@article{benson_christensen_fairchild_davis_2009, title={The mRNA for zona pellucida proteins B1, C and D in two genetic lines of turkey hens that differ in fertility}, volume={111}, ISSN={["1873-2232"]}, DOI={10.1016/j.anireprosci.2008.02.013}, abstractNote={The avian inner perivitelline layer (IPVL) contains zona pellucida protein-B1 (ZPB1), zona pellucida protein-C (ZPC) and zona pellucida protein-D (ZPD). These three proteins may be involved in sperm binding to the IPVL. ZPB1 is produced by the liver and transported to the developing preovulatory follicle, while ZPC and ZPD are synthesized and secreted by the granulosa cells of the preovulatory follicle. The mRNA of ZPB1, ZPC, and ZPD was investigated in two lines of turkey hens selected for over 40 generations for either increased egg production (E line) or increased body weight (F line). Total RNA was extracted from the liver and from 1cm(2) sections of the granulosa layer around the germinal disc and a nongerminal disc area of the F(1) and F(2) follicles of hens from each genetic line. Northern analysis was performed using chicken cDNA probes for all three ZP proteins. Hepatic mRNA for ZPB1 was greater (P<0.05) in turkey hens from the E line than the F line. Although, there was no difference in ZPC mRNA between the germinal disc and nongerminal disc region of the two largest follicles in E line hens, ZPC mRNA was greater in the nongerminal disc region compared to the germinal disc region in the two largest follicles obtained from the F line hens. There were no differences in ZPD mRNA between the germinal disc and nongerminal disc regions of the F(1) and F(2) follicles for either genetic line. The results suggest that the greater rates of fertility previously observed in eggs from the E line hens compared with the F line of hens may be related to differential amounts of the potential sperm binding proteins ZPB1 and ZPC.}, number={2-4}, journal={ANIMAL REPRODUCTION SCIENCE}, author={Benson, A. P. and Christensen, V. L. and Fairchild, B. D. and Davis, A. J.}, year={2009}, month={Apr}, pages={149–159} }
 @article{davis_im_dubin_tomer_boss_2007, title={Arabidopsis phosphatidylinositol phosphate kinase 1 binds F-actin and recruits phosphatidylinositol 4-kinase beta 1 to the actin cytoskeleton (Retracted article. See vol. 284, pg. 16060, 2009)}, volume={282}, ISSN={["1083-351X"]}, DOI={10.1074/jbc.M611728200}, abstractNote={The actin cytoskeleton can be influenced by phospholipids and lipid-modifying enzymes. In animals the phosphatidylinositol phosphate kinases (PIPKs) are associated with the cytoskeleton through a scaffold of proteins; however, in plants such an interaction was not clear. Our approach was to determine which of the plant PIPKs interact with actin and determine whether the PIPK-actin interaction is direct. Our results indicate that AtPIPK1 interacts directly with actin and that the binding is mediated through a predicted linker region in the lipid kinase. AtPIPK1 also recruits AtPI4Kβ1 to the cytoskeleton. Recruitment of AtPI4Kβ1 to F-actin was dependent on the C-terminal catalytic domain of phosphatidylinositol-4-phosphate 5-kinase but did not require the presence of the N-terminal 251 amino acids, which includes 7 putative membrane occupation and recognition nexus motifs. In vivo studies confirm the interaction of plant lipid kinases with the cytoskeleton and suggest a role for actin in targeting PIPKs to the membrane. The actin cytoskeleton can be influenced by phospholipids and lipid-modifying enzymes. In animals the phosphatidylinositol phosphate kinases (PIPKs) are associated with the cytoskeleton through a scaffold of proteins; however, in plants such an interaction was not clear. Our approach was to determine which of the plant PIPKs interact with actin and determine whether the PIPK-actin interaction is direct. Our results indicate that AtPIPK1 interacts directly with actin and that the binding is mediated through a predicted linker region in the lipid kinase. AtPIPK1 also recruits AtPI4Kβ1 to the cytoskeleton. Recruitment of AtPI4Kβ1 to F-actin was dependent on the C-terminal catalytic domain of phosphatidylinositol-4-phosphate 5-kinase but did not require the presence of the N-terminal 251 amino acids, which includes 7 putative membrane occupation and recognition nexus motifs. In vivo studies confirm the interaction of plant lipid kinases with the cytoskeleton and suggest a role for actin in targeting PIPKs to the membrane. Arabidopsis phosphatidylinositol phosphate kinase 1 binds F-actin and recruits phosphatidylinositol 4-kinase β1 to the actin cytoskeleton.Journal of Biological ChemistryVol. 284Issue 23PreviewVOLUME 282 (2007) PAGES 14121–14131 Full-Text PDF Open Access Phosphatidylinositol phosphate kinases (PIPKs) 2The abbreviations used are: PIPK, phosphatidylinositol phosphate kinase; eEF1A, eukaryotic elongation factor 1A; MORN, membrane occupation and recognition nexus; PtdIns4P, phosphatidylinositol-4-phosphate; PtdIns(4,5)P2, phosphatidylinositol-4,5-bisphosphate; PI4K, phosphatidylinositol kinase; PtdOH, phosphatidic acid; MALDI, matrix-assisted laser desorption ionization time; PLC, phospholipase C; GST, glutathione S-transferase; MS, mass spectroscopy; PIPES, 1,4-piperazinediethanesulfonic acid. are a family of enzymes that phosphorylate phosphatidylinositol phosphates (PtdInsP) to phosphatidylinositol bisphosphates (PtdInsP2). Arabidopsis has 11 predicted isoforms of PtdInsP kinases (1Mueller-Roeber B. Pical C. Plant Physiol. 2002; 130: 22-46Crossref PubMed Scopus (322) Google Scholar). AtPIPK1 and AtPIPK10 have been characterized biochemically, and PtdIns4P is the primary substrate making the predominant product phosphatidylinositol (4Perera I.Y. Davis A.J. Galanopoulou D. Im Y.J. Boss W.F. FEBS Lett. 2005; 579: 3427-3432Crossref PubMed Scopus (52) Google Scholar, 5Davis A.J. Perera I.Y. Boss W.F. J. Lipid Res. 2004; 45: 1783-1789Abstract Full Text Full Text PDF PubMed Scopus (12) Google Scholar) bisphosphate (PtdIns(4,5)P2), suggesting that they are similar to the type I PIPKs found in humans (2Hinchliffe K.A. Ciruela A. Irvine R.F. Biochim. Biophys. Acta. 1998; 1436: 87-104Crossref PubMed Scopus (102) Google Scholar, 3Westergren T. Dove S.K. Sommarin M. Pical C. Biochem. J. 2001; 359: 583-589Crossref PubMed Scopus (48) Google Scholar, 4Perera I.Y. Davis A.J. Galanopoulou D. Im Y.J. Boss W.F. FEBS Lett. 2005; 579: 3427-3432Crossref PubMed Scopus (52) Google Scholar, 5Davis A.J. Perera I.Y. Boss W.F. J. Lipid Res. 2004; 45: 1783-1789Abstract Full Text Full Text PDF PubMed Scopus (12) Google Scholar). However, unlike the mammalian, yeast or Caenorhabditis elegans PIPKs, AtPIPK1–9 contain N-terminal putative membrane occupation and recognition nexus (MORN) motifs similar to those first reported in junctophilins (1Mueller-Roeber B. Pical C. Plant Physiol. 2002; 130: 22-46Crossref PubMed Scopus (322) Google Scholar, 6Takeshima H. Komazaki S. Nishi M. Lino M. Kangawa K. Mol. Cell. 2000; 6: 11-22Abstract Full Text Full Text PDF PubMed Google Scholar). MORN motifs are unique to this family of enzymes and have not been reported in any other eukaryotic lipid kinases. In eukaryotic models PtdInsP kinases supply PtdIns(4,5)P2 for many cellular functions, and it is well known that phospholipids play an integral role in regulating the structure and dynamics of the cytoskeleton through the many actin-binding proteins that interact with PtdIns(4,5)P2 (7Simonsen A. Wurmser A.E. Emr S.D. Stenmark H. Curr. Opin. Cell Biol. 2001; 13: 485-492Crossref PubMed Scopus (407) Google Scholar, 8Takenawa T. Itoh T. Biochim. Biophys. Acta. 2001; 1533: 190-206Crossref PubMed Scopus (245) Google Scholar, 9Downes C.P. Gray A. Lucocq J.M. Trends Cell Biol. 2005; 15: 259-268Abstract Full Text Full Text PDF PubMed Scopus (191) Google Scholar, 10Yin H.L. Janmey P.A. Annu. Rev. Physiol. 