Regulation of the epithelial Na+ channel by paraoxonase-2

Paraoxonase-2 (PON-2) is a membrane-bound lactonase with unique anti-oxidative and anti-atherosclerotic properties. PON-2 shares key structural elements with MEC-6, an endoplasmic reticulum–resident molecular chaperone in Caenorhabditis elegans. MEC-6 modulates the expression of a mechanotransductive ion channel comprising MEC-4 and MEC-10 in touch-receptor neurons. Because pon-2 mRNA resides in multiple rat nephron segments, including the aldosterone-sensitive distal nephron where the epithelial Na+ channel (ENaC) is expressed, we hypothesized that PON-2 would similarly regulate ENaC expression. We observed PON-2 expression in aquaporin 2–positive principal cells of the distal nephron of adult human kidney. PON-2 also co-immunoprecipitated with ENaC when co-expressed in HEK293 cells. When PON-2 was co-expressed with ENaC in Xenopus oocytes, ENaC activity was reduced, reflecting a reduction in ENaC surface expression. MEC-6 also reduced ENaC activity when co-expressed in Xenopus oocytes. The PON-2 inhibitory effect was ENaC-specific, as PON-2 had no effect on functional expression of the renal outer medullary potassium channel. PON-2 did not alter the response of ENaC to extracellular Na+, mechanical shear stress, or α-chymotrypsin–mediated proteolysis, suggesting that PON-2 did not alter the regulation of ENaC by these factors. Together, our data suggest that PON-2 regulates ENaC activity by modulating its intracellular trafficking and surface expression.

The mammalian paraoxonase (PON) 2 family consists of three highly conserved genes (pon-1, pon-2, and pon-3) that encode enzymes with hydrolase and lactonase activities (1). PON-1 and PON-3 are secreted proteins that are expressed primarily in liver (1,2). In contrast, PON-2 is a membrane-bound protein that is ubiquitously expressed in many tissue types and organs, including heart, lung, and kidney (3,4). Among different cell types, PON-2 expression has been found within mitochondria, the perinuclear region, the endoplasmic reticulum (ER), and the plasma membrane (5)(6)(7). PON-2 appears to have a protective role against oxidative stressinduced apoptosis and lipid peroxidation (6 -8). PON-2deficient mice showed increased mitochondrial oxidative stress and inflammatory markers, in association with enhanced dietinduced atherogenesis (5,9). A gene-silencing study suggested a role for PON-2 in reducing blood pressure by a dopamine receptor-dependent inhibition of NADPH oxidase activity, preventing excessive reactive oxygen species production within renal proximal tubules (10). Nonetheless, despite growing interest in the PON family, the physiologic functions of PON-2 remain largely unidentified.
PON-2 shares structural features with MEC-6 (mechanosensory abnormality-6), a type-II transmembrane protein found in the nematode Caenorhabditis elegans (11,12). MEC-6 is required for the worm's gentle touch response, an avoidance behavior mediated by an ion channel complex expressed in C. elegans touch-receptor neurons (13,14). At the core of this mechanotransductive complex are MEC-4 and MEC-10, two members of the epithelial Na ϩ channel (ENaC)/degenerin family of ion channel proteins (15)(16)(17)(18). MEC-6 primarily resides in the ER and is required for the proper folding, assembly, and surface expression of MEC-4 (19). Loss-of-function mutations of MEC-6 abolished both the mechanoreceptor current evoked by external forces as well as the worm's response to gentle touch, probably a result of defects in MEC-4 expression (13,20).
The ENaC/degenerin family encodes a group of structurally related ion channels that participate in a large variety of fundamental physiological processes, such as mechanosensation, locomotion, nociception, fear-related behavior, seizure termination, detection of pheromones, airway fluid clearance, salt sensation, and extracellular Na ϩ homeostasis (18,21). ENaC mediates the rate-limiting step of Na ϩ uptake across the apical membrane of specific epithelia. ENaC-dependent Na ϩ absorption in the kidney plays important roles in regulating extracellular fluid volume and blood pressure as well as serum [K ϩ ] by facilitating renal K ϩ secretion. ENaC-dependent Na ϩ absorption in the airway also plays an important role in regulating airway surface liquid volume and mucociliary clearance (22). Functional ENaC complexes in the kidney consist of three homologous subunits, ␣, ␤, and ␥, which form a trimeric ion channel (23). Each subunit contains a highly organized large extracellular domain, two transmembrane helices, and short cytosolic N and C termini (24 -26). Like many other membrane-spanning proteins, this trimeric channel complex inefficiently assembles and folds in the ER, and thus only a small fraction of newly synthesized ENaC subunits exit the ER as assembled channels and traffic to the plasma membrane (27)(28)(29). Indeed, several molecular chaperones have been implicated in key steps during ENaC biogenesis, including subunit folding and channel assembly within the ER (30,31). Other factors regulate intracellular trafficking, plasma membrane insertion, and retrieval as well as both ER-associated and lysosome-mediated degradation of ENaC (32)(33)(34)(35)(36)(37). Together, this multistage machinery impacts the total number of channels (N) at the cell surface.
To date, it is unknown whether mammalian PON proteins regulate ion channel function. As MEC-6 functions as a molecular chaperone to facilitate the biogenesis of MEC-4/MEC-10 channels, we asked whether PON-2 regulates ENaC expression. We found that PON-2 is expressed in aquaporin 2 (AQP2)positive principal cells of the distal nephron, where ENaC resides. PON-2 co-immunoprecipitated with ENaC subunits when expressed in HEK293 cells, and PON-2 reduces ENaC activity when expressed in oocytes. The PON-2-dependent reduction in ENaC activity is associated with a reduction in the number of Na ϩ channels at the cell surface. Our data suggest that a conserved molecular chaperone has co-evolved with specific members of the ENaC/degenerin family to modulate ion channel expression.

