|Home | About | Journals | Submit | Contact Us | Français|
Sac family phosphoinositide phosphatases comprise an evolutionarily conserved family of enzymes in eukaryotes. Our recently determined crystal structure of the Sac phosphatase domain of yeast Sac1, the founding member of the Sac family proteins, revealed a unique conformation of the catalytic P-loop and a large positively charged groove at the catalytic site. We now report a unique mechanism for the regulation of its phosphatase activity. Sac1 is an allosteric enzyme that can be activated by its product phosphatidylinositol or anionic phospholipid phosphatidylserine. The activation of Sac1 may involve conformational changes of the catalytic P-loop induced by direct binding with the regulatory anionic phospholipids in the large cationic catalytic groove. These findings highlight the fact that lipid composition of the substrate membrane plays an important role in the control of Sac1 function.
Phosphoinositides (PIs), a collection of seven biologically active and interconvertable phosphatidylinositol derivatives, are minor yet essential components of the cytosolic leaflet of eukaryotic cell membranes. In response to acute signaling inputs, phosphorylation or dephosphorylation of PIs provides spatial and temporal cues within the cell for PI binding proteins, which in turn promote the assembly of macromolecular complexes and initiate physiological events, including membrane trafficking and actin dynamics. Moreover, in response to extracellular stimuli, PIs can serve as precursors for several types of second messengers in multiple intracellular signaling pathways.1−3 The acute conversion between different PI derivatives is mediated by a large number of PI kinases and phosphatases.
The Sac domain-containing proteins make up an essential family of PI phosphatases. Members of this family include the transmembrane protein Sac1, as well as the cytoplasmic proteins Sac2/Inpp5f, Sac3/Fig4, and several synaptojanin homologues. The founding member of this PI phosphatase family, yeast Sac1p, an integral membrane protein, was identified more than 20 years ago.4,5 Since then, considerable progress on this enzyme has been made from yeast to mammals, as well as plants.6−12 Sac1 is an ER/Golgi-localized PI phosphatase that plays a major role in regulating phosphatidylinositol 4-phosphate [PI(4)P] metabolism.6,13 It has been shown that Sac1 is essential for the development of multicellular organisms,9,10,14 and the deletion of Sac1 in yeast results in a broad spectrum of functional defects, including impaired membrane trafficking, an abnormal actin cytoskeleton, and altered lipid metabolism.15
Sac1 is conserved in all eukaryotes with an N-terminal cytoplasmic Sac homology domain, a central unstructured linker region, and two C-terminal transmembrane motifs. Recently, the crystal structure of the Sac1 cytoplasmic domain has been determined.16 The structure revealed that the Sac homology domain is comprised of two closely packed subdomains: a novel N-terminal subdomain and the catalytic phosphoinositide phosphatase subdomain. Unlike other CX5RT/S-based PI phosphatases, such as PTEN17 or MTM,18 Sac1 has a large positively charged groove at the catalytic site. In addition, the catalytic P-loop of Sac1 has a striking conformation in that the catalytic cysteine is far from the conserved arginine that functions to bind the releasing phosphate group.16 These structural findings suggest that a conformational change of the catalytic P-loop may be required to achieve a functional arrangement of the catalytic residues. It is likely that other protein factors or lipid components are needed for the activation of Sac1. Recent studies have shown that a group of oxysterol binding homology (Osh) proteins activates Sac1 at the ER–plasma membrane contact sites.19 However, it is not known whether Sac1 phosphatase activity can be regulated by the lipid environment of its targeting membrane bilayer.
To understand the intrinsic enzymatic properties and regulation of Sac1, we performed in vitro kinetic studies of the phosphatase domain of Sac1. We found that Sac1 is an allosteric enzyme. The activity of Sac1 can be stimulated by anionic phospholipids, phosphatidylinositol (PtdIns) and phosphatidylserine (PS). We propose that the PI phosphatase activity of Sac1 is intrinsically regulated by the lipid composition of the membrane bilayer where Sac1 acts.
diC16 PI(4)P, diC16 PI(4,5)P2, diC16 PtdIns, diC16 PA, and diC8 PS were obtained from Avanti Polar Lipids, Inc. POPC and POPS were from NOF America Corp. All other diC8 lipids and inositol phosphates were purchased from Cellsignals Inc. (Columbus, OH).
