female	Female
	γWT/WT	γQ4/Q4	γWT/WT	γQ4/Q4
Standard diet
Mice (N)	9	7	12	7
Na+	146 ± 1	146 ± 2	145 ± 2	145 ± 1
K+	4.3 ± 0.2	4.4 ± 0.6	4.1 ± 0.6	4.3 ± 0.4
Cl–	113 ± 3	113 ± 2	113 ± 2	113 ± 3
tCO2	23 ± 2	23 ± 2	22 ± 3	23 ± 2
BUN	29 ± 3	31 ± 4	30 ± 5	28 ± 7
Hb	13.9 ± 0.6	13.8 ± 0.9	13.3 ± 1.0	13.6 ± 1.1
iCa2+	1.27 ± 0.08	1.24 ± 0.09	1.30 ± 0.07	1.34 ± 0.06
Low Na+ diet
Mice (N)	7	7	8	7
Na+	146 ± 1	147 ± 2	147 ± 2	148 ± 2
K+	4.7 ± 0.4	4.2 ± 0.6	4.2 ± 0.2	4.7 ± 0.3
Cl–	117 ± 2	116 ± 3	118 ± 2	118 ± 1
tCO2	21 ± 2	21 ± 1	21 ± 2	21 ± 2
BUN	28 ± 5	26 ± 2	25 ± 5	26 ± 4
Hb	14.1 ± 0.3	14.3 ± 0.4	14.1 ± 0.6	14.6 ± 0.9
iCa2+	1.34 ± 0.04	1.31 ± 0.05	1.32 ± 0.05	1.33 ± 0.07
Aldosteronea	787 ± 189	664 ± 151	789 ± 134	745 ± 256
High K+ diet
Mice (N)	10	7	13	14
Na+	147 ± 2	148 ± 1	147 ± 3	146 ± 2
K+	5.5 ± 1.0	6.1 ± 0.9	6.3 ± 0.7	6.6 ± 0.7
Cl–	120 ± 5	124 ± 1b	121 ± 4	121 ± 4
tCO2	21 ± 2.2	19 ± 1.6	19 ± 2.2	20 ± 0.9
BUN	25 ± 3	25 ± 3	25 ± 5	27 ± 6
Hb	13 ± 1	13 ± 1	13 ± 1	13 ± 1
iCa2+	1.32 ± 0.03	1.32 ± 0.08	1.31 ± 0.06	1.34 ± 0.08

Errors represent standard deviation of the mean. BUN, blood urea nitrogen; Hb, hemoglobin; iCa²⁺, ionized calcium; tCO₂, total carbon dioxide. ^aN for aldosterone measurements = 5 per group. ^bP < 0.05 for comparison with same-sex γ^WT/WT littermate controls by Student’s t test.

Mice with reduced or absent expression of ENaC subunits exhibit impaired body fluid retention and increased sensitivity to a low Na⁺ diet (1, 53, 61). We therefore asked whether γ^Q4/Q4 mice exhibit increased body fluid loss on a low Na⁺ diet. Male or female mice underwent measurement of body weight and total body water via quantitative magnetic resonance. Baseline measurements were performed for three consecutive days on a standard Na⁺ diet before mice were transitioned to the low Na⁺ diet. During this lead-in period on a standard Na⁺ diet, male but not female mice exhibited a slight decline in body weight that was not reflective of a decrease in body fluid (Fig. 3). After transition to the low Na⁺ diet, measurements were performed daily for 10 days. Mouse weights declined only slightly in response to the low Na⁺ diet. In males, body weight plateaued by about day 3, and in γ^WT/WT males, the mean normalized daily body weight from days 3 through 10 was 0.98 ± 0.02, as compared with 0.97 ± 0.02 in γ^Q4/Q4 males (P = NS). In females, the mean normalized daily body weight from days 3 through 10 was 0.95 ± 0.04 in γ^WT/WT and 0.94 ± 0.02 in γ^Q4/Q4 mice (P = NS). Mixed-effects modeling did not reveal any genotype-specific difference in weight as a function of time. Normalized total body water from days 3 to 10 in γ^WT/WT males remained stable at 1.01 ± 0.02, compared with 0.98 ± 0.04 in γ^Q4/Q4 males (P = NS). Over this period in γ^WT/WT females, mean daily normalized total body water declined to 0.96 ± 0.02 as compared with 0.99 ± 0.03 in γ^Q4/Q4 females (P = NS). Mixed-effects modeling revealed a significant day-genotype interaction, for male (P < 0.05) but not female (P = NS) mice, suggesting that male, but not female, γ^Q4/Q4 mice lost body water more rapidly than γ^WT/WT controls.

