|Home | About | Journals | Submit | Contact Us | Français|
With-no-lysine (WNK) kinases are a novel family of protein kinases characterized by an atypical placement of the catalytic lysine. Mutations of 2 family members, WNK1 and WNK4, cause pseudohypoaldosteronism type 2 (PHA2), an autosomal-dominant disease characterized by hypertension and hyperkalemia. WNK1 and WNK4 stimulate clathrin-dependent endocytosis of renal outer medullar potassium 1 (ROMK1), and PHA2-causing mutations of WNK4 increase the endocytosis. How WNKs stimulate endocytosis of ROMK1 and how mutations of WNK4 increase the endocytosis are unknown. Intersectin (ITSN) is a multimodular endocytic scaffold protein. Here we show that WNK1 and WNK4 interacted with ITSN and that the interactions were crucial for stimulation of endocytosis of ROMK1 by WNKs. The stimulation of endocytosis of ROMK1 by WNK1 and WNK4 required specific proline-rich motifs of WNKs, but did not require their kinase activity. WNK4 interacted with ROMK1 as well as with ITSN. Disease-causing WNK4 mutations enhanced interactions of WNK4 with ITSN and ROMK1, leading to increased endocytosis of ROMK1. These results provide a molecular mechanism for stimulation of endocytosis of ROMK1 by WNK kinases.
With-no-lysine (WNK) kinases are a novel family of large serine-threonine protein kinases named for an atypical placement of the catalytic lysine (1). Four family members, WNK1–WNK4, have been identified (1–3). WNK1 spans over 2,100 amino acids and contains an approximately 270–amino acid kinase domain located near the amino terminus (1). WNK2, WNK3, and WNK4 are products of different genes and are 1,200–1,600 amino acids in length (1–3). The 4 WNK kinases share a conserved kinase domain with 85%–90% sequence identity, an autoinhibitory domain, 1–2 coiled-coil domains, and multiple PXXP proline-rich motifs for potential protein-protein interaction (1–3). Little sequence identity exists beyond these conserved domains and motifs.
Pseudohypoaldosteronism type 2 (PHA2), also known as Gordon’s syndrome or familial hyperkalemic hypertension, is an autosomal-dominant disease characterized by hypertension and hyperkalemia (4). Positional cloning has recently identified mutations of WNK1 and WNK4 as causes for PHA2 (3). Mutations of the WNK1 gene in PHA2 are large deletions within the first intron that cause its increased expression. PHA2-causing mutations in the WNK4 gene are missense mutations in the coding sequence outside the protein kinase domain.
WNK1 is widely expressed in multiple spliced forms (1, 5, 6). The full-length WNK1, consisting of over 2,100 amino acids (referred to herein as long WNK1), is ubiquitous (1). A shorter WNK1 transcript encoding a polypeptide lacking the aminoterminal 1–437 amino acids (including most of the kinase domain) of long WNK1 is expressed exclusively in the kidney (referred to herein as kidney-specific WNK1) (5, 6). Mutations in the WNK1 gene in PHA2 cause increased expression of long WNK1 (3). Whether expression of kidney-specific WNK1 is altered in PHA2 is currently unknown.
The discovery of WNK1 and WNK4 gene mutations as causes of PHA2 has led to recent advances in the understanding of the pathogenesis of hypertension and hyperkalemia of PHA2. It was reported that in Xenopus oocytes, WNK4 inhibits the activity of the sodium chloride cotransporter and that PHA2-causing WNK4 mutants fail to inhibit the sodium chloride cotransporter (7, 8). Others reported that in cultured epithelial cells, WNK4 phosphorylates tight-junction proteins claudins 1–4 and regulates the paracellular chloride permeability (9, 10). The paracellular chloride permeability is greater in cells expressing WNK4 mutants than in cells expressing wild-type proteins (9, 10). These results suggest that hypertension in patients with WNK4 mutations is caused by increased NaCl reabsorption through the Na-Cl cotransporter and the paracellular pathway. It has also been reported that expression of long WNK1 abolishes the inhibition of the sodium chloride cotransporter caused by WNK4 (7). Long WNK1 activates serum- and glucocorticoid-inducible kinase, leading to activation of the epithelial Na+ channel (ENaC; ref. 11). Thus, hypertension in PHA2 patients with WNK1 mutations may be caused by increased activity of Na-Cl cotransporter and ENaC.
