Direct modification and regulation of a nuclear receptor-PIP₂ complex by the nuclear inositol-lipid kinase IPMK

Abstract

Phosphatidylinositol (4,5)-bisphosphate (PIP₂) is best known as a plasma membrane-bound regulatory lipid. While PIP₂ and phosphoinositide-modifying enzymes coexist in the nucleus, their roles in the nucleus remain unclear. Here we show that the nuclear inositol polyphosphate multikinase (IPMK), which functions both as an inositol- and a PI3-kinase, interacts with the nuclear receptor SF-1 (NR5A1) and phosphorylates its bound ligand, PIP₂. IPMK failed to recognize SF-1/PIP₂ after blocking or displacing PIP₂ from SF-1’s large hydrophobic pocket. In contrast to IPMK, p110 catalytic subunits of type 1 PI3-kinases were inactive on SF-1/PIP₂. These and other in vitro analyses demonstrated specificity of IPMK for the SF-1/PIP₂ protein/lipid complex. Once generated, SF-1/PIP₃ is readily dephosphorylated by the lipid phosphatase PTEN. Importantly, decreasing IPMK or increasing PTEN expression greatly reduced SF-1 transcriptional activity. This ability of lipid kinases and phosphatases to alter the activity and directly remodel a non-membrane protein/lipid complex such SF-1/PIP₂, establishes a new pathway for promoting lipid-mediated signaling in the nucleus.

Introduction

The coexistence of phosphoinositide phosphates (PIP_n) and their modifying enzymes in the nucleus (1–3) suggests that similar to cytosolic signaling, lipid second messengers are generated to affect nuclear events. In turn, nuclear effector proteins are postulated to interpret or decode nuclear lipid signals by interacting with phosphoinositide headgroups, or in some instances, by binding the lipid tails of phosphoinositides (1). In addition to the presence of phosphoinositides on the inner nuclear envelop (4), others report that phosphoinositides, especially PI(4,5)P or PIP₂ are present in detergent-treated nuclei stripped of their membrane (5). The existence of an intranuclear pool of PIP₂ not associated with a classic lipid bilayer is further supported by colocalization of PIP₂ and lipid kinases to nuclear speckles (6, 7). However, the importance of phosphoinositides in nuclear events remains enigmatic, in part because very few nuclear effector proteins have been identified. Nuclear proteins known to interact with PIP₂ include the poly (A) polymerase Star-PAP that controls expression of select mRNAs (8), ALY, which mediates mRNA export (9), and BRG1, a component of the BAF chromatin remodeling complex (10). Also included in this short list are members of the NR5A subfamily of nuclear receptors, steroidogenic factor-1 (SF-1, NR5A1) and liver receptor homolog (LRH-1, NR5A2). Both are capable of binding PIP₂ as well as other phospholipids, which are thought to act as regulatory ligands (11–15).

SF-1 is a key factor in regulating endocrine transcriptional networks whose targets include a host of steroidogenic enzymes and peptide hormones during both embryonic and adult life. NR5A1 human mutations are frequent, accounting for several monogenic forms of human disorders in sexual development (DSS), male infertility and premature ovarian failure (16). In our previous study, one such missense mutation (R255L) positioned near the entry of the ligand binding pocket diminished PIP₂ binding and reduced SF-1 activity (12, 17). While many proteins recognize phospholipids by either interacting with the headgroup (AKT (18)) or enveloping the entire phospholipid (sfh-1, (19)), SF-1 binds phospholipids with the acyl chains buried deep in the hydrophobic ligand binding domain (LBD) leaving the head group exposed at the mouth of the pocket (Fig. 1A). This protein-lipid arrangement prompted us to ask if phosphoinositide 3-kinases (PI3-kinases) might target the solvent exposed 3 hydroxyl present in an SF-1/PIP₂ complex. Among PI3-kinases, the best studied are p110s, which participate in cytosolic lipid signaling. More recently, the evolutionarily conserved inositol polyphosphate multikinase (IPMK/Ipk2/Arg82) was also shown to convert PI(4,5)P to PI(3,4,5)P in vitro and in vivo (20, 21). IPMK belongs to the large inositol kinase superfamily (22), and is essential for generating higher-ordered phosphorylated inositols (23, 24). Prior to the realization that IPMK possessed kinase activity, yeast genetic screens identified IPMK as a nuclear factor required for adaptive transcriptional responses (23, 25–27). More recently, mouse genetic studies revealed that IPMK is essential for normal embryonic development (21, 28). Here, we assessed whether PI3-kinases and phosphatases directly modify the exposed phospholipid headgroup in an SF-1/PIP_n complex, and asked how such modifications might affect SF-1-mediated transcription.