DNA sequence corresponding to amino acids 1–511 of yeast Sac1 was amplified via polymerase chain reaction and cloned into a pET28-based bacterial expression vector in frame with an N-terminal six-His-sumo tag. The recombinant fusion protein was expressed in BL21 cells. Cells were harvested and lysed by sonication in phosphate-buffered saline (PBS). The lysate was clarified by centrifugation, and the supernatant was collected and passed through a column packed with cobalt affinity beads (Clontech). The column was extensively washed with PBS and eluted with 200 mM imidazole in PBS. The tag was removed by Ulp1 sumo specific protease, and the recombinant protein was further purified by size exclusion chromatography. The purified protein was stored at −80 °C in storage buffer [50 mM Tris (pH 6.8), 150 mM NaCl, 2 mM DTT, and 50% glycerol]. Sac1 mutants were generated by site-directed mutagenesis, and the constructs were verified by sequencing. The mutant proteins were expressed and purified as described above.
Long fatty acid chain lipids were dissolved in a solution containing 90% chloroform and 10% methanol. An appropriate amount of lipids solution was mixed in a glass vial according to the composition of the liposome. The lipids were then dried under a nitrogen stream and rehydrated with enzyme reaction buffer [50 mM Tris (pH 6.8), 150 mM NaCl, and 2 mM DTT] at 37 °C for 1 h. The resulting multilamellar liposome was extruded through a 30 nm pore diameter polycarbonate filter (WaterMan) using a Mini-Extruder device (Avanti) to obtain unilamellar vesicles.
Because the short acyl chain lipids can form micelles at high concentrations, it is critical to ensure the substrate or the activators are below their critical micelle concentration (CMC) in our assay with water-soluble lipids. It has been shown that diC8 PS has a CMC of 2.28 mM.20 The CMC of PtdIns was reported to be ~0.5 mM.21 Although the CMC of diC8 PI(4)P is still unknown, the CMC of diC8 PI(3)P is ~0.7 mM;21 thus, it is safe to estimate that the CMC of diC8 PI(4)P should also be ~0.7 mM. Other phospholipids, such as PI(4,5)P2, should have a CMC value of >0.7 mM because of the larger and highly charged headgroup. The concentrations of the lipids used in our enzymatic assays are all significantly below the CMC values. We expect that the lipid substrates or activators should be monodisperse and do not form micelles in our assays.
Before the enzymatic assay, Sac1 proteins from −80 °C storage were first diluted with reaction buffer to make an enzyme reaction stock (0.1 mg/mL or 1.8 μM). For all the reactions, except for the experiments designed to test the effects of enzyme concentration, the enzyme was further diluted 50 times to a final concentration of 36 nM. All the reaction mixtures were incubated at room temperature. An aliquot of the reaction mixture was taken out and mixed with 50 mM N-ethylmaleimide (NEM) to stop the reaction. Phosphate release was measured by the addition of malachite green reagent and absorbance at OD620. To obtain the initial velocities of the reactions, a series of reaction progress curves was recorded at different substrate concentrations. Initial velocities were derived using Dynafit22 and then plotted against the substrate concentrations using GraphPad Prism version 5 (http://www.graphpad.com/prism).
Sedimentation velocity analytical ultracentrifugation experiments were performed on a Beckman Proteomelab XL-A analytical ultracentrifuge at 10 °C. Protein samples were diluted with 20 mM HEPES, 20 mM NaCl, and 2 mM TCEP to a concentration of 1 or 0.1 mg/mL; 300 scans were collected at 50,000 rpm (200,000 g) and 230 nm with the time interval between scans set to zero. Sedimentation boundaries were analyzed by the continuous distribution method [c(S)] using SEDFIT.23 SEDNTERP24 was used to correct experimental S values (s) to standard conditions at 20 °C in water (S20,w) and to calculate the partial specific volume of Sac1.
