WNK Signaling Is Involved in Neural Development via Lhx8/Awh Expression

WNK kinase family is conserved among many species and regulates SPAK/OSR1 and ion co-transporters. Some mutations in human WNK1 or WNK4 are associated with Pseudohypoaldosteronism type II, a form of hypertension. WNK is also involved in developmental and cellular processes, but the molecular mechanisms underlying its regulation in these processes remain unknown. Here, we identify a new target gene in WNK signaling, Arrowhead and Lhx8, which is a mammalian homologue of Drosophila Arrowhead. In Drosophila, WNK was shown to genetically interact with Arrowhead. In Wnk1 knockout mice, levels of Lhx8 expression were reduced. Ectopic expression of WNK1, WNK4 or Osr1 in mammalian cells induced the expression of the Lhx8. Moreover, neural specification was inhibited by the knockdown of both Wnk1 and Wnk4 or Lhx8. Drosophila WNK mutant caused defects in axon guidance during embryogenesis. These results suggest that WNK signaling is involved in the morphological and neural development via Lhx8/Arrowhead.


Introduction
WNK (with no lysine (K)) is a family of serine/threonine protein kinases that are characterized by an atypical location of the catalytic lysine and are conserved among many species, such as plants, nematode, fly, rat, mouse and human [1][2][3]. There are four mammalian WNK family members, and positional cloning has identified two of them, WNK1 and WNK4, as genes linked to a hereditary form of human hypertension known as Pseudohypoaldosteronism type II (PHAII) [4]. Several groups including our group previously discovered that WNK1 and WNK4 could phosphorylate and activate SPAK or OSR1 kinases, which in turn regulates various ion co-transporters, such as NKCC1, NKCC2 and NCC [5][6][7][8]. We also found that dysregulation of WNK1 and WNK4 in mouse kidney caused phenotypes similar to those of PHAII [9]. These results suggest that the dysregulation of sodium and potassium transport by WNK1 and WNK4 contribute to the pathogenesis of hypertension in PHAII patients.
WNK family members have also been identified in screens of cultured cells for enhanced cell survival and proliferation [10]. WNK1 is required for cell division in cultured cells [11], and proliferation, migration and differentiation of neural progenitor cells [12]. In addition, Wnk1 is ubiquitously expressed in mice, and knockout of the gene is lethal before embryonic day 13 (Zambrowicz et al. and in this report) [13], with the developing mice displaying defects in cardiac development [14]. Furthermore, PHAII patients exhibit other clinical problems in addition to hypertension, such as an intellectual impairment, dental abnormalities and impaired growth [15]. The Drosophila genome contains a single WNK gene (called as CG7177 in Flybase (http://flybase.org)), which was identified in screens for genes involved in cell cycle or neural development [10,16]. These observations suggest that WNK1 plays unknown roles in developmental processes, in addition to its control of ion cotransporters in the kidney.
Here, we demonstrate that the functions of the WNK signaling pathway are conserved between mammals and flies. Mutation of Drosophila WNK (DWNK) caused several morphological defects. Our functional analysis of DWNK identified a new target gene, Arrowhead (Awh), and we found that the mammalian homologue of Awh, Lhx8, is also a target gene of the WNK signaling pathway in mammalian cells. Furthermore, we demonstrated that the WNK signaling pathway modulates Drosophila development via Awh, and modulates neural specification in mammalian cells via Lhx8. These results reveal a novel role for WNK signaling via Lhx8 or Awh in the regulation of morphological and neural development.

Ethics statement
All animal experiments were performed under the ethical guidelines of Tokyo Medical and Dental University, and animal protocols were reviewed and approved by the animal welfare committee of the Tokyo Medical and Dental University. ated by P-mediated germline transformation (injected by BestGene Inc.). DWNK EY18 , a null mutation of DWNK, is a derivative of EY10165. DWNK EY18 has a 1712 bp deletion from the EY10165 insertion point to the middle of exon 3, which includes the translation start site (red line in Fig. S1). However, the 59 region of the P element of EY10165 was retained (1365 bp). We confirmed by RT-PCR analysis that DWNK EY18 produced truncated transcripts, by the presence of several poly-A signal sequences in the retained P element sequences (* in Fig. S1; data not shown).