Fig. 1. IPMK Generates SF-1/PIP₃ From SF-1/PIP₂.

(A) SF-1/PIP₂ model based on the crystal structure of SF-1/PC (PDB ID 3F7D) with SF-1 surface charges represented as: blue = basic, red = acidic, and PIP₂ shown as stick representation. (B) Anion-exchange chromatography of purified mouse SF-1/PIP_n complexes, as indicated. (C) In vitro ³²P-labeled kinase (IVK) assays carried out with rIPMK and p110α/p85α (p110α) on SF-1/PIP₂ complexes or PIP₂-micelle substrates, analyzed by SAX-HPLC. Elution time of the PIP₃ species is indicated (arrow). (D) Coomassie-stained SF-1/PIP_n complexes from (B) resolved by 4–20% native-PAGE. (E) Autoradiography of IVK reactions analyzed by native-PAGE, using indicated SF-1/PIP_n substrates (1 μM) and indicated enzymes (100nM). (F) IVK reactions with SF-1/PIP₂ (1 μM) carried out at indicated times and concentrations of IPMK.

Results

IPMK Specifically Phosphorylates PIP₂ when bound to SF-1

To determine if IPMK and/or p110 PI3-kinases phosphorylate PIP₂ when bound to SF-1, SF-1 LBD-PIP₂ (SF-1/PIP₂) complexes were formed from purified recombinant protein as described previously (12) (Fig. 1B). Using in vitro kinase (IVK) assays and SF-1/PIP₂ as substrate, rat IPMK (rIPMK) was able to generate PI(3,4,5)P (PIP₃) as monitored by SAX-HPLC (Fig. 1C, upper panel). PIP₃ signals were undetectable with p110 PI3-kinases (α or γ) or with a kinase-dead rIPMK (D127A) (Fig. 1C lower panels and Fig. S1), illustrating that IPMK, but not p110 PI3-kinases phosphorylate PIP₂ when bound to SF-1. Using native-PAGE, which preserves the association between SF-1 and PIP_n (Fig 1D), we confirmed that SF-1/PIP₃ could be generated only from SF-1/PIP₂ (Fig. 1E). Moreover, neither p110α nor the kinase-dead IPMK (D144A) exhibited such activity (Fig. 1E). IPMK phosphorylates SF-1/PIP₂ in a dose-dependent manner, with saturation observed in typical reaction times (15 min) (Fig. 1F). Not surprisingly, the structurally related inositol kinase IP₃K lacking PI3-kinase activity (29), failed to act on SF-1/PIP₂ (Fig S1E).

To better define the activity of IPMK on SF-1/PIP₂ we used the newly described SF-1 ligand RJW100, which efficiently displaces PIPs from SF-1 (30). An RJW100-dependent drop in labeled SF-1/PIP₃ after analyzing IVK reactions by native-PAGE (Fig 2A) and by nitrocellulose capture (Fig 2B); this later method detects PIP₃ as well as SF-1/PIP₃ (31). These data establish two important points: first, PIP₂ and not the SF-1 protein is phosphorylated by IPMK, and second, in these IVK conditions, IPMK recognizes PIP₂ only when bound to SF-1. Comparison of IPMK activities on SF-1/PIP₂ and PIP₂-micelles revealed a preference for the protein/lipid substrate, with activity saturating at lower concentrations than observed for PIP₂ micelles (Fig. 2C). For these assays, the PIP₂-micelle and SF-1/PIP₂ substrates contained equal concentrations of PIP₂ (1 μM) and preparation of the PIP₂-micelle followed standard protocols as described in Materials and Methods. Kinetic parameters describing IPMK activity on SF-1/PIP₂ (Fig 2D) are comparable to values obtained for inositol phosphates (32). When compared to PIP₂-micelles, the specificity constant for SF-1/PIP₂ was found to be higher (Fig 2E); for these latter studies the total phospholipid content was held constant at 100 μM and the ratio of PS carrier lipid to PIP₂ adjusted accordingly.