Wild-type WNK4 inhibits the renal outer medullar potassium 1 (ROMK1) channel, and PHA2-causing WNK4 mutants exhibit increased inhibition of ROMK1 (12). Several recent studies have reported that long WNK1 inhibits ROMK1 (13–15). Interestingly, kidney-specific WNK1 by itself does not inhibit ROMK1, but antagonizes the inhibition of ROMK1 caused by long WNK1 (13, 14). These results suggest that inhibition of K+ secretion through ROMK1 may contribute to hyperkalemia caused by mutations of WNK1 and WNK4 in PHA2. ROMK1 channels undergo clathrin-mediated endocytosis, which plays an important role in the regulation of K+ secretion by dietary K+ intake (16–18). The inhibition of ROMK1 by long WNK1 and WNK4 occurs, at least in part, by increasing endocytosis of the channel (12–15). How long WNK1 and WNK4 mediate endocytosis of ROMK1 and how mutations of WNKs increase the endocytosis are unknown.
Endocytosis via clathrin-coated vesicles (CCVs) involves concerted actions of many proteins (19, 20). Among these, endophilin, amphiphysin, and intersectin (ITSN) contain Src-homology 3 (SH3) domains that can interact with PXXP proline-rich motifs (21). In the present study, we found that long WNK1 and WNK4 specifically interacted with the endocytic scaffold protein ITSN. The interaction required specific proline-rich motifs of WNKs, but not their kinase activity. Knockdown of ITSN prevented the stimulation of endocytosis of ROMK1 by long WNK1 and WNK4. WNK4 interacted with ROMK1 as well as with ITSN. Disease-causing mutations enhance interactions of WNK4 with ITSN and ROMK1, leading to increased endocytosis of ROMK1. These results provide a molecular mechanism for stimulation of endocytosis of ROMK1 by WNK kinases.
The abundance of ROMK1 in the plasma membrane is regulated by constitutive clathrin-mediated endocytosis and degradation (16–18). We measured surface abundance of ROMK1 expressed in HEK293 cells using a biotinylation assay. Coexpression of long WNK1 decreased the steady-state surface abundance of ROMK1 (Figure (Figure1A).1A). The effect of long WNK1 was prevented by coexpression of a dominant-negative dynamin (Figure (Figure1A)1A) or by knocking down clathrin heavy chain using siRNA (data not shown). Long WNK1 is ubiquitous (1). We previously reported that knockdown of endogenous long WNK1 increased whole-cell ROMK1 current density in HEK cells (13). Consistent with this observation, we found that knockdown of endogenous long WNK1 increased the surface abundance of ROMK1 (Figure (Figure1B).1B). Surface ROMK1 was detected only when biotin was present (Figure (Figure1C).1C). In addition, biotin did not label clathrin heavy chain, confirming that the assay measured surface protein, not intracellular protein. Furthermore, long WNK1 decreased whole-cell ROMK1 currents, but had no effect on single-channel conductance or open probability of ROMK1 (Figure (Figure1D).1D). Thus, long WNK1 decreased the surface abundance of ROMK1 by stimulating clathrin-mediated endocytosis.
Long WNK1 is unique in that it has multiple PXXP type proline-rich motifs (1, 2) that can interact with SH3 domains (21). We hypothesized that long WNK1 stimulates clathrin-mediated endocytosis of ROMK1 by interacting with SH3-containing endocytic accessory proteins. We examined interactions between long WNK1 and SH3 domains of ITSN, endophilin, and amphiphysin. In humans, rats and mice, there are 2 ITSN genes, ITSN1 and ITSN2 (22–24). Each ITSN gene gives rise to alternatively spliced long and short forms. The short form of ITSN1 (ITSN1s) is ubiquitous (23, 24) and contains 2 Eps15 homology (EH) domains, a central coiled-coil region, and 5 consecutive SH3 domains (SH3A–SH3E; Figure Figure2A).2A). The long form additionally contains Dbl homology, pleckstrin homology, and C2 domains in the C terminus (22–24). Amino acids 1–491 of long WNK1 are sufficient for regulation of ROMK1 (13). We used purified bacterial glutathione S-transferase (GST) fusion proteins containing ITSN1 SH3A–SH3E and the SH3 domains of amphiphysin and endophilin to pull down myc-tagged WNK11–491 from lysates of transfected HEK cells. We found that ITSN1 SH3C pulled down WNK11–491 (Figure (Figure2B).2B). The interaction between ITSN1 SH3C and WNK11–491 was highly specific, as other SH3 domains of ITSN1 and the SH3 domain of amphiphysin and endophilin failed to pull down WNK11–491.