Yeast strains used in this study are described in Table S1 of the Supporting Information. Plasmid pRS415-pGPD-mCherry-2xPHOsh2, which expresses mCherry-labeled tandem PH domains from the yeast Osh2 protein, was used as an in vivo PI(4)P biosensor reporter.19 Microscopy was performed on midlog cultures at room temperature. Images were obtained with a DeltaVision system and were deconvolved with softWoRx. For in vivo PI level analysis, yeast strains incubated at the indicated temperatures were labeled with [3H]-myo-inositol. Lipids were extracted and deacylated for analysis by high-performance liquid chromatography (HPLC) as described previously.25
The phosphatase domain of Sac1 has promiscuous substrate specificity in vitro. In agreement with previous results,7,26,27 we found that Sac1 prefers to act on PI(5)P, PI(4)P, PI(3)P, with low activity on PI(3,5)P2, and with no activity on other phosphoinositides, such as PI(4,5)P2, PI(3,4)P2, and PI(3,4,5)P3, or inositol polyphosphates (Figure S1A of the Supporting Information). Furthermore, unlike other inositol polyphosphate phosphatases,28 divalent ions, such as Mg2+, Ca2+, and Mn2+, have no obvious effect on the activity of Sac1 (Figure S1B of the Supporting Information). Because the major substrate of Sac1 in vivo is PI(4)P,6,8,29 we used water-soluble diC8 PI(4)P, which has two C8 fatty acid chains, or liposome-embedded brain diC16 PI(4)P (Avanti) as the substrate in all of the following kinetic experiments.
The kinetics of hydrolysis of diC8 PI(4)P by Sac1 were evaluated by an end point chromogenic assay based on a modified version of the malachite green method as described previously.30 Initial velocities were derived from series of reaction progress curves at different substrate concentrations with a fixed amount of enzyme (36 nM) and plotted against PI(4)P concentration (Figure (Figure1A).1A). Interestingly, the reaction does not follow simple Michaelis–Menten kinetics; instead, the velocity data assumed a sigmoidally shaped curve, and thus, the data were fit to the Hill equation: v = Vmaxh[S]/[S]h + K0.5h (where v is the initial velocity, h is the Hill coefficient, and K0.5 is the substrate concentration at which v is half-maximal). The calculated Hill coefficient h is 4.5 ± 0.3, and the K0.5 is ~16 ± 2 μM. Sigmoidal kinetics can arise from many possibilities, such as the oligomeric state of the enzyme, multiple catalytic sites, or slow conformational changes.31,32
To test whether Sac1 can form oligomers in solution, we used sedimentation velocity analytic ultracentrifugation (AUC) to examine the solution behavior of Sac1. At 1 mg/mL (18 μM), the enzyme exhibited two peaks with sedimentation coefficients of 3.1 S (3.2 S20,w, 55 kDa) and 4.3 S (4.4 S20,w, 96 kDa), corresponding to monomeric (80%) and dimeric (20%) forms of the protein, respectively (Figure (Figure1B,1B, solid line). However, 95% of the protein is monomeric when the protein concentration is reduced to 0.1 mg/mL (1.8 μM) (Figure (Figure1B,1B, dashed line). In our enzymatic assays, the final enzyme concentration is only 2 μg/mL (36 nM). These data suggest that Sac1 predominantly exists as a monomer under our assay conditions. We further investigated the correlation between the enzyme activity and its concentration. Sac1 enzymatic assays were conducted with the same amount of substrate but different enzyme concentrations (Figure S2A,B of the Supporting Information). Plotting of the initial velocities versus the enzyme concentration resulted in a linear line, indicating that the catalytic activity of Sac1 has no sigmoidal dependence on enzyme concentration. This result further suggests that the sigmoidal kinetics observed for Sac1 is not due to its oligomerization.