Figure 3.

Changes in body fluid and weight in response to a low Na⁺ diet. Normalized body weights (top row) and normalized body water content (bottom row) are shown in male (left) and female (right) mice given a low Na⁺ diet for the number of days indicated on the x-axis. Days before 0 represent days on standard (0.23% Na⁺) mouse chow. N = 6 male γ^WT/WT, 6 male γ^Q4/Q4, 7 female γ^WT/WT, and 8 female γ^Q4/Q4 mice. In both males and females, the change in body weight in response to a low Na⁺ diet was not significantly different in γ^WT/WT compared with γ^Q4/Q4 mice (day-genotype interaction term P = 0.35 or P = 0.96 for males or females, respectively). In male γ^Q4/Q4 mice, the change in total body water over 10 days on the low Na⁺ diet was greater than in male γ^WT/WT controls (*P = 0.04 for day-genotype interaction term). In females, there was no difference between genotypes in change in total body water (P = 0.75, mixed-effects analysis). For clarity, error bars represent standard error. Day-genotype interaction terms were determined by mixed-effects analysis.

In exogenous expression systems, proteolytic removal of the inhibitory tract in ENaC’s γ subunit activates the channel by increasing channel P_O (15). Removal of the γ subunit’s inhibitory tract requires cleavage at sites both proximal to, and distal to, the inhibitory tract (Fig. 1, A–C). We examined whether disruption of the γ subunit’s furin cleavage site reduces P_O of the channel in vivo. Single-channel patch clamp recording was performed on microdissected CNT/CCDs from γ^WT/WT and γ^Q4/Q4 mice on a low Na⁺ diet. A channel with ∼8 pS conductance was observed with Li⁺ as the charge carrier (Fig. 4A), consistent with ENaC (63). The number of channels (N) was not significantly different in patches from γ^WT/WT (1.8 ± 0.4 channels/patch, N = 5), as compared with γ^Q4/Q4 (2.8 ± 1.9 channels/patch, N = 8) mice (P = NS). Open probability (P_O) in γ^WT/WT (0.05 ± 0.03, N = 5) and γ^Q4/Q4 (0.08 ± 0.05, N = 7) mice was also similar (P = NS). NP_O did not differ in patches from γ^WT/WT (0.10 ± 0.06, N = 5) and γ^Q4/Q4 (0.21 ± 0.24, N = 7) mice (P = NS). We also observed a previously described 20 pS channel of unclear identity (Fig. 4B) (56). This channel is noncation selective and is inhibited by high (50 µM) concentrations of benzamil (56). The number of 20 pS channels per patch was similar in γ^WT/WT and γ^Q4/Q4 kidneys (2.1 ± 1.5 channels/patch in γ^WT/WT kidneys, N = 16; vs. 2.3 ± 1.4 channels/patch in γ^Q4/Q4 kidneys, N = 8; P = NS), as were the P_O (0.47 ± 0.36, N = 16 in γ^WT/WT kidneys; 0.49 ± 0.33, N = 8 in γ^Q4/Q4 kidneys; P = NS) and NP_O (0.96 ± 0.83 in γ^WT/WT kidneys, N = 16; 1.09 ± 0.80 in γ^Q4/Q4 kidneys, N = 8; P = NS). Thus, surprisingly, patch clamp electrophysiology provided no evidence that in vivo disruption of the furin cleavage site in ENaC’s γ subunit altered channel activity.

Figure 4.

On-cell patch clamp of CNTs/CCDs dissected from male γ^WT/WT and γ^Q4/Q4 mice on a low Na⁺ diet revealed no significant differences in P_O, channels per patch (N), or NP_O. A: representative on-cell current tracings of 8 pS ENaC recorded from dissected tubules from N = 5 γ^WT/WT or N = 7 γ^Q4/Q4 mice. C, closed state; O1 and O2, currents when one or two channels are open, respectively. Pairwise comparisons of NP_O, P_O, and N (bottom) showed no significant differences in γ^WT/WT or γ^Q4/Q4 mice (via t test). B: current tracings of the 20-pS channel. Pairwise comparisons of NP_O, P_O, and N (bottom) showed no significant differences in 16 γ^WT/WT compared with 8 γ^Q4/Q4 mice (via t test). Error bars represent standard deviation of the mean.