The importance of interaction with ITSN in regulating the surface abundance of ROMK1 by long WNK1 was further examined by using siRNA to knock down endogenous ITSN. Knockdown of endogenous ITSN1s in HEK cells by siRNA was evident from Western blot analysis using an anti-ITSN antibody (Figure (Figure2C).2C). This antibody recognized endogenous ITSN1s and transfected HA-tagged ITSN1s (Figure (Figure2C).2C). The transfected HA-tagged ITSN1s was identified by an anti-HA antibody. Cotransfection with siRNA for ITSN1 (but not control oligonucleotides) increased the surface abundance of ROMK1 (Figure (Figure2D).2D). Furthermore, knockdown of ITSN1 by siRNA increased basal ROMK1 current (i.e., in the absence of exogenous long WNK1) and prevented the inhibition of ROMK1 by exogenous long WNK1 (Figure (Figure2E).2E). Similarly, overexpression of ITSN1 SH3A–SH3E (25, 26) produced a dominant-negative effect to prevent long WNK1 inhibition of ROMK1 (Figure (Figure2F). 2F).
Rat long WNK1 is 2,126 amino acids long (Figure (Figure3A3A and ref. 1). It contains a kinase domain located near the amino terminus (e.g., amino acids 218–491 of rat long WNK1). To further identify determinants of long WNK1 in the regulation of ROMK1, we compared the effects of amino acids 1–119, amino acids 1–491, amino acids 218–491, and full-length long WNK1. We found that WNK11–119 as well as WNK11–491 and full-length long WNK1 inhibited ROMK1 (Figure (Figure3B).3B). The construct containing the kinase domain of WNK1, WNK1218–491, did not cause inhibition of ROMK1 (Figure (Figure3B).3B). Expression and kinase activity of WNK1218–491 were demonstrated (data not shown). Amino acids 1–119 of long WNK1 contain 3 PXXP proline-rich motifs at P94, P103, and P114 (Figure (Figure3A).3A). We examined the role of these PXXP motifs in the regulation of ROMK1. Compared with wild-type WNK11–491, single mutants of P94A, P103A, and P114A each caused partial inhibition of ROMK1 currents (Figure (Figure3C).3C). The triple PXXP mutant of WNK11–491 failed to inhibit ROMK1 (Figure (Figure3C).3C). Biochemical interactions between ITSN1 SH3C and various long WNK1 constructs were examined in pulldown assays using GST-ITSN-SH3C. We found that WNK11–119 interacted with ITSN SH3C and that the triple PXXP mutant failed to interact with ITSN SH3C (Figure (Figure3D).3D). These results support the idea that interactions between the N-terminal proline-rich motifs and ITSN are crucial for long WNK1 regulation of ROMK1. As kidney-specific WNK1 lacks amino acids 1–119 of long WNK1 (5, 6), our current results also support previous reports that kidney-specific WNK1 by itself has no effects on ROMK1 channels (13, 14).
The above results indicate that amino acids 1–119 of long WNK1 are sufficient for regulating ROMK1 via ITSN and that kinase activity is not required, yet mutation of the catalytic K233 or a conserved D368 in the WNK1 kinase domain abolishes its ability to inhibit ROMK1 (13, 14). In addition, we have previously shown that a construct containing amino acids 1–220 of long WNK1 does not inhibit ROMK1 (13). Together, these results suggest that amino acids 120–220 interfere with the ability of amino acids 1–119 of long WNK1 to regulate ROMK1 and that the kinase domain prevents the interference caused by amino acids 120–220. Indeed, we found that coexpression with WNK1120–220 reversed the effect of WNK11–119 on ROMK1 (Figure (Figure4A).4A). By itself, WNK1120–220 had no effect on ROMK1.
The crystal structure of the kinase domain of WNK1 reveals that K233 and D368 are in close proximity (27). Thus, we hypothesized that K233 and D368 form an ion pair and that the ionic interaction is critical for the kinase domain to prevent the interference of the WNK1-ITSN interaction caused by amino acids 120–220 of long WNK1. We tested this hypothesis by making charge reversal mutations of K233 and D368. Consistent with the notion that K233 and D368 are critical for the catalytic activity of long WNK1, WNK11–491 carrying single mutations K233D and D368K and the double mutation K233D/D368K showed no kinase activity (Figure (Figure4B).4B). However, the double mutant decreased ROMK1 currents (Figure (Figure4C)4C) and interacted with ITSN SH3C (Figure (Figure4D).4D). The effect of the double mutant K233D/D368K on ROMK1 was prevented by cotransfection with a dominant-negative dynamin or with siRNA for clathrin heavy chain (data not shown). As positive controls, we found that neither K233D nor D368K regulated ROMK1 (Figure (Figure4C)4C) or interacted with ITSN (Figure (Figure4D).4D). These results provide compelling evidence that long WNK1 regulates ROMK1 endocytosis via ITSN independent of kinase activity.
