PITPs as Targets for Selectively Interfering With Phosphoinositide Signaling in Cells

Abstract

Sec14-like phosphatidylinositol transfer proteins (PITPs) integrate diverse territories of intracellular lipid metabolism with stimulated phosphatidylinositol-4-phosphate production, and are discriminating portals for interrogating phosphoinositide signaling. Yet, neither Sec14-like PITPs, nor PITPs in general, have been exploited as targets for chemical inhibition for such purposes. Herein, we validate the first small molecule inhibitors (SMIs) of the yeast PITP Sec14. These SMIs are nitrophenyl(4-(2-methoxyphenyl)piperazin-1-yl)methanones (NPPMs), and are effective inhibitors in vitro and in vivo. We further establish Sec14 is the sole essential NPPM target in yeast, that NPPMs exhibit exquisite targeting specificities for Sec14 (relative to related Sec14-like PITPs), propose a mechanism for how NPPMs exert their inhibitory effects, and demonstrate NPPMs exhibit exquisite pathway selectivity in inhibiting phosphoinositide signaling in cells. These data deliver proof-of-concept that PITP-directed SMIs offer new and generally applicable avenues for intervening with phosphoinositide signaling pathways with selectivities superior to those afforded by contemporary lipid kinase-directed strategies.

Lipid signaling modulates a wide range of cellular processes, including regulation of G-protein-coupled receptors and receptor tyrosine kinases at the plasma membrane¹, actin dynamics², transcription^3,4, and membrane trafficking⁵. A major pillar of eukaryotic lipid signaling is defined by phosphoinositides and the soluble inositol (Ins) phosphates derived from them^6,7. Phosphatidylinositol (PtdIns) is an essential phospholipid that serves as metabolic precursor for both phosphoinositides and Ins-phosphates. While Ins-phosphates are chemically diverse, the phosphoinositide cabal is simpler. Yeast produce five phosphoinositides (PtdIns-3-phosphate, PtdIns-4-phosphate, PtdIns-5-phosphate, PtdIns-4,5-bisphosphate, and PtdIns-3,5-bisphosphate) while mammals produce seven; those synthesized by yeast as well as PtdIns-3,4-bisphosphate and PtdIns-3,4,5-trisphosphate⁶. This limited phosphoinositide cohort supports a diverse landscape of lipid signaling that modulates the actions of hundreds of proteins⁷.

Specific inactivation of a target enzyme is a desirable instrument for dissecting mechanisms of lipid signaling in cells. This is especially true in the context of phosphoinositide signaling whose very diversification demands highly targeted approaches for clean analysis. However, specific genetic or chemical interventions at the level of individual lipid kinases, or compartment-specific interventions at the level of defined phosphoinositide species using Rapalog technologies^8,9, remain blunt experimental instruments. Such interventions exert pleiotropic effects because many effector activities are impaired upon inhibition of a target Ins-lipid kinase, or upon compartment-specific depletion of a specific phosphoinositide species.

PtdIns-transfer proteins (PITPs) of the Sec14 protein superfamily are key regulators of phosphoinositide signaling that specify discrete biological outcomes of PtdIns kinase action^10,11. Deficiencies in individual Sec14-like PITPs compromise trafficking through the trans-Golgi network (TGN) and endosomal systems¹², phosphatidylserine decarboxylation to phosphatidylethanolamine¹³, fatty acid metabolism¹⁴, polarized growth¹⁵, and fungal dimorphism¹⁶. Mutations in PITPs, or PITP-like proteins, are also root causes of mammalian neurodegenerative and lipid homeostatic diseases^17,18.

Various lines of evidence recommend PITPs as highly discriminating portals for interrogating phosphoinositide signaling, and identify PITPs as unexploited avenues for chemical inhibition of select phosphoinositide signaling pathways in cells. Herein, we exploit the yeast system to make the case. We validate the first chemical inhibitors of a PITP, demonstrate an exquisite in vivo specificity of action for such compounds, and propose a chemical mechanism for how these SMIs exert their inhibitory effects. These studies deliver proof-of-concept that PITP-directed approaches afford powerful advantages for chemically intervening with phosphoinositide signaling, and that the selectivities achieved are superior to those delivered by strategies targeting individual PtdIns-kinase isoforms or individual phosphoinositide species.

RESULTS

Candidate Sec14-directed SMIs

Sec14, the major yeast PITP, is an essential protein required for membrane trafficking through the TGN/endosomal system¹². Chemogenomic profiling of 188 inhibitors of yeast growth identified a candidate for a Sec14-directed SMI¹⁹. This compound, 4130-1278 (1), is a 4-chloro-3-nitrophenyl)(4-(2-methoxyphenyl) piperazin-1-yl)methanone (NPPM). Since 4130-1278 exhibited mediocre potencies, and limited water solubility, we evaluated 13 other NPPM-like SMIs as Sec14 candidate inhibitors (Supplementary Results, Supplementary Fig. 1a). One such derivative, 4130-1276 (2), showed superior water solubility and arrested growth of a sec14Δ/SEC14 heterozygous strain at 10-fold lower concentrations than those observed for 4130-1278 (Supplementary Fig. 1b).

Chemogenomic profiling of ca. 6200 yeast deletion strains correlated gene-dosage with yeast sensitivity to 4130-1278 or 4130-1276 challenge on a genome-wide scale (Supplementary Fig. 2a–f). The profiling identified sec14Δ/SEC14 heterozygous diploid cells as the most sensitive to 4130-1278 and 4130-1276 challenge of all homozygous Δ/Δ and heterozygous Δ/+ diploids tested (non-essential and essential gene queries, respectively; Supplementary Fig. 2a–f). A limited set of other genes was also identified for which dosage reduction decreased fitness in the presence of 4130-1278 and 4130-1276 (Supplementary Fig. 2c, f). Gene functions identified in the more extensive 4130-1276 hit list included Golgi trafficking, sporulation, exocytosis, vacuolar transport, and lipid metabolism. A number of high scoring chemogenomic interactions, include phospholipase D (SPO14), a strong synthetic interactor with sec14-1^{ts 10,20} and the phospholipase D regulator SRF1²¹, were independently recognized in genome-scale synthetic genetic array (SGA) screens that employed sec14-1^ts as query allele^22,23.

Expansion of the candidate Sec14-directed SMI set

In focused searches for additional candidate Sec14-targeted SMIs, an expanded set of 34 compounds was assembled around the 4130-1276 scaffold. For initial rounds of compound selection, the Chembridge library was filtered using Similarity Search (Marvin Applet tool, Hit2Lead – Cambridge database). The filter clamped structural similarity to ca. 80% of the query 4130-1276 molecule and highlighted functional groups as sites for diversification (Supplementary Fig. 3a, b). NPPMs 4130-1276, 67170-49 (3) and 6748-481 (4) were chosen for study because these were potent growth inhibitors. The closely related 5564-701 (5) elicited no growth inhibitory effects and served as negative control. Structures of the five NPPMs of interest are shown in Figure 1a.

Figure 1. NPPMs specifically inactivate Sec14.