Genotypes of all fly lines we used in this study were in figure legends. We used yellow transgene for the clonal marker; the wild type body color represents heterozygous tissue, and yellow body color represents mutant tissue. The mutant tissues were judged by discrimination of the light color compared with the background of wild type color, and the clonal borders were shown by thin black lines.

Histology and staining
All wings were mounted in GMM [17]. Antibody staining, In situ hybridization to fly and mouse embryos were carried out as described previously [18,19]. For in situ hybridization, digoxigeninlabeled RNA probes were prepared by in vitro transcription using Awh and Lhx8 cDNA as a template. Images were obtained using SteREO Discovery, Axioscope and Axio Observer (Carl Zeiss), and processed using Axiovision with extended focus (Carl Zeiss) and Photoshop (Adobe).
We isolated total RNA using RNeasy kit (Qiagen). Microarray was performed by Takara-Bio using GeneChip (Affymetrix).

RT-PCR analysis
Total RNA was isolated by TRIzol (Invitrogen). cDNA synthesis was carried out using Moloney murine leukemia virus reverse transcriptase (Invitrogen). The reaction mixture were denatured at 94 degree for 5 minutes and then cycled at 98 degree/15 seconds and 72 degree/30 seconds, then followed by a final 3 minutes extension at 72 degree (for mouse Wnk1, mouse Wnk4, human WNK1, human WNK4, mouse Osr1, mouse Choline acetyltransferase (ChAT), mouse Glutamic acid decarboxylase 1 (Gad1), mouse Lhx6 and mouse Glyceraldehyde-3-phosphate dehydrogenase (GAPDH)), or at 94 degree for 5 minutes and then cycled at 98 degree/15 seconds, 58 degree/15 seconds and 72 degree/30 seconds, then followed by a final 3 minutes extension at 72 degree (for mouse Lhx8 and mouse Islet-1 (Isl1)). Numbers of cycles are depending on samples (see below, number of cycles are shown after primer sequences).

Quantification and statistical analysis
All results from Western blotting were quantified using Multi Gauge software (GE). Quantitative PCR was performed with an Applied Biosystems 7300 Real-Time PCR Cycler (ABI) using THUNDERBIRD SYBR qPCR Mix (TOYOBO). The sequences of the primer pairs for Lhx8, ChAT, Gad1 and GAPDH were described previously [21][22][23]. GAPDH was used for normalization

WNK-SPAK/OSR1 pathway is conserved between mammals and flies
The Drosophila genome contains only one WNK homologue CG7177, which we hereafter refer to as Drosophila WNK (DWNK; Fig. S1). In contrast, there are four WNK family genes in mammals. Among the mammalian WNK proteins, DWNK was most homologous to WNK1 (Fig. S1). Recent studies have shown that mammalian WNK1-4 interact with and phosphorylate the STE20 kinases, SPAK and OSR1 [5][6][7]. As Drosophila Fray is a homologue of SPAK and OSR1 [24], we investigated whether the biochemical interaction between DWNK and Fray is conserved in Drosophila. We transiently expressed HA-DWNK together with T7-Fray in human embryonic kidney (HEK) 293T cells. When cell extracts were subjected to immunoprecipitation with the HA antibody, followed by immunoblotting, we found that DWNK interacted with Fray (Fig. 1A, lane 3). We next investigated phosphorylation of Fray by DWNK. We produced a glutathione S-transferase (GST)-tagged kinase-negative form of Fray K67M in bacteria and tested its ability to be phosphorylated in vitro. We observed that DWNK phosphorylated Fray in a kinase-dependent manner (Fig. 1B, lanes 4, 8). We attempted to generate a constitutively activate form of Fray by mutating Ser to Asp at amino acid 347, which corresponds to the site of WNK phosphorylation in mouse Osr1 [5,6]. Fray kinase activity was assayed using an amino-terminal fragment of NCC, GST-NCC, as a substrate [5]. This mutant, Fray S347D , exhibited increased phosphorylation of GST-NCC, relative to wild-type Fray ( Fig. 1C lane 4,12), indicating that the mutation of Ser-347 to Asp causes constitutive activation of Fray.
