Wild-type but not mutant HSN2 binds to WNK1, WNK4, SPAK, OSR1 and HSN2.
HSN2 is a neural specific isoform of WNK1, and mutations in HSN2 cause the autosomal recessive neuropathy, HSANII9. WNK1 can form complexes with WNK1 and WNK4 to activate downstream effectors28,29. To analyse whether HSN2 binds to WNK1 and WNK4, we transiently expressed HA-tagged WNK1 and HSN2 with Myc-tagged WNK1 or WNK4 in HEK293T cells. Cell extracts were immunoprecipitated with a Myc antibody and immunoprecipitates were subjected to immunoblotting. As shown in Fig. 1A and B, HSN2 bound similarly to WNK1 and WNK4. In HSANII patients, mutations have been identified in the neural-specific alternatively spliced exon of HSN2, including a 1-bp deletion (2743delA9, referred to as HSN2-delA) and a 1-bp insertion (3237_3238insT14, referred to as HSN2-insT). These mutations cause a frameshift and result in truncated HSN2 proteins at amino acid 916 and 1080, respectively. We also checked the binding of these HSN2 mutants to WNK1 and WNK4 and found that they could not interact with WNK1 or WNK4 (Fig. 1A and B). WNK1 interacts with and phosphorylates the STE20 kinases, SPAK and OSR1, to activate the downstream pathway5,6; therefore, we analysed the binding of HSN2 and its mutants to SPAK and OSR1. We found that HSN2 and WNK1 bound similarly to SPAK and OSR1, but the binding of HSN2 mutants to SPAK and OSR1 was weak compared with that of WNK1 and HSN2 (Fig. 1C and D). Interestingly, HSN2 also bound to itself, but HSN2-delA and HSN2-insT could not bind to wild-type HSN2 (Fig. 1E). These results indicate that HSN2-delA and HSN2-insT mutants are unable to promote activation of downstream signalling because the HSN2 mutants could not interact with HSN2, WNK1, WNK4 or SPAK/OSR1.
HSN2 mutants, HSN2-delA and HSN2-insT, exhibit loss of neural development function.
We have previously reported that WNK1 and WNK4 are involved in induction of neural marker genes and neurite elongation in mouse neuroblastoma Neuro2A cells23,24. HSN2 may, therefore, have similar functions in neural development and HSN2 mutants may exhibit loss of function. To examine whether HSN2 is involved in neural development, we transiently expressed HSN2 in Neuro2A cells and observed that HSN2 induced neurite elongation identically to WNK1 (Fig. 2A upper panel). We then examined the effect of HSN2 mutants on neural development. As shown in Fig. 2A, HSN2-delA and HSN2-insT mutants could not induce neurite elongation. Next, we checked the effects of wild-type and mutant HSN2 on neural differentiation induced by nerve growth factor (NGF). HSN2 constructs were transfected into Neuro2A cells and 24 h later cells were treated with serum-free medium containing 100 ng/ml NGF for 24 h. This treatment induced neurite elongation, and the expression of WNK1 or HSN2 enhanced this induction (Fig. 2A lower panel). In contrast, the expression of HSN2-delA or HSN2-insT suppressed neurite elongation in response to NGF-containing medium (Fig. 2A lower panel). We also analysed the expression of neural marker genes, Lhx8 and Choline acetyltransferase (ChAT) for cholinergic neurons and Glutamine acid decarboxylase 1 (Gad1) for GABAergic neurons. qPCR analysis indicated that NGF treatment greatly induced the expression of Lhx8, ChAT and Gad1, and that exogenous expression of WNK1 or HSN2 induced Lhx8 and ChAT expression (Fig. 2B). WNK1 and HSN2 enhanced NGF-mediated expression of Lhx8 and ChAT, but reduced that of Gad1. In contrast, expression of HSN2-delA and HSN2-insT mutants decreased Lhx8 and ChAT expression and increased Gad1 expression (Fig. 2B). These results indicate that WNK1 and HSN2 are involved in cholinergic neural specification, but that HSN2-delA and HSN2-insT mutants fail in this activity. Interestingly, HSN2 mutants inhibited NGF-independent WNK1- and HSN2-mediated neurite elongation (Fig. 2C), suggesting that these mutants have a dominant negative effect on wild-type WNK1 and/or HSN2.