Fig. 2. SF-1/PIP₂ Is A Bona Fide Substrate Of IPMK In Vitro.

(A) Autoradiography of IVK reactions of IPMK on SF-1/PIP₂ substrate (1 μM) with increasing concentrations of RJW100 (SF-1 ligand, μM) analyzed by native-PAGE. (B) IPMK IVK assays on SF-1/PIP₂ (1 μM) with increasing concentrations of RJW100 ligand added to displace PIP₂, analyzed by nitrocellulose capture assay and quantified by phosphorimaging as described in Materials and Methods. (C) Comparison of IVK reactions using increasing amounts of IPMK and either 1 μM SF-1/PIP₂ or PIP₂-micelles, which were generated with phosphatidylserine carrier lipid as described in Methods. (D) IPMK reaction velocities plotted against increasing concentrations of SF-1/PIP₂ or (E) PIP₂-micelles. Data were fit to both non-linear (Michaelis-Menton) and linear double-reciprocal (Lineweaver-Burke) curves by GraphPad Prism software, refer to Fig S2 for double reciprocal plots. Kinetic parameters describing each enzyme/substrate pair, as indicated, (+/− represents standard error except r², where +/− represents the absolute sum of squares for the curve fit).

We next compared IPMK activity on SF-1/PIP₂ with that of inositol phosphates (InsPs). Inositol phosphates compete with SF-1/PIP₂ for IPMK activity, with the best InsPs substrates competing at a 5-fold molar concentration. IC50 values for four different InsPs correlated well with the rank order of the Vmax/Km values published for human IPMK (32), and competed for SF-1/PIP₂ in the following order from best to worst competitor: Ins(1,3,4,6)P ≈ Ins(1,4,5)P > Ins(1,4,5,6)P ≫ Ins(1,3,4,5)P (Fig 3A). Further kinetic analyses with Ins(1,4,5)P, one of the higher affinity InsP substrates, established that SF-1/PIP₂ and Ins(1,4,5)P bind the identical active site in IPMK (Fig. 3B). Collectively, these in vitro data show that SF-1/PIP₂ is a valid substrate for IPMK.

Fig. 3. SF-1/PIP2 binds to the identical active site in IPMK as phosphorylated inostiols.

(A) IPMK reaction velocities on SF-1/PIP₂ (10 μM) in the presence of competitor phosphorylated inositols (InsP), as indicated. (B) Double-reciprocal competition plots of IPMK reaction velocities determined with indicated concentrations of Ins(1,4,5))P and SF-1/PIP₂, refer to Fig S2 for full data set.

Reducing IPMK Activity Impairs SF-1 Transcriptional Activity

To assess whether recognition of SF-1/PIP₂ by IPMK is biologically relevant in a cellular setting, we asked if expression of SF-1 target genes is altered after knocking-down or chemically inhibiting endogenous IPMK in human embryonic kidney HEK 293 cells. siIPMK resulted in a significant drop in established SF-1 targets including SHP, StAR and other transcripts, as judged by microarray profiling (Fig. 4A) and by RT-qPCR (Fig 4B). Importantly, the residual activity of an SF-1 pocket mutant (A270W, L345F), which prevents PIP₂ from binding the LBD (12), is unaffected by siIPMK treatment (Fig 4C). As expected, overexpressing rIPMK restored SF-1 target gene expression after knocking-down endogenous human IPMK (siIPMK) in HEK 293 cells; SF-1 transcripts were unaffected (Fig S3). However, neither the rIPMK KD mutant nor the Arabidopsis thaliana IPMK (atIPMK) were able to rescue (Fig. 4D). Unlike human or rat IPMK, atIPMK lacks all PI3-kinase activity, as shown here for SF-1/PIP₂ (Fig 4E), and previously for PIP₂ vesicles (21).

Fig. 4. Reducing IPMK Levels Diminishes SF-1 Transcription.