It has been reported that allosteric enzymes can be activated by their products to form a positive feedback loop.33,34 In our reaction system, the product is phosphatidylinositol (PtdIns). We asked whether PtdIns is an activator for Sac1. To examine the effect of PtdIns, the PI(4)P substrate concentration was held constant at 20 μM while the concentration of PtdIns was varied. A series of progress curves were recorded, and the initial velocities extracted from each progress curve were plotted versus PtdIns concentration (Figure (Figure2A).2A). Indeed, the activity of Sac1 was stimulated by the addition of PtdIns, and the activation reached saturation at ~50 μM PtdIns. Furthermore, the activation of Sac1 was not due to enzyme oligomerization because Sac1 remained predominantly as a monomer in the presence of PtdIns, as shown by sedimentation velocity measurements (Figure (Figure2C).2C). These results suggest that PtdIns may function as an allosteric activator. The data further predict that if all the enzyme molecules are fully activated by the addition of PtdIns, the reaction should follow simple Michaelis–Menten kinetics. Indeed, when PtdIns was kept at a fixed concentration (50 μM) in our reaction system, the plot of initial velocity versus PI(4)P substrate concentration changed to a hyperbolic curve (Figure (Figure2B)2B) that is distinct from the sigmoidal curve observed in the absence of an activator. These kinetic results, together with previous structure information, suggest that the enzyme may have heterogeneous conformations in solution and the unique catalytic P-loop may be involved in a slow conformational change sampling between an inactive form (observed in our crystal structure) and the active form (seen in other lipid phosphatases, such as PTEN).16 When activators bind, the balance is favored toward the active form.
Allosteric activation of Sac1 by PtdIns raised the question about the specificity of activators. PtdIns is an anionic lipid with a net charge of −1. We asked whether other negatively charged phospholipids could also activate the enzyme. We performed enzymatic assays in the presence of other water-soluble phospholipids and a fixed amount of PI(4)P. We found that short chain phosphoinositol derivatives, such as PI(4,5)P2 and PI(3,4)P2, did not activate the enzyme. However, Sac1 is strongly activated by phosphatidylserine (PS) (Figure S3A,B of the Supporting Information). To further investigate the activation of Sac1 by PS, we performed kinetic assays similar to the PtdIns activation experiments described above. Interestingly, a sigmoidal PS concentration-dependent rate curve was observed (Figure S4A of the Supporting Information), and the enzyme can be fully activated at a PS concentration of 20 μM. When the enzyme was fully activated by 20 μM PS, a hyperbolic rate curve of initial velocitie versus PI(4)P concentration was obtained (Figure S4B of the Supporting Information). These kinetic data suggest that the activity of Sac1 can be stimulated by anionic phospholipids PtdIns and PS. The biological significance of this activation will be further discussed below.
To further investigate the allosteric mechanism of Sac1, we utilized structure information about Sac1 and created substitutions to perturb the catalytic behavior of Sac1. The surface of Sac1 that contains the catalytic site has a deep cationic groove (Figure (Figure3A).3A). The catalytic P-loop assumes a unique conformation.16 In that conformation, the catalytic C392 is far from R398, which holds the releasing phosphate group on substrates. A conformational change of the P-loop seems to be required to bring C392 and R398 close enough to form a functional catalytic site. We introduced substitutions of several conserved residues, including R207Q, R256Q, C392S, D394N, D397N, and R398Q, that align inside the cationic groove (Figure (Figure3A).3A). R207 and R256 are the major residues that contribute to the cationic properties of the groove. The other residues are components of the catalytic P-loop. Recombinant proteins were expressed and purified as described for the wild-type protein. All mutant proteins were stable and displayed a profile similar to that of the wild-type protein as determined by gel filtration chromatography. Although most of the mutants have no detectable activity in our in vitro enzymatic assays, R207Q is active but with reduced allosteric kinetics compared to that of the wild-type enzyme. Plots of the initial velocity versus PI(4)P concentration were sigmoidal with a Hill coefficient h of 3.6 ± 0.3, and the K0.5 was ~46 ± 5 μM (Figure (Figure3B).3B). The structure of Sac1 revealed that R207 is ~11 Å from the catalytic cysteine and ~20 Å from the arginine (Figure (Figure3A).3A). This large distance suggests that the R207 site cannot be the substrate binding site; instead, it may be an allosteric site. In agreement with this idea, the R207Q mutant can be fully activated by a high concentration (20 μM) of PS (Figure (Figure3C),3C), suggesting R207 does not affect substrate binding. The reduced