2002; 65: 761-789Crossref PubMed Scopus (567) Google Scholar). For example, the absence of PtdInsP kinases in yeast has an adverse affect on the yeast cell morphology (11Homma K. Terui S. Minemura M. Qadota H. Anraku Y. Kanaho Y. Ohya Y. J. Biol. Chem. 1998; 273: 15779-15786Abstract Full Text Full Text PDF PubMed Scopus (139) Google Scholar) and causes them to be unable to properly form actin cables (12Desrivieres S. Cooke F.T. Parker P.J. Hall M.N. J. Biol. Chem. 1998; 271: 15787-15793Abstract Full Text Full Text PDF Scopus (177) Google Scholar). In some animal cell lines overexpression of PtdInsP kinases results in the formation of actin comet tails and stress fibers (13Yamamoto M. Hilgemann D.H. Feng S. Bito H. Ishihara H. Shibasaki Y. Yin H.L. J. Cell Biol. 2001; 152: 867-876Crossref PubMed Scopus (104) Google Scholar, 14Kanzaki M. Furukawa M. Raab W. Pessin J.E. J. Biol. Chem. 2004; 279: 30622-30633Abstract Full Text Full Text PDF PubMed Scopus (53) Google Scholar) and disruption of membrane trafficking (14Kanzaki M. Furukawa M. Raab W. Pessin J.E. J. Biol. Chem. 2004; 279: 30622-30633Abstract Full Text Full Text PDF PubMed Scopus (53) Google Scholar). In these animal cell lines, a functional PtdInsP kinase producing PtdIns(4,5)P2 was necessary to cause the changes in cytoskeletal structure (13Yamamoto M. Hilgemann D.H. Feng S. Bito H. Ishihara H. Shibasaki Y. Yin H.L. J. Cell Biol. 2001; 152: 867-876Crossref PubMed Scopus (104) Google Scholar, 15Tolias K.F. Hartwig J.H. Ishihara H. Shibasaki Y. Cantlry L.C. Carpenter C.L. Curr. Biol. 2000; 10: 153-156Abstract Full Text Full Text PDF PubMed Scopus (220) Google Scholar). In mammalian systems, PIPK activity has been recovered with an F-actin fraction, and the α, β, and γ isoforms have been identified as co-purifying with F-actin. The data support a model where these PIPKs do not directly interact with F-actin but that this interaction is mediated by the presence of Racs, small GTP-binding proteins (16Yang S.A. Carpenter C.L. Abrams C.S. J. Biol. Chem. 2004; 279: 42331-42336Abstract Full Text Full Text PDF PubMed Scopus (62) Google Scholar). In plants, PtdIns kinase activity (17Xu P. Lloyd C.W. Staiger C.J. Drøbak B.K. Plant Cell. 1992; 4: 941-951Crossref PubMed Google Scholar, 18Tan Z. Boss W.F. Plant Physiol. 1992; 100: 2116-2120Crossref PubMed Scopus (73) Google Scholar) and PtdInsP kinase activity co-purify with an F-actin-enriched fraction (18Tan Z. Boss W.F. Plant Physiol. 1992; 100: 2116-2120Crossref PubMed Scopus (73) Google Scholar), but it is not clear from these studies whether plant PtdInsP kinases directly bind F-actin or, like animal PIPKs, associate with a scaffold of actin-binding proteins. Actin remodeling in root hairs and pollen tubes, sites of rapid growth in plants, is sensitive to alterations in PtdIns(4,5)P2 biosynthesis. It was first noted that expression of mutant Arabidopsis Rac2 in tobacco pollen tubes decreased plasma membrane PtdIns(4,5)P2 and disrupted the normal actin filament orientation (19Kost B. Lemichez E. Spielhofer P. Hong Y. Tolias K. Carpenter C. Chua N.H. J. Cell Biol. 1999; 145: 317-330Crossref PubMed Scopus (444) Google Scholar). The GFP-AtRac2 was localized to the apical tip of the pollen tube, and PtdInsP kinase activity was co-immunoprecipitated with Rac2 antibodies. More recent work in pollen tubes suggested that phospholipase C-mediated PtdIns(4,5)P2 turnover also affected actin structure and suggested that PtdIns(4,5)P2 metabolism is necessary in normal pollen growth (20Dowd P.E. Coursol S. Skirpan A.L. Kao T.H. Gilroy S. Plant Cell. 2006; 18: 1438-1453Crossref PubMed Scopus (183) Google Scholar, 21Helling D. Possart A. Cottier S. Klahre U. Kost B. Plant Cell. 2006; 18: 3519-3534Crossref PubMed Scopus (193) Google Scholar). Expression of inactive mutants of PLC isoforms resulted in increased levels PtdIns(4,5)P2 throughout the entire pollen tube and enriched at the apical tip (20Dowd P.E. Coursol S. Skirpan A.L. Kao T.H. Gilroy S. Plant Cell. 2006; 18: 1438-1453Crossref PubMed Scopus (183) Google Scholar, 21Helling D. Possart A. Cottier S. Klahre U. Kost B. Plant Cell. 2006; 18: 3519-3534Crossref PubMed Scopus (193) Google Scholar). In both tobacco (21Helling D. Possart A. Cottier S. Klahre U. Kost B. Plant Cell. 2006; 18: 3519-3534Crossref PubMed Scopus (193) Google Scholar) and petunia (20Dowd P.E. Coursol S. Skirpan A.L. Kao T.H. Gilroy S. Plant Cell. 2006; 18: 1438-1453Crossref PubMed Scopus (183) Google Scholar), disruption of PLC hydrolysis of PtdIns(4,5)P2 using inactive mutants or chemical inhibitors of PLC resulted in tip swelling, suggesting that PLC-mediated PtdIns(4,5)P2 turnover was essential for normal tip-directed growth. In root hairs, decreased tip-localized PtdIns(4,5)P2 resulting from a loss of the inositol phospholipid transfer protein, Atsfh1p, also resulted in unfocused root hair growth and disruption of the tip-directed actin filament orientation (22Vincent P. Chua M. Nogue F. Fairbrother A. Mekeel H. Xu Y. Allen N. Bibikova T.N. Gilroy S. Bankaitis V.A. J. Cell Biol. 2005; 168: 801-812Crossref PubMed Scopus (168) Google Scholar). Both in pollen and root hairs the data suggested that tip-localized PtdIns(4,5)P2 affected tip growth via F-actin-mediated vesicle trafficking (23Boss W.F. Davis A.J. Im Y.J. Galvao R.M. Perera I.Y. Subcell. Biochem. 2006; 39: 181-205Crossref PubMed Scopus (28) Google Scholar). Such a role for the inositol lipids in actin-mediated vesicle trafficking was supported by the work of Preuss et al. (24Preuss M.L. Serna J. Falbel T.G. Bednarek S.Y. Nielsen E. Plant Cell. 2004; 16: 1589-1603Crossref PubMed Scopus (200) Google Scholar). They showed that mutants in PI4Kβ1 and PI4Kβ2 or treatment with latrunculin (24Preuss M.L. Serna J. Falbel T.G. Bednarek S.Y. Nielsen E. Plant Cell. 2004; 16: 1589-1603Crossref PubMed Scopus (200) Google Scholar) disrupted trafficking of specialized RabA4b-associated Golgi vesicles in root hairs. Blocking turnover of PtdIns(4,5)P2 and increasing PtdIns(4,5)P2 pools by a mutation in inositol lipid phosphatases also affected actin filament orientation and cell wall biosynthesis in inflorescent stems of Arabidopsis (25Zhong R. Burk D.H. Morrison W.H. II I Ye Z.H. Plant Cell. 2004; 16: 3242-3259Crossref PubMed Scopus (99) Google Scholar). Although all of these data strongly support a role for PtdIns(4,5)P2 regulating actin cytoskeleton and vesicle trafficking in plants, the underlying mechanism is not known. Because the major family of plant PIPKs have a very different structure to the PIPKs of animals and yeast, it is important to understand how they are regulated and to identify their interacting partners. In this work our approach was to identify proteins that interact specifically with AtPIPK1 and AtPIPK10 and to determine whether one of the PIPKs would associate with F-actin. We show that Arabidopsis PIPK1 directly interacts with actin and recruits PI4Kβ1, suggesting a plant-specific mechanism for influencing cytoskeletal dynamics and lipid signaling. Because AtPIPK1 has an N-terminal MORN domain with the potential for tight membrane adhesion, the presence of AtPIPK1 on F-actin would provide a discrete pool of PtdIns(4,5,)P2 for actin-mediated vesicle trafficking and/or fusion. Production of GST Fusion Proteins in Escherichia coli—GST-AtPIPK1 and GST-AtPIPK10 were cloned and expressed in E. coli as previously described (4Perera I.Y. Davis A.J. Galanopoulou D. Im Y.J. Boss W.F. FEBS Lett. 2005; 579: 3427-3432Crossref PubMed Scopus (52) Google Scholar, 5Davis A.J. Perera I.Y. Boss W.F. J. Lipid Res. 2004; 45: 1783-1789Abstract Full Text Full Text PDF PubMed Scopus (12) Google Scholar). AtPI4Kβ1 was PCR amplified using Pfx DNA polymerase with forward primer 5′-CACCATGCCGATGGGACGCTTT-3′ and reverse primer 5′-CTCACACTCTTCCATTTAAGACCCGTTGGTA-3′. The resulting PCR product was subcloned into the pENTR/SD/D-TOPO destination vectors (Invitrogen) and then into pDEST15 for expression of GST-AtPI4Kβ1 proteins in E. coli. The inactive form of AtPIPK1, AtPIPK1(K468A), was mutated in the ATP binding site using QuikChange XL site-directed mutagenesis kit (Stratagene, La Jolla, CA). The ATP binding site was identified by sequence homology to that of the Human PtdInsP kinase, HsPIPK1α, and mutated as described (26Park S.J. Itoh T. Takenawa T. J. Biol. Chem. 2001; 276: 4781-4787Abstract Full Text Full Text PDF PubMed Scopus (74) Google Scholar). The mutated construct was amplified with forward primer 5′-CTCAAGATGATAGATTTATGATCGCAACGGTGAAGAAATCAGAAGTCAAG-3′ and reverse primer 5′-CTTGACTTCTGATTTCTTCACCGTTGCGATCATAAATCTATCATCTTGAG-3′. The resulting PCR product was subcloned into the pENTR/SD/D-TOPO destination vectors (Invitrogen) and then into pDEST15 for expression of GST-AtPIPK1(K468A). Truncations of AtPIPK1, MORN, ΔMORN, ΔMORN/ΔL, ΔMORN/ΔC, and ΔMORN/ΔL/ΔC were produced as previously described (27Im Y.J. Davis