We next examined whether collecting ducts from γ^Q4/Q4 mice on a low Na⁺ diet exhibit altered electrolyte transport in isolated, microperfused tubules. Net Na⁺ absorption (denoted as a positive flux, J_Na) is stimulated by increased luminal fluid flow rate (59, 64–66). Therefore, we examined J_Na and net K⁺ secretion (denoted as a negative flux, J_K) at a low flow rate (∼1.0 ± 0.1 nL·min⁻¹ mm⁻¹) and a high flow rate (∼5.0 ± 0.2 nL·min⁻¹·mm⁻¹). We predicted that collecting ducts from γ^Q4/Q4 mice would show a reduced flow-stimulated increase in J_Na as compared with collecting ducts from γ^WT/WT mice. At a low flow rate, J_Na did not differ in collecting ducts from γ^WT/WT (10.2 ± 9.2 pmol·min⁻¹·mm⁻¹, N = 5;) and γ^Q4/Q4 (15.7 ± 4.2 pmol·min⁻¹·mm⁻¹, N = 4, P = NS) mice. Increased flow rate stimulated J_Na to a similar degree in γ^WT/WT (50.4 ± 6.4 pmol·min⁻¹·mm⁻¹, N = 5) and γ^Q4/Q4 (55.9 ± 16.7 pmol·min⁻¹·mm⁻¹, N = 4; P = NS) collecting ducts (Fig. 5, left).

Figure 5.

Measurements of J_Na and J_K in isolated perfused cortical collecting tubules. Tubules were dissected from male mice on a low Na⁺ diet. Luminal flow rates were either low (1 nL/min) or high (5 nL/min). J_Na was not different on the basis of mouse genotype at either low (P = 0.31) or high (P = 0.51) flow rates. At a low fluid flow rate, J_K was similar in γ^Q4/Q4 and γ^WT/WT mice (N = 5 γ^WT/WT and 4 γ^Q4/Q4 mice, P = 0.21). However, at a high flow rate, J_K was significantly higher in tubules from γ^WT/WT mice than from γ^Q4/Q4 mice (N = 5 γ^WT/WT, and 4 γ^Q4/Q4 mice, **P < 0.01). Pairwise comparisons were performed via t test. Error bars represent standard deviation of the mean.

Increased luminal flow also stimulates J_K. Tubular K⁺ secretion is inhibited by ENaC blockade, demonstrating its dependence on ENaC activity (67, 68). We predicted that collecting ducts from γ^Q4/Q4 mice would exhibit reduced stimulation of J_K at high flow rate, compared with γ^WT/WT mice. At low flow, J_K was similar in collecting ducts from γ^WT/WT (−0.3 ± 0.4 pmol·min⁻¹·mm⁻¹, N = 5) and γ^Q4/Q4 (−1.1 ± 1.3 pmol·min⁻¹·mm⁻¹, N = 4; P = NS) mice. High flow increased the magnitude of J_K in collecting ducts from both γ^WT/WT (−6.0 ± 1.1 pmol·min⁻¹·mm⁻¹, N = 5) and γ^Q4/Q4 mice (−3.8 ± 0.6 pmol·min⁻¹·mm⁻¹ collecting ducts, N = 4). The flow-induced increase in K⁺ secretion in collecting ducts from γ^WT/WT mice was significantly greater than in collecting ducts from γ^Q4/Q4 mice (P < 0.01) (Fig. 5, right).

ENaC is expressed in mouse distal colon, and its activity is markedly enhanced when mice are placed on a low Na⁺ diet. Distal colons were harvested from mice on a low Na⁺ diet for 14 days and placed in Ussing chambers bathed in Ringer’s solution. Short-circuit current (I_SC) was monitored before and after the addition of 100 μM amiloride to the apical bath. Representative I_SC recordings are shown in Fig. 6A. Amiloride-sensitive I_SC was similar in colons from γ^WT/WT and γ^Q4/Q4 mice (−120 ± 101 µA/cm² in γ^WT/WT mice, N = 6; –198 ± 58 µA/cm² in γ^Q4/Q4 mice, N = 6; P = NS) (Fig. 6B).

