Plasmin Activates Epithelial Na+ Channels by Cleaving the γ Subunit*

Proteolytic processing of epithelial sodium channel (ENaC) subunits occurs as channels mature within the biosynthetic pathway. The proteolytic processing events of the α and γ subunits are associated with channel activation. Furin cleaves the α subunit ectodomain at two sites, releasing an inhibitory tract and activating the channel. However, furin cleaves the γ subunit ectodomain only once. A second distal cleavage in the γ subunit induced by other proteases, such as prostasin and elastase, is required to release a second inhibitory tract and further activate the channel. We found that the serine protease plasmin activates ENaC in association with inducing cleavage of the γ subunit at γLys194, a site distal to the furin site. A γK194A mutant prevented both plasmin-dependent activation of ENaC and plasmin-dependent production of a unique 70-kDa carboxyl-terminal γ subunit cleavage fragment. Plasmin-dependent cleavage and activation of ENaC may have a role in extracellular volume expansion in human disorders associated with proteinuria, as filtered plasminogen may be processed by urokinase, released from renal tubular epithelium, to generate active plasmin.

The epithelial sodium channel (ENaC) 3 transports Na ϩ across the apical membrane of principal cells in the aldosterone-sensitive distal nephron (1). Alterations in ENaC activity disrupt Na ϩ balance, leading to changes in both extracellular volume and blood pressure. Expansion of extracellular volume occurs in a variety of clinical disorders, including nephrotic syndrome. The role of ENaC activation in many of the clinical disorders associated with extracellular volume expansion remains to be defined.
ENaC activity in the cells that line the distal nephron depends on the number of channels in the apical membrane and on channel open probability (P o ) (1). ENaC subunits undergo post-translational processing by specific proteases (2)(3)(4)(5)(6)(7)(8)(9). Cleavage of the ␣ and ␥ subunits by proteases has a key role in activating ENaC, presumably by releasing inhibitory domains within the ectodomains of the ␣ and ␥ subunits (3,5,10,11). We have proposed that multiple proteolytic cleavage events lead to a stepwise activation of ENaC, reflected in a stepwise increase in channel P o (3,5,10,12). Channels that lack proteolytic processing have a low P o (3,10,12,13). Channels that have been cleaved solely by furin, where an ␣ subunit inhibitory tract has been released, exhibit an intermediate P o (3,10,12). Furinprocessed channels likely represent the channels that are observed in Xenopus oocytes at a single channel level. Channels that have released both ␣ and ␥ subunit inhibitory tracts exhibit a high P o , as we observed in oocytes co-expressing ENaC and prostasin (3,5). Both non-cleaved channels and furin-processed channels at the plasma membrane provide a reservoir of channels that can be activated by extracellular proteases (3,14).
One potential activator of ENaC is the serine protease plasmin. Although known for its involvement in fibrinolysis, plasmin has been implicated in other processes, including tumor pathogenesis, inflammation, and atherosclerosis (15). Plasmin cleaves after basic residues (Arg and Lys) similar to trypsin, a known activator of ENaC (16,17). Plasminogen, the inactive precursor to plasmin, is recovered at higher concentrations in the urine of individuals with nephrotic syndrome compared with healthy individuals (18,19). Urokinase, an enzyme that converts plasminogen to plasmin, is released by cells lining both the proximal and distal nephron (20 -22). As active plasmin may be present in the distal nephron in the setting of nephrotic syndrome, it is important to determine whether plasmin can cleave and activate ENaC. We have now shown that extracellular plasmin activates ENaC expressed in Xenopus oocytes. Channel activation by plasmin is dependent on cleavage of the ␥ subunit at a site that is in close proximity to defined sites for prostasin-and elastase-dependent cleavage of the ␥ subunit (5,7,8).
Functional Expression in Xenopus Oocytes-ENaC subunits were expressed in Stage V-VI Xenopus laevis oocytes pretreated with 1.5 mg/ml type IV collagenase (25). ENaC cRNAs were injected at a concentration of 2 ng/subunit/oocyte. Electrophysiological measurements were performed at 24 -32 h post injection using two-electrode voltage clamp (TEV) as described previously (25). The bath solution was 100 mM sodium gluconate, 1.54 mM CaCl 2 , 5 mM BaCl 2 , 10 mM HEPES, 10 mM tetraethylammonium chloride, pH 7.4. Selected oocytes were bathed in TEV buffer with bovine plasmin (10 g/ml; Innovative Research, Inc., Novi, MI) for defined time periods. The amiloride-insensitive component of the whole cell current was determined by perfusion with bath solution supplemented with amiloride (100 M). Currents measured at Ϫ60 mV were recorded at base line and at varying times of exposure to plasmin. Oocytes expressing ENaC containing wild-type or mutant ␥ subunits were incubated with or without plasmin (10 g/ml) for 4 min at room temperature and then placed immediately on ice and washed with ice-cold modified Barth's saline. Oocyte surface proteins were biotinylated, and surface V5-tagged ENaC was recovered with streptavidin beads (Pierce) for immunoblotting with anti-V5 antibodies as described previously (5,7).