A.J. Perera I.Y. Johannes E. Allen N.S. Boss W.F. J. Biol. Chem. 2007; 282: 5443-5452Abstract Full Text Full Text PDF PubMed Scopus (69) Google Scholar). All GST fusion proteins were produced and purified as previously described (5Davis A.J. Perera I.Y. Boss W.F. J. Lipid Res. 2004; 45: 1783-1789Abstract Full Text Full Text PDF PubMed Scopus (12) Google Scholar). Protein concentration was determined using the Bio-Rad protein assay reagent with bovine serum albumin as a standard. Purified recombinant protein bound to the glutathione-Sepharose beads was stored at 4 °C until use in protein binding or activity assays. Plant Material—The recombinant binary plasmid pK7WGF2-HsPIPK1α was transformed into Agrobacterium tumefaciens EHA105 by the freeze-thaw method. For stable transformation, NT-1 cells were transformed using A. tumefaciens-mediated gene transfer following the protocol of Perera et al. (4Perera I.Y. Davis A.J. Galanopoulou D. Im Y.J. Boss W.F. FEBS Lett. 2005; 579: 3427-3432Crossref PubMed Scopus (52) Google Scholar). Cells were subcultured weekly into 25 ml of NT-1 culture medium containing 50 μgml–1 kanamycin as described by Perera et al. (4Perera I.Y. Davis A.J. Galanopoulou D. Im Y.J. Boss W.F. FEBS Lett. 2005; 579: 3427-3432Crossref PubMed Scopus (52) Google Scholar). NT-1 cells expressing ΔMORN were produced as previously described (27Im Y.J. Davis A.J. Perera I.Y. Johannes E. Allen N.S. Boss W.F. J. Biol. Chem. 2007; 282: 5443-5452Abstract Full Text Full Text PDF PubMed Scopus (69) Google Scholar). Protein Pulldown Assays—Protein pulldown assays were performed with purified recombinant protein incubated with precleared Triton-solubilized Arabidopsis membrane fractions for 2 h at 4 °C with continuous mixing in 30 mm β-cyclodextrin, phosphate-buffered saline (0.1 m KH2PO4, 0.1 m K2HPO4, 135 mm NaCl, and 2.7 mm KCl, pH 7.3), and final concentrations of 3 mm ATP or 0.5 mm GTP where indicated. Triton-solubilized Arabidopsis membranes were prepared by incubating a 40,000 × g pellet isolated as described previously in 1% (v/v) Triton X-100 10 min at 4 °C then centrifuging at 10,000 × g for 10 min to obtain a Triton-solubilized supernatant (5Davis A.J. Perera I.Y. Boss W.F. J. Lipid Res. 2004; 45: 1783-1789Abstract Full Text Full Text PDF PubMed Scopus (12) Google Scholar). Membrane fractions were precleared by incubating the solubilized membranes with purified GST immobilized on glutathione-Sepharose beads. The cleared supernatant was used for protein-protein interactions. After incubation with the membrane proteins, the beads were washed extensively with phosphate-buffered saline. Direct interactions of proteins were investigated using purified recombinant proteins either coupled to beads, released from the beads with reduced glutathione (Novagen, manufacturers instructions), or cleaved from the GST tag with thrombin (Novagen, manufacturers instructions) in phosphate-buffered saline or where indicated by using purified native protein. After washing unbound proteins from the beads, the bound proteins and beads were either added directly into 4× SDS-PAGE sample buffer, heated at 100 °C for 5 min for separation by SDS-PAGE, or were used immediately for lipid kinase activity analysis. Mass Spectrometry—After SDS-PAGE, the gels were stained with Coomassie Brilliant Blue, and stained bands were selected for excision. Digestion was carried out with bands excised from 10% polyacrylamide gel stained with Coomassie in 20 ml of 100 mm ammonium bicarbonate buffer, pH 8.0, 1 mm CaCl2. 1 ml of modified trypsin (sequencing grade, Promega) solution (2 mg/ml in 50 mm acetic acid) was added to yield a final enzyme protein ratio of 1:10. After incubation at 37 °C for 2 h, 2 ml of 1 n HCl was added to bring the mixture to pH 3.0 and inactivate the trypsin. One-half of the sample (10 ml) was moved to another tube and mixed with 2 ml of 0.1 m Tris-(2-carboxyethyl)phosphin (Pierce) in 0.1 m citrate buffer, pH 3.0. The mixture was incubated at 37 °C for 30 min for reduction of the disulfide bond-containing tryptic peptides. 1-ml aliquots of reduced or non-reduced tryptic peptide mixtures were used for MALDI-MS analysis without further purification. MALDI time-of-flight MS analyses were carried out using a Voyager-DE STR mass spectrometer (Applied Biosystems Inc.) equipped with a pulsed UV nitrogen laser (337 nm, 3-ns pulse) and a dual microchannel plate detector. For molecular mass determination of peptides, spectra were acquired at linear-delayed extraction (DE) mode, acceleration voltage set at 25 kV, grid voltage at 95% of the acceleration voltage, delay time at 320 ns, and low mass gate set at 1000 Da. The mass to charge ratio was calibrated with the molecular mass of a mixture of proteins (mass, 5,734.58–16,952.56). For analysis of tryptic peptides, the spectra were acquired at reflectron-DE mode with acceleration voltage set to 20 kV, grid voltage at 72% of the acceleration voltage, delay time at 200 ns, and low mass gate at 250 Da. The mass to charge ratio was calibrated with the mass of a mixture of standard peptides (mass, 904.46–5,734.58). Saturated α-cyano-4-hydroxycinnamic acid in 70% acetonitrile containing 0.1% trifluoroacetic acid was used as the matrix for analysis of tryptic peptides, and saturated sinapinic acid in 50% acetonitrile containing 0.1% trifluoroacetic acid was used as the matrix for protein analysis. 1 μl of a solution of reduced or non-reduced tryptic peptide mixture was applied on the MALDI plate followed by 1 μl of saturated matrix solution. Spectra were recorded after evaporation of the solvent and processed using Data Explorer software for data collection and analysis. Predicted masses were calculated by the ExPASy Peptide Mass program. Immunoblotting Blotting—After separation by SDS-PAGE, proteins were transferred to polyvinylidene difluoride membranes by electroblotting. Membranes were blocked in 5% (w/v) powdered milk in Tris-buffered saline. Blots were incubated with the primary antibody for 1 h followed by incubation in the secondary antibody for 1 h. Secondary antibodies were coupled to horseradish peroxidase or to IRDye800 (Pierce). For antibodies coupled to horseradish peroxidase, immunoreactivity was visualized by incubating the blot in SuperSignal West Pico Chemiluminescent substrate (Pierce) and exposure to x-ray film. For the fluorescently labeled secondary antibodies, immunoreactivity was detected with Odyssey infrared imaging system (Licor, Lincoln NE) according to the manufacturer’s instructions. After immunoreactivity detection, total protein was visualized by staining the blots with Amido Black (Sigma). Lipid Kinase Assays—PtdInsP kinase activity and phosphatidylinositol kinase activity was assayed in duplicate as described by (5Davis A.J. Perera I.Y. Boss W.F. J. Lipid Res. 2004; 45: 1783-1789Abstract Full Text Full Text PDF PubMed Scopus (12) Google Scholar) with a final reaction volume of 50 μl. Each assay contained 10 μg of purified recombinant protein on glutathione-Sepharose beads that had been incubated with solubilized Arabidopsis membrane fraction or with purified proteins and washed once with 50 mm Tris, pH 7.5. Lipid substrate was prepared using PtdIns4P or PtdIns (porcine brain; Avanti Polar Lipids, Alabaster, AL) from 1 or 5 mg/ml stocks, respectively. Lipids were dried under an N2 atmosphere and solubilized for use in the lipid kinase assays in the presence of cyclodextrins as described previously (5Davis A.J. Perera I.Y. Boss W.F. J. Lipid Res. 2004; 45: 1783-1789Abstract Full Text Full Text PDF PubMed Scopus (12) Google Scholar). The lipid kinase assay was performed as described (5Davis A.J. Perera I.Y. Boss W.F. J. Lipid Res. 2004; 45: 1783-1789Abstract Full Text Full Text PDF PubMed Scopus (12) Google Scholar). Lipid extraction was performed using an acid solvent system. Extracted lipids were separated by TLC on silica gel plates (LK5D; Whatman, Clifton, NJ) using a CHCl3:MeOH: NH4OH:water (90:90:7:22, v/v) solvent system. The 32P-labeled phospholipids were quantified with a Bioscan System 500 imaging scanner. Actin Polymerization—Polymerization buffer contained 20 mm PIPES, 2 mm EGTA, 2 mm MgCl2, 1 mm ATP, 50 mm KCl, pH 6.5 (28Carraway K.L. Carraway C.A.C. The Cytoskeleton: A Practical Approach. 