To examine the functional conservation of human and Drosophila WNK, we used the UAS-Gal4 system to express human WNK in Drosophila. We made an expression construct of human WNK1 (hWNK1) and generated UAS-transgenic flies (UAS-hWNK1). As a UAS-DWNK line, we used the EY10165 line, in which the pEY construct has been inserted into the 1 st exon of the DWNK gene (Fig. S1B). Although any wing phenotypes were not obtained from either the heterozygous UAS-DWNK or hedgehog (hh)-Gal4 driver line ( We also generated UAS-transgenic flies of mouse Osr1 (mOsr1; UAS-mOSR1) and its fly homologue, fray (UAS-fray) [24]. Wing phenotypes of the flies overexpressing mOsr1 were similar to those overexpressing fray (Fig. 2F-G and Fig. S2). To analyze the genetic interaction between DWNK and Fray, we examined the phenotypes of DWNK overexpression in a heterozygous fray mutant (fray 07551 ) background. Heterozygous fray 07551 has a normal-looking wing. The penetrance of wing phenotypes induced by DWNK overexpression were partially rescued in a heterozygous fray 07551 mutant background ( Fig. 2H and Fig. S2). Moreover, the wing phenotypes induced by fray S347D , the constitutively active form of Fray driven by hh-Gal4, were more frequent, but similar phenotypes to those seen in flies overexpressing mOsr1or fray ( Fig. 2I and Fig. S2). We also generated an inactive mutant of DWNK, DWNK EY18 , by imprecise P element excision ( Fig. S1; see also Materials and Methods for detail characterization of DWNK mutant). Heterozygous DWNK EY18 mutant did not cause any phenotype in wing. The penetrance of the phenotypes of fray S347D overexpression could not be rescued in a heterozygous DWNK EY18 background ( Fig. 2J and Fig. S2). Taken together, these results suggest that Fray functions downstream of DWNK, and that the WNK-SPAK/OSR1/Fray pathway is conserved among many species.

DWNK-Fray pathway functions in abdominal development
To investigate the developmental function of DWNK, we generated UAS system constructs for the expression of the kinasedead form of DWNK (UAS-DWNK D420A ). As the expression of DWNK D420A by hh-Gal4 caused lethal, we switched to sd-Gal4 driver. The expression of DWNK D420A by sd-Gal4 driver induced the loss of wing margin (Fig. S3). In addition, the expression of DWNK D420A by sd-Gal4 driver caused the complete disruption of abdominal differentiation in the pharate adults (compared Fig. 3B with Fig. 3A). To confirm that these phenotypes were caused by the loss of DWNK, we examined the adult phenotypes of DWNK EY18 . We employed a mosaic analysis since homozygous DWNK EY18 flies die between late embryonic stage and early 2 nd instar larva. Various phenotypes, such as extra veins in the wing and loss of macrocheate or microcheate bristles in the notum, were observed in the large DWNK EY18 clones generated by Minute methods (Fig. S3). Moreover, the defects in abdominal development were observed in the mosaic clones (Fig. 3C). We also observed that the abdominal phenotypes were almost rescued by the overexpression of DWNK (Fig. S4), suggesting that the abdominal phenotypes were the results of DWNK deficiency. These observations indicate that phenotypes induced by the kinase-dead form of DWNK, DWNK D420A , were indeed caused by loss of DWNK function, suggesting that DWNK D420A worked as a dominant negative.