PON-2 is expressed in principal cells of the distal nephron
Deep RNA sequencing revealed that pon-2 transcripts are present in most nephron segments of rat kidney, including collecting ducts (38). The presence of PON-2 protein has been reported in renal proximal tubules (10), but its presence in other nephron segments is less well-determined.
We evaluated the expression of PON-2 in human renal tubules by staining paraformaldehyde-fixed and cryopreserved human kidney sections with a rabbit polyclonal anti-human PON-2 antibody (Fig. 1A). As shown in Fig. 1B, the antibody recognizes two specific bands (ϳ40 kDa) of the endogenous PON-2 in human whole kidney lysate as well as in mouse cortical collecting duct (mCCD c1 ) cells overexpressing human PON-2. The antibody did not recognize endogenous mouse PON-2 in either the mCCD c1 cells or mouse whole-kidney lysate (Fig. 1B), suggesting a high degree of species specificity. The human kidney sections were co-stained for AQP2, a marker for principal cells of the latter aspects of the distal nephron (39). We detected tubule-specific staining of PON-2 in human kidneys. PON-2 is detected in multiple nephron segments, including the AQP2-positive tubules (Fig. 1A). Notably, PON-2 (red) co-localized with AQP2 (green) in principal cells within the vicinity of the apical membrane (Fig. 1A). In addition, PON-2 was also detected in intercalated cells (which were negative for AQP2 staining) of the CCD (Fig. 1A). These results suggest that PON-2 is expressed in the distal nephron, where ENaC resides.

PON-2 interacts with ENaC subunits
AS PON-2 is expressed in principal cells of the distal nephron, we examined whether PON-2 is in a complex with ENaC subunits. Mouse PON-2 and V5 epitope-tagged mouse ENaC subunits were expressed in HEK293 cells. Each ENaC subunit (␣, ␤, or ␥) was individually expressed. Alternatively, the three subunits were co-expressed, with only one subunit bearing an epitope tag. Whole-cell lysates were immunoprecipitated with an anti-V5 antibody, and immunoblots were analyzed for the presence of PON-2 and ENaC subunits. As shown in Fig. 2, three ϳ40-kDa species of mouse PON-2 were detected in V5 precipitates as well as in whole-cell lysates (top). No coimmunoprecipitating proteins were observed in mock-transfected controls or when HEK293 cells were transfected with PON-2 alone. Mouse PON-2 co-immunoprecipitated with each of the three ENaC subunits when all three ENaC subunits A, human kidney sections were co-stained for PON-2 (red) and for AQP2 (green), a marker for principal cells of the aldosterone-sensitive distal nephron. Nuclei were stained with To-Pro3. The co-localization of PON-2 and AQP2 within collecting tubules is indicated by arrows in the overlay image. No PON-2 antibody (Ab) controls are shown with images taken under the same conditions. Experiments were performed with samples from three adult human kidneys. B, PON-2 antibody characterization. We assessed antibody specificity in mCCD c1 cells transiently transfected with human PON-2 and control mCCD c1 cells. This antibody recognized two bands around 40 kDa (the predicted size for PON-2) in mCCD c1 cells expressing human PON-2. Additionally, the antibody recognized endogenous PON-2 protein in the whole human kidney lysate but not mouse kidney lysate. Blot were stripped and blotted for ␤-actin as loading control (denoted with an asterisk). The immunoblots (IB) were performed three times.

PON-2 inhibits ENaC
were present in the channel complex. In contrast, when PON-2 was co-expressed with individual ENaC subunits, the interaction between PON-2 and the ␤ subunit was much weaker than with the ␣ or ␥ subunit of ENaC (top), although the immunoprecipitation of ␤ subunit was as efficient as that of the other two subunits (bottom). The expression of mouse PON-2 and ENaC subunits was detected in whole-cell lysates as expected (Fig. 2). We did not observe cleavage products of ␣ and ␥ subunits, as the experiments were performed under non-reducing conditions, and the cleaved products are linked by a disulfide bond and will migrate with non-cleaved ␣ and ␥ subunits (24 -26).

PON-2-dependent ENaC inhibition does not depend on its lactonase activity
PON-2 has a short cytoplasmic N terminus followed by a single transmembrane domain and a large C-terminal enzymatic domain that resides within the lumen of intracellular secretory organelles or in the extracellular space (6). The resolved structure of rabbit PON-1 resembles a six-bladed ␤-propeller with two calcium atoms in its central tunnel (12). Previous studies have shown that the conserved calcium-bind- Mouse PON-2 and mouse ENaC subunits were co-expressed in HEK293 cells as indicated. A C-terminal V5 epitope-tagged ENaC subunit (␣, ␤, or ␥) was expressed alone (single subunit) or with the other two untagged subunits to produce ␣␤␥ channels. Cell extracts were immunoprecipitated with anti-V5 antibodies. Both the immunoprecipitates (IP) (left) and total cell lysate (right) were subjected to SDS-PAGE and immunoblotting (IB) with either an anti-mouse PON-2 antibody (top) or the V5 antibody (bottom). The experiment was performed under non-reducing conditions (no ␤-mercaptoethanol in the sample buffer). The mobility of the Bio-Rad molecular weight standards (MW) is shown to the right of all blots. Three independent experiments were performed for each condition.