1992; (Oxford University Press, New York): 47-71Google Scholar). Buffer was added to 30 μg of protein from the depolymerized actin fraction from Arabidopsis membranes, 5 μg of plasma membrane isolated from NT-1 cells as described previously (29Heilmann I. Perera I.Y. Gross W. Boss W.F. Plant Physiol. 1999; 229: 1331-1339Crossref Scopus (43) Google Scholar), or to 3 μg of pure actin containing purified recombinant proteins eluted from glutathione-Sepharose beads with reduced glutathione. Polymerization reactions were incubated for 1 h at 25 °C, then the F-actin was pelleted by centrifugation. For centrifugation at 20,000 × g for 1 h at 4 °C, polymerization volume was 30 μl. For centrifugation at 100,000 × g for 30 min, polymerization volume was 100 μl (28Carraway K.L. Carraway C.A.C. The Cytoskeleton: A Practical Approach. 1992; (Oxford University Press, New York): 47-71Google Scholar). Latrunculin treatments were performed by adding 10 μm latrunculin B to 2 gof 4-day-old NT-1 cells for 1 h (30Van Gestel K. Kohler R.H. Verbelen J.P. J. Exp. Bot. 2002; 53: 659-667Crossref PubMed Scopus (186) Google Scholar). An equal volume of Me2SO was used as a control. Cells were harvested, and the plasma membrane was isolated as previously described (29Heilmann I. Perera I.Y. Gross W. Boss W.F. Plant Physiol. 1999; 229: 1331-1339Crossref Scopus (43) Google Scholar). Insect Cell Protein Production—Serum-free Spodoptera frugiperda (Sf9) cells were obtained from Invitrogen and maintained at 28 °C at a concentration of 2.5 × 105 to 5 × 106 cells ml–1 in Sf-900 II insect cell serum-free medium (Invitrogen). The expression system used was the Bac-to-Bac baculovirus expression system (Invitrogen). Recombinant baculoviruses were generated from the recombinant expression vectors according to the manufacturer’s recommendations. Optimal production of AtPI4Kβ1 was 2 days after infection, and optimal production of AtPI4Kα1 (31Stevenson-Paulik J. Love J. Boss W.F. Plant Physiol. 2003; 132: 1053-1064Crossref PubMed Scopus (32) Google Scholar) and AtPI4Kγ7(ΔN/ΔC) was on the third day after infection. 3R. M. Galvąo and W. F. Boss, unpublished results. All assays were performed using cells optimally producing the respective polypeptide. Infected cells were then harvested and lysed as previously described (31Stevenson-Paulik J. Love J. Boss W.F. Plant Physiol. 2003; 132: 1053-1064Crossref PubMed Scopus (32) Google Scholar). The cleared lysate was analyzed for protein concentration by the Bradford method (Bio-Rad) and used for protein-protein interactions. Cyclodextrin Enhanced Recovery of Proteins in Pulldown Reactions—Our first goal was to identify proteins that interacted specifically with AtPIPK1. For this purpose, constructs were developed to produce GST fusion proteins of AtPIPK1 and AtPIPK10 in E. coli to pulldown-interacting proteins from Arabidopsis cell fractions. Fig. 1 shows the production of the recombinant GST-tagged proteins in E. coli as detected by a GST antibody (Fig. 1A) and with antibodies raised against AtPIPK1 or AtPIPK10 (Fig. 1B). Although the antibodies raised against full-length AtPIPK1 and AtPIPK10 readily detected the recombinant E. coli-expressed proteins, these antibodies were unable to detect PtdInsP kinases from an Arabidopsis cell fraction (data not shown), suggesting that the antibodies were not robust and/or that the protein levels were very low. For this reason, we used GST antibodies or activity assays to monitor the enzymes. Because PtdInsP 5-kinase activity was associated with F-actin as well as plasma membranes from plants (18Tan Z. Boss W.F. Plant Physiol. 1992; 100: 2116-2120Crossref PubMed Scopus (73) Google Scholar, 29Heilmann I. Perera I.Y. Gross W. Boss W.F. Plant Physiol. 1999; 229: 1331-1339Crossref Scopus (43) Google Scholar, 32Sommarin M. Sandelius A.S. Biochim. Biophys. Acta. 1988; 958: 268-278Crossref Scopus (81) Google Scholar), to increase the recovery of potential interacting proteins, cells were homogenized using a buffer developed to enhance the recovery of F-actin with the membranes (33Abe S. Ito Y. Davies E. J. Exp. Bot. 1992; 43: 941-949Crossref Scopus (36) Google Scholar). Recovery of interacting proteins was compared using a 40,000 × g pellet and the soluble fraction from 4-day-old Arabidopsis cells grown in suspension culture. As described under “Experimental Procedures,” all fractions isolated from the suspension culture cells were pre-cleared by incubation with purified GST immobilized on glutathione-Sepharose beads to reduce the number of GST binding proteins. No interacting proteins were evident based on Coomassie and silver staining when the 40,000 × g supernatant was used for the pulldown assays with either GST-AtPIPK1 or GST-AtPIPK10 (data not shown); however, several proteins were evident if the Triton X-100-solubilized membrane fraction was used. Although there were interacting proteins present for both isoforms from the solubilized membrane fraction, there was not enough protein present on the stained gels to identify the proteins by mass spectrometry. To enhance the recovery of the solubilized proteins, we developed a protocol that involved adding cyclodextrins to the pulldown reaction mixture. Cyclodextrins have been used to aid in protein refolding and in sequestering lipids (34Fauvelle F. Debouzy J.C. Crouzy S. Goschl M. Chapron Y. J. Pharm. Sci. 1997; 86: 935-943Abstract Full Text PDF PubMed Scopus (66) Google Scholar, 35Karuppiah N. Sharma A. Biochem. Biophys. Res. Commun. 1995; 211: 60-66Crossref PubMed Scopus (134) Google Scholar). α-, β-, and γ-cyclodextrins (0–50 mm) were added after pre-clearing the Triton-solubilized membrane proteins to see which if any would enhance the recovery of interacting proteins. γ-Cyclodextrin did not improve the interaction of the Arabidopsis proteins with the recombinant proteins at any of the concentrations tested, and α-cyclodextrin enhanced the interactions of a few of the proteins from the membranes with the recombinant proteins (data not shown); however, when 30 mm β-cyclodextrin was added, there was a 6-fold increase in the recovery of the peptides detected (Fig. 1C). One possible explanation for the enhanced protein recovery is that the cyclodextrin helps the recombinant protein fold properly to allow for interaction with the Arabidopsis peptides (35Karuppiah N. Sharma A. Biochem. Biophys. Res. Commun. 1995; 211: 60-66Crossref PubMed Scopus (134) Google Scholar). In addition, cyclodextrin will sequester the Triton or lipids present in the membrane fraction and thereby may enhance the interaction of the protein partners during the pulldown reaction. Because β-cyclodextrin enhanced recovery and did not appear to affect the distribution or number of the proteins bands, it was used for all subsequent pulldown experiments using microsomal proteins unless indicated otherwise. AtPIPK1 and AtPIPK10 Recovered Different Proteins Based on Mass Spectrometry Analysis—Using the optimized pull-down conditions for experiments, we were able to recover sufficient amounts of interacting peptides to identify them by mass spectrometry. To eliminate any false positives, we selected only for Coomassie-}, number={19}, journal={JOURNAL OF BIOLOGICAL CHEMISTRY}, author={Davis, Amanda J. and Im, Yang Ju and Dubin, Joshua S. and Tomer, Kenneth B. and Boss, Wendy F.}, year={2007}, month={May}, pages={14121–14131} }
 @article{im_perera_brglez_davis_stevenson-paulik_phillippy_johannes_allen_boss_2007, title={Increasing plasma membrane phosphatidylinositol(4,5)bisphosphate biosynthesis increases phosphoinositide metabolism in Nicotiana tabacum}, volume={19}, ISSN={["1532-298X"]}, DOI={10.1105/tpc.107.051367}, abstractNote={Abstract A genetic approach was used to increase phosphatidylinositol(4,5)bisphosphate [PtdIns(4,5)P2] biosynthesis and test the hypothesis that PtdInsP kinase (PIPK) is flux limiting in the plant phosphoinositide (PI) pathway. Expressing human PIPKIα in tobacco (Nicotiana tabacum) cells increased plasma membrane PtdIns(4,5)P2 100-fold. In vivo studies revealed that the rate of 32Pi incorporation into whole-cell PtdIns(4,5)P2 increased >12-fold, and the ratio of [3H]PtdInsP2 to [3H]PtdInsP increased 6-fold, but PtdInsP levels did not decrease, indicating that PtdInsP biosynthesis was not limiting. Both [3H]inositol trisphosphate and [3H]inositol hexakisphosphate increased 3-and 1.5-fold, respectively, in the transgenic lines after 18 h of labeling. The inositol(1,4,5)trisphosphate [Ins(1,4,5)P3] binding assay showed that total cellular Ins(1,4,5)P3/g fresh weight was >40-fold higher in transgenic tobacco lines; however, even with this high steady state level of Ins(1,4,5)P3, the pathway was not saturated. Stimulating transgenic cells with hyperosmotic stress led to another 2-fold increase, suggesting that the transgenic cells were in a constant state of PI stimulation. Furthermore, expressing Hs PIPKIα increased sugar use and oxygen uptake. Our results demonstrate that PIPK is flux limiting and that this high rate of PI metabolism increased the energy demands in these cells.}, number={5}, journal={PLANT CELL}, author={Im, Yang Ju and Perera, Imara Y. and Brglez, Irena and Davis, Amanda J. and Stevenson-Paulik, Jill and Phillippy, Brian Q. and Johannes, Eva and Allen, Nina S. and Boss, Wendy F.}, year={2007}, month={May}, pages={1603–1616} }