allosteric effect of the R207Q mutation is likely due to the reduced affinity for allosteric activators. Taken together, our structural observation and kinetic data suggest that R207 is an allosteric site, through which the binding of potential activators induces conformational changes in the catalytic P-loop to activate the enzyme. We thus propose a general mechanism for the kinetic behavior of Sac1 (Figure (Figure3D).3D). Sac1 is an allosteric enzyme and is predominantly in its inactive form in the absence of allosteric regulators. Upon binding of activators, the catalytic P-loop changes its conformation to allow the catalytic cysteine and arginine residues to be positioned in its proximity as seen in other active PI phosphatases, such as PTEN, leading to active catalysis.17
The physiological substrates of Sac1 are embedded in the membrane bilayer. We then asked whether the observed activation of Sac1 by PtdIns or PS applies to membrane-embedded substrates. To answer this question, liposomes were prepared as described (see Materials and Methods). Reaction progress curves were evaluated using liposomes with different lipid compositions (Figure (Figure4A,B).4A,B). When liposomes with a PC:PI(4)P lipid ratio of 4:1 or a PC:PI(4,5)P2:PI(4)P lipid ratio of 3:1:1 were used in the assay, the hydrolysis of PI(4)P was barely detectable. However, when liposomes with a PC:PtdIns:PI(4)P lipid ratio of 3:1:1, a PC:PS:PI(4)P lipid ratio of 3:1:1, or a PC:PA:PI(4)P lipid ratio of 3:1:1 were tested, significant PI(4)P hydrolysis was detected in comparison to liposomes lacking PtdIns, PS, or PA (Figure (Figure4A,B).4A,B). Because PS and PI are relatively abundant phospholipids in eukaryotic cells, we further investigated the dependency of the enzyme activity on the mole fraction of PS and PtdIns in liposomes. Liposomes were prepared with a fixed molar ratio of diC16 PI(4)P. The amount of PS or PtdIns was sampled from 0 to 20%, and the neutral lipid PC was adjusted accordingly to make the rest of the fraction. The progress of the reaction was recorded, and initial velocities were plotted (Figure (Figure4C,D4C,D and Figure S5 of the Supporting Information). Interestingly, Sac1 activity on PI(4)P-containing liposomes can be stimulated in a PS or PtdIns mole fraction-dependent fashion. These data demonstrate that the activity of Sac1 can be modulated by the lipid composition of the membrane bilayer.
The allosteric activation of Sac1 by anionic lipids, particularly by PS, led us to predict that PI(4)P levels should increase in cells with reduced levels of PS. We took the advantage of the available yeast W303-1A strain, which carries the cho1 gene deletion.35 Yeast Cho1 is a CDP-diglyceride serine phosphatidyltransferase and is responsible for the de novo synthesis of PS (Figure (Figure5A).5A). We first used the PI(4)P FLARE (fluorecent lipid-associated reporter) mCherry-2xPHOsh2 to examine the localization of PI(4)P in vivo. In wild-type cells, the PI(4)P FLARE was found at the Golgi apparatus and the inner plasma membrane surface. However, in cho1Δ yeast cells, which lack PS,35 the amount of PI(4)P FLARE was increased at the PM (Figure (Figure5B),5B), similar to the localization of the PI(4)P FLARE in cells lacking Sac1.19,36 Furthermore, analysis of [3H]inositol-labeled phospholipids indicated that PI(4)P levels in cho1Δ cells are nearly 2-fold higher than in wild-type cells (Figure (Figure5C).5C). This effect was specific for PI(4)P, as levels of other phosphoinositide isoforms were reduced in cho1Δ cells (Table S2 of the Supporting Information). Taken together, our results suggest that the anionic lipid PS modulates the localization and levels of PI(4)P in vivo, likely through the regulation of the phosphatase activity of Sac1.
In this study, we report that Sac1 is an allosteric enzyme. Although sigmoidal kinetics often indicate multimeric enzymes with several interacting active sites, monomeric enzymes have also been reported to display allosteric regulation.37 In the case of Sac1, equilibrium between the dimer and monomer was observed in our AUC experiment at a relatively high concentration. However, at the concentration of our enzymatic assays, Sac1 did not exhibit any sigmoidal correlation between its activity and concentration. Although dimerization of Sac1 may provide a regulatory mechanism,11 the allosteric kinetics of Sac1 that we recorded is likely not due to the oligomerization of the enzyme. On the basis of our mutagenesis studies and the previously published Sac1 structure,16 we propose that Sac1 is an allosteric enzyme. Binding of activators in the deep cationic groove induces a conformational switch of the catalytic P-loop, thereby activating Sac1. This intrinsic structural switch is reflected macroscopically by the sigmoidal kinetic behavior of the enzyme. Our studies further revealed that Sac1 can be activated by anionic phospholipids PtdIns and PS. The activation of Sac1 seems to be specific as other lipids with negatively charged headgroups, such as PI(4,5)P2 and PI(3,4)P2, did not activate Sac1.