WNK4 spans 1,223 amino acids (Figure (Figure5A5A and refs. 2, 3). The kinase domains of WNK1 and WNK4 are conserved (1–3). Other conserved domains include an autoinhibitory domain, 2 coiled-coil domains, multiple PXXP proline-rich motifs, and the highly acidic region, where disease-causing mutations of WNK4 are clustered (Figure (Figure5A).5A). To identify region(s) of WNK4 involved in the regulation of ROMK1, we examined the effects of several N-terminal WNK4 fragments. We found that amino acids 1–584, but not amino acids 1–535 or amino acids 1–444, of WNK4 decreased ROMK1 currents (Figure (Figure5A).5A). The region of amino acids 473–584 decreased ROMK1 currents (Figure (Figure5A).5A). The inhibition of ROMK1 current density by WNK41–584 correlated with decreased surface abundance of the channel (Figure (Figure5B).5B). Experiments using GST pulldown assays revealed that ITSN1 SH3A, SH3B, and SH3C interacted with WNK41–584, whereas even at much higher concentrations, ITSN SH3D and SH3E and the SH3 domains of amphiphysin and endophilin, had no interaction with WNK41–584 (Figure (Figure5C).5C). In contrast to WNK41–584, WNK41–444 did not interact with ITSN (Figure (Figure5D).5D). It has previously been reported that WNK4 interacts with the C terminus of ROMK1 (12). As a positive control for the pulldown assay, we found that WNK41–444 interacted with a GST fusion protein containing the C terminus of ROMK1 (Figure (Figure5D).5D). The functional importance of WNK4-ITSN interaction in the regulation of ROMK1 was demonstrated by the finding that overexpression of ITSN1 SH3A–SH3E prevented WNK4 inhibition of ROMK1 (Figure (Figure5E).5E). Similarly, cotransfection with siRNA for ITSN1 prevented WNK4 inhibition of ROMK1 (Figure (Figure5F). 5F).
The above results indicate that amino acids 473–584 of WNK4 are important for interaction with ITSN and inhibition of ROMK1. This region contains 3 conserved PXXP motifs (which begin at P545, P552, and P555) adjacent to disease-causing mutations E559K, D561A, and Q562E (Figure (Figure6A).6A). Disease-causing mutations increase the inhibition of ROMK1 by WNK4 (12). We hypothesized that PXXP motifs within this region interact with SH3 domains of ITSN and that disease-causing mutations enhance the interaction to increase endocytosis of ROMK1. To test this hypothesis, we first examined the role of PXXP motifs within this region in the regulation of ROMK1. Triple mutations of PXXP motifs (Figure (Figure6A)6A) abolished the decrease of surface abundance of ROMK1 mediated by WNK41–584 (Figure (Figure6B).6B). In addition, the triple PXXP mutant failed to interact with ITSN SH3A, SH3B, or SH3C (Figure (Figure6C). 6C).
We next examined the effects of disease-causing mutations on the surface abundance of ROMK1. We found that disease-causing WNK4 mutants E559K, D561A, and Q562E each decreased the surface abundance of ROMK1 more than wild-type WNK41–584 (Figure (Figure6D).6D). The effects of disease-causing mutations on the surface abundance correlated with the effects on ROMK1 current: E559K decreased ROMK1 current more than did wild-type WNK41–584 (Figure (Figure6E).6E). Similarly, the D561A and Q562E mutants decreased ROMK1 current more than did wild-type WNK41–584 (data not shown). Triple mutations of PXXP motifs abolished the inhibition of ROMK1 by WNK41–584 and the enhanced inhibition caused by E559K (Figure (Figure6E).6E). Furthermore, disease-causing mutant E599K pulled down more of the SH3 domain of ITSN than did wild-type WNK4 (Figure (Figure6F),6F), as did D561A and Q562E (data not shown).
WNK4 interacts with the C terminus of ROMK1 (12). We examined potential roles of the interaction in the regulation of ROMK1. The inward-rectifying K+ 1 (IRK1) channel has approximately 40% amino acid sequence identity to ROMK1 (28). In contrast to ROMK1, IRK1 is not present in kidney and is not known to be regulated by clathrin-mediated endocytosis. WNK41–584 interacted with the hexahistidine-tagged (H6-tagged) C terminus of ROMK1 but not of IRK1 (Figure (Figure7A).7A). WNK41–584 had no effect on IRK1 currents (Figure (Figure7A).7A). However, a shorter WNK4 construct, WNK41–444, interacted with the C terminus of ROMK1 (Figure (Figure5D)5D) despite the lack of ROMK1 inhibition (Figure (Figure5A).5A). Thus, interaction with ROMK1 by itself is not sufficient for WNK4 to inhibit the channel.
