Dishevelled coordinates phosphoinositide kinases PI4KIIIα and PIP5KIγ for efficient PtdInsP₂ synthesis

This paper investigates in living cells the organization and cooperative action of phosphoinositide kinases and Dishevelled (Dvl) proteins, a family of scaffolding proteins that organize effectors of Wnt signaling pathways (Sharma et al., 2018). Phosphoinositides are minority membrane lipids with important functions in cellular signaling. The heterogeneous distribution of phosphoinositides gives each cellular membrane a distinct lipid signature, and this distribution is kept in homeostasis by regulation of lipid kinases and phosphatases. Phosphatidylinositol (PtdIns) can be phosphorylated on the 3-, 4-, or 5-hydroxyl positions of its myo-inositol ring to generate seven species of phosphoinositides, including phosphatidylinositol(4,5)-bisphosphate (PtdInsP₂), which resides on the inner, cytoplasmic leaflet of the plasma membrane. PtdInsP₂ is an important signaling molecule with diverse cellular functions including regulating ion channels and transporters, remodeling the cytoskeleton, and regulating vesicle movement and fusion (Balla, 2013; Mandal, 2020).

Our focus is the generation of PtdInsP₂ from PtdIns via phosphorylation by phosphatidylinositol 4-kinase (PI4K) and phosphatidylinositol 4-phosphate 5-kinase (PIP5K). The synthesis of PtdInsP₂ from PtdIns is typically considered as two sequential and physically separate reactions catalyzed by PI4K and PIP5K (Fig. 1A, top). A recent kinetic model of phosphoinositide dynamics (Olivença et al., 2018) found it necessary to postulate direct transformation of PtdIns into PtdInsP₂ to match experimental rates of PtdInsP₂ production, especially under conditions of low phosphatidylinositol (4)-phosphate (PtdInsP). Previous evidence for a ternary complex that includes both kinases and a scaffolding protein (Fig. 1A, bottom) comes from co-immunoprecipitation experiments showing physical association amongst PI4K type IIα (also known as PI4K2A), PIP5K and a Dvl scaffolding protein during Wnt signaling (Qin et al., 2009). We assessed the degree to which a different type of PI4K and PIP5KIγ (also known as PIP5K1C) associate and the functional consequences of this partnership in the context of two receptor signaling cascades.

Here, we compared signaling from two types of receptors. The M₁ muscarinic acetylcholine receptor (CHRM1, referred to here as M₁ receptor) couples to Gαq and phospholipase C beta (PLCβ) to hydrolyze PtdInsP₂ to inositol (1,4,5)-trisphosphate (InsP₃) and diacylglycerol, leading to several downstream events including an increase in Ca²⁺ concentration, membrane depolarization and regulation of several ion channels (Hille et al., 2015). The Ror2 receptor is a tyrosine kinase-like non-canonical Wnt receptor known to associate with Dvl proteins (Nishita et al., 2010; Witte et al., 2010). The early signaling events initiated by Ror2 are less well defined. In neurons, activation of Ror2 leads to an increase in intracellular Ca²⁺, membrane depolarization and increased trafficking of NMDA receptors (Cerpa et al., 2011; 2015; McQuate et al., 2017). Dvl proteins regulate the cytoskeleton, modify synaptic activity and shuttle to the nucleus (Ahmad-Annuar et al., 2006; Castro-Piedras et al., 2021; Sharma et al., 2018; Sheldahl et al., 2003). Genetic mutations in DVL1, DVL3, ROR2 or WNT5A cause impaired development, particularly affecting the skeleton, classified as Robinow syndrome (White et al., 2015, 2016, 2018). We tested the role of the Dvl protein Dvl3 in PtdInsP₂ metabolism downstream of both receptors and compared it to the role of phosphoinositide kinases.