Next, we examined whether Fray also worked at the downstream of WNK in abdominal development. While the expression of the constitutively active form of fray, fray S347D , did not cause any phenotype in abdomen (Fig. 3D), the abdominal defects caused by DWNK D420A could be rescued by the expression of fray S347D (Fig. 3E). We also confirmed that this rescue was not due to the titration of Gal4 expression by adding another UAS construct (Fig. S5). Thus, these results suggest that DWNK-Fray pathway plays important roles in Drosophila abdominal development.

Genetic interaction of DWNK with Awh
Both the DWNK minute clones and flies expressing the dominant-negative form of DWNK exhibited defects in abdominal development ( Fig. 3B-C). Similar abdominal phenotypes have been previously described for the Arrowhead (Awh) mutant (Fig. 3F) [25], a transcription factor of the Lim homeobox type. We therefore investigated whether DWNK genetically interacts with Awh. We generated UAS-Awh lines and induced Awh expression by sd-Gal4 driver, together with or without DWNK D420A expression. Overexpression of Awh did not affect the abdominal development (Fig. 3G). However, the defect in abdominal development caused by expression of DWNK D420A was rescued in flies co-expressing Awh ( Fig. 3H and 3B). We also tested whether expression of Awh rescued the abdominal phenotype of DWNK minute clones. Using a combination of the FLP/FRT mosaic system and Gal80 suppression, we could express Awh locally in DWNK minute clones. As shown in Fig. 3I, overexpression of Awh partially rescued the abdominal phenotype of DWNK minute clones. These results indicate that DWNK genetically interacts with Awh.

WNK regulates the expression of the Awh/Lhx8 gene
The mammalian homologue of Awh is Lhx8 (also called L3 or Lhx7) [26][27][28]. To analyze how Lhx8 functions in the WNK signaling pathway, we first examined whether WNK1 binds to Lhx8 or regulates expression of the Lhx8 gene. We were unable to detect an interaction between ectopically expressed WNK1 and Lhx8 in cultured cells (data not shown). We performed microarray analyses using embryos of wild-type or Wnk1 knockout mice at embryonic day 9. These microarray data revealed that Lhx8 expression was reduced in Wnk1 knockout mice (Fig. 4A). We also examined Lhx8 expression in developing mice embryos by in situ hybridization. As previously reported [27], Lhx8 was expressed in the craniofacial region of wild-type embryos at embryonic day 10.5 (Fig. 4B). In contrast, Lhx8 expression was very weak in the similar region of Wnk1 knockout mice embryos (Fig. 4C). Moreover, we examined whether DWNK controls Awh expression in Drosophila embryos. Awh was expressed in a striped pattern at stages 11 and 13 (Fig. 4D-E), and in the histoblast nest (which is the primordia of abdominal tissue) at stage 16 (arrows in Fig. 4F). We could not detect expression of Awh in the histoblast nests of embryos homozygous for DWNK EY18 at stage 16 (Fig. 4I). However, expression of Awh in embryos homozygous for DWNK EY18 at stages 11 and 13 was not changed (data not shown). We expected that the zygotic mutant embryos of DWNK EY18 would be rescued by maternal transcripts of DWNK, since DWNK is maternally expressed according to the High Throughput Expression Data in Flybase (http://flybase.org). Instead, we found that expression of Awh at stages 11 and 13 was reduced in the embryos overexpressing DWNK D420A driven by da-Gal4 (Fig. 4G-H). Together, these results suggest that the expression of the Awh/ Lhx8 gene is regulated by WNK.