We next asked whether interaction with ROMK1 is essential for WNK4 to inhibit the channel. We took advantage of the possibility that WNK41–444 may be used as a dominant-negative inhibitor of ROMK1-WNK4 interaction to examine this question. A 3-fold overexpression of WNK41–444 (relative to the amount used in Figure Figure5A)5A) but without exogenous WNK1 or WNK4 increased ROMK1 currents (Figure (Figure7B).7B). ROMK1 expressed in HEK cells is partially inhibited by endogenous WNKs (13). The cause of the above finding is likely that WNK41–444 reversed the inhibition by endogenous WNKs. Consistent with this idea, overexpression of WNK41–444 reversed the inhibition of ROMK1 caused by exogenous WNK11–491 or WNK41–584 (Figure (Figure7B).7B). Overall, our results suggest that interaction with ROMK1 is necessary — although not alone sufficient — for WNK4 regulation of the channel. We further compared the interaction of ROMK1 with wild-type and disease-causing mutants of WNK4. H6-tagged wild-type WNK41–584, E559K, D561A, and Q562E were used to pull down full-length ROMK1. Each mutant interacted with ROMK1 much more strongly than did wild-type WNK41–584 (Figure (Figure7C).7C). Thus, PHA2-causing mutations of WNK4 increase its interaction with ROMK1 as well as with ITSN.
ROMK1 channels are present in the distal nephron, including the distal convoluted tubule (DCT), the connecting tubule (CNT), and the cortical collecting duct (CCD), and are important for basal K+ secretion in these segments (29). We examined the distribution of ITSN1 in rat kidney by immunofluorescent staining. The staining for ITSN1 by a specific anti-ITSN1 antibody was strong in the distal nephron, including DCT and CCD (Figure (Figure8A,8A, arrowhead and arrow, respectively), but was not apparent in the proximal tubule or glomerulus. No staining was observed without the anti-ITSN1 antibody (Figure (Figure8B).8B). Sodium-calcium exchanger isoform 1 (NCX1) is present in the basolateral membrane of DCT and CNT (30). Aquaporin-2 (AQP2) is present in the apical membrane of CNT and CCD (31). Staining for ITSN was present in NCX1-positive tubules (Figure (Figure8C)8C) and in AQP2-positive tubules (Figure (Figure8D).8D). The patterns of cellular and subcellular staining observed for ITSN1, NCX1, and AQP2 are consistent with the idea that ITSN1 is predominantly distributed intracellularly and near the apical membrane of DCT, CNT, and CCD. WNK1 and WNK4 are also present in the distal nephron (3). These results support the hypothesis that ITSN1 in important for endocytosis of ROMK1 by WNKs.
ITSN1s and the ITSN1 long form are alternative splice variants containing identical EH, coiled-coil, and SH3 domains. Our antibody raised against the SH3 domains recognizes both long and short forms. ITSN1s is ubiquitously expressed whereas the long form is predominantly in neuronal tissues (22–24). Thus, ITSN1s is likely the ITSN1 isoform that is important for regulation of ROMK1 by WNKs in the kidney.
PHA2 is characterized by hypertension, hyperkalemia, and hyperchloremic metabolic acidosis. Recently, several studies have reported that long WNK1 and WNK4 inhibit ROMK1 by stimulating clathrin-mediated endocytosis of the channel (12–15). PHA2-causing mutations of WNK4 increase the inhibition of ROMK1 by WNK4 (12), suggesting that the increased endocytosis of ROMK1 may contribute to hyperkalemia in PHA2. How long WNK1 and WNK4 stimulate clathrin-mediated endocytosis of ROMK1 is unknown. Endocytosis via CCVs requires concerted actions of many accessory proteins (19, 20). ITSN is a multimodular scaffold protein that recruits endocytic proteins to the clathrin-coated pits (22, 32–34). There are 2 ITSN genes, each producing alternatively spliced long and short variants (22–24). Both long and short forms have EH and SH3 domains. The EH domains of ITSN bind epsin, which targets ITSN to the clathrin coat (22, 32). The SH3 domains of ITSN bind synaptojanin and dynamin (22, 32, 33). Dynamin further recruits amphiphysin, endophilin, and syndapin (19, 20). Consistent with the notion that the scaffolding role of ITSN is crucial for endocytosis, flies with mutation of Dap160, the Drosophila homology of ITSN1s, are defective in synaptic vesicle endocytosis (34, 35). The levels of dynamin, endophilin, and synaptojanin at nerve terminals are markedly reduced in mutant flies (34, 35).