 @article{im_davis_perera_johannes_allen_boss_2007, title={The N-terminal membrane occupation and recognition nexus domain of Arabidopsis phosphatidylinositol phosphate kinase 1 regulates enzyme activity}, volume={282}, ISSN={["1083-351X"]}, DOI={10.1074/jbc.M611342200}, abstractNote={The type I B family of phosphatidylinositol phosphate kinases (PIPKs) contain a characteristic region of Membrane Occupation and Recognition Nexus (MORN) motifs at the N terminus. These MORN motifs are not found in PIPKs from other eukaryotes. To understand the impact of the additional N-terminal domain on protein function and subcellular distribution, we expressed truncated and full-length versions of AtPIPK1, one member of this family of PIPKs, in Escherichia coli and in tobacco cells grown in suspension culture. Deletion of the N-terminal MORN domain (amino acids 1–251) of AtPIPK1 increased the specific activity of the remaining C-terminal peptide (ΔMORN) >4-fold and eliminated activation by phosphatidic acid (PtdOH). PtdOH activation could also be eliminated by mutating Pro396 to Ala (P396A) in the predicted linker region between the MORN and the kinase homology domains. AtPIPK1 is product-activated and the MORN domain binds PtdIns(4,5)P2. Adding back the MORN peptide to ΔMORN or to the PtdOH-activated full-length protein increased activity ∼2-fold. Furthermore, expressing the MORN domain in vivo increased the plasma membrane PtdInsP kinase activity. When cells were exposed to hyperosmotic stress, the MORN peptide redistributed from the plasma membrane to a lower phase or endomembrane fraction. In addition, endogenous PtdInsP kinase activity increased in the endomembrane fraction of hyperosmotically stressed cells. We conclude that the MORN peptide can regulate both the function and distribution of the enzyme in a manner that is sensitive to the lipid environment.}, number={8}, journal={JOURNAL OF BIOLOGICAL CHEMISTRY}, author={Im, Yang Ju and Davis, Amanda J. and Perera, Imara Y. and Johannes, Eva and Allen, Nina S. and Boss, Wendy F.}, year={2007}, month={Feb}, pages={5443–5452} }
 @article{perera_davis_galanopilou_im_boss_2005, title={Characterization and comparative analysis of Arabidopsis phosphatidylinositol phosphate 5-kinase 10 reveals differences in Arabidopsis and human phosphatidylinositol phosphate kinases}, volume={579}, ISSN={["1873-3468"]}, DOI={10.1016/j.febslet.2005.05.018}, abstractNote={Arabidopsis phosphatidylinositol phosphate (PtdInsP) kinase 10 (AtPIPK10; At4g01190) is shown to be a functional enzyme of the subfamily A, type I AtPtdInsP kinases. It is biochemically distinct from AtPIPK1 (At1g21980), the only other previously characterized AtPtdInsP kinase which is of the B subfamily. AtPIPK10 has the same K(m), but a 10-fold lower V(max) than AtPIPK1 and it is insensitive to phosphatidic acid. AtPIPK10 transcript is most abundant in inflorescence stalks and flowers, whereas AtPIPK1 transcript is present in all tissues. Comparative analysis of recombinant AtPIPK10 and AtPIPK1 with recombinant HsPIPKIalpha reveals that the Arabidopsis enzymes have roughly 200- and 20-fold lower V(max)/K(m), respectively. These data reveal one explanation for the longstanding mystery of the relatively low phosphatidylinositol-(4,5)-bisphosphate:phosphatidylinositol-4-phosphate ratio in terrestrial plants.}, number={16}, journal={FEBS LETTERS}, author={Perera, IY and Davis, AJ and Galanopilou, D and Im, YJ and Boss, WF}, year={2005}, month={Jun}, pages={3427–3432} }
 @article{davis_perera_boss_2004, title={Cyclodextrins enhance recombinant phosphatidylinositol phosphate kinase activity}, volume={45}, ISSN={["1539-7262"]}, DOI={10.1194/jlr.D400005-JLR200}, abstractNote={Inositol lipid kinases have been studied extensively in both plant and animal systems. However, major limitations for in vitro studies of recombinant lipid kinases are the low specific activity and instability of the purified proteins. Our goal was to determine if cyclodextrins would provide an effective substrate delivery system and enhance the specific activity of lipid kinases. For these studies, we have used recombinant Arabidopsis thaliana phosphatidylinositol phosphate kinase 1 (At PIPK1). At PIPK1 was produced as a fusion protein with glutathione-S-transferase and purified on glutathione-Sepharose beads. A comparison of lipid kinase activity using substrate prepared in α-, β-, or γ-cyclodextrin indicated that β-cyclodextrin was most effective and enhanced lipid kinase activity 6-fold compared with substrate prepared in Triton X-100-mixed micelles.We have optimized reaction conditions and shown that product can be recovered from the cyclodextrin-treated recombinant protein, which reveals a potential method for automating the assay for pharmacological screening. Inositol lipid kinases have been studied extensively in both plant and animal systems. However, major limitations for in vitro studies of recombinant lipid kinases are the low specific activity and instability of the purified proteins. Our goal was to determine if cyclodextrins would provide an effective substrate delivery system and enhance the specific activity of lipid kinases. For these studies, we have used recombinant Arabidopsis thaliana phosphatidylinositol phosphate kinase 1 (At PIPK1). At PIPK1 was produced as a fusion protein with glutathione-S-transferase and purified on glutathione-Sepharose beads. A comparison of lipid kinase activity using substrate prepared in α-, β-, or γ-cyclodextrin indicated that β-cyclodextrin was most effective and enhanced lipid kinase activity 6-fold compared with substrate prepared in Triton X-100-mixed micelles. We have optimized reaction conditions and shown that product can be recovered from the cyclodextrin-treated recombinant protein, which reveals a potential method for automating the assay for pharmacological screening. Inositol lipid kinases function at the interface of the lipid bilayer and selectively phosphorylate the head group of inositol phospholipids (1Hurley J.H. Tsujishita Y. Pearson M.A. Floundering about at cell membranes: a structural view of phospholipid signaling.Curr. Opin. Struct. Biol. 2000; 10: 737-743Google Scholar). One of the limitations of in vitro lipid kinase assays is that the recombinant lipid kinases are often unstable and exhibit low activity when presented with lipid substrate as sonicated or Triton-mixed micelles. Our goal was to determine if cyclodextrins could increase recombinant lipid kinase activity by more effectively delivering lipid substrate. We used recombinant Arabidopsis thaliana phosphatidylinositol phosphate kinase 1 (AtPIPK1) (2Mueller-Roeber B. Pical C. Inositol phospholipid metabolism in Arabidopsis. Characterized and putative isoforms of inositol phospholipid kinase and phosphoinositide-specific phospholipase C.Plant Physiol. 