Figure 3. PON-2 inhibits ENaC expression in Xenopus oocytes.
A, representative traces of whole-cell Na ϩ current measured in oocytes expressing mouse ENaC (mENaC; 2 ng of cRNA/subunit) alone or co-expressing mouse PON-2 (ϩmPON-2, 2 ng of cRNA), with the holding potential set at Ϫ100 mV. After reaching a steady current, 10 M amiloride (gray bar) was added to the bath. B, summary of the amiloride-sensitive Na ϩ currents of mouse ENaC expressed alone or in the presence of mouse PON-2. Data were pooled from three batches of oocytes. C, summary of amiloride-sensitive Na ϩ currents of mouse ENaC alone or co-expressed with human PON-2. Data were pooled from five batches of oocytes. D, summary of the BaCl 2 -sensitive K ϩ currents of oocytes expressing ROMK alone or ROMK with mouse PON-2. Data were pooled from three batches of oocytes. E, dose-dependent effect of PON-2 on ENaC activity. Oocytes were injected with mouse ENaC cRNAs alone (2 ng/subunit) or co-injected with the indicated amount of human PON-2 cRNA (0.4, 2, or 10 ng). Whole-cell Na ϩ currents were measured 24 h after injection at Ϫ100 mV. Data were pooled from four batches of oocytes. The average whole-cell Na ϩ current in oocytes expressing ENaC alone was Ϫ2.1 Ϯ 1.8 A (n ϭ 78). Statistical comparisons were analyzed with the unpaired Student's t test (B-D) (***, p Ͻ 0.001) or with one-way ANOVA followed by a Bonferroni's post hoc test (E) (*, p Ͻ 0.05; ** p Ͻ 0.01). The numbers of oocytes (n) assayed are indicated in each panel. Whiskers, 10th and 90th percentiles. ϩ, mean.

PON-2 inhibits ENaC
ing sites within PON-2 (His 114 , His 133 , and Asp 268 ) are essential for its lactonase activity (40). In addition, there are two N-glycosylation sites (Asn 254 and Asn 323 ) that are required for PON-2 maturation and activity (40,41). To identify key sites within PON-2 that mediate its inhibitory effect on ENaC activity, we introduced single mutations at specific sites: H114Q, H133Q, D268A, N254A, and N323A (see Fig. 4A). In our experiments, we also introduced a Cys at Ser 311 , which represents the most common PON-2 polymorphism (42). Whole-cell Na ϩ currents were measured in oocytes expressing mouse ENaC alone or mouse ENaC plus either wild-type mouse PON-2 or an individual PON-2 mutant. The inhibitory effect of PON-2 on ENaC activity was retained with the H133Q, N254A, and S311C mutants (Fig. 4B). ENaC currents in oocytes expressing the H114Q and N323A mutants also appeared to be reduced, although the reductions in current did not reach statistical sig-nificance. In contrast, the PON-2 D268A mutant did not inhibit ENaC activity.
We next examined whether the lack of an inhibitory effect of the D268A mutant was due to impaired interaction between ENaC subunits and the PON-2 mutant. ENaC containing a V5-tagged ␣ subunit was expressed in HEK293 cells with either wild-type mouse PON-2 or the D268A mutant. Both wild-type and D268A PON-2 co-immunoprecipitated ENaC (Fig. 4C), suggesting that the interaction between the channel complex and PON-2 was unaffected by the D268A mutation. We noticed that the expression level of the D268A mutant in HEK293 whole-cell lysates was lower than that of wild-type PON-2 (49 Ϯ 16% of wild-type PON-2, n ϭ 7, p Ͻ 0.001 when expressed alone or 48 Ϯ 12% of wild-type PON-2, n ϭ 3, p Ͻ 0.05 when co-expressed with ENaC; Fig. 4D). These observations suggest that the loss of the inhibitory effect seen with the showing key residues required for enzymatic activity (His 114 , His 133 , and Asp 268 ) and N-glycosylation (Asn 254 and Asn 323 ) as well as a common polymorphism (Ser 311 ). B, Xenopus oocytes were injected with cRNAs encoding mouse ENaC (2 ng/subunit) in the presence or absence of the cRNA (2 ng) of an individual PON-2 construct. Amiloride-sensitive Na ϩ currents measured in individual oocytes were normalized to the average whole-cell Na ϩ current of oocytes expressing mENaC alone from the same batch. The overall average whole-cell Na ϩ current in oocytes expressing mouse ENaC alone was Ϫ3.4 Ϯ 2.7 A (open bar, n ϭ 76). Data were collected from two batches of oocytes for each PON-2 mutant and four batches of oocytes for mENaC alone or mENaC plus WT PON-2. The numbers of oocytes (n) assayed are indicated in parentheses. Whiskers, 10th and 90th percentiles. ϩ, mean. Statistical comparisons were analyzed with one-way ANOVA followed by a Bonferroni post hoc test (*, p Ͻ 0.05; **, p Ͻ 0.01; ***, p Ͻ 0.001). C, HEK293 cells were transfected with mouse PON-2 (WT or D268A) alone or co-transfected with mouse ENaC containing a C-terminal V5 epitope-tagged ␣ subunit and non-tagged ␤ and ␥ subunits, as indicated. Cell extracts were incubated with anti-V5 antibodies, and immunoprecipitates (IP) were immunoblotted (IB) with anti-PON-2 antibodies (top). Whole-cell lysates were immunoblotted with anti-PON-2 antibodies, anti-V5 antibodies, or anti-␤-actin antibodies. SDS-PAGE was performed under non-reducing conditions. The mobility of the Bio-Rad molecular weight standards is shown to the right of the blots. D, PON-2 expression was normalized to the ␤-actin level in each experiment. The expression of the D268A mutant relative to WT PON-2 was assessed in seven or three independent experiments, respectively. Bar, mean. Statistical comparisons were analyzed with an unpaired Student's t test (*, p Ͻ 0.05; ***, p Ͻ 0.001).