Detection of the Surface Pool and Total Cellular Pool of ENaC in Madin-Darby Canine Kidney (MDCK) Cells-Type I MDCK cells from Barry Gumbiner (University of Virginia, Charlottesville, VA), grown on plastic wells, were transiently transfected with epitope-tagged ␣␤␥ENaC (␣-Myc, ␤-FLAG, and either HA-␥-V5 or HA-␥RKRK 186 /QQQQ-V5) cDNAs using Lipofectamine 2000 (Invitrogen) as described previously (23). Twenty-four h post-transfection, ENaC was immunoprecipitated from cell extracts with anti-V5 antibodies and Protein G conjugated to Sepharose as previously described (23) prior to incubation for 0, 6, and 18 min with 15 l of purified plasmin (20 g/ml) in phosphate buffered saline (137 mM NaCl, 2.6 mM KCl, 15.2 mM Na 2 HPO 4 , 1.47 mM KH 2 PO 4 , 0.5 mM MgCl 2 , and 0.7 mM CaCl 2 ). The beads were washed, and the immunoprecipitate was subjected to immunoblotting with anti-V5 antibodies. Alternatively, MDCK cells expressing ENaC were treated with 1 ml of 3, 9, or 27 g/ml plasmin in phosphatebuffered saline for 6 min at 37°C and then treated with EZ-Link sulfo-NHS-SS-biotin (Pierce) to biotinylate surface proteins as described previously (23). Cells were solubilized, and ENaC was immunoprecipitated with anti-V5 antibodies and Protein G conjugated to Sepharose as described previously (23). Proteins from washed beads were eluted with 50 l of 1% (w/v) SDS in HEPES-buffered saline (150 mM NaCl, 10 mM HEPES, pH 7.4) by heating to 95°C for 2 min. Samples were diluted with 1 ml of 1% (v/v) Triton X-100 in HEPES-buffered saline, and biotinylated surface ENaC subunits were recovered with 30 l of streptavidin-conjugated beads (Pierce). The beads were washed, eluted into Laemmli sample buffer (Bio-Rad) containing 10% ␤-mercaptoethanol and 2% SDS by heating to 95°C for 3.5 min, and immunoblotted as described previously (23).
Statistical Analysis-Data are presented as the mean Ϯ S.E. The S.E. for base line normalized currents was calculated by dividing the S.E. of the amiloride-sensitive current by the mean current. Unpaired Student's t test was used to determine significance between groups, with p Ͻ 0.05 considered to be statistically significant.

Plasmin Treatment Increases Amiloride-sensitive ENaC Currents and
Cleaves the ␥ Subunit-We and others have previously shown that ENaC expressed in Xenopus oocytes is activated by furin-dependent cleavage of the ␣ and ␥ subunits. Co-expression of ENaC and prostasin in oocytes further activates the channel by inducing cleavage of the ␥ subunit at a second site distal to the furin site (5). ENaC expressed in oocytes can also be activated by treating oocytes with extracellular proteases, such as trypsin or elastase (4,7,8). Neutrophil elastase also induces cleavage of the ␥ subunit at a second distal site (7,8). To determine whether plasmin activates ENaC, whole cell currents in oocytes expressing mouse ␣␤␥ENaC, with or without treatment with 10 g/ml plasmin, were measured by TEV (Fig. 1). A time-dependent increase in whole cell amiloride-sensitive current was observed using TEV within 1 min, and the current continued to increase over 6 min (Fig. 1A), similar to the time course for activating ENaC by external trypsin (27,28). The current increase reached a maximum by 10 min and then remained stable for an additional 20 min (Fig. 1B). In the absence of plasmin treatment, whole cell currents remained relatively stable. Plasmin did not activate currents in non-injected oocytes (data not shown). Exposure of ENaC-expressing oocytes to plasmin (10 g/ml) for 4 min significantly increased amiloride-sensitive ENaC currents by 1.95-fold (p Ͻ 0.005) (Fig. 1C).