2002; 130: 22-46Google Scholar) fused to glutathione-S-transferase (GST) to investigate the effects of using cyclodextrins to deliver lipid substrate to the recombinant lipid kinase. GST-AtPIPK1 phosphorylates phosphatidylinositol-4-phosphate (PtdIns4P) to form phosphatidylinositol-(4,5)-bisphosphate PtdIns(4,5)P2 (3Mikami K. Katagiri T. Iuchi S. Yamaguchi-Shinozaki K. Shinozaki K. A gene encoding phosphatidylinositol-4-phosphate 5-kinase is induced by water stress and abscisic acid in Arabidopsis thaliana.Plant J. 1998; 15: 563-568Google Scholar, 4Westergren T. Dove S.K. Sommarin M. Pical C. AtPIP5K1, an Arabidopsis thaliana phosphatidylinositol phosphate kinase, synthesizes PtdIns(3,4)P2 and PtdIns(4,5)P2 in vitro and is inhibited by phosphorylation.Biochem. J. 2001; 359: 583-589Google Scholar). Cyclodextrins are cyclic oligomers of α-d-glucopyranose that are produced naturally in bacteria. Their ring structures form a cone shape that has a hydrophilic outer surface and a hydrophobic inner core. There are three naturally occurring cyclodextrins: α-cyclodextrin (αCD) contains six glucopyranose units; β-cyclodextrin (βCD) contains seven glucopyranose units; and γ-cyclodextrin (γCD) contains eight glucopyranose units. In addition, naturally occurring cyclodextrins have been modified with various substitutions on the glucopyranose subunits to increase their efficacy in specific industrial and scientific applications (5Singh M. Sharma R. Banerjee U.C. Biotechnological applications of cyclodextrins.Biotechnol. Adv. 2002; 20: 341-359Google Scholar). Industrial applications of cyclodextrins include use in pharmaceuticals to enhance drug stability and delivery and in food additives to preserve flavors and enhance shelf-life (5Singh M. Sharma R. Banerjee U.C. Biotechnological applications of cyclodextrins.Biotechnol. Adv. 2002; 20: 341-359Google Scholar). Recent studies in polymer sciences have used cyclodextrins to facilitate the formation of polymers and enhance the intercalation of small molecules into the polymer matrices for potential drug delivery (6Wei M. Shuai X.T. Tonelli A.E. Melting and crystallization behaviors of biodegradable polymers enzymatically coalesced from their cyclodextrin inclusion complexes.Biomacromolecules. 2003; 4: 783-792Google Scholar). Cyclodextrins are also used in the cosmetics industry to create longer lasting fragrances and prevent the oxidation and degradation of oils (5Singh M. Sharma R. Banerjee U.C. Biotechnological applications of cyclodextrins.Biotechnol. Adv. 2002; 20: 341-359Google Scholar). Laboratory applications include using cyclodextrins as size-exclusion columns, as artificial chaperones to remove detergents and facilitate the refolding of recombinant proteins (7Nomura Y. Ikeda M. Yamaguchi N. Aoyama Y. Akiyoshi K. Protein refolding assisted by self-assembled nanogels as novel artificial molecular chaperone.FEBS Lett. 2003; 553: 271-276Google Scholar, 8Rozema D. Gellman S.H. Artificial chaperone-assisted refolding of carbonic anhydrase B.J. Biol. Chem. 1996; 271: 3478-3487Google Scholar), as a vehicle to develop molecular machines (9Harada A. Cyclodextrin-based molecular machines.Acc. Chem. Res. 2001; 34: 456-464Google Scholar), and as a means for the delivery and removal of lipids from membranes to study bilayers and lipid rafts (10Leventis R. Silvius J.R. Use of cyclodextrins to monitor transbilayer movement and differential lipid affinities of cholesterol.Biophys. J. 2001; 81: 2257-2267Google Scholar, 11Peres C. Yart A. Perret B. Salles J.P. Raynal P. Modulation of phosphoinositide 3-kinase activation by cholesterol level suggests a novel positive role for lipid rafts in lysophosphatidic acid signalling.FEBS Lett. 2003; 534: 164-168Google Scholar, 12Pike L.J. Miller J.M. Cholesterol depletion delocalizes phosphatidylinositol bisphosphate and inhibits hormone-stimulated phosphatidylinositol turnover.J. Biol. Chem. 1998; 273: 22298-22304Google Scholar, 13Tanhuanpaa K. Somerharju P. γ-Cyclodextrins greatly enhance translocation of hydrophobic fluorescent phospholipids from vesicles to cells in culture: importance of molecular hydrophobicity in phospholipid trafficking studies.J. Biol. Chem. 1999; 274: 35359-35366Google Scholar). We have taken advantage of the ability of cyclodextrins to bind lipids (11Peres C. Yart A. Perret B. Salles J.P. Raynal P. Modulation of phosphoinositide 3-kinase activation by cholesterol level suggests a novel positive role for lipid rafts in lysophosphatidic acid signalling.FEBS Lett. 2003; 534: 164-168Google Scholar, 12Pike L.J. Miller J.M. Cholesterol depletion delocalizes phosphatidylinositol bisphosphate and inhibits hormone-stimulated phosphatidylinositol turnover.J. Biol. Chem. 1998; 273: 22298-22304Google Scholar, 14Fauvelle F. Debouzy J.C. Crouzy S. Goschl M. Chapron Y. Mechanism of alpha-cyclodextrin-induced hemolysis. 1. The two-step extraction of phosphatidylinositol from the membrane.J. Pharm. Sci. 1997; 86: 935-943Google Scholar) and deliver them to cells (13Tanhuanpaa K. Somerharju P. γ-Cyclodextrins greatly enhance translocation of hydrophobic fluorescent phospholipids from vesicles to cells in culture: importance of molecular hydrophobicity in phospholipid trafficking studies.J. Biol. Chem. 1999; 274: 35359-35366Google Scholar) and asked whether cyclodextrin could be used to deliver inositol phospholipids to recombinant lipid kinases for in vitro lipid kinase assays. We found that using cyclodextrin for substrate delivery increased AtPIPK1 specific activity 4- to 6-fold compared with sonicated substrate alone or Triton-mixed micelles, respectively. In addition, when cyclodextrin was used for substrate delivery, the product PtdIns(4,5)P2 could be recovered with the recombinant GST-AtPIPK1 beads. Long, arduous lipid extractions in organic solvents are an additional challenge when performing lipid kinase assays. The data presented here provide a basis for developing an environmentally friendly method that does not require organic solvents for the recovery of phosphorylated lipid products and uses a procedure that would be readily applicable for large-scale screening of kinase inhibitors. The cDNA of At PIPK1 (At1g21980) was amplified by PCR using the sense primer AAACCCATGGGAATGAGTGATTCAGAAGAAG and the antisense primer GTTAAAAACTCGAGCCTTCTTGTCTTTAGCC to create NcoI and XhoI restriction sites, respectively, for directional cloning into pET-41a vector (Novagen, Madison, WI). The PCR product was digested and ligated into the pET-41a vector that had been digested with NcoI and XhoI. The sequence of the resulting construct, pET-41a5K1, was confirmed by DNA sequencing. The Escherichia coli expression strain BL21(DE3)pLysS (Novagen) was transformed with pET-41a5K1 and used to express the fusion protein GST-AtPIPK1. For recombinant protein expression, an overnight culture of BL21(DE3)pLysS carrying pET-41a5K1 was diluted 1:500 with fresh Lennox L broth (Invitrogen, Carlsbad, CA) medium and grown at 37°C with shaking until an OD600 of 0.3 was reached. At this point, isopropyl-β-d-thiogalactoside was added to a final concentration of 1 mM, and incubation continued at 25°C for 4 h with shaking. After 4 h, the cells were collected by centrifugation and the bacterial pellets were frozen at −20°C until the recombinant protein was to be used. Bacterial pellets were thawed and resuspended in ice-cold PBS buffer (0.1 M KH2PO4, 0.1 M K2HPO4, 135 mM NaCl, and 2.7 mM KCl, pH 7.3) and sonicated on ice for 20 s. The sonicated solution was centrifuged at 12,000 g for 10 min. The supernatant was removed and combined with glutathione-Sepharose beads (Amersham Pharmacia Biotech, Pitscataway, NJ) or with magnetic glutathione-agarose beads (Novagen) preequilibrated with PBS. The mixture was incubated at 4°C for 2 h with continuous mixing followed by extensive washing of the beads with PBS. Protein concentration was determined using the Bio-Rad protein assay reagent (Bio-Rad, Hercules, CA) with BSA as a standard. Purified recombinant proteins bound to the glutathione-Sepharose beads or the magnetic glutathione-agarose beads were stored at 4°C until use in lipid kinase assays. The purified enzyme was not stored longer than 12 h before use. The storage time of the purified lipid kinase was correlated with a decrease in the specific activity of the enzyme, as seen when comparing Figs. 2 and 3. In Fig. 2, the purified lipid kinase was used immediately, and in Fig. 3, the purified enzyme was stored for 12 h.Fig. 3Purified recombinant GST-AtPIPK1 (10 μg) was assayed using increasing concentrations of PtdIns4P and a constant concentration of 5 mM βCD (black squares) or 30 mM βCD (black triangles). The results are reported as averages of two independent experiments.View Large Image Figure ViewerDownload (PPT) Microsomes were isolated from A. thaliana cells grown in suspension culture. Cells were harvested at 4 days after subculture by gravity filtration and immediately homogenized in an equal volume of ice-cold buffer containing 5 mM HEPES, pH 7, 10 mM MgCl2, 2 mM EGTA, 