PON-2 does not affect ENaC gating in response to extracellular factors
The inhibitory effect of PON-2 on ENaC activity suggests a reduction in channel open probability (P o ) and/or in the number (N) of functional channels at the cell surface. ENaC activity is tightly regulated by a variety of endogenous or external factors that affect channel biogenesis or gating (43,44). Inhibition of ENaC by extracellular Na ϩ and activation of ENaC by laminar shear stress (LSS), a mechanical force, both reflect changes in channel P o (45)(46)(47). To investigate whether PON-2 inhibits ENaC activity by affecting channel gating, we examined the Na ϩ self-inhibition response of mouse ENaC in oocytes in the presence or absence of PON-2 (Fig. 5). A rapid increase in bath [Na ϩ ] from 1 to 110 mM resulted in a 26 Ϯ 5% reduction in the whole-cell steady-state current, relative to the peak current in oocytes expressing ENaC alone (Fig. 5B). We observed a similar degree of current reduction in oocytes co-expressing mouse PON-2 (27 Ϯ 6% (Fig. 5B)). We also examined the LSS response of ENaC by exposing oocytes to flow-mediated shear stress delivered through vertical perfusion (Fig. 6). LSS of 0.12 dynes/ cm 2 evoked a 2.2 Ϯ 0.5-fold increase in whole-cell Na ϩ currents in oocytes expressing ENaC alone (Fig. 6B). The co-expression of PON-2 with ENaC did not change the magnitude of the LSS response (2.0 Ϯ 0.5 increase in Na ϩ currents; Fig. 6B).

ENaC activation by an exogenous protease is unaffected by PON-2
Both the ␣ and ␥ subunits of ENaC are cleaved by proteases (48,49). Proteases, such as trypsin and chymotrypsin activate ENaC by releasing embedded inhibitory tracts, thus increasing channel P O at the cell surface (50 -53). This phenomenon allows for a measure of whether a protein or condition affects channel density (N) or P O at the plasma membrane. ␣-chymotrypsin treatment (2 g/ml for 2 min) increased whole-cell Na ϩ currents by 1.9 Ϯ 0.6-fold in oocytes expressing mouse ENaC (Fig. 7). We observed a similar activating effect of ␣-chymotrypsin in oocytes co-expressing ENaC and PON-2 (1.8 Ϯ 0.9fold increase in whole-cell Na ϩ currents (Fig. 7B)). As ENaC activity was still lower in the presence of PON-2 after ␣-chymotrypsin treatment (52 Ϯ 27% reduction, p Ͻ 0.001), our results suggest that ENaC density at the cell surface in oocytes was reduced by PON-2. In summary, the effects of several factors that regulate ENaC P o are not altered by PON-2.

Discussion
PON-2 shares structural features with MEC-6, an ER-resident chaperone in C. elegans that enhances the biogenesis of MEC-4/MEC-10 channels in worms (19). In this paper, we  PON-2. A, superimposed traces from oocytes expressing mENaC (black) or co-expressing mPON-2 (gray) are shown. At t ϭ 70 s, the bath perfusion with 1 mM Na ϩ (gray bar) was switched to 110 mM Na ϩ to initiate the Na ϩ self-inhibition response. Amiloride was added to the bath at the end of each recording (black bar). B, the percentage of channel inhibition by extracellular Na ϩ was estimated as described under "Experimental procedures." Data were pooled from three experiments using different batches of oocytes. The numbers of oocytes (n) assayed are indicated in parentheses. Whiskers, 10th and 90th percentiles. ϩ, mean. No statistical significance was found between the two groups with an unpaired Student's t test.

PON-2 inhibits ENaC
examined whether PON-2 plays a role in regulating ENaC, an ion channel that is structurally related to the MEC-4/MEC-10 channel.
ENaC subunits are expressed in the latter aspect of the aldosterone-sensitive distal nephron, a key site for the highly regulated reabsorption of filtered Na ϩ as well as tubular K ϩ secretion (55). We found that PON-2 is expressed in AQP2-positive principal cells in the distal nephron of human kidney. It is also expressed in AQP2-negative intercalated cells of the distal nephron as well as other nephron segments (e.g. AQP2-negative tubules). This finding is not surprising, as pon-2 transcripts were detected in multiple segments of the rat nephron (38), and PON-2 protein expression has been reported in mouse and human proximal tubules (10).  ). B, the magnitude of channel activation by LSS was estimated as described under "Experimental procedures." Data were pooled from three experiments using different batches of oocytes. The number of oocytes (n) assayed is indicated in parentheses. Whiskers, 10th and 90th percentiles; ϩ, mean. No statistical significance was found between the two groups with an unpaired Student's t test.

Figure 7. PON-2 does not affect the magnitude of ENaC activation by
␣-chymotrypsin. A, superimposed traces from oocytes expressing mENaC (black) or co-expressing mPON-2 (gray). ␣-Chymotrypsin (2 g/ml) was added to oocytes for 2 min (gray bar) to reach the peak current, followed by 1-min perfusion with the bath solution. Amiloride was added to the bath at the end of each recording (black bar). B, relative increase in whole-cell Na ϩ currents in response to ␣-chymotrypsin treatment. Whole-cell Na ϩ currents were measured prior to (Ϫ) and following (ϩ) ␣-chymotrypsin treatment in individual oocytes. The relative increases in whole-cell Na ϩ currents were estimated by normalizing the basal currents prior to stimulation by ␣-chymotrypsin in the same oocyte. The average basal current (prior to the ␣-chymotrypsin treatment) was 1.9 Ϯ 1.5 A in oocytes expressing mENaC and 1.1 Ϯ 1.3 A in oocytes co-expressing mENaC and mPON-2. Data were pooled from four experiments using different batches of oocytes. The numbers of oocytes (n) assayed are indicated in the graph. Whiskers, 10th and 90th percentiles; ϩ, mean. Statistical comparisons were analyzed with two-way ANOVA followed by a Bonferroni post hoc test (**, p Ͻ 0.01; ***, p Ͻ 0.001).