As shown in Fig. 2D, siWnk1 and siWnk4 treatment suppressed NGF-containing serum-free medium-mediated neurite elongation in Neuro2A cells. Moreover, knockdown of Wnk1 and Wnk4 suppressed Lhx8 and ChAT expression, and induced Gad1 expression (Fig. 2E), indicating that WNK proteins are required for neural differentiation of Neuro2A cells, as previously reported23,24. We next examined whether expression of HSN2 constructs could rescue these effects. We found that expression of WNK1 or HSN2 rescued neurite elongation and expression of neural marker genes (Fig. 2D and E). In contrast, these rescue effects were not observed with HSN2-delA or HSN2-insT mutants (Fig. 2D and E). These data indicate that HSN2 mutants fail to exert the normal functions of WNK1/HSN2 in neural development.
NGF activates SPAK/OSR1 through WNK1 and HSN2.
To analyse whether HSN2 regulates activation of SPAK and OSR1 in neural differentiation, we checked phosphorylation of SPAK and OSR1 in Neuro2A cells treated with NGF-containing serum-free medium. Phosphorylation of SPAK and OSR1 was observed at 10 to 15 minutes after treatment with NGF-containing medium (Fig. 3A). In contrast, knockdown of either Wnk1 or Wnk4 suppressed the phosphorylation of SPAK and OSR1 by NGF (Fig. 3B). Moreover, knockdown of both Wnk1 and Wnk4 completely abolished SPAK and OSR1 phosphorylation (Fig. 3B), indicating that NGF activates the WNK-SPAK/OSR1 pathway in Neuro2A cells. We next analysed the effect of HSN2 constructs on the phosphorylation of SPAK and OSR1 mediated by NGF. We found that HSN2 expression increased the level of SPAK and OSR1 phosphorylation identically to that following WNK1 expression (Fig. 3C). By contrast, the expression of HSN2-delA and HSN2-insT mutants suppressed the NGF-induced phosphorylation of SPAK and OSR1 (Fig. 3C). These results indicate that HSN2 mutants cannot activate SPAK and OSR1 and work in a dominant negative manner. These data are consistent with the neurite elongation and neural marker gene expression results (Fig. 1A–C).
We previously demonstrated that the WNK1-OSR1 pathway is involved in neural specification23. To confirm whether NGF-induced neural development is mediated by the HSN2-OSR1 pathway, we examined the effect of Osr1 knockdown on neural development. As expected, WNK1 expression enhanced neurite outgrowth induced by NGF treatment, and knockdown of Osr1 inhibited this elongation (Fig. 3D). Similar results were obtained with HSN2 expression and knockdown of Osr1 (Fig. 3D). In contrast, expression of HSN2 mutants suppressed NGF-induced neurite elongation, and knockdown of Osr1 completely inhibited neurite outgrowth (Fig. 3D). RT-PCR and qPCR analysis showed that NGF-induced Lhx8 and ChAT expression was completely suppressed by Osr1 knockdown even though WNK1 or HSN2 was expressed in Neuro2A cells (Fig. 3E). Moreover, although NGF-induced Gad1 expression was reduced by WNK1 or HSN2 expression, knockdown of Osr1 rescued this reduction (Fig. 3E). We also examined the effect of a kinase-negative form of OSR1 (OSR1K46M), and similar results were obtained to those with Osr1 knockdown (Fig. 3F and G). These results demonstrate that NGF-induced neural development via WNK1 and HSN2 require OSR1 activity.
The HSN2-OSR1-LHX8 pathway is important for NGF-induced neural development.