(A) Heat map of selective SF-1 targets after profiling HEK 293 SF-1 cells transfected with control siRNA (siCON) or siIPMK followed by induction of SF-1 by tetracycline (TET). (B) Changes in transcript levels of indicated genes after siRNA treatment of HEK 293 cells expressing SF-1 wild type (WT) or (C) an SF-1 pocket mutant (A270W, L345F). (D) Relative expression of SHP and StAR transcripts as in (B), following transfection wild type or kinase dead (KD) rat IPMK (rIPMK), or Arabidopsis IPMK (atIPMK). (E) Phosphorimage and quantification of IPMK IVK reactions on SF-1/PIP₂ analyzed by nitrocellulose capture, using different IPMK enzymes as indicated. Data represent mean +/− SEM, *** = p < 0.001 here and in all figures.

Activation of SF-1 targets was also monitored after adding a known inhibitor of IPMK, aurintricarboxcylic acid (ATA) and the related polyphenolic compound, epigallocatechin gallate (EGCG) (33), which acts as a broad-spectrum inhibitor of IP₃ inositol kinases. Consistent with their pharmacological profiles, ATA but not EGCG inhibited IPMK in IVK assays (Fig. 5A, right panels), and selectively decreased SF-1 target genes (Fig 5B). Similar to results with siPMK, activity of the SF-1 pocket mutant was unaffected by ATA (Fig 5C). The PI3-kinase inhibitor wortmannin failed to inhibit IPMK in vitro (Fig S4C) and generated a significantly different pattern of SF-1 target gene expression compared to ATA or siIPMK (Fig S4D). ATA exposure also decreased recruitment of SF-1 to the StAR promoter by chromatin immunoprecipitation (ChIP, Fig. 5D). Levels of total SF-1 protein, phospho-S203 SF-1, and SF-1 transcripts, as well as SF-1 subcellular localization were unchanged following ATA exposure (Figs 5E and S4A, B). Taken together, these results suggest strongly that generation of SF-1/PIP₃ from SF-1/PIP₂ by IPMK enhances SF-1-mediated transcription.

Fig. 5. Chemical Inhibition Of IPMK Diminishes SF-1 Transcription.

(A) Structures of aurintricarboxcylic acid (ATA) and epigallocatechin gallate (EGCG) and their effect on IPMK IVK activity. (B) qPCR analysis of SHP and StAR transcripts following 50 μM ATA (red) or EGCG (Blue) treatment of HEK 293 cells expressing wild type or (C) pocket mutant SF-1. (D) Recruitment of SF-1 to the endogenous promoter of StAR as measured by ChIP-qPCR in identical cellular conditions as (B). (E) Western analysis of total SF-1 (anti-FLAG) and phosphorylated SF-1 (anti-P-SF-1) after ATA or EGCG treatment as in (B).

IPMK Phosphorylates and Interacts with Cellular SF-1/PIP₂

We next determined whether full-length (FL) SF-1 purified from cultured cells is associated with PIP₂, and like the SF-1 LBD/PIP₂ substrate, is recognized by IPMK. Indeed, wild type SF-1 isolated from HEK 293 cells (Fig. S5A) is readily phosphorylated by IPMK (Fig 6A). Significantly less phosphorylation was detected when the IPMK KD mutant, the SF-1 pocket mutant, or excess RJW100 ligand are used in IVK reactions (Fig 6A). These results, coupled with the fact that IPMK fails to phosphorylate SF-1 protein itself (Fig S5B), imply that SF-1/PIP₂ is present in vivo, and is available for phosphorylation by IPMK. Finally, co-immunoprecipitation experiments showed that IPMK and SF-1 are capable of interacting in cells (Fig. 6B), further facilitating IPMK recognition of SF-1/PIP₂.

Fig. 6. IPMK Interacts With And Phosphorylates Cellular SF-1.

(A) Nitrocellulose capture assays and quantification of IVK reactions carried out with SF-1 WT or pocket mutant purified from HEK 293 cells, and using wild type or kinase-dead IPMK (100nM). RJW100 SF-1 ligand (10 μM) was added to indicated reactions 10 min prior to addition of IPMK. (B) Anti-Myc immunoprecipitation (IP) of MYC-tagged-IPMK transiently transfected into HEK 293 cells expressing Flag-tagged SF-1, with control IgG shown in last lane. Corresponding input of total SF-1 or IPMK protein is shown in lower panel.