The localization and regulatory context of phosphoinositide kinases is becoming apparent through structural and functional studies. The substrate of PI4K, PtdIns, is abundant on the endoplasmic reticulum, where colocalization with the PtdInsP 4-phosphatase Sac1 (also known as SACM1L) prevents the accumulation of PtdInsP. PtdIns is transported from the endoplasmic reticulum to the plasma membrane by PtdIns transfer proteins. PI4KIIIα (also known as PI4KA) – which is cytosolic when expressed alone – is anchored to the plasma membrane by cognate proteins, several pairs of which are sufficient plasma membrane tethers: EFR3A/B and TTC7A/B, EFR3A/B and TMEM150A, TTC7A/B and FAM126A (Chung et al., 2015; Lees et al., 2017; Nakatsu et al., 2012). PIP5K localizes to the plasma membrane, and the catalytic activity of PIP5K1A is enhanced by binding the DIX domain of Dvl1 (Hu et al., 2015). We build on this context to show how these proteins work together in living cells. In this paper, we show that Dvl3 organizes PI4K and PIP5K for enzymatic synergy in synthesizing PtdInsP₂, that this mechanism is essential for timely PtdInsP₂ resynthesis following M₁ receptor activation, and that Ror2 increases PtdInsP₂, likely via interaction with Dvl3 and PIP5K.

This study tests the hypothesis that Dvl proteins organize PI4K and PIP5K to facilitate synthesis of PtdInsP₂ (Fig. 1A). Immunoblots for PI4KIIIα, PIP5KIγ, and Dvl2 and Dvl3 confirmed that these proteins are expressed endogenously in untransfected human tsA-201 cells and that overexpression increases their levels as expected (Fig. 1B). Endogenous PI4KIIIα was imperceptible in whole-cell lysates but could be detected following immunoprecipitation with an anti-PI4K antibody.

We assessed PtdInsP₂ levels by measuring potassium currents upon overexpression of KCNQ2 and KCNQ3 (referred to hereafter as KCNQ2/3). Due to their high sensitivity to plasma membrane PtdInsP₂ levels (Li et al., 2005), KCNQ2/3 potassium channels (M current) are specific real-time indicators of plasma membrane PtdInsP₂ in living cells. KCNQ2/3 currents are virtually eliminated by strong activation of PLCβ and subsequent depletion of PtdInsP₂ (Horowitz et al., 2005), whereas increased levels of PtdInsP₂ are correlated with faster activation gating and slower deactivation of the channel (Dai et al., 2016). Changes to baseline PtdInsP₂ levels could be ascertained from activation and deactivation kinetics of KCNQ2/3 currents elicited by a depolarizing pulse to −20 mV. Overexpression of the scaffolding protein Dvl3 accelerated activation gating and slowed deactivation compared to control cells, indicating an increase in steady-state levels of plasma membrane PtdInsP₂ (Fig. 1C–F, blue traces). As expected, overexpressing either PI4KIIIα or PIP5KIγ also accelerated activation and slowed deactivation (Fig. 1C–F, orange and green traces, respectively).

To establish the cellular consequences of Dvl proteins on PtdInsP₂, we studied the effects of Dvl3 overexpression on the kinetics of PtdInsP₂ depletion and resynthesis in the dynamic environment of receptor activation. Activation of Gq-coupled receptors such as the M₁ receptor activates PLCβ, causing hydrolysis of PtdInsP₂ (Horowitz et al., 2005). We assessed PtdInsP₂ localization using PH-PLCδ1, a fluorophore-tagged probe that serves as a specific visual indicator of PtdInsP₂ (Varnai et al., 2006). In control cells, PH-PLCδ1 accumulated primarily at the plasma membrane, as expected, and transiently translocated to the cytosol upon activation of the M₁ receptor, reflecting hydrolysis of PtdInsP₂ by PLCβ (Fig. 2A–C; control in A, black trace in B and C). Overexpressing Dvl3 reduced the baseline cytosolic intensity of PH-PLCδ1 (Fig. 2B, blue trace), indicating increased resting plasma membrane PtdInsP₂. When M₁ receptors were activated in cells overexpressing Dvl3, much of the PH-PLCδ1 was retained at the plasma membrane; translocation to the cytosol was reduced (Fig. 2G), and the probe returned to its baseline distribution much faster than in control cells (Fig. 2H). The reduced initial translocation to the cytosol and the faster recovery of the probe to the plasma membrane suggest larger resting pools and accelerated synthesis of PtdInsP₂ in the presence of overexpressed Dvl3.