MDCK cells expressing epitope-tagged ␣␤␥ENaC (␣-Myc, ␤-FLAG, and HA-␥-V5) were treated with increasing amounts of plasmin (0, 3, 9, and 27 g/ml) for 6 min to determine whether plasmin cleaves ␥ at a site distinct from that cleaved by furin. After plasmin treatment, surface proteins were biotinylated. Following serial precipitation with anti-V5 antibodies and streptavidin-conjugated beads, immunoblots were probed with anti-V5 antibodies (Fig. 2). Both the non-cleaved (93 kDa) and furin-cleaved (75 kDa) forms of ␥ were present at the cell surface in the absence of plasmin exposure. However, an additional band at 70 kDa was observed when cells were treated with increasing concentrations of plasmin. The size of the new fragment was consistent with plasmin cleaving ␥ at a site distal to the furin cleavage site at Arg 143 , and was similar in size to the prostasin-dependent cleavage fragment resulting from cleavage at RKRK 186 (5). Oocytes injected with cRNAs for non-tagged ␣ and ␤ and HA-␥-V5 were analyzed by TEV before and after exposure to plasmin (10 g/ml) for the indicated times. A and B, time course for activation of amiloride-sensitive Na ϩ currents in oocytes treated with external plasmin (10 g/ml) is shown. A, for control oocytes, amiloride-sensitive currents at base line were Ϫ2330 Ϯ 390 nA (mean and S.E., n ϭ 17). For oocytes exposed to 10 g/ml plasmin, amiloride-sensitive currents at base line were Ϫ2020 Ϯ 300 nA (mean and S.E., n ϭ 16 -17). Normalized currents are also shown. *, p Ͻ 0.005, plasmin-treated versus control normalized currents at each time point following the initiation of plasmin treatment. B, for control oocytes, amiloride-sensitive currents at base line were Ϫ1560 Ϯ 310 nA (mean and S.E., n ϭ 8 -10). For oocytes exposed to 10 g/ml plasmin, amiloride-sensitive currents at base line were Ϫ1320 Ϯ 330 nA (mean and S.E., n ϭ 11). Normalized currents are also shown. *, p Ͻ 0.005, plasmin-treated versus control normalized currents at each time point following the initiation of plasmin treatment. C, for control oocytes, amiloride-sensitive currents at base line were Ϫ1700 Ϯ 260 nA (mean and S.E., n ϭ 27). For oocytes exposed to 10 g/ml plasmin, amiloride-sensitive currents at baseline were Ϫ1810 Ϯ 260 nA (mean and S.E., n ϭ 23). Normalized currents are also shown. *, p Ͻ 0.005, plasmintreated versus control normalized currents at each time point following the initiation of plasmin treatment. Treatment of oocytes for 4 min with plasmin (10 g/ml) increased amiloride-sensitive sodium currents by 1.95-fold. Oocytes expressing ␣␤␥ENaC (␣, ␤, and HA-␥-V5) were also treated with plasmin (0 or 10 g/ml) for 4 min to determine whether plasmin cleaves ␥ at a site distinct from that cleaved by furin. After plasmin treatment, surface proteins were biotinylated and precipitated with streptavidin beads, and immunoblots were probed with anti-V5 antibodies. As observed with ENaC expressed in MDCK cells, both the non-cleaved (93 kDa) and furin-cleaved (75 kDa) forms of ␥ were present at the cell surface in the absence of plasmin treatment. The additional band at 70 kDa was observed when oocytes were exposed to plasmin (Fig. 3A, lanes 2 and 3).
Mutation of the Prostasin Cleavage Site Does Not Block Plasmindependent Activation of ENaC or ␥ Cleavage-The prostasin consensus site mutation ␥RKRK 186 /QQQQ blocks both ENaC activation by prostasin when co-expressed in oocytes and prostasin-dependent ␥ cleavage when ENaC and prostasin are co-expressed in MDCK cells (5). In contrast, we observed that ENaC containing ␥RKRK 186 /QQQQ was significantly activated 2.48-fold in oocytes exposed to 10 g/ml plasmin for 4 min (Fig. 4). Plasmin cleavage of the ␥ subunit was also not affected by the ␥RKRK 186 /QQQQ mutation (Figs. 3 and 5). ENaC containing either wild-type ␥ or mutant ␥RKRK 186 / QQQQ was immunoprecipitated from transfected MDCK cells and subsequently treated with plasmin (20 g/ml) for 0, 6, or 18 min. In both cases, immunoblot analysis revealed the noncleaved (93 kDa) and furin-cleaved (75 kDa) forms of ␥ in the absence of plasmin (t ϭ 0), whereas incubation with plasmin for 6 or 18 min revealed the plasmin-dependent 70-kDa fragment.