(a) Chemical structures of 4130-1278, 4130-1276, 67170-49, 6748-481, and 5564-701. (b) Growth inhibition of wild-type (WT; CTY182, gray) and SEC14^P-136 (CTY374; black) strains by 6748-481 (●,■; IC₅₀s=2.87±0.6 and 0.44±0.16μM, respectively) or 5564-701 (▲,◆; IC₅₀s>200μM). Relative growth compares growth rate with compound relative to “no-drug” control (DMSO). Data are plotted as function of NPPM concentration (x-axis). Doubling times were measured at λ_610nm. Values show mean ± s.e.m of normalized doubling times for each drug concentration from 3 independent experiments. IC₅₀s represent 95% confidence interval from 3 independent experiments. (c) Yeast expressing physiological [YCp(SEC14)], or elevated [YEp(SEC14)], Sec14 levels were spotted in 10-fold dilution series onto YPD agar with DMSO or 6748-481 (20μM) and incubated at 30°C (48h). Bar, 1cm. (d) [³H]-PtdIns-transfer was monitored in vitro using purified Sec14 with 5564-701 (IC₅₀>100μM), 4130-1276 (IC₅₀=283 ± 30nM), 6748-481 (IC₅₀=211 ± 19nM) or 67170-49 (IC₅₀=175 ± 26nM). Relative [³H]-PtdIns transfer relates activity in presence of NPPM relative to vehicle control (DMSO) in assays where Sec14 concentration was clamped (287nM). Values indicate mean ± s.e.m of triplicate determinations from 3 independent experiments. IC₅₀ values fall into 95% confidence intervals. [³H]-PtdIns input ranged from 12790–16800 cpm, with background ranging from 478–751 cpm. (e) Purified Sfh proteins (10μg) were assayed for [³H]-PtdIns transfer in the presence and absence of indicated NPPM (40μM). Values are the mean ± s.d of triplicate determinations from 3 independent experiments. P*=1.8202E⁻¹¹ and P**=4.31133E⁻¹² relative to DMSO control (two-tailed t-test with heteroscedastic variance, Microsoft Excel 2010). [³H]-PtdIns input ranged from 14441–15101 cpm, with background ranging from 640–657 cpm.

Sensitivity to NPPM is a function of Sec14 levels

Dose response experiments showed NPPM sensitivity was related to cellular Sec14 levels. NPPM 6748-481 served as representative SMI in these assays as it was the most water soluble bioactive compound and the most potent growth inhibitor. The half-maximal inhibitory concentration (IC₅₀) for 6748-481 was 2.9 ± 0.6 μM for WT haploid cells (Fig. 1b). When endogenous Sec14 levels were reduced ~7-fold by driving SEC14 expression from a truncated promoter (SEC14^P-136) ²⁴, the IC₅₀ fell proportionately. The effect was specific as SEC14^P-136 haploids were not sensitized to the inactive 5564-701 (Fig. 1b), and elevated Sec14 levels enhanced resistance to 6748-481 (Fig. 1c).

NPPMs inhibit Sec14 in vitro

As Sec14 is not an enzyme, its activity is operationally defined by exchange of PtdIns and PtdCho between membranes in vitro¹⁰. NPPMs 67170-49, 6748-481 and 4130-1276 displayed potent and dose-dependent inhibitions of Sec14-catalyzed [³H]-PtdIns transfer in a purified system. The IC₅₀s ranged from 175–283nM (Fig. 1d). At these concentrations, the Sec14:NPPM molar stoichiometries were ca. 1:1.

Two results excluded trivial possibilities that NPPMs interfered with Sec14 activities via non-specific membrane-active effects. First, biologically inactive NPPMs (5564-701) failed to diminish Sec14-catalyzed PtdIns-transfer at concentrations 500-fold above the IC₅₀s measured for Sec14-active NPPMs (Fig. 1d; Supplementary Fig. 4). Second, neither of the Sec14-active NPPMs tested (6748-481 and 4130-1276; 40μM) interfered with [³H]-PtdIns transfer activities of other yeast Sec14-like PITPs (Sfh1–Sfh5²⁵) -- even at concentrations ~200-fold greater than the corresponding Sec14 IC₅₀s (Fig. 1e). Sfh protein indifference to NPPMs demonstrated selectivity of the SMIs for Sec14 as Sfh PITPs share 23%–64% primary sequence identity with Sec14 (43%–89% similarity). That Sfh1-mediated transfer activities were not inhibited underscored the Sec14 selectivity of these NPPMs as Sfh1 is the closest known Sec14 homolog²⁶.

NPPM structure activity relationships

Forty-six NPPM-like molecules were evaluated for structure-activity relationships (SAR) for Sec14 inhibition. The cumulative SAR data assigned the relative importance of each NPPM functional group to activity as Sec14 inhibitor (IC₅₀<10μM in vitro activity threshold). SAR reported an obligatory requirement for ortho-Cl in the Cl-nitrophenyl (i.e. activated aryl halide) NPPM moiety (Fig. 2a; Supplementary Figs. 3,4). Either Cl- removal (4130-1276 vs. 5355-152 [6]), or Cl- shift to the para position (4130-1276 vs. 5658-722 [7]), increased the Sec14 IC₅₀ >500-fold (Fig. 2b). The linker ketone that connects the Cl-nitrophenyl and piperazinyl groups was key. Its modification reduced NPPM potency in vitro by 30-fold (6748-481 vs. 5263-433 [8]). The NO₂-group also contributed to NPPM potency. Its removal resulted in a ~10-fold increase in IC₅₀ (6748-481 vs. 5357-399 [9]; Supplementary Fig. 4). Addition of hydrophobic functional groups to the NPPM fluorobenzene tail enhanced potency ca. 4-fold (compare 5348-909 [10] with 6748-481, 7276-196, [11] and 6828-980 [12]). However, extending the linker that bridges the piperazinyl and fluorobenzene groups by a single carbon reduced NPPM potency (5348-909 vs. 5356-684 [13]; Supplementary Figs. 3,4).

Figure 2. NPPM SAR relationships.

(a) A summary of the contributions of each highlighted functional group of NPPM 6748-481 to potency as Sec14 inhibitor is depicted. Data represent a compilation, and superimposition onto the 6748-481 scaffold, of IC₅₀ measurements for Sec14-catalyzed [³H]-PtdIns transfer activity for each SMI listed in Supplementary Figure 4. (b–e) Chemical-induced inhibition of Sec14-catalyzed [³H]-PtdIns transfer activity was monitored in the presence of the indicated SMI. Chemical identities for functional groups X, Y and Z are identified for each SMI tested, and are highlighted in red. Relative [³H]-PtdIns transfer compares activity in presence of compound relative to the “no-drug” control (DMSO) in assays where Sec14 concentration was clamped at 287nM. IC₅₀ values: 5355-152 (>100,000 nM),5658-722 (>100,000 nM), 5348-723 (11,060 ± 1,130 nM), 6748-481 (211 ± 19 nM), BBV34896-755 (36,800 ± 5,000 nM), BBV34846-247 (173 ± 35 nM), BBV34846-244 (100 ± 6 nM), Z1082669-326 (174 ± 17 nM), 5356-628 (>100,000 nM), 5567-782 (56,500 ± 7,680 nM), 5356-350 (22,000 ± 3,100 nM), 5357-399 (2,550 ± 230 nM), 5353-036 (768 ± 71 nM), 7329-906 (589 ± 51 nM), 5656-333 (>100,000 nM), 6031-920 (>100,000 nM), 5353-383 (>100,000 nM), 5653-520 (>100,000 nM). [³H]-PtdIns input ranged from 7,542-13,002 cpm per assay, with background ranging from 282-1,317 cpm and transfer efficiency (% of total input) ranged from 11% to 32%. Values indicate the mean ± s.e.m of triplicate determinations from three independent experiments. IC₅₀ values represent the 95% confidence interval or for single concentration points a predicted IC₅₀. Statistical comparisons of values used the “unpaired two-tailed t-test” where P***<0.0001, P**=0.0024 and P*=0.0089.

Sec14 inhibition and the NPPM halide

The ortho-Cl requirement on the activated aryl halide raised the question of how halide identity influences NPPM potency. A variable aryl halide series was assembled [(BBV34896-755 [14], BBV34846-244 [15], Z1082669-326 [16]), and 6748-481 was resynthesized for inclusion in the series (BBV34846-247 [17]). The Cl-, Br- and I- derivatives (4130-1276 and 6748-481; BBV34846-244; Z1082669-326) exhibited potencies superior to those of the H-, CH₃- or F-derivatives (5355-152, 5348-723 [18], BBV34896-755). In rank order: Br-NPPM > I-NPPM ≥ Cl-NPPM (6748-481; BBV34846-247) ≫ F-NPPM, CH₃-NPPM, H-NPPM (Fig. 2c, Supplementary Fig. 4). Consistent with in vitro data, Cl-, Br- and I- versions were potent growth inhibitors, whereas H-, CH₃- and F- derivatives were not (Supplementary Fig. 3).