Figure 8. PON-2 reduces ENaC cell surface expression in Xenopus oocytes.
A, oocytes were injected with 2 ng of cRNA/subunit of mENaC (lacking a FLAG tag), mENaC with a FLAG-tagged ␤ subunit (mENaC ␤FLAG), or mENaC ␤FLAG ϩ mouse PON-2 at a cRNA ratio of 1:1. Surface expression of ENaC was measured 48 h after injection and is shown as the average of relative light units from individual oocytes. Data were pooled from three independent experiments using different batches of oocytes. The numbers of oocytes (n) assayed are indicated in the graph. Whiskers, 10th and 90th percentiles; ϩ, mean. Statistical comparisons were analyzed with one-way ANOVA followed by a Bonferroni post hoc test (***, p Ͻ 0.001). B, Xenopus oocytes were injected with 2 ng of cRNA/subunit of mENaC containing a V5-tagged ␣ subunit alone or co-injected with an equal amount of mouse PON-2 cRNA. An equal number of non-injected oocytes were used as a negative control. The following day, oocytes were homogenized and immunoblotted (IB) with either an anti-V5 antibody or an anti-GAPDH antibody. The mobility of the Bio-Rad molecular weight standards is shown to the right of the immunoblots. C, ENaC ␣ subunit expression (full-length polypeptide ϩ cleavage product) was normalized to the GAPDH level in each experiment. The effect of PON-2 on ␣ subunit expression was assessed in four independent experiments. Bar, mean. No significant changes were observed in ␣ subunit expression when analyzed with an unpaired Student's t test.

PON-2 inhibits ENaC
We found that PON-2 and ENaC subunits are components of a protein complex when co-expressed in HEK293 cells (Fig. 2). Mouse PON-2 co-immunoprecipitated each mouse ENaC subunit when all three ENaC subunits were co-expressed in HEK293 cells. When individual mouse ENaC subunits were expressed in HEK293 cells, the ␣ and ␥ subunits co-immunoprecipitated with mouse PON-2. Weak co-immunoprecipitation was seen with the ␤ subunit. These subunit-specific ENaC/ PON-2 interactions suggest that conformational differences exist among ENaC subunits that depend on whether one or three ENaC subunits are expressed, which in turn impacts interactions within the complex. Functional ENaCs comprise subunits that are 30 -40% identical at the amino acid level. Subunit-specific interactions have also been shown with ENaC and Lhs1/GRP170, an ER-resident Hsp70-like chaperone protein (33). Lhs1/GRP170 -mediated ENaC degradation was specific for the ␣ subunit expressed alone and was blocked when the other two subunits were present in the channel complex (33,56). Nonetheless, our data show that PON-2 interacts with ENaC subunits in a heterologous expression system and raise the possibility that these interactions occur in the distal nephron and perhaps in other epithelia.
MEC-6 functions as a chaperone, increasing MEC-4/MEC-10 -dependent whole-cell Na ϩ currents when expressed in Xenopus oocytes (11,19). Interestingly, PON-2 did not alter MEC-4/MEC-10 currents in oocytes. In contrast, both PON-2 and MEC-6 reduced ENaC currents in oocytes (Figs. 3 and 9). Our observations suggest that the chaperone function of C. elegans and mammalian paraoxonase-like proteins is conserved, although they have distinct effects on different members of the ENaC/degenerin ion channel family.
It is well-known that chaperones exert different effects on functionally related proteins, and their effects may be cell typespecific. For example, whereas Hsc70 enhances ASIC2 expression in vascular smooth muscle cells (57), it reduces ASIC2 expression in glioma cells as well as ENaC expression in Madin-Darby canine kidney cells (35,58). The adaptability of molecular chaperones is well-documented (59). For example, homologous Hsp90␣ and Hsp90␤ have different effects on endothelial nitric oxide signals (60). Moreover, the related heat shock proteins, Hsc70 and Hsp70, exert different effects on ENaC expression (34). Even minor changes in a chaperone substrate (in many cases a single missense mutation) can dramatically affect substrate specificity and chaperone function (61).
It has been shown that PON-2 lactonase activity is impaired by mutating the calcium-binding sites (His 114 , His 133 , and Asp 268 ) or residues that impact protein maturation (Asn 254 and Asn 323 ) (40,41). To determine whether the lactonase activity of PON-2 is required for ENaC regulation, we introduced mutations at these sites in mouse PON-2 (Fig. 4A). The reduction in ENaC activity observed with both wild-type PON-2 and some mutants (H133Q, N254A, and S311C) was significant (Fig. 4B). For other mutants, reductions in ENaC currents were also evident but did not achieve significance (H114Q and N323A). There was only one mutant (D268A) where we observed levels of functional ENaC expression that were similar to levels seen with ENaC alone (i.e. in the absence of PON-2). Whole-cell expression levels of the PON-2 D268A mutant were ϳ50% of wild-type PON-2 (Fig. 4D). The reduced expression levels of this mutant could explain the lack of an inhibitory effect of the D268A mutant on ENaC activity. Assuming that these mutants result in a loss of PON-2 lactonase activity (40,41), our results suggest that lactonase activity is dispensable for PON-2-dependent inhibition of ENaC. ENaCs are exposed to multiple extracellular factors, including ions, proteases, and mechanical forces, which regulate channel activity by altering P o , (44,53). We examined whether PON-2 affects ENaC gating in response to these extracellular stimuli. The relative magnitude of ENaC activation by LSS or chymotrypsin was unaffected by PON-2 (Figs. 6 and 7). Furthermore, the extent of channel inhibition by extracellular Na ϩ was unaffected by PON-2 (Fig. 5). Together, these data suggest that PON-2 does not alter the regulation of ENaC P o by these A, whole-cell Na ϩ currents were measured in oocytes expressing mouse ENaC (2 ng cRNA/subunit) alone or co-expressing MEC-6 (2 ng of cRNA) 24 h after cRNA injection at a holding potential of Ϫ100 mV. Amiloride (10 M) was added to the bath to determine leak currents at the end of each recording. The numbers of oocytes (n) assayed for each group are indicated in parentheses. Whiskers, 10th and 90th percentiles; ϩ, mean. Statistical comparison was analyzed with an unpaired Student's t test (***, p Ͻ 0.001). B, oocytes were injected with cRNAs encoding MEC-4 A713T, MEC-10, and MEC-2 (5 ng/each) alone or with the addition of cRNA (1 ng) encoding either MEC-6 or mouse PON-2. Whole-cell Na ϩ currents were measured 4 -8 days after cRNA injection by clamping oocytes at Ϫ60 mV. At the end of each recording, 50 M benzamil was added to the bath to determine leak currents. The experiment was performed with four individual batches of oocytes. For the latter time points, data were pooled from at least two batches of oocytes. Results were expressed as the mean Ϯ S.D. with the numbers of oocytes (n) assayed indicated in parentheses. Statistical comparisons of benzamil-sensitive whole-cell currents of oocytes injected with MEC-4 A713T, MEC-10, and MEC-2 versus MEC-4 A713T, MEC-10, and MEC-2, and either MEC-6 or mouse PON-2 at each time point was analyzed with one-way ANOVA followed by a Bonferroni post hoc test (***, p Ͻ 0.001).