Plasmin Activation of ENaC and Plasmin Cleavage of ␥ Are
Blocked by the Mutation ␥K194A-The results of a microarraybased proteolytic profiling assay suggested that plasmin has a strong preference for cleaving after Arg or Lys when the preceding residue (at the P2 position) is an aromatic residue Phe or Tyr, in addition to His (i.e. FR, HR, YR, FK, HK, YK) (17). These results are consistent with cleavage of the known and postulated substrates of plasmin (17). The assay also provided data for preferred residues at the P3 position. Several of these preferred tripeptide sequences are present in the ␥ subunit: ␥PYK 106 , ␥IHK 194 , ␥WYK 237 , ␥SFK 370 , and ␥SFK 476 . On the basis of the plasmin-dependent 70-kDa form of ␥, we predicted that plasmin cleaved at ␥IHK 194 , just distal to the prostasin cleavage site at RKRK 186 . To address this possibility, we examined whether plasmin could activate channels with a ␥K194A mutant. Whereas plasmin treatment of oocytes expressing wild-type ENaC significantly activated amiloride-sensitive currents, ENaCs containing a ␥K194A mutant were not activated by plasmin (Fig. 6). Consistent with this observation, we observed that plasmin treatment of oocytes expressing wildtype ENaC, but not ENaC containing the ␥K194A mutant, produced the plasmin-dependent 70-kDa fragment of ␥ when oocytes were exposed to plasmin (10 g/ml) for 4 min (Fig. 3A,  lanes 2 and 3 (wild-type ␥) and lanes 4 and 5 (␥K194A)). The non-cleaved 93-kDa and furin-cleaved 75-kDa forms of ␥ were present when ENaC with either wild-type ␥ or mutant ␥K194A was expressed in oocytes.
We next examined whether plasminogen is present in the urine of rats with proteinuria. Urine was collected from obese, FIGURE 3. Plasmin cleaves at ␥Lys 194 when ENaC is expressed in Xenopus oocytes. A, Xenopus oocytes were injected with cRNA for wild-type ␣ and ␤ and either wild-type ␥ (␥WT) or ␥ with mutation of the prostasin cleavage site (␥RKRK 186 QQQQ) or the plasmin cleavage site predicted at ␥IHK 194 (␥K194A). NA, no addition CRNA. All ␥ constructs contained N-terminal HA and C-terminal V5 epitope tags. The next day, oocytes were treated with plasmin (10 g/ml) for 4 min prior to surface biotinylation. Biotinylated proteins were precipitated with streptavidin beads and eluted for analysis by immunoblotting with anti-V5 antibodies. Appearance of the 70-kDa band (Ͻ) due to plasmin cleavage is blocked by the ␥K194A mutation. B, model of ␥ENaC denoting the cleavage sites determined for furin (␥RKRR 143 ), prostasin (␥RKRK 186 ), pancreatic elastase (␥Ala 195 ), neutrophil elastase (␥Val 198 ), and now, plasmin (␥IHK 194 ) (4,5,7,8). See "Results" for details. Numbers to the right of the gel represent the mobility of Bio-Rad Precision Plus protein standards in kDa on a 10% gel. Oocytes injected with cRNAs for non-tagged ␣ and ␤ and mutant HA-␥RKRK 186 /QQQQ-V5 were analyzed by TEV before and after exposure to plasmin (10 g/ml) for 4 min. For control oocytes, amiloridesensitive base-line currents were Ϫ2410 Ϯ 490 nA (mean and S.E., n ϭ 12). For oocytes exposed to plasmin, amiloride-sensitive currents at base line were Ϫ1740 Ϯ 320 nA (mean and S.E., n ϭ 15). Application of extracellular plasmin (10 g/ml) for 4 min increased mutant ENaC currents by 2.48-fold. *, p Ͻ 0.005, plasmin-treated versus control normalized currents at 4 min. diabetic, and hypertensive ZSF1 rats and lean, nondiabetic, and hypertensive littermate controls. Obese ZSF1 rat urine contained 400 Ϯ 52 mg of protein/kg/day (n ϭ 5) compared with 53 Ϯ 11 mg/kg/day (n ϭ 5) for lean littermate controls (p Ͻ 0.0005). Equal volumes of urine were concentrated and subjected to SDS-PAGE and immunoblotting with an anti-plasminogen antibody. Although plasminogen and plasmin L-chain were readily detected in the urine of obese ZSF1 rats, they were largely absent in urine from lean littermate controls (Fig. 7).