Six ‘NO₂-less’ derivatives, differing in halide chemistry and/or position, were evaluated (5356-628 [19], 5356-350 [20], 5567-782 [21], 5357-399, 5353-036 [22], 7329-906 [23], 5355-139 [24], 5528-054 [25]). The relative potencies of these compounds as Sec14 inhibitors were also directly proportional to the atomic mass and lipophilicities of the ortho-halide, and inversely proportional to halide electronegativity (Fig. 2d, Supplementary Fig. 4). The ortho-halide requirement for SMI activity was conserved (Fig. 2e, Supplementary Fig. 4). Parallel in vivo experiments demonstrated Cl-, Br- and I- compounds potently inhibited growth, whereas H-, CH₃- and F- derivatives did not (Supplementary Fig. 3).

NPPM binding models

NPPMs are amphipathic molecules that partition into apolar environments (C_logP ~ 3.0), suggesting NPPMs load into the Sec14 hydrophobic pocket during phospholipid exchange. To address how Sec14 binds NPPMs, a 1400Ų virtual pocket surface was modeled to form a boundary for unconstrained docking routines using 6748-481 as ligand (Supplementary Fig. 5). Multiple independent simulations, using several docking platforms, produced >3000 binding solutions that reduced themselves into 6 representative modes. These modes shared certain features with regard to Sec14::6748-481 interactions, but exhibited unique features as well (Supplementary Figs. 6,7). Two pose classes (modes 1,2 and modes 3,4) represented mirror images where orientations of bound 6748-481 were rotated 180° around the NPPM long axis. Another pose class (modes 5,6) shifted the 6748-481 binding site down the hydrophobic pocket. Sec14::NPPM interaction fingerprints highlighted how modes 5,6 forced NPPM polar groups into hydrophobic environments (Supplementary Figs. 6,7). This unappealing feature led us to reject mode 5,6 poses from consideration.

Fingerprints common to modes 1,2 and 3,4 included intercalation (in flipped orientation) of the 6748-481 fluorobenzene tail into the narrow hydrophobic cleft lined by Sec14 residues L₂₃₂, F₂₂₈, F₂₂₅, I₂₄₀, F₂₂₁, L₁₇₉, I₂₁₄, and M₁₇₇ (Supplementary Figs. 6,7), and wedge of the fluoro-benzyl moiety between the side chains of F₂₂₈ on one side and F₂₁₂ and M₁₇₇ on the other; thereby consolidating a network of hydrophobic and π-π stacking interactions. A second common fingerprint inter-digitated the 6748-481 chloro-nitrophenyl group in a hydrophilic subdomain of the Sec14 pocket where it set within H-bonding distance (1.5–3.5Å) to residues framing the Sec14 PtdCho headgroup-coordinating substructure (S₁₇₃, Y₁₁₁, Y₁₂₂, Y₁₅₁; Supplementary Figs. 6,7). All modes predicted NPPM-binding was sterically incompatible with PtdIns/PtdCho-binding (Supplementary Fig. 7b).

Modes 1,2 were distinguished from modes 3,4 by orientation of the 6748-481 ketone. Modes 1,2 anchored the ketone via H-bond interactions with S₂₀₁, whereas modes 3,4 assigned ketone interactions with Y₁₅₁. Modes 1 and 2 were further distinguished by their mirror-image orientation of the activated aryl halide. Mode 1 anchored the nitro- group by interactions with residues S₁₇₃/Y₁₂₂ and the halide with Y₁₁₁/S₁₇₃. Mode 2 featured nitro- and halide interactions with S₁₇₃ and Y₁₁₁/Y₁₂₂, respectively. Mode 3 assigned interactions of the nitro- and halide groups with residues S₁₇₃ and Y₁₁₁, respectively, whereas mode 4 projected both the nitro- and halide groups engage S₁₇₃/Y₁₁₁. In sum, the solution sets projected Sec14 S₁₇₃, Y₁₁₁, Y₁₂₂, and Y₁₅₁ interactions with activated aryl halide (albeit in flipped orientations), and anchor of the ketone to S₂₀₁ or Y₁₅₁.

NPPM-resistant Sec14

The NPPM docking models made distinguishing predictions regarding which missense substitutions should diminish NPPM binding without affecting Sec14 activity. The corresponding Sec14 proteins were purified and assayed for PtdIns transfer activity. Residue S₁₇₃ was of particular interest as it was consistently identified as highest scoring NPPM-interacting residue in docking simulations. Indeed, S₁₇₃C rendered Sec14 resistant to inhibition by NPPM 6748-481 in vitro -- even at concentrations approaching NPPM solubility limits (>736-fold more resistant than Sec14: Supplemental Figs. 8a). Specificity of effect was evident by the more modest NPPM-resistance exhibited by Sec14^T175C(Supplemental Figs. 8a), and Sec14^S173C resistance translated across the spectrum of Sec14-active NPPMs (Supplementary Fig. 8b, c). Strong reductions in 6748-481 sensitivity were also scored for Sec14^Y111A, Sec14^Y151A and Sec14^S201C (Fig. 3a, Supplemental Fig. 9a–c). The corresponding PtdIns-transfer activities were 337-, 140- and 22-fold less sensitive to 6748-481 inhibition (Fig. 3a, Supplemental Fig. 9a–c). Sec14^Y111F, Sec14^Y122A, Sec14^M177C and Sec14^Y122F showed only ~2- to 5-fold reductions in the same (Fig. 3a, Supplemental Fig. 9a, b, d, e.), whereas Sec14^Y151F displayed modestly enhanced sensitivity to 6748-481 (Fig. 3a, Supplemental Fig. 9a, b, e).

Figure 3. Sec14 is the essential cellular target of bioactive NPPMs.

[³H]-PtdIns transfer was measured using purified recombinant Sec14 proteins (287nM) in the presence of 6748-481. (a) Relative [³H]-PtdIns transfer compares activities in presence of 6748-481 relative to the “no-drug” control (DMSO). Input ranged from 9,227–16985 cpm per assay (background; 160–724 cpm). Transfer efficiencies: Sec14 (24–32%), Sec14^Y111A (5–13%), Sec14^Y111F (38–41%), Sec14^Y122A (13–15%), Sec14^Y122F (18–27%), Sec14^Y151A (31–40%), Sec14^Y151F (31–40%), Sec14^S173C (22–24%), Sec14^T175C (5–16%), Sec14^S201C (6–11%), Sec14^M177C (8–10%), and Sec14^F228A (18–27%). IC₅₀ values represent 95% confidence interval of triplicate determinations from three independent experiments. Statistical comparisons of WT to mutant IC₅₀ values used “extra sum-of-squares F-test” (P***<0.0001). (b) Expression of NPPM-resistant Sec14 derivatives renders cells insensitive to NPPM-mediated growth arrest. Yeast expressing physiological levels of Sec14, YCp(SEC14), or the indicated Sec14 variant, as the sole source of Sec14 in the cell, were diluted in 10-fold series onto YPD agar containing DMSO, 20μM or 40μM 6748-481. Plates were incubated at 30°C (48h). (c) ‘Bypass Sec14’ mutants are NPPM-resistant. WT (CTY182), SEC14^P-136 (CTY374), and sec14Δ cki1Δ (CTY303) and sec14-1^ts kes1Δ (CTY159) ‘bypass Sec14’ mutants were grown as in (b). Bar, 1cm.

Sec14 is the sole essential NPPM target

Two lines of evidence demonstrated that Sec14 represents the sole essential NPPM target in cells. First, yeast expressing Sec14^S173C were resistant to challenge with 6748-481 (Fig. 3b). The second came from exploitation of ‘bypass Sec14’ mutants that no longer require Sec14 for viability. Genetic inactivation of the CDP-choline pathway for PtdCho biosynthesis (cki1Δ), or of the oxysterol binding protein homolog Kes1 (kes1Δ), effects ‘bypass Sec14’^27–30. Both cki1Δ and kes1Δ cells were indifferent to 6748-481, 67170-49 or 4130-1276 (Fig. 3c).