8% (w/v) sucrose, 1 mM PMSF, 1 mg/100 ml leupeptin, and polyvinylpolypyrrolidone (0.1 g/g cells). The homogenate was centrifuged twice at 2,000 g for 6 min at 4°C. The resulting supernatant was centrifuged at 40,000 g for 60 min at 4°C to obtain the microsomal fraction. The microsomes were washed in 50 mM Tris, pH 7.5, and centrifuged at 40,000 g for 30 min at 4°C, and the final pellet was resuspended in 50 mM Tris, pH 7.5. Microsomes were placed on ice and used immediately for lipid kinase assays. Protein concentration was determined using the Bio-Rad protein assay reagent with BSA as a standard. Phosphatidylinositol phosphate (PtdInsP) kinase activity was assayed in duplicate as described by Cho and Boss (15Cho M.H. Boss W.F. Transmembrane signaling and phosphoinositides.Methods Cell Biol. 1995; 49: 543-554Google Scholar) with a final reaction volume of 50 μl. Each assay contained either 30 μg of microsomal protein or 10 μg of purified recombinant protein on glutathione-Sepharose beads washed once with 50 mM Tris, pH 7.5. Lipid substrate was prepared using PtdIns4P (porcine brain; Avanti Polar Lipids, Alabaster, AL), PtdIns3P (Matreya, Inc., Pleasant Gap, PA), or PtdIns5P (Echelon Biosciences, Inc., Salt Lake City, UT) from 1 mg/ml stocks. Lipids were divided into aliquots and dried under an N2 stream to form a thin, even film in the bottom of the test tube. Dried lipid films were solubilized for use in the lipid kinase assays in the presence and absence of cyclodextrins. In the absence of cyclodextrins, lipids were sonicated for 10 s in 50 mM Tris, pH 7.5, or in a solution of Triton X-100 resulting in a final concentration of 0.1% Triton X-100 (v/v) in the final reaction volume and then incubated on ice for 10 min. Triton (0.1%) was determined to give optimal enzyme activity (D. Galanopolou, I. Y. Perera, and W. F. Boss, unpublished results). Lipids were also solubilized by vortexing in the presence of deoxycholate to give a final concentration of 1% in the final reaction volume, as described by Westergren et al. (4Westergren T. Dove S.K. Sommarin M. Pical C. AtPIP5K1, an Arabidopsis thaliana phosphatidylinositol phosphate kinase, synthesizes PtdIns(3,4)P2 and PtdIns(4,5)P2 in vitro and is inhibited by phosphorylation.Biochem. J. 2001; 359: 583-589Google Scholar). Cyclodextrin solubilization was accomplished by adding αCD, βCD, or γCD (all from Sigma) from a 150 mM (saturated) stock solution to achieve the desired concentration in the 50 μl reaction volume. The final concentrations of cyclodextrin solutions produced from the stock solution were confirmed by comparison with the refractive indices of cyclodextrin solutions of known concentrations. The cyclodextrin solution was added to the dried lipid film, vortexed for 5 s, and incubated on ice for 10 min before use. The lipid concentration for each lipid kinase assay was 125 μM, except where noted. The reaction mixture contained final concentrations of 50 mM Tris, pH 7.5, 10 mM MgCl2, 1 mM EGTA, 1 mM Na2MoO4, and 50 μM ATP (9 μCi of [32P]ATP per reaction). In experiments performed at varying pH values, all proteins, lipids, and reaction mixtures were prepared in 50 mM Tris at the appropriate pH. Reactions were incubated at room temperature for 10 min with shaking, stopped by adding 1.5 ml of CHCl3/methanol (1:2, v/v), and stored at 4°C until the lipids were extracted. Lipid extraction was performed using an acid solvent system as described previously (15Cho M.H. Boss W.F. Transmembrane signaling and phosphoinositides.Methods Cell Biol. 1995; 49: 543-554Google Scholar). Extracted lipids were separated by TLC on silica gel plates (LK5D; Whatman, Clifton, NJ) using a CHCl3/methanol/NH4OH/water (90:90:7:22, v/v) solvent system. The 32P-labeled phospholipids were quantified with a Bioscan System 500 imaging scanner. Fluorescence spectroscopy was used to monitor lipid distribution during substrate preparation and during the lipid kinase assays. For the substrate preparation, 6.25 μg of d(+)-sn-1-O-[1-[6′-[6-[(7-nitrobenz-2-oxa-1,3-diazol-4-yl)amino]hexanoyl]amino]hexanoyl]2-O-hexadecanoylglyceryl d-myo-phosphatidylinositol 4-phosphate (NBD-PtdIns4P; Echelon Biosciences, Inc.) was divided into aliquots for each reaction and dried under an N2 stream to form a thin, even film in the bottom of the test tube. Lipids were solubilized as described above either by sonication in 50 mM Tris, pH 7.5, or in a Triton X-100 solution or by incubation with βCD and incubated on ice for 10 min. Triton X-100 and βCD were added to correspond to 0.1% Triton X-100 or 5 mM βCD in the 50 μl volume of a lipid kinase assay. The supernatant was removed. The supernatant and the residual, nonsolubilized lipid adhering to the test tube were extracted as described above. The extracts were dried under vacuum and reconstituted in 1 ml of chloroform. All samples were analyzed in a Hitachi F-2000 fluorescence spectrophotometer with an excitation wavelength of 470 nm and an emission wavelength of 530 nm. The relative amount of lipid solubilized with each method was calculated by comparing the fluorescence recovered in the supernatant with the total fluorescence recovered (supernatant plus residue). Recovery of lipid from GST-AtPIPK1 beads treated with cyclodextrin was also monitored by fluorescence spectroscopy. NBD-PtdIns4P was divided into aliquots and dried under an N2 stream for each experiment. Lipid was prepared by adding 50 mM Tris, pH 7.5, and sonicating or by vortexing in 5 mM βCD to yield a final concentration of 6.25 μg of NBD-PtdIns4P per 10 μl of solution. For each experiment, 5 μl magnetic glutathione-Sepharose beads, 2 μg of purified GST on magnetic glutathione-Sepharose beads, or 2 μg of purified GST-AtPIPK1 on magnetic glutathione-Sepharose beads was incubated with 1.25 μg of the prepared lipid, either in Tris buffer or in βCD. To assay for lipid kinase activity, purified GST-AtPIPK1 and prepared lipid were mixed, ATP was added to a final concentration of 0.5 mM, and the reaction was incubated with mixing for 1 h. To stop the reaction, 2 ml of 50 mM Tris, pH 7.5, was added. The beads were retained with a magnet, and the solution was removed and discarded. This washing procedure was repeated once. After the final wash, the fluorescence was monitored microscopically or the lipids were extracted as above. After the extraction, lipids were reconstituted in CHCl3:methanol:water (2:1:0.01, v/v) and spotted on a TLC plate. The plate was developed in the same solvent system as described above. After the TLC plate was dry, the regions where PtdIns4P and PtdIns(4,5)P2 migrated were scraped and the lipids were eluted from the silica gel with two washes of 500 μl of CHCl3:methanol:NH4OH:water (90:90:7:22, v/v). The eluted lipids were analyzed in a Hitachi F-2000 fluorescence spectrophotometer with an excitation wavelength of 470 nm and an emission wavelength of 530 nm. To determine whether adding PtdIns4P in the presence of cyclodextrins increased enzyme activity, we compared the specific activity of the purified recombinant GST-AtPIPK1 in the presence and absence of cyclodextrins. αCD, βCD, or γCD was added to the lipid substrate as described in Methods to achieve a concentration of 0–30 mM cyclodextrin in the final reaction mixture. The specific activity of the lipid kinase was compared with that obtained using PtdIns4P solubilized in Triton X-100 or by sonication. The results of this experiment (Fig. 1)indicate that under identical reaction conditions, the PtdInsP kinase activity was approximately four to six times greater when the substrate was delivered in a solution of βCD compared with sonication or Triton, respectively. The optimal cyclodextrin concentration was 5 mM βCD. Although 5 mM αCD or γCD also enhanced enzyme activity compared with Triton or sonication, neither of these cyclodextrins enhanced activity to the extent of βCD, and at higher concentrations they decreased enzyme activity. Because βCD gave the highest enzyme activity and because it is the most cost-effective delivery system, we focused on using βCD to optimize conditions for enzyme activity. AtPIPK1 had previously been shown to have a much greater preference for PtdIns4P than for PtdIns3P or PtdIns5P as the substrate (4Westergren T. Dove S.K. Sommarin M. Pical C. AtPIP5K1, an Arabidopsis thaliana phosphatidylinositol phosphate kinase, synthesizes PtdIns(3,4)P2 and