PON-2 inhibits ENaC
factors. These results are also consistent with our findings that reduced ENaC activity by PON-2 reflects reduced channel density at the cell surface (Fig. 8) and suggest that PON-2 modifies ENaC biogenesis and/or intracellular trafficking.
PON-2 can counteract lipid peroxidation at the plasma membrane (6). Under oxidative stress, endogenous PON-2 is recruited to the plasma membrane and protects the membrane from lipid peroxidation (6). ENaC subunits comprise two membrane-spanning domains and are believed to interact with membrane phospholipids as well as membrane-anchored proteases, which then modulate channel activity (62)(63)(64)(65)(66). We cannot rule out the possibility that PON-2 might modify membrane lipid composition in the setting of oxidative stress and alter ENaC gating.
In summary, we identified PON-2 as a novel regulator of ENaC that acts by altering the number of channels at the cell surface. As ENaC performs critical functions in many epithelia, its regulation by molecular chaperones represents an essential checkpoint for ENaC quality control in health and disease.

Site-directed mutagenesis and in vitro transcription
Human PON-2 was cloned in the pCDNA3.1 vector (a gift from Dr. Srinivasa T. Reddy, UCLA). Mouse Pon-2 in the pCMV6 vector was purchased from OriGene (Rockville, MD). PON-2 mutants tested in the study were generated with a Q5 site-directed mutagenesis kit (New England Biolabs, Ipswich, MA). Nucleotide sequencing was performed to confirm the targeted mutations. cRNAs of mouse and human ENaC ␣, ␤, and ␥ subunits; mouse and human PON-2 constructs; or C. elegans MEC-2, MEC-4 A713T, MEC-6, and MEC-10 cRNAs were synthesized following the manufacturer's protocols with the T3 or T7 mMESSAGE mMACHINE, respectively (Thermo Fisher Scientific). Previously described ENaC subunits with an N-terminal HA epitope tag and a C-terminal V5 epitope tag (48) were used for detecting protein-protein interaction in HEK293 cells.

Oocyte expression and whole-cell current measurement
Oocytes were harvested from Xenopus laevis following a protocol approved by the University of Pittsburgh Institutional Animal Care and Use Committee. Stage V-VI oocytes were injected with 2 ng of cRNA/subunit of mouse ENaC or 1 ng of rat ROMK cRNA. Equal amounts of mouse or human PON-2 (or mutant mouse PON-2) cRNAs were injected for the co-expression assay, unless otherwise noted. The injected oocytes were incubated in modified Barth's saline (MBS: 88 mM NaCl, 1 mM KCl, 2.4 mM NaHCO 3 , 15 mM HEPES, 0.3 mM Ca(NO 3 ) 2 , 0.41 mM CaCl 2 , 0.82 mM MgSO 4 , 10 g/ml sodium penicillin, 10 g/ml streptomycin sulfate, and 100 g/ml gentamycin sulfate, pH 7.4) at 18°C. All experiments were performed at room temperature 24 -48 h following injections. Oocytes were placed in a recording chamber from Warner Instruments (Hamden, CT) and perfused with a solution containing 110 mM NaCl, 2 mM KCl, 1.6 mM CaCl 2 , and 10 mM HEPES, pH adjusted to 7.4 (NaCl-110). Voltage clamp was performed using a Dagan TEV-200 amplifier (Dagan Corp., Minneapolis, MN) and Digi-Data 1440A interface (Molecular Devices, Sunnyvale, CA). Whole-cell Na ϩ currents were measured by clamping oocytes at Ϫ100 mV. To examine the effect of proteases on ENaC activity, oocytes were perfused with NaCl-110 containing 2 g/ml ␣-chymotrypsin (C4879, Sigma) for 2 min. At the end of each recording, oocytes were perfused with 10 M amiloride to determine ENaC-independent leak currents. To measure the PON-2 effect on ROMK activity, we replaced the NaCl in the perfusion solution with 100 mM KCl. Whole-cell K ϩ currents were measured by clamping oocytes at Ϫ100 mV. At the end of each recording, 5 mM BaCl 2 was delivered to the oocyte to determine ROMK-independent leak currents (67). To examine C. elegans degenerin channel expression, oocytes were injected with cRNAs encoding MEC-4 A713T (MEC-4d, degenerin mutant), MEC-10, and MEC-2 (5 ng/each) alone or with the addition of cRNA (1 ng) encoding either MEC-6 or mouse PON-2. 4 -8 days after cRNA injection, Na ϩ currents were measured by clamping oocytes at Ϫ60 mV. At the end of each recording, 50 M benzamil was added to the bath to determine leak currents.