To understand the relationship of Dishevelled proteins with PI4KIIIα and PIP5KIγ, we compared the overexpression of Dvl3 with overexpression of the phosphoinositide kinases alone or together. We found that neither PI4KIIIα nor PIP5KIγ overexpressed alone changed the baseline cytosolic distribution of PH-PLCδ1, but that in combination they increased resting plasma membrane PtdInsP₂, as reflected by decreased baseline cytosolic intensity of PH-PLCδ1 (Fig. 2E, purple trace). Activation of M₁ receptors in cells overexpressing either PI4KIIIα or PIP5KIγ depleted PtdInsP₂ from the plasma membrane, seen as a transient increase in cytosolic PH-PLCδ1 (Fig. 2D,E). Overexpression of either kinase altered the kinetics of PH-PLCδ1 movement in response to activation of M₁ receptors, with PIP5KIγ delaying the release of PH-PLCδ1 from the plasma membrane during M₁ receptor activation and PI4KIIIα accelerating the return of PH-PLCδ1 to the plasma membrane during resynthesis of PtdInsP₂ (Fig. 2H). When both kinases were co-expressed, translocation of PH-PLCδ1 to the cytosol was suppressed by 62%, similar to the effect of overexpression of Dvl3 (Fig. 2G). In the few cells with evident translocation of PH-PLCδ1, kinetic rates matched those for overexpression of PIP5KIγ during PtdInsP₂ hydrolysis and those for PI4KIIIα during PtdInsP₂ resynthesis (Fig. 2F,H). These data suggest that overexpression of Dvl3 recapitulates the co-overexpression of PI4KIIIα and PIP5KIγ and support the idea that Dishevelled proteins coordinate PI4KIIIα and PIP5KIγ to act synergistically to synthesize PtdInsP₂. Our data also suggest that PI4KIIIα is rate limiting in PtdInsP₂ synthesis, in agreement with our previous work (Falkenburger et al., 2010).

Interpretation of these results must be tempered with acknowledgement of the constraints of the probe used. Although PH-PLCδ1 allows us to observe net changes in plasma membrane PtdInsP₂, it may not be a fully linear indicator of PtdInsP₂ (Myeong et al., 2020). PH-PLCδ1 mobility is likely saturated at high PtdInsP₂ levels (such that no further movement to the plasma membrane is possible even with increasing PtdInsP₂), for instance when Dvl3 is overexpressed or both kinases are overexpressed together.

Since PI4KIIIα and PIP5KIγ appear to work synergistically to synthesize PtdInsP₂ and overexpressed Dvl3 recapitulates the effects of PI4K and PIP5K co-expression, we hypothesized that these proteins might colocalize. The localization of PI4K, PIP5K and Dvl3 was probed using high-resolution confocal microscopy and fluorescence resonance energy transfer (FRET). Overexpressed PI4KIIIα was found in the cytosol, plasma membrane and a Golgi-like structure (Fig. 3A), whereas PIP5KIγ was localized primarily in the plasma membrane (Fig. 3B). To facilitate plasma membrane docking of PI4KIIIα, throughout this paper this kinase was always co-expressed with the complementary proteins TTC7B and EFR3B (Chung et al., 2015). Notably, overexpressing Dvl3 increased the plasma membrane component of PIP5KIγ (Fig. 3B) but not of PI4KIIIα (Fig. 3A). Overexpressing PI4KIIIα also increased the plasma membrane component of PIP5KIγ (Fig. 3D), whereas PIP5KIγ only marginally increased the proportion of PI4KIIIα at the plasma membrane (Fig. 3C). Knocking down Dvl3 decreased the plasma membrane component of PIP5KIγ (Fig. 3F) but had no effect on PI4KIIIα (Fig. 3E). To quantify the magnitude of interaction between overexpressed PI4KIIIα–GFP and PIP5KIγ–CFP, we calculated the Pearson's r coefficient from confocal micrographs of control cells, those additionally overexpressing Dvl3, and those with knockdown of Dvl3 using siRNA. Co-expression of Dvl3 increased the Pearson's coefficient (mean Pearson's r value=0.37) compared to control cells (mean Pearson's r value=0.27), and knocking down Dvl3 reduced colocalization (mean Pearson's r value=0.18; Fig. 3G). Measurements of FRET between PIP5KIγ–CFP and Dvl3–YFP support the hypothesis that Dvl3 organizes PIP5KIγ via physical interaction (Fig. 3H).