The WNK-SPAK/OSR1 pathway induces expression of the Lhx8 gene We further attempted to examine Lhx8 gene expression in the WNK-SPAK/OSR1 pathway. A previous study has reported that WNKs are activated by hypertonic stimulation [29]. We first performed Western blotting analysis using anti-phospho OSR1 antibody, which recognizes Ser325 of mOsr1 phosphorylated by WNK kinases 5 , whether WNKs are activated by hypertonic stimulation in NIH3T3 cells. We found that WNKs were immediately activated by hypertonic stimulation in NIH3T3 cells (Fig. 5A lanes 1-5). On the other hand, knockdown of either Wnk1 or Wnk4 using siRNA significantly reduced the phophorylation level of mOsr1 (Fig. 5A lanes 6-10 or 11-15). Moreover, the knockdown of both Wnk1 and Wnk4 synergistically caused the loss of the phosphorylation of mOsr1 (Fig. 5A lanes 16-20). Next, we performed RT-PCR analysis to ask whether the expression of Lhx8 gene is activated under hypertonic conditions. We found that Lhx8 expression was induced in NIH3T3 cells by hypertonic stimulation for 8 hours (Fig. 5B lanes 1-4). Knockdown of either Wnk1 or Wnk4 using siRNA resulted in significantly reduced induction of Lhx8 ( Fig. 5B lanes 5-8 or 9-12). In addition, Lhx8 activation was completely suppressed by knockdown of both Wnk1 and Wnk4 (Fig. 5B lanes 13-16). Conversely, Lhx8 expression was induced by the expression of hWNK1 or hWNK4 (Fig. 5C lanes 2 and 4). Although expression of a kinase-dead form of either hWNK1 or hWNK4 (hWNK1 D368A or hWNK4 K186M ) also weakly activated Lhx8 expression (Fig. 5C lanes 3 and 5), co-expression of both hWNK1 D368A and hWNK4 K186M did not activate Lhx8 expression (Fig. 5C lane 6). Phosphorylation of Osr1 was also confirmed to correlate with the expression of Lhx8 (Fig. 5C bottom two rows). Moreover, expression of wild-type mOsr1 also induced Lhx8 expression (Fig. 5D lane 2). The expression of mOsr1 K46M , the kinase-dead form of mOsr1, did not induce Lhx8 expression (Fig. 5D lane 3). Furthermore, expression of mOsr1 S325D , the constitutively active form of mOsr1, strongly induced Lhx8 expression (Fig. 5D lane 4). On the other hand, induction of Lhx8 induced by expression of WNK1 was completely suppressed when Osr1 was knocked down using siRNA (Fig. 5E lane 4). In addition, the prior treatment of cycloheximide (CHX), an inhibitor of protein biosynthesis, did not inhibit Lhx8 expression by hypertonic stimulation (Fig. 5F), indicating that Lhx8 is a direct target gene in WNK activation through hypertonic stimulation.
These results suggest that the WNK-OSR1 pathway regulates Lhx8 expression, and that Lhx8 is a downstream target gene in the WNK signaling pathway.

The WNK signaling pathway is involved in neural specification
It is known that Lhx8 is involved in the determination of cholinergic neural fate in the forebrain [30,31], and the specification of neural fate in Neuro2A cells that are able to differentiate into cholinergic or GABAergic neurons [32]. Moreover, WNK1 is known to play an important role in the proliferation, migration and differentiation of neural progenitor cells [12]. Thus, we speculated that the WNK signaling pathway might be involved in neural differentiation via expression of the Lhx8 gene. Since retinoic acid (RA) is known to induce differentiation of Neuro2A cells, we investigated whether RA is able to activate WNK signaling. We performed Western blotting analysis using anti-phospho OSR1 antibody. We found that phosphorylation of mOsr1 increased at around 2 to 4 hours after RA stimulation (Fig. 6A), indicating that WNK kinases were activated by RA in Neuro2A cells. We also found that expression of the Lhx8 gene was induced by RA in Neuro2A cells (Fig. 6B). After treatment with RA for 24 hours, Neuro2A cells were clearly differentiated and had generated several elongated neurites ( Fig. 6C; see also Fig. S6). RA-induced neurite elongation was not suppressed by siRNA knockdown of either Wnk1 or Wnk4 alone in Neuro2A cells (Fig. 6D-E), but was inhibited by the combined knockdown of both Wnk1 and Wnk4 (compare Fig. 6F with Fig. 6C). Moreover, knockdown of Lhx8 also inhibited neurite elongation (Fig. S6). These results suggest that the WNK signaling pathway is involved in the elongation of neurites.