LSS response assay
Oocytes were placed in a recording chamber (20-mm diameter and 6 mm deep) that was constantly perfused with the NaCl-110 solution at a rate of 3.5 ml/min. LSS of 0.12 dynes/ cm 2 was applied by perfusion (1.5 ml/min) through a vertical pipette submerged near the top of the oocyte (68). A final concentration of 5 M benzamil was added to the bath at the end of each experiment to determine the ENaC-independent leak current. The magnitude of the LSS response was expressed as a ratio between the LSS stimulated current (I LSS ) and the basal current (I basal ). When determining I LSS and I basal , whole-cell currents were corrected for the benzamil-insensitive component.

Na ؉ self-inhibition assay
To measure the inhibition of ENaC by extracellular Na ϩ , oocytes were perfused with the NaCl-110 solution for the first 60 s and then with NaCl-1, a low-[Na ϩ ] bath solution (NaCl-1) that contains 1 mM NaCl, 109 mM N-methyl-D-glucamine, 2 mM KCl, 1.6 mM CaCl 2 , and 10 mM HEPES, pH adjusted to 7.4 with HCl for another 40 s. To initiate Na ϩ self-inhibition, the NaCl-1 solution was rapidly replaced by the NaCl-110 solution while whole-cell currents were continuously recorded. At the end of each experiment, the oocyte was perfused with 10 M amiloride prepared in NaCl-110. The maximal inward current immediately after the switch from NaCl-1 to NaCl-110 (I peak ) and the steady-state current (I ss ) measured 40 s after the I peak were obtained to assess the percentage of current loss due to Na ϩ self-inhibition. Whole-cell currents were corrected for the amiloride-insensitive component when determining I peak and I ss .

ENaC surface expression and whole-cell expression in oocytes
A chemiluminescence assay was performed to assess ENaC surface expression in oocytes as described previously (54). Mouse ENaC ␤ subunit with an extracellular FLAG epitope tag (DYKDDDDK) was included in the cRNA mixture. A nontagged wild-type ␤ subunit was used as a control. Surface expression was assayed 2 days after cRNA injection. Briefly,

PON-2 inhibits ENaC
oocytes were initially incubated in MBS supplemented with 1% BSA (MBS-BSA, without antibiotics) for 30 min. Oocytes were then labeled with a monoclonal anti-FLAG M2 antibody (1 g/ml; F3165, Sigma) in MBS-BSA for 2 h on ice. Following extensive wash with MBS-BSA, oocytes were incubated with an HRP-coupled goat anti-mouse IgG (1 g/ml; 115036072, Jackson ImmunoResearch, West Grove, PA) in MBS-BSA for 1 h on ice. To remove excess antibodies, oocytes were extensively washed with MBS-BSA followed by MBS. Only intact oocytes were then transferred into a 96-well plate individually, and 100 l of SuperSignal ELISA Femto maximum sensitivity substrates (Thermo Fisher Scientific) was added to each well. After incubation at room temperature for 1 min, the chemiluminescence of each oocyte was quantified as relative light units with a Glo-Max-Multi Detection System (Promega, Madison, WI).
ENaC whole-cell expression in the presence and absence of mouse PON-2 was assessed with Western blotting. Oocytes were injected with 2 ng of mouse ENaC cRNAs containing a V5-tagged ␣ subunit with or without an equal amount of mouse PON-2 cRNA. 24 h after injection, 15-20 oocytes of each group were lysed in a detergent solution (100 mM NaCl, 40 mM KCl, 1 mM EDTA, 10% glycerol, 1% Nonidet P-40, 0.4% deoxycholate, 20 mM HEPES, pH 7.4) supplemented with protease inhibitor mixture III (Calbiochem). The homogenate was vortexed for 30 s and subsequently centrifuged at 200 ϫ g and then 20,800 ϫ g for 10 min at 4°C. The supernatant (whole-cell lysate) was heated for 5 min at 95°C with an equal amount of Bio-Rad Laemmli sample buffer containing 0.14 M ␤-mercaptoethanol and subjected to SDS-PAGE and immunoblotting with either the anti-V5 antibody (0.2 g/ml; Invitrogen) or anti-GAPDH antibody (0.2 g/ml, ProteinTech, Rosemont, IL). The immunoblots were developed using a chemiluminescence reagent (Pierce TM ECL Western blotting substrate, Thermo Fisher Scientific) and imaged with the Bio-Rad ChemiDoc TM system. Experiments were repeated in four batches of oocytes.