In the present study, we found that long WNK1 and WNK4 specifically interacted with the endocytic scaffold protein ITSN1 by binding to its SH3 domains. The interaction with ITSN1 was crucial for the endocytosis of ROMK1 stimulated by long WNK1 and WNK4. Mutations of WNK1 in PHA2 are large deletions of intron-1 that result in increased expression of long WNK1 (3), which would likely lead to increased endocytosis of ROMK1 via ITSN1 and contribute to hyperkalemia. This is likely mediated by ITSN1s, which is ubiquitously expressed (22–24). The long form is predominantly in neuronal tissues. Most WNK4 mutations in PHA2 are missense mutations of amino acids localized immediately distal to the first coiled-coil domain (3). We found that missense mutations of WNK4 enhanced the ability of WNK4 to interact with ITSN1 as well as with ROMK1. WNK4 interacts with ROMK1 (12). We found that the interaction with ROMK1, though not in itself sufficient, was essential for WNK4 stimulation of endocytosis of ROMK1. These results suggest that the increased endocytosis of ROMK1 caused by PHA2 mutations of WNK4 may be due to enhanced interactions of WNK4 with ROMK1 and ITSN1. ROMK1 is targeted to coated pits via a C-terminal NPXY motif (16). Interaction with WNK4 may also help localization of ROMK1 to coated pits.
WNK1 and WNK4 regulate a growing list of proteins. Regulation of some of these proteins, including claudins, synaptotagmin 2, and oxidative stress-response kinase 1 (OSR1) and related kinases, require kinase activity of WNKs (9, 10, 36–38). Regulation of ENaC by long WNK1 is kinase independent (11, 39). With respect to ROMK1, Kahle et al. reported that the regulation by WNK4 is independent of its kinase activity (12). However, mutation of the catalytic lysine or the conserved D368 within the kinase domain of WNK1 prevents its regulation of ROMK1, suggesting that the kinase domain plays some roles (13, 14). In the present study, we confirmed that the kinase domain of long WNK1 was important for regulation of ROMK1. The regulation of ROMK1 by long WNK1, however, was clearly independent of its catalytic activity. The kinase core of long WNK1, like many protein kinases crystallized to date, adopts a bilobal architecture: an N-terminal lobe composed of mainly β strands and a C-terminal lobe of predominantly α helices (27). The interface of the 2 lobes forms the catalytic site. Our results suggest that in long WNK1 an ionic interaction between K233 and D368 in the catalytic site stabilizes the kinase domain fold, allowing the PXXP motifs preceding the kinase domain to interact with ITSN.
Long WNK1 and WNK4 each contain more than 15 PXXP type proline-rich motifs distributed in several clusters along the length of the peptide chain (1–3). Yet we found that only PXXP proline-rich motifs clustered between amino acids 94 and 117 of long WNK1 and between amino acids 545 and 558 of WNK4 are critical for interaction with SH3 domains of ITSN. Interestingly, the PXXP motifs of WNK4 critical for interaction with ITSN were located right in front of the region of 4 amino acids (EADQ, 559–562) where PHA2-causing mutations are clustered. We found that mutations of the PXXP motifs between amino acids 545 and 558 of WNK4 abolished the enhanced interaction with ITSN and the increased endocytosis of ROMK1 elicited by disease-causing mutations of WNK4. Thus, the ability of these PXXP motifs of WNK4 to interact with ITSN SH3 domains was influenced by amino acids 559–562, where disease-causing mutations are clustered. Mutations of amino acids 559–562 enhanced WNK4-ITSN interaction.
Amino acids 559–562 of the disease-causing region of WNK4 are conserved in WNK1 (3). To date, there are no reports of missense mutations in this region of WNK1 in patients of PHA2. Notably, the 3 conserved PXXP motifs located in front of the conserved EADQ amino acids 559–562 of WNK4 are not present in long WNK1. Rather, we found that in long WNK1, the PXXP motifs in the N terminus preceding the kinase domain were critical for interaction with ITSN. The general principle governing context-specific interactions between PXXP motifs and SH3 domains remains elusive.
Regarding context-specific interactions between PXXP motifs and SH3 domains, the SH3 domains of endocytic proteins have different ligand specificities (22, 33) and participate at different steps of CCV formation (24). For example, ITSN SH3A, SH3C, and SH3E interact with dynamin stronger than do SH3B and SH3D (22, 26, 33). ITSN SH3A is required for early events leading to formation of constricted pits, whereas other SH3 domains are preferentially involved in the late events leading to CCV fission (26). We found that long WNK1 interacted with ITSN SH3C, whereas WNK4 interacted with SH3A, SH3B, and SH3C, raising the possibility that WNKs may be important in coordinating the recruitment of endocytic proteins by ITSN at different stages of CCV formation. The differential interactions with ITSN may also underlie the synergism between long WNK1 and WNK4 stimulating endocytosis of ROMK1 (13).
Long WNK1 is ubiquitous (1). WNK4 also has broad tissue distribution (40). A recent siRNA-based genome-wide screen using virus entry as a measure of clathrin-mediated endocytosis identifies WNK4 as a target (41). Knockdown of long WNK1 also appears to decrease the endocytosis (41). Thus, WNKs may be general players in the endocytic cascades acting upstream of ITSN.