NPPM challenge phenocopies genetic inactivation of Sec14

Sec14 promotes membrane trafficking through the TGN/endosomal system by coordinating PtdCho and PtdIns-4-phosphate (PtdIns-4-P) metabolism^22,28,30. Thermal inactivation of Sec14 (by shift of sec14-1^ts mutants to 37°C) provoked intracellular accumulation of cargo-engorged TGN/endosomes. These morphological phenotypes were recapitulated in WT and SEC14^P-136 cells intoxicated with NPPMs 6748-481, 67170-49 or 4130-1276 (Supplementary Fig. 10). Mock challenge with DMSO, or with inactive 5564-701 control, had no effect. The kes1Δ and cki1Δ ‘bypass Sec14’ mutants were unaffected by 6748-481, 67170-49, or 4130-1276 (Supplementary Fig. 11), and cells expressing Sec14^S173C were similarly unperturbed by NPPM challenge (Supplementary Fig. 12).

The trafficking defects induced by NPPM-poisoning were also on display when transport and/or recycling of cargo through the TGN/endosomal system were analyzed. Endocytic transport of the bulk membrane tracer FM4-64 to the limiting vacuole membrane (Supplementary Fig. 13a, b), and recycling of endocytosed Snc1 v-SNARE from TGN/endosomes to the cell surface, were both retarded by 6748-481, 67170-49 or 4130-1276 (Supplementary Fig. 14a, b). Mock challenge with DMSO or 5564-701 was without effect. Trafficking of secretory invertase to the cell surface was also impaired upon 6748-481, 67170-49, or 4130-1276 challenge, but not by 5564-701 or DMSO (Supplementary Fig. 15). NPPM-induced FM4-64, GFP-Snc1, and invertase trafficking defects were corrected by kes1Δ (Supplementary Figs. 16a–d). Sec14^S173C expression similarly rendered invertase secretion resistant to inhibition by Sec14-active NPPMs, (Supplementary Fig. 17).

The effects of NPPM challenge on transit of vacuolar carboxypeptidase Y (CPY) through the TGN/endosomal system were monitored by pulse-chase radiolabeling. Impaired trafficking from ER or TGN compartments was diagnosed by accumulation of CPY precursor forms (p1 and p2)³¹. Pools of p1CPY, and particularly of p2CPY, accumulated when sec14-1^ts yeast were shifted to restrictive temperatures (37°C) and upon treatment of WT or SEC14^P-136 cells with NPPMs 6748-481, 67170-49 or 4130-1276 (Supplementary Fig. 18a, b). Neither DMSO, nor 5564-701, impaired CPY transport from ER or TGN/endosomal compartments in WT or SEC14^P-136 cells. The inhibitory effects of Sec14-active NPPMs on CPY trafficking were dose-and time-dependent (Supplementary Fig. 19a, b), and were poorly reversible. Precursor CPY forms failed to chase to mCPY – even after a 2hr washout period with cycloheximide (Supplementary Fig. 20). Finally, NPPM 6748-481, 67170-49, and 4130-1276-mediated inhibition of CPY transit through the secretory pathway was remediated in kes1Δ and cki1Δ ‘bypass Sec14’ mutants (Supplementary Fig. 21, a, b), and in cells expressing Sec14^S173C as sole Sec14 source (Supplementary Fig. 21c).

NPPMs discriminate between phosphoinositide pools

Sec14 potentiates activities of both essential yeast PtdIns 4-OH kinases in vivo (i.e. Stt4 and Pik1)^10,32–34. We therefore tested whether bulk PtdIns-4-P production was sensitive to inhibition by Sec14-active NPPMs. In these experiments, sac1 mutants inactivated for the major yeast PtdIns-4-P phosphatase were employed because these mutants accumulate PtdIns-4-P to high levels^33,35. This accumulation simplifies PtdIns-4-P biosynthetic rate measurements, and increases statistical power of PtdIns-4-P quantification. Challenge with 6748-481, 67170-49, 4130-1276, reduced bulk PtdIns-4-P levels by ~40% compared to levels recorded when cells were challenged with DMSO or 5564-701 (Supplementary Fig. 22a). Neither challenge with 6748-481, nor with 5564-701, affected PtdIns-3-P or PtdIns-4,5-bisphosphate (PtdIns-4,5-P₂) levels when compared to vehicle alone (Supplementary Fig. 22b, c).

NPPM selectivity on phosphoinositide homeostasis in WT yeast was also on display in vital imaging assays using isomer-selective biosensors. To probe status of PtdIns-4,5-P₂ pools, we monitored intracellular distribution of the GFP-2xPH^PLC_δ1 PtdIns-4,5-P₂ biosensor. This reporter localized to the plasma membrane in a PtdIns-4,5-P₂-dependent manner -- as evidenced by biosensor release into the cytoplasm upon inactivation of a temperature-sensitive version of the single yeast PtdIns-4-P 5-OH kinase (Mss4; Supplementary Fig. 22d, e). In accord with [³H]-inositol radiolabeling data, GFP-2xPH^PLC_δ1 remained bound to plasma membrane after 3h incubation of cells with Sec14-active SMIs. Release of GFP-2xPH^PLC_δ1 from plasma membrane was not observed upon inactivation of other PtdIns kinases (Supplementary Fig. 22d, e).

Status of TGN/endosomal PtdIns-4-P pools was queried using two biosensors; GFP-GOLPH3 and GFP-2xPH^{Osh2 36–38}. Optimal association of both reporters with TGN/endosomal compartments was unaffected in cells devoid of PtdIns-3-phosphate and PtdIns-3,5-P₂ (vps34Δ), but was Pik1-dependent as evidenced by release from these organelles upon shift of pik1^ts mutants to 37°C (Supplementary Fig. 23a–d). Thermal inactivation of Sec14 (sec14^ts mutant, 37°C condition), or intoxication of yeast with 6748-481, 67170-49, or 4130-1276, similarly released GFP-2xPH^Osh2 from TGN/endosomal structures (Fig. 4a, b). A compartmental pool-specificity was evident for NPPM-mediated interference of PtdIns-4-P signaling in these experiments. In addition to the Sec14- and Pik1-dependent TGN/endosomal PtdIns-4-P pool, GFP-2xPH^Osh2 registered an Stt4-dependent plasma membrane pool of this phosphoinositide (Fig. 4a, b; Supplementary Fig. 23a, b). Challenge of WT cells with 6748-481, 67170-49 or 4130-1276 (but not 5564-701 or DMSO) released GFP-2xPH^Osh2 from TGN/endosomal compartments -- without compromising biosensor targeting to the plasma membrane (Fig. 4a, b). In that regard, we previously showed that kes1Δ rescues growth defects of pik1^ts, but not stt4^ts, mutants at semi-permissive temperatures²⁹ – identifying Kes1 as an antagonist of Pik1-dependent PtdIns-4-P signaling. The kes1Δ ‘bypass Sec14’ mutant retained GFP-2xPH^Osh2 on TGN/endosomal membranes in the presence of 6748-481, 67170-49, or 4130-1276 (Supplementary Fig. 24a, b).

Figure 4. Sec14-active NPPMs exhibit compartment-specific inhibition of PtdIns-(4)-phosphate signaling.

(a) WT yeast (CTY182) or the indicated temperature sensitive mutants containing GFP-2xPH^Osh2 were treated with indicated SMIs (20μM) for 3h at 25°C, or 37C. GFP-2xPH^Osh2 profiles were imaged, and profiles scored. (b) GFP-2xPH^Osh2 distributions were classified as PM, punctate, or both. ND – not determined. At least 300 cells were scored for each condition in (a). (c) Experimental conditions were as in (a) except GFP-GOLPH3 distribution was monitored. The mss4-5^ts mutant was incubated for 30 min. at non-permissive temperature. (d) GFP-GOLFP3 distribution was scored (>300 cells for each condition), and GFP-GOLPH3 distributions were classified as PM, punctate, or both. Bar, 5μm.

Compartmental pool-specificities for NPPM-mediated interference of PtdIns-4-P signaling were also evident in GFP-GOLPH3 imaging experiments. The effects of genetic and NPPM-mediated Sec14 inactivation on GFP-GOLPH3 localization differed from those observed for GFP-2xPH^Osh2 as these conditions effected morphological derangement of GFP-GOLPH3-positive TGN/endosomal compartments. Significant, but more limited, release of the reporter from TGN/endosomal membranes was observed (Fig. 4c, d; Supplementary Fig. 23c, d).