To determine the necessity of Dvl3 for PtdInsP₂ resynthesis, we tested the effects of Dvl3 loss of function on steady-state PtdInsP₂ levels and on the kinetics of PtdInsP₂ resynthesis. An siRNA against Dvl3 (siDvl3) effectively eliminated Dvl3 expression (Fig. 4A). Quantification of PI4KIIIα and PIP5KIγ showed that knocking down Dvl3 had no effect on the endogenous expression of PI4KIIIα or PIP5KIγ (Fig. 4B). Since the voltage activation of KCNQ2/3 channels is strongly sensitive to PtdInsP₂ levels (Kim et al., 2016), we measured the current–voltage relationship in cells with and without expression of siDvl3. With siDvl3, KCNQ2/3 currents required more depolarized potentials to activate, consistent with a decrease in PtdInsP₂ (Fig. 4C). An increase in PtdInsP₂ induced by overexpression of PIP5KIγ had the opposite effect (Fig. 4C), permitting channel opening at less depolarized potentials. Tail currents elicited at a range of voltages were fitted with sigmoid curves yielding a mid-point voltage V_1/2 of −11.3 mV for control cells, −28.1 mV for cells overexpressing PIP5KIγ and +5.0 mV for cells with siDvl3 (Fig. 4C).

Consistent with a decrease in steady-state plasma membrane PtdInsP₂, siDvl3 reduced the resting fraction of PH-PLCδ1 at the plasma membrane (Fig. 4D). In addition, PH-PLCδ1 returned only partially to the plasma membrane following activation of M₁ receptors (Fig. 4E), indicating a compromised capacity for PtdInsP₂ resynthesis. Treating cells for 15 min with GSK-A1, an inhibitor of PI4KIIIα, elicited a strikingly similar reduction of plasma membrane PtdInsP₂ at rest (Fig. 4F) and impaired resynthesis after receptor activation (Fig. 4G). These data indicate that both PI4KIIIα and Dvl3 are necessary for efficient resynthesis of PtdInsP₂.

Ror2 is a non-canonical Wnt receptor activated by Wnt5a whose partially characterized downstream signaling includes Dvl proteins. We overexpressed Ror2 receptors and tested whether application of Wnt5a changes the localization of Dvl3 or PIP5KIγ. Using total internal reflection fluorescence (TIRF) microscopy, we found that Wnt5a induced a small and transient translocation of Dvl3 to the plasma membrane, peaking ∼30 s after agonist application and decreasing before agonist was removed (Fig. 5A,B). During this time, the Ror2 receptor remained stable in the plasma membrane. Likewise, confocal microscopy showed a brief recruitment of PIP5KIγ to the plasma membrane, followed by formation of cytosolic bodies labeled for PIP5KIγ (Fig. 5C,D). Thus, addition of Wnt5a to cells overexpressing Ror2 receptors was sufficient to recruit Dvl3 and PIP5KIγ.

We performed immunoblots on cell homogenates from cells overexpressing Ror2 to establish whether Dvl3 is phosphorylated upon application of Wnt5a. Phosphorylation of Dvl proteins is thought to be part of their mechanism of activation (Nishita et al., 2010; Witte et al., 2010). Dvl3 phosphorylation was quantified from immunoblots using an anti-phosphotyrosine antibody, and blot intensity was normalized to control non-stimulated cells (Fig. 5E). We found that stimulation with Wnt5a increased Dvl3 tyrosine phosphorylation, whereas Wnt7a, a canonical agonist that fails to activate Ror2 (Cerpa et al., 2015), had no effect (Fig. 5E).