LHX8 is a key regulator of cholinergic neural function and is involved in the determination of cholinergic and GABAergic cell fate19,20. We previously demonstrated that the WNK1-OSR1 pathway is important for Lhx8 expression23. To determine whether Lhx8 expression is mediated by the HSN2-OSR1 pathway, we examined the effects of HSN2 constructs on Neuro2A cells in the absence of NGF stimulation. We confirmed that expression of WNK1 or HSN2 mediated Lhx8 induction, and that knockdown of Osr1 suppressed this induction (Fig. 4A). We also examined the effect of a kinase-negative OSR1K46M mutant on Lhx8 expression, and found that OSR1K46M also suppressed Lhx8 expression induced by WNK1 or HSN2 expression (Fig. 4B). In contrast, HSN2-delA and HSN2-insT did not induce Lhx8 expression (Fig. 4A and B). We next analysed whether HSN2-induced neurite outgrowth and neural specification in Neuro2A cells is mediated by LHX8. Lhx8 knockdown inhibited not only WNK1- but also HSN2-induced neurite outgrowth (Fig. 4C). Furthermore, although the expression of HSN2 caused an increase in ChAT expression and a decrease in Gad1 expression, knockdown of Lhx8 suppressed ChAT expression and enhanced Gad1 expression by the expression of HSN2 (Fig. 4D). These results indicate that HSN2-induced neural differentiation, including neurite elongation and expression of cholinergic neuron marker genes, was mediated by LHX8 through OSR1 activity.
HSN2 mutants suppress neural development via GSK3β.
GSK3β is a positive effector of WNK signalling in neural development24. To confirm whether HSN2 mediates NGF-induced neural specification through GSK3β, we examined the effect of Gsk3β knockdown on neural development of Neruo2A cells. As shown in Fig. 5A, WNK1- and HSN2-mediated neurite elongation was suppressed by siGsk3β treatment. Moreover, HSN2 mutants suppressed NGF-induced neurite elongation, and knockdown of Gsk3β completely inhibited neurite outgrowth (Fig. 5A). We also analysed expression of neural marker genes in Neuro2A cells and found that Gsk3β knockdown suppressed NGF-induced Lhx8 and ChAT expression, even though WNK1 or HSN2 was expressed (Fig. 5B). Furthermore, although WNK1 or HSN2 expression decreased NGF-induced Gad1 expression, knockdown of Gsk3β rescued this reduction (Fig. 5B). These results demonstrate that NGF-induced neural development via WNK1 and HSN2 requires not only OSR1 activity but also GSK3β.
We next analysed the interaction between GSK3β and HSN2, and found that HSN2 interacted more strongly with GSK3β than with WNK1 (Fig. 5C). Interestingly, HSN2 mutants associated more robustly with GSK3β than with wild-type HSN2 (Fig. 5D). From these results, we predicted that HSN2 mutants might suppress the binding of WNK1 and HSN2 to GSK3β by covering the binding sites of GSK3β. To confirm this hypothesis, we transiently expressed Flag-GSK3β and HA-WNK1 or HA-HSN2 with or without Myc-HSN2-delA and Myc-HSN2-insT mutants in HEK293T cells and then immunoprecipitated cell extracts with a Flag antibody. The binding of WNK1 and HSN2 to GSK3β was inhibited by expression of each HSN2 mutant (Fig. 5E and F), indicating that HSN2 mutants suppress the function of GSK3β in the WNK1/HSN2 pathway by preventing the interaction. We then examined the effect of HSN2 mutants on GSK3β function in neural development. As shown in Fig. 5G, GSK3β expression weakly induced neurite elongation of Neuro2A cells and WNK1 and HSN2 enhanced this induction under treatment with vehicle or NGF. In contrast, HSN2 mutants inhibited the GSK3β-induced neurite elongation (Fig. 5G). Consistent with these results, RT-PCR and qPCR analysis showed that GSK3β elevated NGF-induced Lhx8 and ChAT expression and decreased Gad1 expression (Fig. 5H). In addition, although wild-type HSN2 enhanced the expression of Lhx8 and ChAT, HSN2 mutants suppressed the induction of these genes and facilitated Gad1 expression (Fig. 5H). These results indicate that HSN2 mutants suppress the neural development function of GSK3β by preventing the interaction of WNK1 and HSN2 with GSK3β.