NPPMs discriminate between local PtdIns-4-P pathways

The GFP-2xPH^Osh2 imaging results demonstrated Sec14-active SMIs disrupted PtdIns-4-P signaling in a compartment-specific manner (TGN/endosomes vs. plasma membrane). The distinct biological activities of Sec14 and the Sec14-like Sfh4 provided a more discerning test of whether Sec14-active NPPMs distinguished between parallel PtdIns-4-P signaling pathways that operate in the same endomembrane system. Sec14 and Sfh4 both control PtdIns-4-P production in TGN/endosomes. Unlike Sec14, Sfh4-dependent PtdIns-4-P signaling supports phosphatidylserine (PtdSer) decarboxylation to phosphatidylethanolamine (PtdEtn) in those compartments (Fig. 5a) ¹³. As PtdEtn is an essential lipid, loss of Sfh4 activity resulted in Etn auxotrophy when the PtdSer decarboxylase 1 pathway was incapacitated (psd1Δ; Fig. 5b).

Figure 5. NPPMs discriminate between Sec14- and Sfh4-mediated PtdIns-4-P signaling.

(a) Sec14 and Sfh4 control distinct PtdIns-4-P signaling pathways in TGN/endosomal compartments. Sec14 couples PtdIns 4-OH kinase action with vesicle biogenesis, whereas Sfh4 pathway couples PtdIns 4-OH kinase action with PtdSer decarboxylation to PtdEtn. Decarboxylation is catalyzed by Psd2 and is posited to involve membrane contact sites that bridge the endoplasmic reticulum (ER) with TGN/endosomes. (b) psd1Δ (PYY23), sfh4Δ (PYY40), or sfh4Δ psd1Δ (PYY30) yeast expressing the NPPM-resistant Sec14^S173C [YCp(SEC14^S173C)] were spotted in 10-fold dilution series on uracil-free agar ± ethanolamine (1mM), with or without 6748-481 (20μM). Plates were incubated at 30°C (96h). (c) WT (CTY182), sec14-1^ts (CTY1-1A), psd1Δ (PYY23), sfh4Δ (PYY40), or sfh4Δ psd1Δ (PYY30) cells expressing Sec14 or Sec14^S173C were cultured in uracil-free media containing ethanolamine (2mM) at 30°C. Mid-logarithmic growth phase cultures (λ₆₀₀=0.3) were incubated with [³H]-serine (3.3μCi/ml) for six hours. At hour three, cells were presented with 6748-481 (20μM), DMSO, or shifted to non-permissive temperature (37°C). Lipids were extracted and resolved by thin layer chromatography. PtdSer and PtdEtn species were harvested, quantified by liquid scintillation counting, and data expressed as precursor/product ratios. Values represent mean ± s.e.m from at least 3 independent experiments. Defects in Sfh4-dependent conversion of PtdSer to PtdEtn present high PtdSer/PtdEtn ratios – e.g. psd1Δ sfh4Δ control strains. Statistical comparisons of values used the ‘unpaired two-tailed t-test” relative to DMSO control, where # (P = 0.0495), ## (P = 0.002), ** (P = 0.0051), *** (P = 0.0004). Bar, 1 cm.

Because Sfh4 was not inhibited in vitro by Sec14-active NPPMs, we examined whether Sec14-active SMIs respected the PtdIns-4-P pool-selectivities of these distinct PITP-dependent metabolic circuits. Thus, psd1Δ yeast were reconstituted for Sec14^S173C expression (to circumvent NPPM-induced growth defects associated with Sec14 inactivation), and cells were intoxicated with 6748-481. NPPM challenge failed to impose Etn auxotrophy onto the PSD1 control, or onto the isogenic psd1Δ derivative (Fig. 5b) – thereby demonstrating that Sfh4 retained biological activity in the face of Sec14-active NPPM.

These phenotypic results were confirmed by [³H]-serine radiolabeling experiments that measured Sfh4-dependent decarboxylation of PtdSer to PtdEtn in TGN/endosomes. NPPM 6748-481 challenge of psd1Δ mutants did not compromise PtdSer decarboxylation in vivo. This result was observed regardless of whether psd1Δ mutants expressed Sec14 (TGN/endosomal trafficking was blocked) or Sec14^S173C (cells remained competent for TGN/endosomal trafficking; Fig. 5c). Thus, NPPM challenge failed to interfere with Sfh4-dependent PtdIns-4-P signaling in TGN/endosomes even as it strongly impaired Sec14-dependent PtdIns-4-P signaling in the same endomembrane system.

DISCUSSION

Using a rigorous set of genetic and biochemical criteria, we identify NPPMs as the first validated Sec14-directed inhibitors, and establish PITPs as pharmacological targets for high resolution dissection of PtdIns-4-P signaling in cells. Our findings emphasize the exquisite PITP selectivities of NPPMs, and demonstrate these compounds do not exert their effects via non-specific membrane-active mechanisms. Rather, the data indicate NPPMs load into the Sec14 hydrophobic pocket during a phospholipid exchange cycle, and invade PtdIns- and PtdCho-acyl chain space while simultaneously engaging Sec14 residues essential for PtdCho headgroup coordination. NPPM engagement of Sec14 residues essential for coordinating the PtdCho headgroup rationalizes the PITP-selectivities of NPPMs. Other Sec14-like PITPs do not conserve the PtdCho headgroup coordinating substructure (Supplementary Fig. 25), and therefore lack the structural elements required for NPPM binding. In this regard, Sfh1 presents a conundrum as it is highly homologous to Sec14 and conserves the functional PtdCho-coordinating unit critical for NPPM binding. Subtle differences in Sec14 vs Sfh1 pocket geometry likely account for the NPPM-resistance of Sfh1. However, given Sec14-like PITPs conserve the PtdIns-binding ‘barcode’^10,18, the structural engineering of these proteins also offers prospects for development of broad range inhibitors which survey the PtdIns-headgroup binding site.

The chemical properties of activated aryl halides suggest several mechanisms for how NPPMs inhibit Sec14 activity. One involves aromatic nucleophilic substitution³⁹, but the potency vector of the ortho-halide NPPM series excludes such mechanisms. Rather, we favor non-covalent halogen bonding mechanisms^40,41 (Fig. 6a). Halogen bond interactions require formation of an electropositive ‘σ-hole’ on the charge surface of the halogen. This organization is enhanced by vicinal e- withdrawing groups (e.g. -NO₂), and by propensities of larger halide atoms to adopt anisotropic distributions of electrostatic potentials^41,42. Correlations of Sec14-active NPPM potencies with electrostatic surface potentials of cognate activated aryl halides support halogen bonding mechanisms for NPPM-mediated Sec14 inhibition. Our calculations project NPPM activated aryl halides form σ-holes, and that σ-hole magnitudes are inversely proportional to halide electronegativity (Fig. 6a). This relationship is congruent with experimentally determined NPPM potencies as Sec14 inhibitors. We propose NPPM occupancy in the Sec14 pocket is anchored by S₁₇₃ engagement with ortho-halide via halogen bond, NO₂-group engagements with Y₁₁₁ and Y₁₅₁, H-bond interaction of the ketone with S₂₀₁, and intercalation of the NPPM fluorobenzene between F₂₂₈ of the helical gate and F₂₁₂/M₁₇₇ of the pocket floor (Fig. 6b, c). Proof of a halogen-bonding mechanism requires structural data, however, as the cardinal signature is a ‘short’ halide-oxygen bond whose length is less than the sum of the halide and oxygen van der Waals radii^40–42.

Figure 6. Model for NPPM-mediated inhibition of Sec14.