Co-immunoprecipitation of PON-2 and ENaC
HEK293 cells grown on 6-well size dishes were transfected with the indicated amount of plasmids encoding PON-2 and ENaC subunits using Lipofectamine 2000 (Invitrogen). An epitope-tagged ENaC subunit (ϩPON-2) was transfected alone or with non-tagged subunits to express ␣␤␥ ENaC. In each case, only one ENaC subunit had a C-terminal V5 epitope tag. The next day, cells were extracted with 0.25 ml of detergent solution (20 mM HEPES, 100 mM NaCl, 40 mM KCl, 1 mM EDTA, 10% glycerol, 1% Nonidet P-40, 0.4% deoxycholate, pH 7.4) supplemented with protease inhibitor mixture III (Calbiochem) for 10 min on ice. Insoluble material was removed by centrifugation at 14,000 rpm for 7 min at 4°C, and a 5% aliquot of the supernatant was retained as a whole-cell lysate. The supernatant was incubated for 2 h at 4°C with end-over-end mixing with 40 l of anti-V5 antibodies conjugated to beads (Invitrogen). The isolated beads were washed once with HEPES-buffered saline (HBS: 10 mM HEPES, 150 mM NaCl with phosphatase inhibitors, pH 7.4) containing 1% Triton X-100 and then once with HBS before proteins were eluted into 30 l of Laemmli sample buffer (Bio-Rad) by heating at 90°C for 5 min. Both the immunoprecipitated and whole-cell lysate were subjected to immu-noblotting by SDS-PAGE on a Bio-Rad Criterion 4 -15% gel and transferred to Millipore nitrocellulose. The membrane was incubated with a rabbit anti-PON-2 antibody (1 g/ml; ab40969, Abcam) overnight at 4°C, followed by a 1-h incubation with an HRP-conjugated goat anti-rabbit antibody (KPL-474-1506, Gaithersburg, MD). The expression of V5-tagged ENaC subunits was detected in the whole-cell lysate by blotting the membranes with a mouse anti-V5 antibody (0.2 g/ml; R96025, Invitrogen) overnight at 4°C, followed by a 1-h incubation with an HRP-conjugated goat anti-mouse (KPL-474-1806). The immunoblots were developed using a chemiluminescence reagent (Pierce TM ECL Western blotting substrate, Thermo Fisher Scientific) and imaged with the Bio-Rad ChemiDoc TM system. Experiments were performed a minimum of three times.

Kidney tissue immunostaining and confocal microscopy
Adult human kidneys were obtained from fresh cadavers by the Center for Organ Recovery and Education (Pittsburgh, PA) through a protocol approved by the University of Pittsburgh Committee for Oversight of Research and Clinical Training Involving Decedents. Kidney transverse sections (5-10 mm) were fixed in 4% paraformaldehyde overnight at 4°C, followed by three 15-min washes in cold PBS. Fixed tissues were quenched in 200 mM NH 4 Cl for 15 min at 4°C, followed by three 15-min washes with cold PBS, and stored in PBS with 0.02% NaN 3 at 4°C. Fixed kidney wedges were soaked in 30% sucrose overnight at 4°C and frozen by embedding in OCT medium at Ϫ20°C. Kidney tissues were then cut into 4 -5-m sections on a Leica CM1950 cryostat. Tissues were stained using a SDS-mediated antigen retrieval method adapted from Brown et al. (69). Briefly, tissues were rehydrated in PBS and blocked with 1% BSA in PBS for 15 min. Slides were then incubated with a rabbit anti-human PON-2 antibody (8 g/ml; ab183710, Abcam) and a goat anti-AQP2 antibody (0.2 g/ml; sc-9882, Santa Cruz Biotechnology, Inc., Dallas, Texas) for 75 min at room temperature in a humid box. After extensive washes with PBS, slides were then incubated with secondary antibodies (donkey anti-rabbit Cy3 for PON-2 and donkey antigoat FITC for AQP2) for 1 h at room temperature in a humid box. All antibody dilutions were performed in a backgroundreducing reagent (DAKO, Carpinteria, CA). Slides were then washed with PBS and covered with a droplet of To-Pro3 nuclear stain (0.33 M) for 5 min. After washing twice in PBS, the slides were mounted on coverslips with VectaShield medium (Vector Laboratories, Burlingame, CA). Slides were imaged on a Leica TCS SP5 CW-STED confocal imaging system with a ϫ63 glycerol immersion lens with a 1.25 numerical aperture.
The specificity of the anti-PON-2 antibody (ab183710, Abcam) was validated by immunoblotting. mCCD c1 cells were transiently transfected with the human PON-2 cDNA using Lipofectamine 2000 (Invitrogen). 24 h after transfection, cells were lysed directly in Laemmli sample buffer containing 0.14 M ␤-mercaptoethanol. Control non-transfected (mock) mCCD c1 cells were also lysed for immunoblotting. Kidneys from three C57BL/6 mice and three human samples were individually homogenized in the detergent buffer in the presence of protease and phosphatase inhibitors. Both mCCD c1 cell lysates and

PON-2 inhibits ENaC
whole-kidney lysates were subjected to SDS-PAGE under reducing conditions and blotted with the rabbit anti-PON-2 antibody (0.8 g/ml; ab183710, Abcam) overnight at 4°C, followed by a 1-h incubation with an HRP-conjugated goat antirabbit antibody (KPL-474-1506). The blot was then stripped and probed with a mouse anti-␤-actin antibody (0.21 g/ml; A1978, Sigma) overnight at 4°C, followed by a 1-h incubation with an HRP-conjugated goat anti-mouse antibody (KPL-474-1806). Immunoblots were developed using a chemiluminescence reagent (Pierce TM ECL Western blotting substrate, Thermo Fisher Scientific) and imaged with the Bio-Rad ChemiDoc TM system.

Statistical analyses
Two-electrode voltage-clamping experiments were repeated with a minimum of two batches of oocytes obtained from different frogs, and the numbers of replicates are indicated in the figure legends. Data are expressed as the mean Ϯ S.D. throughout. Box-and-whisker diagrams are used to show the distribution of data: median (middle line), mean (cross), 25th to 75th percentile (box), 10th to 90th percentile (whisker). The number of oocytes examined for each group is indicated in the figures. Electrophysiological data were analyzed with Clampfit version 10.5 and plotted with Origin 2015 (OriginLab, Northampton, MA). Statistical comparisons were made using Origin 2015. A p value of Ͻ0.05 was considered statistically significant.