GFP-ROMK1 was described previously (13). Full-length cDNAs encoding rat long WNK1 and WNK4 (gifts of M. Cobb and B. Xu, University of Texas Southwestern Medical Center) were cloned in pCMV5-myc vector (1, 11, 13, 36, 38, 39, 42). WNK1 and WNK4 fragments were amplified by PCR and subcloned into pCMV5-myc. Point mutations were generated by site-directed mutagenesis (QuickChange kit; Stratagene) and confirmed by sequencing. GST- and H6-tagged bacterial fusion proteins containing the C terminus of ROMK1 and IRK1 have been described previously (43). Plasmids encoding GST fusion proteins containing SH3 domains of ITSN (Xenopus laevis), endophilin (human), and amphiphysin (human; all gifts of P. McPherson, McGill University, Montreal, Quebec, Canada) have also been described previously (26). Sense and antisense oligonucleotides (Dharmacon RNA Technology) for long WNK1 siRNA were 5′-UGUCUAACGAUGGCCGCUU dT dT and 5′-AAGCGGCCAUCGUUAGACA dT dT, respectively (13). Sense and antisense oligonucleotide for ITSN siRNA were 5′-GGACAUAGUUGUACUGAAAUU and 5′-PUUUCAGUACAACUAUGUCCUU, respectively.
HEK293 cells were cultured and transfected as described previously (13). For pulldown of WNK1 and WNK4 fragments by SH3 domains, cells transfected with myc-tagged WNK plasmids were lysed in an isotonic PBS buffer (1 ml per 6-well dish) containing phosphatase and protease inhibitors. Cell lysates (500 μg) were incubated with GST fusion proteins of interest (1 μg in 0.5 ml final volume), and precipitated by glutathione-4B sepharose beads (43). Pulldown of WNKs was examined using an anti-myc antibody (13). Fusion proteins were detected by an anti-GST antibody. For pulldown of full-length ROMK1 by H6-tagged WNK4 fragments, cells transfected with GFP-ROMK1 were lysed in PBS containing 1% Triton X-100. Pulldown of GFP-ROMK1 was examined using anti-ROMK1 antibody (43). Pulldown of HA-tagged mouse ITSN1 SH3A–SH3E (gift of J. O’Bryan, University of Illinois at Chicago, Chicago, Illinois, USA) by H6-tagged WNK4 fragments was examined using an anti-HA antibody. WNK4 fragments were detected by an anti-H6 antibody. Endogenous long WNK1 and ITSN1s were detected using anti-WNK1 antibody (1) (gift of M. Cobb) and anti-ITSN1 antibody (gift of J. O’Bryan), respectively. The anti-ITSN1 antibody was raised against SH3 domains of mouse ITSN1.
For biotinylation of surface ROMK1, cells were washed twice with ice-cold PBS before incubation with 1 ml PBS containing 1 mg/ml EZ-link-NHS-SS-Biotin (Pierce) for 1 hour at 4°C. After quenching with glycine (100 mM), cells were lysed in PBS containing 1% Triton X-100. Lysates (500 μg in 1 ml reaction volume) were mixed with 100 μl streptavidin beads (Pierce) and incubated for 1 hour. Beads were subsequently washed 3 times with PBS plus 1% Triton X-100. Biotin-labeled proteins were eluted in sample buffer, separated by SDS-PAGE, and transferred to nitrocellulose membranes for Western blotting. ROMK1 proteins on the membrane were detected with a rabbit polyclonal anti-ROMK1 antibody (43). Each experiment shown was performed at least 3 times with similar results. The intensity of protein bands in Western blot analysis was digitized and quantified using Scion Image analysis software (alpha 22.214.171.124; http://www.scioncorp.com/frames/fr_download_now.htm; Scion Corporation).
Myc-tagged long WNK1 fragments were immunoprecipitated from lysates of transfected cells using an anti-myc antibody (13) and washed 3 times with a solution containing 1 M NaCl and 20 mM Tris-HCl (pH 7.4) and once with a solution containing 10 mM MgCl2 and 10 mM HEPES (pH 8.0). To measure kinase activity (37, 38), immunoprecipitates were incubated in 50 μl kinase buffer containing 10 mM HEPES (pH 8.0), 10 mM MgCl2, 50 μM ATP (5 μCi of [γ-32P]ATP), 1 mM benzamidine hydrochloride, 1 mM dithiothreitol, and 3 μg purified (kinase-dead) GST-OSR1/K44R (gift of M. Cobb and A. Anselmo, University of Texas Southwestern Medical Center; ref. 38) at 30°C for 30 minutes.