(a) Simulations of the electrostatic potentials of the indicated NPPM activated aryl halides predict formation of progressively larger electropositive σ-holes in the -Cl, -Br and -I activated aryl halides, respectively. The -CH₃ and -F derivatives are forecast to be incapable of forming significant σ-holes. (b) NPPM binding mode 2 is most consistent with the various data and is depicted within the Sec14 hydrophobic pocket. H-bonds (orange), carbon (blue), oxygen (magenta), nitrogen (dark blue), chlorine (teal) and fluorine (orange). (c) A focused view of NPPM binding mode 2 highlighting Sec14 residues which form the PtdCho headgroup coordinating substructure, and whose alteration renders Sec14 resistant to NPPM inhibition. Hydrogen bonds (orange), carbon (blue), oxygen (red), nitrogen (dark blue), chlorine and fluorine (green).

How does non-covalent NPPM binding in the Sec14 pocket exert a poorly reversible inhibition of PITP activity? NPPM-mediated bridging of the pocket floor to gate sub-structures might tether these elements too strongly for Sec14 to spring the gate for ligand exchange upon membrane association. Alternatively, Sec14-active NPPMs may decouple conformational switch elements required for gate opening by disturbing H₂O networks that lubricate the pocket surface⁴³ -- thereby locking Sec14 in a ‘closed’ NPPM-bound state.

The availability of powerful yeast genetic technologies notwithstanding, we forecast significant utility for Sec14-active NPPMs as tool compounds. For example, these SMIs circumvent the tedium of incorporating sec14^ts mutations into large sets of isogenic yeast strains for running genome-scale screens. Sec14-active SMIs also enable manipulation of in vitro membrane trafficking assays reconstituted with wild-type components – a significant advantage given productive exploitation of genetic tools in cell-free systems is often frustrated by misbehavior of mutant biochemical fractions.

Finally, as PITPs are ubiquitously distributed across the Eukaryota⁴⁴, PITP-directed screening platforms offer new applications for drug discovery. Because expression of heterologous Sec14-like proteins, or structurally-unrelated mammalian StART-like PITPs, rescue Sec14 defects in yeast⁴⁵, such platforms can be purposed for discovery of SMIs directed against target PITPs derived from many biological sources. PITP inhibitors, whether directed against Sec14-like or StART-like PITPs, will enable study of phosphoinositide signaling in organisms intractable to genetic approaches. Many such organisms are significant pathogens. In that regard, we have identified new classes of PITP-directed SMIs that inhibit dimorphic transitions of pathogenic fungi – i.e. inhibit the developmental processes essential to success of these organisms as infectious agents.

ONLINE METHODS

Statistical analyses

Curve fitting and t-test were performed using GraphPad Prism version 5.00 for Windows, GraphPad Software, La Jolla California USA (www.graphpad.com) unless otherwise noted. General data handling was carried out in Excel 2010 (v14.0.4734.1000, 32-bit; Microsoft Corporation). Statistical comparisons of [³H]-PtdIns transfer activities and growth inhibition were calculated using the “extra sum-of-squares F-test” in Graphpad prism v 5.00.

Molecular graphics and chemical drawing

Molecular graphics and analyses were performed with the UCSF Chimera package (version 1.8; http://www.cgl.ucsf.edu/chimera/). Chemical structures in main figures were constructed using ChemBioDraw (PerkinElmer, Cambridge, MA). Chemical structures, substructures and reactions in Supplementary Figures were drawn and displayed using Marvin (5.10.0 and 5.11.4, 2012) and ChemAxon (http://www.chemaxon.com). Docking poses and cavity surfaces were generated using MOE suite (2011.10; Chem. Comp. Group Inc., Montreal, Canada).

Yeast strains, media and reagents

Yeast media and transformation methods are described ⁴⁶. Restriction endonucleases were from New England Biolabs (Ipswich, MA), standard reagents from Sigma (St. Louis, MO) or Fisher Scientific (Norcross, GA), and all phospholipids were purchased from Avanti Polar Lipids Inc. (Alabaster, AL). [³⁵S]-Translabel was purchased from MP Biomedicals (Irvine, CA). Yeast strain genotypes are listed in Supplementary Table 1 and plasmids in Supplementary Tables 2 and 3.

Growth rate analyses

Growth assays were conducted in 96 well microtiter plates. In Supplemental Figures 1 and 3a, optical densities were measured every 15 min over the course of 20 hours using a GENios microplate reader (Tecan). All other growth rates were determined as follows. Cells were cultured to mid-logarithmic growth phase in YPD medium (2% glucose) and diluted to λ_600nm=0.1 in media appropriately supplemented with SMI or DMSO. Cultures were incubated in 96 well plates in a final volume of 250μl of YPD (2% glucose) for 10–16 hours between 30°C and 32°C. ODs were measured every 15 min. at λ_610nm (BioTek Synergy 2) or λ_595nm (PerkinElmer VictorX3 3030 Multilabel Plate Reader). Doubling times were calculated and normalized to an internal DMSO control. Nonlinear regression was applied to the dataset to calculate the best fit equation using [Y=100/(1+10^((LogIC₅₀−X)*HillSlope))] in Graphpad Prism v5.0. IC₅₀ values represent the 95% confidence interval from at least three independent experiments unless otherwise noted. Statistical comparison of SMI-mediated growth inhibition were determined using the “extra sum-of-squares F-test” in Graphpad prism v 5.00.

Site-directed mutagenesis

Site-directed mutations were generated using QuickChange^™ (Stratagene) as recommended by the manufacturer. Primer sequences are available from the authors by request.

Small molecule inhibitors

The compounds shown in Supplementary Figure 1 were from ChemDiv (San Diego, CA). SMIs BBV34896-755, BBV34846-244, Z1082669-326, BBV34846-247, BBV34847-734 were synthesized by UORSY/Ukrorgsyntez Ltd. (Riga Latvia). Unless otherwise noted, all other compounds were purchased from ChemBridge Chemical Store, San Diego, CA (www.hit2lead.com). SMIs were dissolved in DMSO (Fisher, D128-500) to a final stock concentration of 20mM and stored in the dark at room temperature. Quality control ¹H NMR spectra and LC-MS data are provided in the Supplementary note.

Chemogenomic screening

Pools of bar-coded homozygous and heterozygous deletion strains were grown in YPD + 25mM HEPES (pH 6.8) supplemented with 4130-1278 (35.5μM) and 4130-1276 (6.7μM) for 5 and 20 generations, respectively. Genomic DNA extraction, PCR amplification of molecular barcodes, Genflex tag16k array hybridization/scanning (Affymetrix), and analysis of chemogenomic data are described¹⁹. Quantile normalized fluorescence values for each tag were log₂-transformed, and z-scores calculated: Tag z-score = [(average of controls)−(experimental value)]/(std. dev. of controls); where the controls were 12 replicate samples of pools treated with DMSO. The z-score for each strain is the average of the 2 tags associated with that strain, and represents the sensitivity value.

NPPM chemogenomic interactions

Interactions were described according to their Gene Ontology (GO) descriptors⁴⁷. Data sets from chemogenomic profiling were analyzed and enriched gene sets were chosen that had Z-scores greater than 4. Gene-sets that did not pass this enrichment threshold are not shown.

Protein purification

Recombinant proteins were purified as described¹⁰. In summary, pET28b-His₈-Sec14, pET28b-His₈ -Sfh1, pET28b-His₈ –Sfh3 and pET28b-His₈–Sfh4 were grown in E.coli BL21 (DE3; New England BioLabs Inc, Ipswich, MA). Sfh2 and Sfh5 expression was driven by pQE30-His₆-Sfh2 and pQE30-His₆-Sfh5 vectors in E.coli strain KK2186²⁵. Recombinant proteins of interest were bound to TALON metal affinity beads (Clontech, Mountain View, CA), and eluted with imidazole (10mM–200mM gradient) and dialyzed (Prod # 68100, Thermo Scientific, Rockford, IL). In the case of Sec14, dialysis was against 300mM NaCl, 25mM Na₂HPO₄ (pH=7.5), 5mM β-mercapthoethanol. Purified Sfh proteins were dialyzed against the same buffer with the exception that 50mM Na₂HPO₄ was used. Proteins mass was quantified by SDS-PAGE with BSA standard and A₂₈₀.