PTEN Hydrolyzes SF-1/PIP₃ and Attenuates SF-1 Activity

The emerging roles of PTEN in the nucleus (34) raised the possibility that this lipid phosphatase might regenerate SF-1/PIP₂ by hydrolyzing PIP₃ bound to SF-1. Indeed, PTEN efficiently removed the radiolabeled 3′-phosphate from IPMK IVK reactions on SF-1/PIP₂ resulting in an upward shift on native-PAGE, whereas incubation with phospholipase D had no effect (Fig. 7A). Kinetic studies of PTEN activity on SF-1/PIP₃ revealed apparent Km and Vmax values of 1.0 μM and 0.7 μmol/min/mg (Fig S2), compared with the published values of 5 μM and 1.8 μmol/min/mg for PIP₃ in phosphatidylcholine vesicles (35). Moreover, coexpression of wild type PTEN and SF-1 in human PTEN-negative Ishikawa cells severely compromised SF-1 activity (Fig. 7B), suggesting that IPMK and PTEN oppose each other to enhance or reduce the functional output of SF-1, respectively. Thus, similar to membrane bound phospholipids, the headgroups of PIP₂/PIP₃ bound to SF-1 appear perfectly posed for modification by both lipid kinases and lipid phosphatases.

Fig. 7. PTEN Opposes IPMK Activity on SF-1/PIP₂.

(A) Radiolabeled SF-1/PIP₃, analyzed by nitrocellulose capture or native-PAGE after incubating with phosphatase PTEN or phospholipase D (PLD). (B) Aromatase-luciferase (ARO-Luc) reporter activity after transfecting Ishikawa cells with wild type SF-1 or control vector, plus PTEN or an inactive PTEN mutant (G129R). (C) Schematic depicting the multiple functions of IPMK as a soluble InsPs kinase, and as a PI3-kinase acting on both membrane bound PIP₂ and SF-1/PIP₂. SF-1 transcriptional output is affected after IPMK and PTEN modify the solvent-exposed phospholipid head group present in the bound PIP₂ or PIP₃ ligands.

Discussion

Our findings suggest that in addition to regulating inositol phosphate and phosphoinositide metabolism, IPMK also modulates gene expression by selective phosphorylation of PIP₂ when bound to a nuclear protein, such as SF-1 (Fig. 7C). IPMK was found to efficiently and selectively phosphorylate PI(4,5)P when complexed with recombinant SF-1 LBD, or when bound to purified full length SF-1. The ability of IPMK to act on cell-derived SF-1 illustrates for the first time that a substantial fraction of SF-1 is normally bound to PIP₂ in cells. By contrast, IPMK no longer recognizes SF-1/PIP₂ after displacing the phospholipid ligand with a synthetic ligand for NR5A receptors, or after introducing bulky amino acid substitutions that fill the large hydrophobic ligand binding cavity of SF-1. Thus, in the context of SF-1, IPMK targets SF-1/PIP₂ only when PIP₂ is incorporated into the LBD of SF-1. In vitro competition studies also showed that the solvent exposed PIP₂ headgroup in a SF-1/PIP₂ complex competes with inositol polyphosphates for the same binding site on IPMK. When taken together with our cellular data, we posit that protein-phosphoinositide complexes, such as SF1/PIP₂ represent a new bona fide class of substrates for the multikinase, IPMK.

Our results add to the remarkable versatility of this kinase, which is proposed to be the oldest member of the inositol kinase superfamily (36). Based on the previously elucidated NR5A/phospholipid structures, we suggest that the 3′ hydroxyl in the phospholipid headgroup is optimally positioned for recognition by IPMK. In contrast to IPMK, we found that other kinases known to phosphorylate the 3′-hydroxy group were inactive on the SF-1/PIP₂ substrate; these include human IP₃-kinase, Arabidopsis IPMK, and p110 PI3-kinases. Thus, the 3′ hydroxyl group in SF-1/PIP₂ cannot be the sole determinant for IPMK recognition of this protein/lipid substrate. Alignment of human IPMKs with other members of inositol phosphate kinase superfamily identified an IPMK-specific loop (S260-E375) that resides outside of the highly conserved core catalytic domain (32, 36). Interestingly, this loop is absent in Arabidopsis IPMK. Structure-function studies on yeast IPMK also found that this loop is required for protein-protein interactions and normal transcriptional responses (26). Clearly, it will be of interest to determine if these IPMK-specific regions mediate interaction between IPMK and SF-1 and facilitate recognition of the SF-1/PIP₂ substrate.