HEK293 cells were cotransfected with cDNAs encoding GFP-ROMK1 and full-length or fragments of WNK1 or WNK4 plus additional constructs as indicated (ITSN siRNA in Figure Figure2E2E and Figure Figure5F5F and ITSN-DN in Figure Figure2F2F and Figure Figure5E).5E). In each experiment, the total amount of DNA for transfection was balanced using an empty vector. About 36–48 hours after transfection, whole-cell currents were recorded using an Axopatch 200B amplifier (Molecular Devices) as previously described (13). Transfected cells were identified using epifluorescent microscopy. The pipette solution contained 140 mM KCl and 10 mM HEPES (pH 7.2); the bath solution contained 140 mM KCl, 1 mM MgCl2, 1 mM CaCl2, 10 mM HEPES (pH 7.4). Capacitance and access resistance were monitored and 75% compensated. The voltage protocol consisted of 0 mV holding potential and 400-ms steps from –100 to 100 mV in 20-mV increments. Current density was calculated by dividing current at –100 mV (pA; measured at 25°C) by capacitance (pF). Results were shown as mean ± SEM (n = 5–10). For cell-attached single-channel recording of ROMK1, the pipette solution contained 140 mM KCl, 1 mM MgCl2, 1 mM CaCl2, and 10 mM HEPES (pH 7.4). Single-channel recording and analysis of single-channel conductance and open probability were performed as described previously (44).
Sprague-Dawley rats (approximately 250 g, of either sex) were anesthetized and perfused via the heart with 100 ml PBS followed by 50 ml 4% paraformaldehyde in PBS. Kidneys were harvested, fixed in 4% paraformaldehyde at 4°C for 4 hours, dehydrated by immersion in 30% sucrose in PBS at 4°C overnight, and mounted in OCT (Tissue-Tek; Sakura) for sectioning. Sections (3–5 μm) were dried at 37°C for 30 minutes, treated with 0.1% Triton for 10 minutes at room temperature, and quenched in 0.1% NaBH4 for 30 minutes at room temperature followed by extensive washing in PBS. After blocking with 10% goat or donkey serum in PBS for 2 hours, sections were incubated overnight at 4°C with primary antibodies, which included rabbit polyclonal antibodies against ITSN1 (1:200; gift of J. O’Bryan), mouse monoclonal antibodies against NCX1 (1:400; gift of K. Philipson, University of California at Los Angeles, Los Angeles, California, USA), and goat polyclonal antibodies against AQP2 (1:200; Santa Cruz Biotechnology Inc.). After washing, specimens were incubated with secondary antibodies Alexa Fluor 568–conjugated goat anti-rabbit (diluted 1:500; Invitrogen) or fluorescein-conjugated donkey anti-goat (diluted 1:500; Jackson ImmunoResearch Laboratories Inc.) for 30 minutes at room temperature. Fluorescent images were obtained using a Nikon Eclipse TE2000-U fluorescent microscope and overlaid with differential interference contrast images, as we have previously described (16, 18). All procedures were performed in compliance with relevant laws and institutional guidelines and were approved by the University of Texas Southwestern Medical Center at Dallas Institutional Animal Care and Use Committee.
Statistical comparisons between 2 groups of data were made using 2-tailed unpaired Student’s t test. Multiple comparisons were made using 1-way ANOVA followed by 2-tailed Student’s t test adjusted for multiple comparisons. P values less than 0.05 and 0.01 were considered significant for single and multiple comparisons, respectively.
We thank Melanie Cobb, Peter McPherson, John O’Bryan, and Kenneth Philipson for plasmids and antibodies; Melanie Cobb, Donald Hilgemann, and Peter Igarashi for discussion and comments; and Beverley Huet for advice on statistical analysis. This work was supported by NIH grants DK-59530 and DK-54368 (to C.-L. Huang).
Nonstandard abbreviations used: AQP2, aquaporin-2; CCD, cortical collecting duct; CCV, clathrin-coated vesicle; CNT, connecting tubule; DCT, distal convoluted tubule; EH, Eps15 homology; GST, glutathione S-transferase; H6, hexahistidine; IRK1, inward-rectifying K+ 1; ITSN, intersectin; ITSN1s, ITSN1 short form; NCX1, sodium-calcium exchanger isoform 1; OSR1, oxidative stress-response kinase 1; PHA2, pseudohypoaldosteronism type 2; ROMK1, renal outer medullar potassium 1; SH3, Src-homology 3; WNK, with-no-lysine.
Conflict of interest: The authors have declared that no conflict of interest exists.
Citation for this article: J. Clin. Invest. 117:1078–1087 (2007). doi:10.1172/JCI30087
Guocheng He and Hao-Ran Wang contributed equally to this work.