PtdIns-transfer assays

[³H]-PtdIns-transfer assays were performed using established methods¹⁰. In assays involving SMI, purified recombinant PITP was pre-incubated in the presence of acceptor membranes, buffer (300mM NaCl, 25mM Na₂HPO₄, pH 7.5) and SMI for 30 min. at 37°C prior to initiating the assay by addition of radiolabeled donor membranes. Fractional transfer of [³H]-PtdIns was normalized to mock DMSO controls. Nonlinear regression was applied to the dataset to calculate the best fit equation using [Y=100/(1+10^((LogIC₅₀−X)*HillSlope))] in Graphpad Prism v5.0. Statistical comparisons of inhibition was calculated using “extra sum-of-squares F-test” in Graphpad prism v 5.00.

Metabolic labeling and immunoprecipitation

Strains were grown in minimal media lacking methionine and cysteine to mid-logarithmic phase (OD_600nm ~ 0.5). Where indicated, cultures were treated with 20μM SMI or shifted to 37°C for 2h and radiolabeled with 20μCi/ml [³⁵S]-amino acids (Translabel; MP Biomedicals). Chase was initiated by addition of unlabeled methionine and cysteine (2 mM each, final concentration) and terminated with trichloroacetic acid (5% wt/vol, final concentration). CPY immunoprecipitation, SDS-polyacrylamide gel electrophoresis (PAGE), and autoradiography were performed as described⁴⁸. In washout experiments, cultures were pulse-radiolabeled and subjected to chase. Cells were then pelleted (30sec at 4,000rpm), washed 2X with fresh YPD medium, resuspended in YPD containing cyclohexamide (100μg/ml), and further incubated for the indicated times at 30°C. Subsequently, cells were poisoned with trichloroacetic acid (5% wt/vol, final concentration) and samples further processed as described above.

Phosphoinositide analyses

Samples were prepared as described ^32,33. Strain CTY100 (sec14-1^ts, sac1Δ) was grown overnight in uracil-free minimal media containing 2% glucose, 1% case amino acids and labeled to steady state for at least 20h with 10μCi/ml [³H]-myo-inositol (ART 0116A; American Radiolabeled Chemicals Inc., St. Louis, MO). Cells were either shifted to 37°C, or challenged with NPPM or DMSO vehicle for 3h, as appropriate. Labeling was terminated with trichloroacetic acid (5% final concentration) and samples incubated on ice for 30 min. Cells were pelleted (10,000 rpm for 1 min), washed twice in 500μl of cold ddH₂O, and resuspended in 500μl 4.5% perchloric acid. Approximately, 300μl of 0.5mm glass beads were added and cells disrupted by vigorous agitation for 10 min in 1 min bursts with 1 min rest on ice.

In experiments where inositol-glycerophospholipids were deacylated and resolved by strong anion exchange HPLC, bulk lipids were extracted as previously described with modification⁴⁹. Bulk lipids were extracted in 2x 250μl of CH₃CH₂OH:ddH₂O: (C₂H₅)₂O:C₄H₉OH (15:15:5:1 vol/vol), dried under N₂ gas, and deacylated by resuspension in 300μl of CH₃OH:ddH₂O: C₄H₉OH:CH₃NH₂ (0.8:0.6:0.2:0.35 vol/vol) and incubation for 30 min. at 53°C ⁵⁰. 100μl of cold CH₃CH₂CH₂OH was added to the solution, the liquid centrifuged to dryness under vacuum, and dessicate resuspended in 400μl ddH₂O. The solution was extracted 2X with 750μl 1-butanol:petroleum ether:ethyl formate (20:4:1 v/v/v), adjusted to 10mM (NH₄)₃PO₄, (pH 3.5), and soluble glycero-phosphoinositols resolved and quantified by HPLC.

In experiments where phosphoinositides were quantified by thin layer chromatography, lysates from disrupted cells was collected and centrifuged at 13,000 rpms for 10 min., the pellet washed with 500μl of 100mM EDTA (pH 7.4), and resuspended in 500μl of CHCl₃:CH₃OH:HCL (2:1:0.007). A two-phase system was produced by addition of 100μl of 0.6M HCl, the sample vortexed for 5 min., and sample centrifuged for 5 min. (13,000 rpm). The organic phase was collected, washed 2X with 250μl of CH₃OH:0.6M HCl:CHCl₃ (1:0.94:0.06), dried under N₂ gas and resuspended in 50μl CHCl₃. Samples were resolved by thin layer chromatography on Partisil LK6DF 60Å silica gel plates (Whatman, Cat# 4866-821) using a CHCl₃:CH₃OH:ddH₂O:NH₄OH (1:0.83:0.15:0.1) solvent system. Lipids were visualized by autoradiography and quantified with ImageJ (version 1.47t, National Institute of Health).

[³H]-Serine labeling of yeast

[³H]-Serine radiolabeling of yeast was performed as described¹³. The indicated strains were grown overnight at 30°C in uracil- and serine-free minimal media containing glucose (3%), ethanolamine (2mM), and sub-cultured to a λ_600nm=0.3. Cells were metabolically radiolabeled for 3h with 3.33μCi/ml L-[3-³H]-serine (ART 0246; American Radiolabeled Chemicals Inc., St. Louis, MO), and either shifted to 37°C, or challenged with 6748-481 (20μM) or DMSO for 3h, as appropriate. Labeling was terminated upon addition of ice-cold trichloroacetic acid (10% final concentration), and samples were incubated on ice for 30 min. Pellets were washed 2X with cold ddH₂O and re-suspended in ddH₂O:absolute ethanol (1:4, v/v) at 100°C for 45 min. The aqueous phase was re-extracted with CHCl₃:CH₃OH:0.2M KCl (4:4:3.3 v/v/v). The organic phase was washed 2X with PBS: CH₃OH (9:10, v/v), dried under N₂ gas, and the lipid film re-suspended in CHCl₃:CH₃OH (2:1, v/v) with 1mg/ml of butylated hydroxytoluene. Lipids were resolved by Silica Gel H thin layer chromatography (Analtech, Newark, DE) in a CHCl₃:2-propanol:0.25% KCl:triethylamine (30:9:6:18, v/v/v/v) solvent system. Plates were sprayed with 0.2% (w/v) 8-anilino-1-napthalenesulfonic acid, and lipids visualized under UV illumination. Individual lipid species were identified by internal standards (Avanti), collected, and radioactivity quantified by liquid scintillation counting. Sample loads were normalized by total cpm.

Homology modeling of the Sec14 closed conformer

A homology model for the closed Sec14 conformer was generated using Modeller⁵¹ based on the templates of the open conformer of Sec14 (PDB ID 1AUA) ⁵² and the closed conformer of Sfh1 bound to PtdIns (PDB ID 3B7N) ¹⁰. Gate residues in the Sec14 open conformation (I₂₁₅ – Y₂₄₇) were removed from that template structure prior to modeling whereas the corresponding gate residues in the closed conformation in Sfh1/PtdIns were retained. In addition, residues A₈₄ – Q₁₁₁ on the far side of the binding pocket from the gate were removed from the Sfh1 template prior to modeling since they were structurally divergent from the corresponding Sec14 residues.

Docking simulations

Several independent docking platforms were used. These included; GOLD [CCDC]⁵³; Glide⁵⁴; QM Polarized Ligand Docking [QM-PLD]⁵⁵. Details are presented below.

GOLD docking

Computational docking used the genetic algorithm-based ligand docking program GOLD (version 3.0.25)⁵³. GOLD exhaustively explores ligand conformations and provides limited flexibility to protein side chains with -OH groups by reorienting the H-bond donor and acceptor groups. GOLD scoring function is based on favorable conformations found in Cambridge Structural Database, and on empirical results of weak chemical interactions. The active site was defined by a single solvent accessible point near the center of the protein active site, a radius of ~10 Å, and the GOLD cavity detection algorithm. GOLD docking was unconstrained to obtain unbiased results and to explore all possible ligand binding modes. Ligand was docked in independent runs, 50 solutions were produced for each run, (except for one where 20 were generated), as opposed to the default of 10, and early termination of ligand docking was switched off. All other parameters were defaults. All ligands were docked using the same parameters.