To confirm the neural fate of the differentiated Neuro2A cells, we examined the expression of marker genes by RT-PCR analysis. Choline acetyltransferase (ChAT) or Glutamic acid decarboxylase 1 (Gad1) was used as a marker for cholinergic or GABAergic neurons, respectively. After treatment of RA for 24 hours, Neuro2A cells differentiated into neurons and expressed the marker genes, ChAT and Gad1 (Fig. 6G lane 3). As previously reported [32], we confirmed that knockdown of Lhx8 expression caused a decrease in ChAT expression and increase in Gad1 expression (Fig. 6G lane 4). Knockdown of Wnk1 or Wnk4 caused a decrease in ChAT and increase in Gad1 expression, compared to cells treated with control siRNA (Fig. 6H lanes 9,11,13 and 6I lanes 9,11,13). In addition, knockdown of both Wnk1 and Wnk4 completely reduced ChAT expression (Fig. 6H lane 15 and 6I lane 15). These results suggest that the WNK kinases are involved in the specification of neural fate.  These results indicate that Lhx8 expression correlates with the activities of the WNK kinases. Thus, we first examined whether expression of the constitutively active form of mOsr1 (mOsr1 S325D ) rescues the neural specification phenotype induced by knockdown of both Wnk1 and Wnk4. While the expression of mOsr1 S325D did not affect and could rescue the elongation of neurite (Fig. S6), mOsr1 S325D overexpression significantly induced Lhx8 expression even in undifferentiated Neuro2A cells (Fig. 6H lane 2,4,6,8) and also affected an enhancement in ChAT expression and a reduction in Gad1 expression with RA treatment in Neuro2A cells (Fig. 6H  lane 9-10). Under the condition of the knockdown of both Wnk1 and Wnk4, mOsr1 S325D overexpression could rescue Lhx8 expression, the decreasing of ChAT and increasing of Gad1 (Fig. 6H lane [15][16]. On the other hand, Lhx8 overexpression did not affect the elongation of neurites (Fig. S6), but could not rescue the elongation of neurites caused by the knockdown of both Wnk1 and Wnk4 (Fig.   S6). Lhx8 overexpression also caused an enhancement in ChAT expression and a reduction in Gad1 expression (Fig. 6I lane 9-10). However, Lhx8 overexpression could not rescue the decrease in ChAT or increase in Gad1 expression caused by knockdown of both Wnk1 and Wnk4 (Fig. 6I lane 15-16). These results suggest that WNK-OSR pathway is involved in the neural development, and that Lhx8 is necessary but not sufficient for the specification of the neural fate.

The WNK signaling pathway is involved in neural development in Drosophila
Since Fray was required for axonal ensheathment [24] and Awh was expressed in neuroblasts in stage 9 embryos in Drosophila [26], we wondered whether the WNK signaling pathway was also involved in neural development during Drosophila development. To analyze this possibility, we examined the formation of the peripheral nervous system in fly embryos by staining with 22C10 monoclonal antibodies, which can visualize neuronal morphology and axonal projections [33,34] (Fig. 7A). We could find only a few DWNK EY18 embryos exhibiting minor defects of axon guidance (arrows in Fig. 7B). These are expected that the zygotic mutant embryos of DWNK EY18 would be rescued by maternal transcripts of DWNK, as explained above. Therefore to better confirm this result, we examined the effects of a dominant negative form of DWNK. We found that the expression of a dominant negative form of DWNK (DWNK D420A ) driven by the 1407-Gal4 driver, which expressed in neuroblast and nervous system [35], also caused severe defects in the peripheral nervous system (Fig. 7C). We next tested whether Fray or Awh could rescue the phenotypes. While overexpression of the constitutively active form of fray (fray S347D ) did not cause any phenotypes (Fig. 7D), the constitutively active form of fray could rescue the effects of the dominant negative form of DWNK (Fig. 7E). On the other hand, overexpression of Awh did not cause any defects of axon guidance (Fig. 7F), and could not rescue the phenotypes by overexpression of DWNK D420A (Fig. 7G). We also confirmed that the rescue by the co-expression of fray S347D was not due to the titration of Gal4 expression (Fig. S5). These results suggest that DWNK-Fray pathway plays an important role in neural development in Drosophila, but that Awh is not sufficient for determinant in neural development.