The multi-faceted nature of IPMK raises the possibility that the transcriptional effects of IPMK on SF-1 reflect indirect mechanisms of IPMK on phospholipid and/or inositol metabolism. However, as mentioned above, PI3-kinase inhibitor wortmannin elicited a significantly different transcriptional response than ATA or siIPMK, consistent with its failure to inhibit IPMK. The possibility that siIPMK simply alters inositol metabolism to lower SF-1 activity is a valid suggestion, especially in light of the recent discovery that Ins(1,4,5,6)P₄ functions as an modulatory bridging factor for HDAC3 and the histone deacetylase activation domain of the NCOR2/SMRT corepressor (37). However, despite the fact that atIPMK retains the inositol phosphate kinase activity of mammalian IPMKs, this plant IPMK fails to restore SF-1 activity after human siIPMK knockdown. Moreover, SF-1 activity is inert to the pan IP₃-kinase inhibitor EGCG, again showing that SF-1 is fully functional even after disturbing general inositol metabolism. Finally, given our data with the SF-1 pocket mutant and the fact that IPMK interacts with SF-1, we favor a model whereby IPMK regulates SF-1 activity by directly phosphorylating its bound phospholipid ligand, PIP₂.

This new type of IPMK activity might help explain the more intricate roles of IPMK in transcription, especially in aspects of chromatin remodeling (38). That IPMK and PTEN directly modify a protein-bound lipid to affect transcription is a clear departure from their established roles in membrane phosphoinositide and inositol metabolism (Fig 7C). As outlined by Anderson (1), these new activities would be predicted to compartmentalize nuclear PIP₂ and PIP₃ lipid signals, and analogous to the docking roles of membrane-bound phosphoinositides, may help to bridge SF-1 to other PIP_n-interacting proteins, such as the BAF remodeling complex (39), or chromatin associated PH-domain proteins (40). Thus, we speculate that the exposed inositol moiety in SF-1/PIP_n plays a similar role as inositol tetraphosphate does in bridging corepressor complexes. Such interactions may account for the observed decrease in chromatin recruitment of SF-1 after chemically inhibiting IPMK with ATA. Interestingly, silencing DAG kinase θ also reduced SF-1 activity and recruitment to target gene promoters (41). Beyond the phosphoinositide cycle, others have shown that nuclear sphingolipids and their kinases also affect histone acetylation and chromatin recruitment (41, 42). It remains to be determined whether these and other lipid modifying enzymes regulate nuclear factors by directly remodeling their bound lipid cargo, as shown here for IPMK and SF-1/PIP₂. Although further research is needed to understand how nuclear proteins, such as SF-1 decode nuclear lipid signaling to affect gene expression, our study begins to unravel the longstanding enigma of nuclear phosphoinositides.

Materials and Methods

Purification of LBD SF-1/PIP_n or Cell-Derived FL SF-1/PIP_n

SF-1 ligand binding domain (LBD) protein was purified to homogeneity as previously described (11). Briefly, mouse SF-1 constructs spanning amino acids 219–462 were expressed in E. coli, purified and complexed with dipalmitoyl (C16, fully saturated) PI, PI(4,5)P (PIP₂), or PI(3,4,5)P (PIP₃) to generate SF-1/PIP_n complexes. Protein/lipid complexes were purified from free PIPs and unbound SF-1 LBD using Mono Q chromatography on an AKTA FPLC system (Pharmacia, GE biosciences), as described (12). All PIPs were purchased from Cayman Chemical (Ann Arbor, MI), and stored under a vacuum as lyophilized powder or under N₂ as a 1g/L solution in ddH₂O.