Activation of an enzyme by its product has been observed in other lipid phosphatases, including PTEN,33,38 myotubularin phosphatases,34 and a secreted salivary inositol polyphosphate 5-phosphatase.28 PtdIns is most enriched in the endoplasmic reticulum (ER), up to 15% of total phospholipids in ER.39 Moreover, PtdIns synthesis occurs in the ER where Sac1 resides. This may provide a mechanism for efficiently maintaining low PI(4)P levels at the ER. PA is a minor component of cell membranes. Production of PA by receptor-stimulated phospholipase D has been shown to activate PIP 5-kinase that converts PI(4)P to PI(4,5)P2 at the plasma membrane.40−42 Our results showed that PA can activate Sac1 in vitro. However, whether PA can also stimulate Sac1 activity in vivo and the biological consequence of this potential regulation by PA remains to be investigated. The allosteric activation of Sac1 by PS may have interesting biological implications. The requirement of PS for Sac1 to hydrolyze its substrates presented on liposomes suggests that the lipid composition of the membrane on which Sac1 acts is critical for its enzyme activity. Consistent with this, we found that PI(4)P levels are increased in cho1Δ mutant cells that are defective in PS synthesis. This further implies that Sac1 may prefer to act in a membrane environment enriched with PS. While PS is an abundant negatively charged phospholipid in eukaryotic cells, it is distributed unevenly within cellular membrane-bound organelles. PS is found preferentially in the inner leaflet of the plasma membrane and less abundant in other membranes, such as the ER and Golgi compartments.39,43 Genetic and biochemical data have shown that Sac1 regulates a pool of PI(4)P on the plasma membrane.6,13 The crystal structure of Sac1 revealed a long flexible linker between the catalytic domain and the transmembrane domains. This long linker physically allows the ER-anchored Sac1 to function in trans on the plasma membrane.16 Our in vitro kinetic results support the idea that the PS-enriched plasma membrane would be a favorable membrane environment in which Sac1 can function. The regulation of Sac1 activity by the anionic phospholipid PS may also be part of a homeostatic regulatory loop that controls the overall surface charge of the plasma membrane. In agreement with this notion, we found that cho1Δ cells with reduced PS levels in turn have increased levels of PI(4)P on the inner leaflet of the plasma membrane. Interestingly, cells with impaired Sac1 activity accumulate PI(4)P and in turn have reduced levels of PS.8,44
Allosteric enzymes possess a physiological advantage in their ability to respond to changes in the concentration of their substrates and products. The sigmoidal behavior of Sac1 suggests an intrinsic mechanism for the tight and acute control of PI(4)P levels in vivo. However, control of Sac1 activity also requires additional regulatory protein factors. Recent data have shown that a group of oxysterol binding homology (Osh) proteins activate Sac1 at ER–plasma membrane contact sites.19 Our finding of the activation of Sac1 by anionic lipids adds another layer of regulation to the maintenance of PI(4)P homeostasis in vivo.
We thank Dr. Hong Qian (University of Washington, Seattle, WA) for advice on enzyme kinetics. We thank Dr. Susan Henry (Cornell University) for cho1 mutant yeast strains. We thank Dr. Scott Emr and Fenghua Hu (Cornell University) for discussion about the experiment and reading of the manuscript.
This work is supported by National Institutes of Health Grants 1R01GM094347 (to Y.M.) and 1R01CA14052 (to M.S.C.).
The authors declare no competing financial interest.
§ School of Forensic Medicine, Kunming Medical University, Kunming 650500, China.
Supporting materials, including Figures S1–S5 and Tables S1 and S2. This material is available free of charge via the Internet at http://pubs.acs.org.