Overexpressing Ror2 receptors accelerated activation gating of KCNQ2/3 current (Fig. 5F) and slowed down the deactivation of the channel, consistent with an increase in resting plasma membrane PtdInsP₂. Ror2 receptor overexpression also blunted the translocation of PH-PLCδ1 to the cytosol in response to M₁ receptor activation (Fig. 5G), reflective of increased PtdInsP₂ at the plasma membrane and/or accelerated PtdInsP₂ resynthesis. Interestingly, the recovery of PH-PLCδ1 to the plasma membrane appears to be biphasic when Ror2 receptor is overexpressed (Fig. 5G). Confocal micrographs of PH-PLCδ1 taken in the same period show rod-shaped regions adjacent to the plasma membrane (denoted with red arrows in Fig. 5G). It will be interesting to see whether these possibly intracellular pools of PtdInsP₂ contribute to the slow, second phase of plasma membrane PtdInsP₂ recovery in these cells. This experiment was repeated with Tubby_C^R332H, another optical probe for PtdInsP₂. Similar kinetics for PtdInsP₂ depletion and recovery upon M₁ receptor activation were observed (Fig. 5H). Even in the absence of exogenous agonist, Ror2 overexpression seems to increase the plasma membrane pool of PtdInsP₂.

We conclude with a quantitative assay for phosphoinositides. Mass spectrometry was used to measure levels of PtdInsP, PtdInsP₂, and PtdInsP₃. The conditions of transient transfection matched those for which protein expression was detected in Fig. 1B. In cell populations overexpressing PIP5KIγ, we found as expected that PtdInsP, the substrate of the PIP5K, declined by 29% and PtdInsP₂, the product, increased by 41% compared to levels in control cells. However, we were surprised to find that overexpression of PI4KIIIα reduced the prevalence of its immediate product, PtdInsP, by 22% while still increasing PtdInsP₂ by 29% (Fig. 6A,B). What could account for the diminution of PtdInsP when PI4K is overexpressed? We hypothesize that an interaction with PI4KIIIα increased the enzymatic activity of PIP5K. Efficient phosphorylation, without the accumulation of intermediate PtdInsP, could be accomplished by coupling of PI4K and PIP5K. PIP5KIγ also increases the prevalence of PtdInsP₃ by 54% (Fig. 6C).

We repeated these mass spectrometry measurements in cells overexpressing Dvl3 and found a 27% decrease in PtdInsP (Fig. 6D), a 21% increase in PtdInsP₂ (Fig. 6E) and a 66% increase in PtdInsP₃ (Fig. 6F). For all three phosphoinositides, overexpression of Dvl3 reproduced the effect of overexpressing PI4KIIIα or PIP5KIγ. Dvl3 seems able to mobilize endogenous PI4K and PIP5K in a manner similar to overexpression of those kinases. Given the increase in PtdInsP₃, it could be worth investigating whether the activity of phosphatidylinositol 3-kinases also depends on Dvl3. Our functional and quantitative assays show that Dvl3 increases baseline PtdInsP₂, is required for efficient synthesis of PtdInsP₂, localizes PIP5KIγ to the plasma membrane and promotes colocalization of PI4KIIIα and PIP5KIγ.

This report shows that PI4K and PIP5K act in coordination to synthesize PtdInsP₂, and that the Dishevelled protein Dvl3 is necessary to organize the two kinases for this task (Fig. 1A). This coordination is important both for maintenance of basal levels of PtdInsP₂ and for replacement of PtdInsP₂ following receptor-induced depletion.

Dvl3 organizes the synergistic activity of PI4KIIIα and PIP5KIγ. Overexpression of Dvl3 in the presence of endogenous phosphoinositide kinases was sufficient to reproduce the changes in PtdInsP₂ observed upon simultaneous overexpression of PI4KIIIα and PIP5KIγ. Dvl3 overexpression recapitulated the increase in resting PtdInsP₂ (Figs 1,6) and acceleration of PtdInsP₂ resynthesis (Fig. 2) present with overexpression of PI4KIIIα and PIP5KIγ. By simply organizing the low-abundance endogenous kinases, Dvl3 is sufficient to increase their efficacy substantially. Indeed, when endogenous Dvl3 was knocked down, resting plasma membrane PtdInsP₂ was reduced and recovery from receptor-induced PtdInsP₂ hydrolysis was incomplete (Fig. 4). Blocking PI4KIIIα in control cells had a markedly similar effect on both resting PtdInsP₂ and PtdInsP₂ resynthesis (Fig. 4G) as Dvl3 knockdown, underscoring both the importance of Dvl and the notion that PI4K is rate limiting in these cells.