Figure 6.

Amiloride-sensitive currents in colonic epithelium. A: representative short-circuit current tracings from colonic epithelia dissected from male γ^WT/WT and γ^Q4/Q4 mice are shown. B: quantitative summary of amiloride-sensitive current amplitudes showed no significant difference in colonic epithelia from N = 6 γ^WT/WT mice compared with N = 6 γ^Q4/Q4 mice (P = 0.13 via t test). Error bars represent standard deviation of the mean.

DISCUSSION

This study confirmed that interruption of the mouse ENaC γ subunit furin cleavage site prevents activation of ENaC by the serine protease prostasin in channels expressed in Xenopus oocytes, as previously reported (14). Somewhat surprisingly, we also observed a lack of proteolytic activation by the more promiscuous protease, trypsin. These results suggest that the furin site mutation generally suppresses activation of ENaC by γ subunit extracellular proteolysis. This study also demonstrated that γ^Q4/Q4 mice lacking the γ subunit furin cleavage site exhibit loss of the 57-kDa γ subunit C-terminal fragment noted following PNGase F treatment, which corresponds to the C-terminal γ subunit furin cleavage fragment. These results confirmed that the γ subunit in γ^Q4/Q4 mice is resistant to furin-mediated proteolysis. A band migrating slightly slower than 57 kDa was noted in male kidney lysates from both WT (without PNGase F treatment) and γ^Q4/Q4 mice (with and without PNGase F treatment), suggesting that this is a nonspecific band. Although we included this band in our analysis of the 57-kDa band intensity in PNGase F-treated lysates from the γ^Q4/Q4 mice (Fig. 2), analyses of the immunoblots excluding this band suggested there was virtually no γ subunit furin cleavage in kidneys from males and female γ^Q4/Q4 mice. Given these observations and the dramatic difference in proteolytic activation of the channel in vitro, we expected to see large and significant phenotypic differences between WT and γ^Q4/Q4 mice, but this was not what we observed.

The γ^Q4/Q4 mice were viable and appeared healthy. At baseline, these mice exhibited no difference in body weight, body fluid, or plasma electrolytes. A high K⁺ diet uncovered no difference in whole blood electrolyte values, apart from a higher plasma Cl^– in blood from γ^Q4/Q4 males on a high K⁺ diet. On a low Na⁺ diet, no differences in plasma electrolytes or aldosterone levels were evident. Although ENaC is expressed in the colonic epithelium, amiloride-sensitive short-circuit currents in colonic epithelia from γ^Q4/Q4 and γ^WT/WT mice receiving a low Na⁺ diet were similar.

In isolated, microperfused tubules from male γ^Q4/Q4 mice on a low Na⁺ diet, no difference in flow-induced stimulation of J_Na was observed, compared with γ^WT/WT controls. However, tubules from γ^Q4/Q4 mice did exhibit attenuated flow-induced stimulation of J_K. We recently demonstrated that BK channels in intercalated cells mediate flow-induced K⁺ secretion (59). The reduced flow-induced J_K seen in tubules from γ^Q4/Q4 mice suggests an altered lumenal potential, even though differences in J_Na were not observed.

Perhaps the most perplexing findings were that a loss of the γ subunit furin cleavage site did not demonstrably change either activity of the 8 pS channel (measured with Li⁺ as the charge carrier) or J_Na in isolated tubules. In exogenous expression systems, proteolytic processing clearly activates ENaC, increasing P_O without changing unitary conductance (69–71). It is notable that the P_O of the 8 pS channels on a low Na⁺ diet was low. It may reflect our use of 129/Sv as a background strain and is consistent with our other ENaC single-channel recordings in CNTs/CCDs from this mouse strain (56). The low P_O suggests that, in vivo (at least in this mouse strain), factors in addition to the γ subunit inhibitory tract and not present in vitro have a role in activating ENaC (increasing channel P_O) in the setting of a low Na⁺ diet.