To purify full length SF-1 from cells, 3X-FLAG tagged full-length WT SF-1 was immunopurified from two-15 cm plates, each containing 15×10⁶ HEK 293 WT or pocket mutant (A270W, L345F) SF-1 cells using M2-Anti-FLAG agarose resin (Sigma). Cells were harvested at 4°C in native lysis buffer (50 mM HEPES pH7.5, 400 mM NaCl, 0.2 mM EDTA, 5 mM MgCl2, 2 mM CHAPS, 50 mM DTT, and protease inhibitors [Roche]). Cleared lysates were incubated with M2 FLAG resin (Sigma, 30 μL of 50% slurry) for 45 minutes at 4°C, washed and eluted (50 mM HEPES pH7.5, 100 mM NaCl, 2 mM CHAPS, 50 mM DTT and protease inhibitors) with 150 μg/mL 3X FLAG peptide (Sigma). Eluates (~47 ng SF-1) were used in IPMK IVK assays, as described below, except that 1 μM cold ATP was used. For RJW100-treated IVK reactions, 20 μM RJW100 SF-1 ligand was added to IVK reactions for 15 minutes at 37°C, followed by addition of ATP to initiate IVK reactions, which were analyzed by nitrocellulose capture and by SDS-PAGE (BioRad, Hercules, CA).

IPMK purification and IVK reactions

Wild type human (accession Q8NFU5) and rat (accession AY014898) 6XHis-tagged IPMK constructs in pTRCb (gift of A. Resnick, U Penn) were expressed in E. coli, and purified as described (20). Kinase dead (KD) mutants included rat IPMK (rIPMK, D127A), and human IPMK (IPMK, D144A) (20). IPMK (EC = 2.7.1.151) IVK reactions were carried out in 50 mM HEPES pH 7.5, 5 mM MgCl₂, 10 μCi of 32P-γATP (Perkin Elmer, Boston, MA) and 10 μM cold ATP (Roche, Mannheim, Germany) at 37°C in either 25 or 50 μL reactions, except in kinetic studies (see below). IVK using human p110α/p85α (Millipore, Billerica, MA) or human p110γ (Sigma, St. Louis, MO) were performed according to manufacturer’s recommendations. PIP₂ micelles substrates were prepared using phosphatidylserine (PS) carrier lipid with the total lipid content equal to 100 μM. Lyophilized C16 saturated (dipalmitoyl) PIPs (Cayman) were kept under vacuum in borosilicate glass and directly resuspended in ddH₂O to 1 g/L, sonicated and stored under N₂ at 4°C.

For HPLC analysis of IVK reactions, lipids were extracted using methanol:chloroform (2:1), de-acylated in methylamine:methanol:n-butanol:H₂O (26:45:11:16), then polar glyceroinositol head-groups subjected to strong anion exchange HPLC using a Whatman PartiSphere SAX column over a gradient of NH₄H₂PO₄, monitored in real time using a Packard flow scintillation detector. Reference retention times included the product of p110α/p85α and p110γ activity on PIP₂ micelles and free radiolabeled ³²P-γATP.

For native-PAGE analysis, total IVK reactions were loaded and resolved on TBE-buffered 4–20% PAGE polyacrylamide gels (Lonza, Basel, Switzerland) followed by autoradiography or phosphorimaging. Nitrocellulose capture analysis of IVK reactions was performed essentially as described (31) with 10% of the entire IVK reaction spotted onto nitrocellulose (0.2 micron pore size), followed by 1M NaCl and 1% phosphoric acid washes. Membranes were phosphorimaged to quantify signal.

For PTEN assays, IPMK IVK reactions were stopped with excess ATP (10 mM) and PTEN (0.3μg) and 10 mM DTT added, and incubated for 15 minutes at 37°C. Reactions were analyzed by nitrocellulose capture assay and by native-PAGE. For kinetic analyses, PTEN reactions were coupled to p110α to quantify PTEN generation of PIP₂ from SF-1/PIP₃. PTEN reactions on SF-1/PIP₃ were stopped by a methanol:chloroform (2:1) extraction, dried down, and the PIP₂ content indirectly quantified by standard p110α IVK reactions, followed by nitrocellulose capture analyses.

IPMK kinetic parameters were measured in the linear range with respect to time and enzyme concentrations as described above using the indicated substrate concentrations and 10 mM cold ATP, analyzed by nitrocellulose capture, then subjected to both linear and non-linear curve fits using GraphPad Prism; both fits yielded similar a value. Competition with IP_ns (Cayman Chemical) freshly suspended in 50 mM HEPES (7.5) and 5 mM MgCl₂, were analyzed by nitrocellulose capture; IP_n are not retained.