Although previous studies have suggested that PI4K is rate limiting for resynthesis of PtdInsP₂ (Falkenburger et al., 2010; Myeong et al., 2020), we provide new details about the specific roles of two kinases. In cells overexpressing PIP5KIγ (but not PI4KIIIα), the translocation of PH-PLCδ1 in response to PLCβ activation is smaller and slower than in control cells. This is presumably because PIP5KIγ continues to generate PtdInsP₂ from PtdInsP, counteracting the ongoing PtdInsP₂ hydrolysis and slowing down release of PH-PLCδ1 from the plasma membrane. However, PtdInsP is eventually depleted and becomes rate limiting during PtdInsP₂ resynthesis, as evidenced by PI4KIIIα overexpression advancing the start of PtdInsP₂ resynthesis by 10 s relative to that in control cells (Fig. 2F, orange trace versus black trace after Oxo-M). PIP5KIγ appears to be rate limiting for PtdInsP₂ synthesis during PLCβ activation, and PI4KIIIα is rate limiting for PtdInsP₂ resynthesis after agonist withdrawal. Interestingly, once begun, the rate of resynthesis is similar among control cells and those overexpressing either PI4KIIIα or PIP5KIγ.

What evidence do we have that PI4KIIIα, PIP5KIγ and Dvl3 associate in a physical complex? First, both Dvl3 and PI4KIIIα increase the presence of PIP5KIγ at the plasma membrane (Fig. 3). Second, overexpressing Dvl3 increases colocalization of PI4KIIIα and PIP5KIγ, whereas knocking down Dvl3 decreases PI4K–PIP5K colocalization (Fig. 3G). Third, FRET between PIP5KIγ and Dvl3 shows proximity at a molecular level between these proteins (Fig. 3H). Future biochemical and structural experiments will be important in clarifying how these proteins interact.

Following application of Wnt5a in cells overexpressing Ror2, we observed sequential and transient recruitment of Dvl3 and PIP5KIγ to the plasma membrane and increased phosphorylation of Dvl3 (Fig. 5A–E). In addition, Ror2 overexpression increases basal PtdInsP₂ and blunts the translocation of PH-PLCδ1 (Fig. 5G) comparably to overexpression of PIP5KIγ (compare Fig. 1C, Fig. 2D–H). These effects may be due to Ror2 interacting with Dvl3 and recruiting PI4K and PIP5K to the plasma membrane. It will be interesting to investigate these signaling events in neurons, where the depolarization and Ca²⁺ concentration increase downstream of Wnt5a appear to depend on Gαq and PLC (McQuate et al., 2017). For instance, does Ror2 coordinate PI4K and PIP5K activity in hippocampal neurons, and how does that impact events downstream of PLC? Ror2 signaling appears to be context specific depending on co-receptors and accessory proteins (Green et al., 2014), so signaling pathways may vary depending on cell type and receptor expression.

In conclusion, we have shown that Dvl3 organizes PI4KIIIα and PIP5KIγ to synthesize PtdInsP₂ efficiently, especially in response to receptor-induced PtdInsP₂ depletion. Ror2 overexpression increases plasma membrane PtdInsP₂, and Wnt5a stimulates Dvl3 phosphorylation and plasma membrane recruitment of Dvl3 and PIP5KIγ. It will be interesting to determine whether kinase organization plays a role in the etiology of Robinow syndrome arising from mutations in Dvl3.

Oxotremorine-M (Oxo-M), GSK-A1, sodium formate, HCl and trimethylsilyl-diazomethane (TMS-DM; 2.0 M in diethyl ether or hexanes) were from Sigma-Aldrich, and Wnt5a was from BD Biosciences. Mass spectrometry-grade methanol, chloroform, dichloromethane and acetonitrile were from Fisher Scientific.