Recent evidence showed that impaired ENaC activity due to a γ subunit mutation that prevents its palmitoylation was associated with increased activity of a higher conductance (20 pS), nonselective cation channel that is blocked by a high (50 µM) concentration of benzamil (56). We determined NP_O, N, and NP_O of the 20 pS channel in CNTs/CCDs from WT and γ^Q4/Q4 mice on a low Na⁺ diet. We also detected no differences in the activity of this higher conductance channel. The similar ENaC P_O we observed in CNTs/CCDs from WT and γ^Q4/Q4 mice agrees with the similar levels of amiloride-sensitive Na⁺ currents that we observed in distal colonic epithelia from either γ^Q4/Q4 or WT mice maintained on a low Na⁺ diet.

Despite that lack of a difference in ENaC P_O, we observed a small, but significant, impairment in body fluid conservation in the context of dietary Na⁺ restriction in male γ^Q4/Q4 mice compared with WT. This observation suggests that there may be modest differences in ENaC activity that we were unable to discern in our patch clamp or microperfused tubule studies. Previous studies suggested that factors associated with enhanced proteolytic processing of ENaC subunits also lead to an increase in Na⁺ reabsorption in the aldosterone-sensitive distal nephron. Examples of this include dietary Na⁺-deprivation (11), aldosterone infusion (12), and direct tubular application of proteases (8). Conversely, pharmacologic inhibition of proteolysis in the kidney enhances Na⁺ excretion (72–74). We did not observe impaired body fluid conservation in female γ^Q4/Q4 mice on a low Na⁺ diet.

Previous reports described similar ENaC expression and γ subunit proteolytic processing in male versus female mice (75), and we do not have an explanation for these sex-specific differences. It is possible that other experimental perturbations capable of stimulating ENaC subunit proteolysis, renal Na⁺ retention, and volume expansion (i.e., nephrotic syndrome and aldosterone administration) could result in phenotypic differences between WT and γ^Q4/Q4 mice.

After more than 40 years of studies exploring the influence of proteases on ENaC function and Na⁺ handling, this report represents the first study examining the in vivo importance of furin-mediated proteolysis of ENaC’s γ subunit upon bodily Na⁺ and K⁺ homeostasis. Despite the large effects on ENaC activity that have been described in association with γ subunit proteolytic processing and release of its inhibitory tract in vitro, the effects that we observe in vivo on ENaC activity (i.e., channel P_O), transepithelial Na⁺ transport in CCDs and distal colon, and volume regulation were modest, at best. Our results suggest that other factors have important roles in modulating ENaC activity and compensate for a lack of γ subunit furin processing in the γ^Q4/Q4 mice placed on a low Na⁺ or high K⁺ diet. These findings are consistent with a recent report showing that deletion of the gene encoding the serine protease prostasin, a known activator of ENaC (see Fig. 1), in kidney tubules was not associated with reduced ENaC function (76). Numerous reports in rodents use γ subunit proteolysis as a marker of ENaC activation. Although we agree that assessing the extent of ENaC subunit proteolysis is important, our work suggests that proteolytic processing should not be relied upon in isolation as a primary indicator of in vivo ENaC activity.

DATA AVAILABILITY

Data will be made available upon reasonable request.

GRANTS

This work was supported by National Institutes of Health Grants K08DK110332, R01HL147181, R01DK129285, T32DK061296, P30DK079307, U54DK137329, and UL1TR001857 and by an American Society of Nephrology Carl W. Gottschalk Research Scholar Grant.

DISCLOSURES

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

AUTHOR CONTRIBUTIONS

E.C.R., S.S., S.G., T.R.K., and A.N. conceived and designed research; E.C.R., A.N., S.S., R.C.-G., T.L., A.M., L.Z., A.J., C.B., A.W., Z.K., and S.G. performed experiments; E.C.R., A.N., S.S., R.C.-G., A.J., S.G., and L.M.S. analyzed data; E.C.R., A.N., S.S., R.C.-G., L.M.S., and T.R.K. interpreted results of experiments; E.C.R., A.N., and S.S. prepared figures; E.C.R. drafted manuscript; E.C.R., A.N., R.C.-G., A.K., and T.R.K. edited and revised manuscript; E.C.R., A.N., S.S., R.C.-G., T.L., A.M., L.Z., A.J., C.B., A.W., Z.K., S.G., A.K., L.M.S., and T.R.K. approved final version of manuscript.