DISCUSSION
Our data show that extracellular plasmin activates ENaC currents in oocytes. Channel activation by plasmin is dependent on cleavage of the ␥ subunit at or near Lys 194 . These observations are consistent with our proposed paradigm for the activation of ENaC by proteases, where activation occurs in a stepwise manner from (i) non-cleaved near silent channels to (ii) furin-processed channels that have an intermediate P o , and potentially to (iii) channels that have been cleaved by furin and a second protease and have a high P o (3,5). We propose that extracellular plasmin activates furin-processed channels by  Oocytes injected with cRNAs for non-tagged ␣ and ␤ and either the HA-␥K194A-V5 mutant or wild-type HA-␥-V5 were analyzed by TEV before and after exposure to plasmin (10 g/ml) for 4 min. Oocytes expressing the ␣␤␥K194A mutant or wild-type ␣␤␥ had stable amiloride-sensitive currents over 4 min of TEV recording in both the absence and presence of 10 g/ml plasmin. For control oocytes expressing the ␣␤␥K194A mutant, amiloride-sensitive currents at base line were Ϫ2130 Ϯ 400 nA (mean and S.E., n ϭ 19). For oocytes exposed to plasmin, amiloride-sensitive currents at base line were Ϫ3380 Ϯ 530 nA (mean and S.E., n ϭ 23). Extracellular plasmin (10 g/ml) for 4 min did not increase ␣␤␥K194A ENaC currents. Treatment of oocytes expressing wild-type ␣␤␥ENaC for 4 min with plasmin (10 g/ml) was used as a positive control. Amiloride-sensitive currents for control oocytes (Ϫplasmin) at baseline were Ϫ2020 Ϯ 360 nA (mean and S.E., n ϭ 18). For oocytes exposed to plasmin, amiloride-sensitive currents at base line were Ϫ2020 Ϯ 340 nA (mean and S.E., n ϭ 17). Extracellular plasmin (10 g/ml) for 4 min did increase wild-type ␣␤␥ENaC currents by 1.8-fold. *, p Ͻ 0.005, plasmin treated versus control normalized currents at 4 min.  5) and obese (lanes 6 -10) rats were separately concentrated 10-fold and subjected to SDS-PAGE under reducing conditions for analysis by immunoblotting (IB) with anti-human plasminogen antibodies (Ab) as described under "Experimental Procedures." Note the plasminogen (ϳ100 kDa) and plasmin L-chain (ϳ37 kDa) in the urine of the obese rats (asterisks). Numbers to the right of the gel indicate the mobility of Bio-Rad Precision Plus protein standards in kDa on a 4 -15% gel.
In addition to activating ENaC, purified plasmin induces cleavage of ENaC expressed at the surface of both oocytes and MDCK cells. As purified plasmin also induces cleavage of immunopurified ENaC (Fig. 5), it is likely that plasmin directly cleaves the ␥ subunit and activates the channel. In addition to cleaving the ␥ subunit, it is possible that plasmin cleaves and activates other proteases that could process the ␥ subunit. Whereas plasmin may cleave the channel at other sites in addition to ␥Lys 194 , the lack of ENaC activation with the ␥K194A mutant suggests that plasmin-dependent cleavage at the site is required for channel activation in oocytes. Whether plasmin cleaves the channel at other sites and whether cleavage at these additional sites would affect channel activity in other cell types are questions that we have not addressed. For example, ␥PYK 106 is a potential plasmin cleavage site. If plasmin cleaves the ␥ subunit at both Lys 106 and Lys 194 , it would potentially release the ␥ inhibitory tract we previously identified (5), providing a means of activating channels that have not been processed by furin.
Plasmin-dependent cleavage and activation of ENaC may be particularly relevant in human disorders associated with proteinuria and extracellular volume expansion. The pathogenesis of renal sodium retention in nephrotic syndrome is unclear. Recent studies suggest that enhanced ENaC activity has an important role in the renal Na ϩ retention that has been observed in nephrotic syndrome (29 -30). Plasminogen and its cleavage products are present in urine from a proteinuric rat model and not detected in urine from lean littermate controls. This observation is consistent with previous reports of differences in urinary plasminogen concentration found between humans with nephrotic syndrome and controls (18,19). In the setting of nephrotic syndrome, plasminogen filtered by the glomerulus could be processed by urokinase released from cells that line the proximal and distal nephron to generate active plasmin in the region where ENaC resides (18 -22), providing a mechanism for ENaC activation and renal Na ϩ retention.
Note Added in Proof-While this manuscript was in press, we learned that Svenningsen and colleagues also have a manuscript in press in J. Am. Soc. Nephrol. demonstrating that plasmin cleaves and activates ENaC (31).