Cells of the tsA-201 cell line (RRID:CVCL_2737, Sigma-Aldrich), originally derived from HEK-293 human embryonic kidney cells and certified by the vendor by short tandem repeat profiling, were cultivated in DMEM (Fisher Scientific) with 10% fetal bovine serum (Sigma-Aldrich) and 2% penicillin-streptomycin (Gibco), and were used between passages 15 and 50. Cells were transfected at 75–85% confluency using Lipofectamine 2000 or 3000 (Invitrogen) with the following cDNA plasmids: Ror2–mCherry (Cerpa et al., 2015); PI4KIIIa–GFP, EFR3B and TTC7B (from Pietro de Camilli, Yale University, USA); human EGFP–PIP5KIγ_i2 and CFP–PIP5KIγ_i2 (from Rosa Ana Lacalle and Santos Mañes, Centro Nacional de Biotecnología/Consejo Superior de Investigaciones Científicas, Madrid, Spain; Lacalle et al., 2015); human Dvl3–EYFP and Dvl3–HcRed (from Randall Moon, University of Washington, USA); mouse M1 muscarinic receptor (from Neil Nathanson, University of Washington, USA); human eCFP–PH-PLCδ1, eYFP–PH-PLCδ1 and CFP–CAAX (from Kees Jalink, the Netherlands Cancer Institute, Amsterdam, The Netherlands); RFP–PH-PLCδ (from Ken Mackie, Indiana University, USA), Tubby_C^R332H–YFP (from Andrew Tinker, University College London, UK); human KCNQ2 and rat KCNQ3 (from David McKinnon, SUNY Stony Brook, USA); and siRNA for Dvl3 (Santa Cruz Biotechnology). TYE 563 dye (Integrated DNA Technologies) was included with siRNA transfection to screen for successfully transfected cells.

Mass spectrometry was performed as previously described (de la Cruz et al., 2020). In brief, lipids were precipitated from pelleted cells treated with trichloroacetic acid and extracted with chloroform–methanol. Dried extracts were methylated with TMS-DM and quantified by targeted analysis as described previously (Traynor-Kaplan et al., 2017). For each phosphoinositide type, peaks were analyzed for the three most predominant species: 38:4, 36:2 and 36:1. Peak areas were converted to mg per 10⁶ cells as described using 37:4 lipid internal standards (Avanti Lipids) and DNA-per-sample measurements.

KCNQ2/3 current was recorded by patch clamp in whole-cell configuration at room temperature with a HEKA EPC 9 amplifier (HEKA Elektronik). The resistance of the borosilicate glass pipettes was 3–5 MΩ. Series resistance was ≤10 MΩ and was compensated ≥70%. The bath solution consisted of 160 mM NaCl, 2.5 mM KCl, 2 mM CaCl₂, 1 mM MgCl₂, 10 mM HEPES and 8 mM glucose adjusted to pH 7.4 with NaOH. The pipette solution consisted of 175 mM KCl, 5 mM MgCl₂, 5 mM HEPES, 0.1 mM BAPTA K₄ (Invitrogen), 3 mM Na₂ATP and 0.1 mM Na₃GTP adjusted to pH 7.4 with KOH.

Confocal images were obtained with an inverted confocal system (Zeiss LSM 880) run by ZEN black v2.3. A Plan Apochromat 63×/1.40 NA oil immersion objective was used. Fluorescent proteins were excited with three lasers tuned to 405 nm, 458–514 nm and 561 nm. Emission light was detected using an Airyscan 32 GaAsP detector and appropriate emission filter sets. The point spread functions were calculated using ZEN black software using 0.1 μm fluorescent microspheres. After deconvolution, the point spread functions were: 124 nm in x–y and 216 nm in z (for 488 nm excitation); 168 nm in x–y and 212 nm in z (for 594 nm excitation). Temperature inside the microscope housing was 27–30°C. Time series were taken with an interval of 2–5 s. Drug perfusion while imaging used a gravity system set to flow at 2 ml min⁻¹. Images were analyzed with ImageJ (NIH, Bethesda, MD, USA). Colocalization analysis was performed using the Fiji (https://fiji.sc/) plugin Coloc 2 to measure the Pearson correlation coefficient.