Cell Lines, siRNA and Luciferase Assays

3XFLAG N-terminally tagged mSF-1 cDNA (wild type or pocket mutant (A270W, L345F) constructs were expressed in Flip-in, T-Rex tetracycline (TET)-inducible human embryonic kidney 293 (HEK 293) cells (Invitrogen Corporation, Carlsbad, CA), maintained at 37°C in 5% CO₂ in DMEM supplemented with 10% tetracycline-certified-free FBS (Hyclone, Logan, UT), SF-1 protein was induced for 18 hrs using 50 ng/mL TET in 70% EtOH.

For siRNA experiments, HEK 293 WT or pocket mutant SF-1 cells were reverse transfected with 20 nM SmartPool Silencer siRNA directed against human IPMK, or control siRNA #1, (Dharmacon, Lafayette, CO), using RNAiMAX transfection reagent (Invitrogen), according to the manufacturer’s protocol. Fifty-four hours post-siRNA transfection, SF-1 was induced for 18 hours, cells harvested and analyzed as described below. For rescue experiments, cells were transfected with 20 ng of Rattus norvegicus or Arabidopsis thaliana IPMK constructs (gift of A. Resnick) using RNAiMAX transfection reagent. HEEBO microarrays or RT-qPCR was carried out using cDNA as described below.

For cellular PTEN experiments, PTEN-negative Ishikawa cells (human endometrial carcinoma) were cotransfected with an ARO-Luc reporter (250 ng), 3X FLAG-mSF-1 (pCINEO) and WT or phosphatase dead (G129R) human PTEN (in pCDNA3) using FuGENE 6 (Roche) and analyzed 24 hrs after transfection for luciferase activity using a Ventas Microplate Luminometer (Turner Biosystems, Sunnyvale, CA). PTEN vectors were a generous gift from D. Stokoe.

Small Molecules

RJW100 SF-1 ligand was generously provided by Richard J. Whitby (Southampton University, UK) (30), and was used in a 4% final DMSO (Sigma Ultrapure) solution for all IVK reactions; PIP₂ binding to SF-1 is unaffected by 4% DMSO (30). SF-1/PIP₂ complexes (1 μM) were incubated with serially diluted RJW100 for 30 minutes at 37°C in IPMK assay buffer (see above), followed by addition of IPMK (141 nM) plus 10 μCi ³²P-ATP. After 15 minutes IVK reactions quenched by spotting onto nitrocellulose and washing, as described above. Light-sensitive aurintricarboxcylic acid (ATA, Sigma) was kept in the dark. Epigallocatechin gallate (EGCG) and wortmannin were also from Sigma. Four hours after SF-1 induction with TET (50 ng/mL), ATA or EGCG was added to HEK 293 SF-1 cells in 6 well plates for 14 hours at indicated concentrations, then harvested and analyzed.

Other Methods

HEEBO (Human Exonic Evidence Based Open-source) microarray analysis and qPCR were carried out as described previously (43). All primer sets were validated and designed to cross an intron/exon boundary, primer sequences are provided in Table S1. ChIP was carried out as described (43) using HEK 293 SF-1 cells treated with TET (50 ng/ml) followed by addition of DMSO vehicle, 50 μM ATA or EGCG for 14 hours. Primer sets are listed in Table S2.

For Co-IP studies, HEK 293 SF-1 cells grown in a 10 cm dish were transfected with 5 μg of Myc-tagged human IPMK expression vector using Fugene HD (Roche), and 36 hrs later SF-1 was induced with TET (50 ng/ml) for 16 hours. Cells were lysed in 50 mM Tris-Cl (8.0), 150 mM NaCl, 1mM EDTA, 0.5% NP-40 with protease inhibitors, cleared, and protein lysate (1 μg) was immunoprecipitation used for anti-c-Myc agarose (Sigma) and analyzed by Western blotting (anti-Myc, 1:2,500 [Bishop lab, UCSF]; anti-Flag M2, 1:10,000 [Sigma]). Other antibodies for Western blots included anti-phospho S203 SF-1 (custom antibody) at 1:1000 and anti-actin (Sigma) at 1:4000 using the Pierce Super Signal West Pico Enhanced chemiluminescence detection kit (Thermo Scientific, Waltham, MA).