Hydropathic scoring

HINT (Hydropathic INTeractions) scoring function was used to analyze docking solutions (version 3.11S b)^56,57. HINT evaluates each atom-atom interaction in a biomolecular complex using a parameter set derived from solvation partition coefficients for 1-octanol/water. The thermodynamic parameter Log P_o/w is directly correlated with free energy. HINT describes specific interactions between two molecules as:

$$\mathrm{B}=\mathrm{\sum }\mathrm{\sum }{\mathrm{b}}_{\text{ij}}=\mathrm{\sum }\mathrm{\sum }({\mathrm{a}}_{\mathrm{i}}{\mathrm{S}}_{\mathrm{i}}{\mathrm{a}}_{\mathrm{j}}{\mathrm{S}}_{\mathrm{j}}{\mathrm{R}}_{\text{ij}}{\mathrm{T}}_{\text{ij}}+{\mathrm{r}}_{\text{ij}})$$

where a is the hydrophobic atom constant derived from Log_o/w, S is the solvent accessible surface area, T is a function that differentiates polar-polar interactions (acid–acid, acid–base or base–base), and R, r are functions of the distance between atoms i and j as previously described ⁵⁷. The binding score, b_ij, describes the specific atom–atom interaction between atoms i and j, whereas B describes the total interaction. For selection of the optimum docked conformation and to further differentiate the relative binding efficacy of the NPPM ligands, interaction scores were calculated for each pose found by docking. The protein and ligands were partitioned as distinct molecules. ‘Essential’ hydrogen atoms, that is, only those attached to polar atoms (N, O, S, P), were explicitly considered in the model and assigned HINT constants. The inferred solvent model, where each residue is partitioned based on its hydrogen count, was applied. The solvent accessible surface area for the amide nitrogens was corrected with the ‘+20’ option.

Glide docking

Protein and ligands were prepared using Protein Preparation Wizard and LigPrep module of Maestro 9.2 Interface of Schrodinger Suite (Schrodinger Suite 2012; Glide version 5.8). Receptor Grids were generated without using any constraints and standard settings were used. Docking was performed using Standard Glide and QM-PLD modules with SP and XP scoring function respectively. No similarity, torsional and inter-molecular interaction (hydrogen bonding or hydrophobic) constraints were used. Ligand was docked flexibly with nitrogen inversions and ring sampling turned on with post-docking minimization.

PLIF

Protein-Ligand Interaction Fingerprint (PLIF) was calculated within MOE suite (2011.10; Chem. Comp. Group Inc., Montreal, Canada). PLIF was calculated between a closed Sec14 conformer and 6 representative binding modes produced by the docking runs. The protein Ligand Interaction Fingerprint (PLIF) is a method to encapsulate the interaction between ligands and proteins using a fingerprint scheme. To generate PLIF within MOE, maximum 250 bits were used with Min Score 1 turned off and keeping the Min Score 2 to its default value.

Simulation of charge distribution on activated aryl halides

Wave function calculations were carried out using the PC Spartan package (Wavefunction Inc. Irvine, CA; version 10 1.1.0). Starting geometries were obtained using Spartan’s interactive building mode, and pre-optimized using the MMFF force field. Wave functions were approximated using the Hartree-Fock method at the 3–21G⁽*⁾ gaussian basis set. Electrostatic potentials were generated onto surfaces of molecular electron densities (0.002 electrons per A³).

Fluorescence imaging

N-[3-Triethylammoniumpropyl]-4-[p-diethylaminophenylhexatrienyl] pyridinium dibromide (FM4-64; Invitrogen, Carlsbad, CA) staining was performed as described below. Cells were grown to mid-logarithmic phase (OD_{600 nm}=0.5) in YPD medium at 30°C, then either shifted to 37°C for 2h or treated with 20μM SMI at 25°C for 2 h. Subsequently, cells were pulsed with 10μM FM4-64 (Invitrogen) for 15 min. washed 2X in YPD media matched to the appropriate drug or temperature condition. Labeling was terminated at indicated times by washing cells in NaN₃/NaF (1mM final concentration of each) and placing samples on ice prior to imaging.

Cultures for GFP-Snc1 imaging were grown in synthetic defined medium lacking uracil at 30°C. Cells processed for imaging of phosphoinositide biosensors were collected from liquid cultures grown in uracil-free YNB supplemented with 3% glucose and 1% casamino acids at 25°C by centrifugation at 4,000 rpm for 1 min, and resuspended into fresh uracil-deficient medium prior to analysis. Cells were immobilized onto a thin layer of growth medium with 20% gelatin (G-2500, Sigma-Aldrich), sealed under a coverslip with Valap, and examined at 25°C. The imaging system employed a CFI plan apochromat lambda 100x oil immersion objective lens NA 1.45 mounted on a Nikon Ti-U microscope base (Nikon, Melville, NY) interfaced to a Photometrics CoolSNAP HQ2 high sensitivity monochrome CCD camera (Roper Scientific, Ottobrunn, Germany) or an Andor Neo sCMOS CCD camera (Andor Technology, Belfast, UK). A Lumen 200 Illumination System (Prior Scientific Inc., Rockland, MA.) was used in conjunction with a B-2E/C (465–495nm/515–555nm;EX/EM) or G-2E/C (528–553nm/590–650nm;EX/EM) filter set (Nikon, Melville, NY). Images were captured using the Nikon NIS Elements software package (Nikon, Melville, NY, version 4.10) and exported as .TIF files. Image analyses were performed using ImageJ (version 1.47t, National Institute of Health) and figures were constructed using Adobe Illustrator and Adobe Photoshop CS6 (version 15.0.0).

Transmission electron microscopy

Yeast were grown to an OD_{600 nm}=0.5 and cultures were either shifted to 37°C for 2 hours or challenged with 20μM SMI for 2 hours at 30°C. Cells were fixed in 3% glutaraldehyde, converted to spheroplasts, stained with 2% OsO₄ and 2% uranyl acetate, dehydrated in a 50%, 70%, 90% ethanol series, and washed in 100% ethanol and 100% acetone, respectively. Cell pellets were embedded into Spurr’s resin at 60°C for 48h and sectioned⁵⁸. Thin sections produced from strains in the SEC14^p-136 (CTY374) background were imaged at 80 kV on a Tecnai 12 electron microscope (FEI, Hillsboro, OR), and images captured using Gatan micrograph with version 3.9.3 software (Gatan, Pleasanton, CA). All other samples were visualized on a Jeol 1200 EX TEM operated at an accelerating voltage of 100 kV. Images were captured at calibrated magnifications using an optically coupled 3k slow scan CCD camera (model 15C, SIA, Duluth, GA) and Maxim DL imaging software.

Invertase secretion assays

Cells were grown in YPD (2% glucose) at 30°C, cultures were split, and cells incubated at 30°C ± NPPM (20μM) or DMSO or 37°C for 1h in YPD (2% glucose). Cells were then pelleted (2000g), washed twice with pre-warmed YPD (0.1% glucose), resuspended in the low glucose YPD medium, and incubated as before for 1.5h. To halt trafficking, samples were adjusted to 10mM NaN₃, and incubated on ice. The samples were washed 3X with 500μl ice-cold 10mM NaN₃ and re-suspended in 500μl of the same. The samples were split into 10mM NaN₃ buffers ± 0.2 % Triton X-100 (final) with the Triton-solubilized fractions also being subjected to one cycle of freeze-thaw to generate the permeabilized cell fraction. The partner non-permeabilized and permeabilized samples were used to determine extracellular and total invertase activities, respectively⁵⁹. Invertase units were expressed as nmoles of glucose produced per min. at 30°C.

Sequence alignment

Protein sequences were acquired from the Universal Protein Resource, aggregated using UGENE (version 1.10.1; http://ugene.unipro.ru/), and aligned with the T-Coffee module using the default settings. Homologous sequences were superimposed onto structural models (PDB IDs 1AUA, 1OLM, 3B7Z, 4FMM) to highlight the PtdIns/PtdCho lipid binding barcode.