TIRF microscopy was performed on cells plated on #0 glass coverslips fitted to a recording chamber. TIRF footprints were acquired using a Nikon Eclipse Ti-E microscope equipped with a 60×/1.25 oil-immersion objective and Photometrics QuantEM EMCCD camera (Nikon). Time-series images were taken every 10 s at 23°C and analyzed with ImageJ.

Cells were homogenized in 500 µl of lysis buffer [50 mM Tris-HCl (pH 7.4), 150 mM NaCl, 5 mM EDTA, 1 mM EGTA, 0.5% protease inhibitor (Millipore Sigma), 1% Triton X-100 and 1% (w/v) deoxycholate] and centrifuged at 11,200 g for 10 min. Supernatant was extracted, proteins were quantified using the BCA method, and 20 μg of protein was run on a 7.5% polyacrylamide gel. Proteins were transferred to nitrocellulose membrane, blocked with 5% milk for 2 h, and incubated with the primary antibody: rabbit anti-PI4K antibody (#4902; Cell Signaling Technology), mouse anti-PIP5KIγ antibody (H-9, sc-377061; Santa Cruz Biotechnology) or rabbit anti-Dvl3 antibody (#3218; Cell Signaling Technology) for PI4KIIIΑ, PIP5KI and Dvl3, respectively.

Samples were incubated in primary antibody (1:5000 dilution) overnight and in secondary antibody (1:10,000 dilution) for 2 h [mouse anti-rabbit IgG–HRP (sc-2357, Santa Cruz Biotechnology) or m-IgGκ BP–HRP (sc-516102, Santa Cruz Biotechnology)]. Antibody complexes were detected by chemiluminescence using SuperSignal West Femto Kit (Thermo Fisher Scientific, USA). Chemiluminescence was captured using C-DiGit^® Blot Scanner (LI-COR Biosciences, USA).

Cells from three 35 mm dishes were lysed using 500 µl of lysis buffer and centrifuged at 11,200 g for 10 min. The supernatant was extracted for immunoprecipitation by incubation with 0.5 μg of the rabbit anti-PI4K antibody (#4902; Cell Signaling Technology) and 20 µl Protein A–Sepharose™ 4B (Invitrogen) at 4°C for 2 h. The immune complexes obtained were precipitated for 1 min at 11,200 g and washed with 1× phosphate-buffered saline (PBS) three times before being dissolved in 15 µl 6× SDS loading buffer (Thermo Fisher Scientific, USA). The sample was run in denaturing electrophoresis with standard running buffer containing SDS and transferred to nitrocellulose membrane, blocked with 5% milk for 2 h, and incubated with the primary antibody (1:5000, rabbit anti-PI4K; #4902, Cell Signaling Technology) at 4°C overnight. A secondary incubation with m-IgGκ BP–HRP (sc-516102; Santa Cruz Biotechnology) at 1:10,000 was performed, and the antibody complexes were detected by chemiluminescence using SuperSignal West Femto Kit (Thermo Fisher Scientific, USA). Chemiluminescence was captured using a C-DiGit^® Blot Scanner (LI-COR Biosciences, USA).

Immunoprecipitation of cells was performed as above, except that supernatant was extracted by incubation with anti-Dvl3 antibody (1:2000; #3218; Cell Signaling Technology) and membranes were incubated with 1:5000 mouse anti-phosphotyrosine primary antibody (#05-321; EMD Milipore Corp.).

Mean values in text and figures are given with s.e.m. calculated using GraphPad Prism. P-values were determined using two-tailed unpaired Student's t-test to compare two groups or ANOVA with Dunnett's multiple comparisons test for three or more groups.

We thank Lea Miller for administrative support, Alexis Traynor-Kaplan for assistance with mass spectrometry experiments, and Bertil Hille for invaluable mentorship, critical discussion and feedback on the manuscript.

The peer review history is available online at https://journals.biologists.com/jcs/article-lookup/doi/10.1242/jcs.259145.