Discussion
The WNK-SPAK/OSR1 pathway is known to regulate various ion co-transporters and is widely conserved among many species [1,2]. Wnk1 knockout mice die before embryonic day 13 (Zambrowics et al. and in this report) [13], and display defects in cardiac development [14]. WNK1 is also required for cell division in cultured cells [11], and proliferation, migration and differentiation of neural progenitor cells [12]. Furthermore, PHAII patients display a number of other clinical features, such as an intellectual impairment, dental abnormalities and impaired growth in addition to hypertension [15]. Accordingly, the new role of the WNK signaling pathway described here may provide further insight into the development and pathogenesis of PHAII. In this set to 100. (H-I) Neuro2A cells were transfected with various combinations of siRNAs and expression plasmids and gene expressions by RT-PCR or quantitative RT-PCR analysis were examined. Cells were treated with control siRNA (Control) in lanes 1,2,9,10, with siWnk1in lanes 3,4,11,12, with siWnk4 in lanes 5,6,13,14, and with both siWnk1 and siWnk4 in lanes 7,8,15,16. Cells were also transfected with control expression vector pRK5 in lanes 1,3, 5,7,9,11,13,15   study, we have identified Lhx8/Awh as a new downstream molecule in the WNK-SPAK/OSR1 pathway and discovered a novel function for the WNK-Lhx8 pathway in neural development.
There are four mammalian WNK family members, and WNK1 and WNK4 genes are linked to a hereditary form of human hypertension known as Pseudohypoaldosteronism type II (PHAII) [4]. In Drosophila, only one WNK gene, DWNK, has been identified. We found that both the wild-type and kinase-dead forms of WNK1 or WNK4 caused the up-regulation of Lhx8 gene expression in NIH3T3 cells (Fig. 5D lanes 2-5). Similarly, our previous study showed that SPAK, a substrate of WNK1, was weakly phosphorylated by the kinase-dead form of WNK1 following a long incubation [5]. These results are inconsistent with the idea that the kinase-dead form of DWNK functions as a dominant-negative mutant in Drosophila. Studies of WNK1 and WNK4 suggest that these molecules phosphorylate each other and coordinated to regulate NaCl cotransport [36]. Therefore, these results raised the possibility that the kinase-dead forms of WNK1 and WNK4 coordinate with their respective endogenous WNK1 and WNK4 counterparts in mammalian cells. In fact, we found that co-expression of both kinase-dead forms of WNK1 and WNK4 did not cause either induction of Lhx8 gene expression or phosphorylation of mOsr1 ( Fig. 5D lane 6). These results suggest that the kinase activity of WNKs is required for induction of Lhx8 gene expression and the activation of SPAK/OSR1, and that the kinase-dead form of WNK acts as an actual dominant-negative form in the signaling pathway. Furthermore, the expression of Lhx8 by either hypertonic or RA stimulation was required for the expression of both WNK1 and WNK4 (Fig. 5B lanes 13 A previous report demonstrated that WNK1 might control the formation of microtubules in developing neurons [12]. On the other hand, other studies suggested that Lhx8 plays an important role in the development of basal forebrain cholinergic neurons [30,31,37], that Fray is required for axonal ensheathment [24], and that Awh is expressed in neuroblasts in stage 9 embryos in Drosophila [26]. In this study, we showed that the WNK-OSR1 pathway regulates Lhx8 gene expression, that knockdown of both Wnk1 and Wnk4 in Neuro2A cells caused a shortening of neurites, as well as reduced Lhx8 expression (Fig. 6G-J and Fig. S6), and that the expression of the constitutively active form of mOsr1, mOsr1 S325D , could rescue the phenotype caused by the knockdown of both Wnk1 and Wnk4 (Fig. 6I and Fig. S6). In addition, mutation of DWNK or expression of a dominant-negative form of DWNK in fly embryos caused defects in axon guidance in the peripheral nervous system (Fig. 7B,C), and the constitutively active form of fray, fray S347D , expression could rescue the phenotypes by the expression of the dominant negative form of DWNK (Fig. 7E). Furthermore, ubiquitous expression of Awh by da-Gal4 showed severe defects of axon guidance as similar to DWNK D420A expression by da-Gal4 (Fig. S7), although neural specific expression of Awh did not showed any phenotype (Fig. 7F). Taken together, these findings clearly indicate that the WNK-OSR1/Fray-Lhx8/ Awh pathway is involved in neural development. However, the phenotypes caused by knockdown of both Wnk1 and Wnk4, such as the shortening of neurites and the reduction in ChAT expression, were not rescued by the expression of Lhx8 in Neuro2A cells ( Fig. 6J and Fig. S6). In addition, the expression of Awh could not rescue the defects in the peripheral nervous system by the expression of the dominant-negative form of DWNK (Fig. 7G). Previous reports showed that Lhx8 might work with other factors, such as Lhx6 or Isl1 [27,37,38]. However, we also found that the expression of Lhx6 and/or Isl1 with Lhx8 could not rescue the defects by knockdown of both Wnk1 and Wnk4 in Neuro2A cells (Fig. S8). These results suggest that other molecule(s) are involved in neural differentiation induced by WNK signaling. Our studies may provide the first evidence identifying a target gene that acts downstream in the WNK-SPAK/OSR1 pathway, and demonstrate the significance of the WNK-OSR1-Lhx8 pathway in neural development. However, the details of how other unknown molecules controlled by WNK signaling specifically contribute to neural developmental remain to be determined and will require additional study.
Genetic mutations of WNK1 or WNK4 in PHAII patients result in abnormal expression of the WNK1 gene or WNK4 kinase activity, respectively [4]. Abnormal activation of the WNK signaling pathway caused by these mutations result in the misregulation of NCCs in the kidney, which in turn causes hypertension [5][6][7][8][9]39]. However, PHAII patients display other clinical features, such as an intellectual impairment, dental abnormalities and impaired growth [15]. Although these features are also thought to be caused by WNK1 or WNK4 mutations, the details of how these pathologies occur are unknown except for hypertension. In this study, we identified Lhx8 as a downstream target of the WNK signaling pathway (Fig. 5). We also found evidence that the WNK-Lhx8 pathway is involved in neural development (Fig. 6). Previous studies have shown that knockdown of Lhx8 using antisense oligodeoxynucleotides caused the loss of tooth germ [40], and Lhx8 and Lhx6 are key regulators of mammalian dentition [41]. Furthermore, Lhx8 knockout mice show a reduction in the number of cholinergic neurons in the ventral forebrain [30,31,37] and exhibit a severe deficit in spatial learning and memory [42]. These observations indicate that Lhx8 has essential functions in the formation of the tooth development, the specification of the cholinergic neurons and the processing of the spatial information in mice. Therefore, the similarities between the clinical features of PHAII and the phenotypes of Lhx8 knockdown or knockout mice strongly suggest that the WNK-Lhx8 pathway is involved in the pathogenesis of PHAII, aside from hypertension. Further investigation will be needed to prove this hypothesis.  Figure 2. We could observe delta phenotype at the tip of vein 4 (arrowheads in Figure 2) with or without extra veins around vein 5 (arrows in Figure 2). (TIF)