PtdIns(4,5)P2 in vitro and is inhibited by phosphorylation.Biochem. J. 2001; 359: 583-589Google Scholar). To determine if the addition of βCD to the lipid kinase assays altered the substrate specificity of AtPIPK1, the activity of the recombinant enzyme was compared using PtdIns4P, PtdIns3P, and PtdIns5P prepared in 5 mM βCD. The specific activity of the lipid kinase with PtdIns4P was 22 times greater than that of the enzyme with PtdIns3P (data not shown). The low level of activity when PtdIns3P was used as the substrate confirmed previous results (4Westergren T. Dove S.K. Sommarin M. Pical C. AtPIP5K1, an Arabidopsis thaliana phosphatidylinositol phosphate kinase, synthesizes PtdIns(3,4)P2 and PtdIns(4,5)P2 in vitro and is inhibited by phosphorylation.Biochem. J. 2001; 359: 583-589Google Scholar). There was no significant difference in lipid kinase specific activity when PtdIns3P was prepared by sonication or in 5 mM βCD, and there was no detectable activity when PtdIns5P was used as the substrate with either method under our assay conditions (data not shown). These results indicate that PtdIns4P is the preferred substrate for AtPIPK1 and that βCD does not affect the substrate specificity of the enzyme. To determine the optimal concentration of substrate, we altered the concentration of substrate, keeping the ratio of substrate to cyclodextrin constant. At 5 mM βCD, the PtdIns4P-to-βCD ratio was 1:40 (w/w); therefore, for each concentration of PtdIns4P, the molar ratio of PtdIns4P to βCD was kept at 1:40 (w/w). The lipid kinase activity increased sharply from 0 to 125 μM PtdIns4P but did not increase further at 250 μM PtdIns4P (Fig. 2). By keeping the concentration of βCD constant at 5 mM and changing the concentration of lipid, we were able to determine that the optimal PtdIns4P concentration is between 125 and 250 μM (Fig. 3). The Km and Vmax values for PtdIns4P were calculated using two different concentrations of βCD (Fig. 3), Triton X-100, and deoxycholate (data not shown). The calculations indicate a Km of 69 μM and a Vmax of 600 pmol PtdIns(4,5)P2/mg·min at 5 mM βCD. At 30 mM βCD, the Km did not change significantly but the Vmax was decreased to 340 pmol PtdIns(4,5)P2/mg·min. The solubilization of PtdIns4P in the detergents Triton X-100 and deoxycholate did not significantly change the Km from that of 5 mM βCD, but the Vmax was decreased to 100 and 79 pmol PtdIns(4,5)P2/mg·min, respectively (data not shown). These results are consistent with the idea presented by Harper, Easton, and Lincoln (16Harper J.B. Easton C.J. Lincoln S.F. Cyclodextrins to increase the utility of enzymes in organic synthesis.Curr. Org. Chem. 2000; 4: 429-454Google Scholar) that cyclodextrins can be used as a reservoir of substrate and to facilitate substrate delivery for enzymes. However, at 30 mM βCD, cyclodextrins may be complexing with the inositol head group and forming aggregates (14Fauvelle F. Debouzy J.C. Crouzy S. Goschl M. Chapron Y. Mechanism of alpha-cyclodextrin-induced hemolysis. 1. The two-step extraction of phosphatidylinositol from the membrane.J. Pharm. Sci. 1997; 86: 935-943Google Scholar). If aggregates formed at the higher concentrations and made the head group less accessible for modification, this would contribute to the reduction in Vmax of GST-AtPIPK1 with 30 mM βCD. Because a lipid-to-cyclodextrin molar ratio of 1:40 (w/w) and a substrate concentration of 125 μM were optimal for AtPIPK1, 5 mM βCD and 125 μM PtdIns4P were used in subsequent experiments. To determine if the optimum pH was altered with the addition of cyclodextrins, lipid kinase assays were performed at pH 6.5, 7.0, 7.5, and 8.0 with lipid substrate prepared in 5 mM βCD or sonicated. When cyclodextrin was used for substrate delivery, the specific activity increased up to pH 7.5 and then decreased at higher pH (Fig. 4). The decrease in specific activity at higher pH may reflect increased aggregation of βCD-PtdIns4P complexes. The decrease in specific activity was not observed using sonicated PtdIns4P. A fluorescent PtdIns4P analog, NBD-PtdIns4P, was used to monitor the relative efficiency of substrate solubilization. Lipids solubilized by sonication in buffer, with Triton X-100, or with βCD were removed after a 10-min incubation on ice. The lipids in solution and the lipids that remained adhering to the glass were extracted and quantified. The results (Table 1) indicate that each method solubilized similar amounts of lipid. Therefore, the increase in enzyme activity was not attributable to an increase in solubilized substrate with cyclodextrin but rather to the presentation of the substrate to the enzyme.TABLE 1Comparison of lipid solubilization methods using NBD-PtdIns4PMethodTotal LipidSolubilized LipidPercent SolubilizedSonication2,868 ± 46.32,248 ± 65.178.4 ± 1.2Sonication with Triton X-1003,681 ± 48.23,009 ± 32.381.7 ± 1.3Incubation with β-CD3,240 ± 38.82,590 ± 28.179.9 ± 0.8βCD, β-cyclodextrin. Solubilization of d(+)-sn-1-O-[1-[6′-[6-[(7-nitrobenz-2-oxa-1,3-diazol-4-yl)amino]hexanoyl]amino]-hexanoyl]-2-O-hexadecanoylglyceryl d-myo-phosphatidylinositol-4-phosphate (NBD-PtdIns4P) was quantified by extracting the solubilized and nonsolubilized lipids from each method of lipid preparation. Extracted lipids were analyzed by measuring the fluorescence in a spectrofluorometer with an excitation wavelength of 470 nm and an emission wavelength of 530 nm. The values shown are percentages of the fluorescent lipid solubilized relative to the total amount of fluorescent lipid recovered from the supernatant (solubilized) plus the residual (nonsolubilized) lipid adhering to the tube. Plus-minus values indicate the SD of four values from two independent experiments. Open table in a new tab βCD, β-cyclodextrin. Solubilization of d(+)-sn-1-O-[1-[6′-[6-[(7-nitrobenz-2-oxa-1,3-diazol-4-yl)amino]hexanoyl]amino]-hexanoyl]-2-O-hexadecanoylglyceryl d-myo-phosphatidylinositol-4-phosphate (NBD-PtdIns4P) was quantified by extracting the solubilized and nonsolubilized lipids from each method of lipid preparation. Extracted lipids were analyzed by measuring the fluorescence in a spectrofluorometer with an excitation wavelength of 470 nm and an emission wavelength of 530 nm. The values shown are percentages of the fluorescent lipid solubilized relative to the total amount of fluorescent lipid recovered from the supernatant (solubilized) plus the residual (nonsolubilized) lipid adhering to the tube. Plus-minus values indicate the SD of four values from two independent experiments. In the presence of cyclodextrin, fluorescent lipids bound to GST-AtPIPK1 could be readily detected using fluorescence microscopy (data not shown). The recovery of fluorescent lipids from glutathione-Sepharose beads was quantified after TLC separation. NBD-PtdIns4P binds more effectively to the purified recombinant protein when added with βCD (Table 2). Four percent of the NBD-PtdIns4P added was recovered from the AtPIPK1 beads after an aqueous wash. There was no detectable lipid binding to glutathione-Sepharose beads and very little binding to purified GST. Because the data suggested that cyclodextrins delivered substrate more effectively to the lipid kinase, we investigated whether the product of the reaction, PtdIns(4,5)P2, also might be trapped in the cyclodextrin. When ATP was added to the mixture of βCD plus NBD-PtdIns4P and GST-AtPIPK1, NBD-PtdIns(4,5)P2 was recovered even after washing the beads with buffer (Table 2). The fact that product could be recovered from the beads suggested that cyclodextrins could be used in a method for high-throughput analysis of lipid kinases that would not require extensive organic extractions.TABLE 2Cyclodextrin enhances the recovery of NBD-PtdIns4P and NBD-PtdIns(4,5)P2 from GST-AtPIPK1 beadsSampleRecovered FluorescenceNBD-PtdIns4PNBD-PtdIns(4,5)P2Glutathione-Sepharose beads 11Glutathione-Sepharose beads + βCD 22Purified GST 31Purified GST + βCD 41Purified GST-AtPIPK1 161Purified GST-AtPIPK1 + βCD 782Purified GST-AtPIPK1 + ATP 1212Purified GST-AtPIPK1 + βCD + ATP 7347Fluorescence was quantified by extracting the lipids from beads after a 1 h incubation with the fluorescent lipid preparations. Extracted lipids were separated by TLC. The fluorescent lipids were quantified by eluting the lipids from the silica matrix in the regions where PtdIns4P and phosphatidylinositol-(4,5)-bisphosphate PtdIns(4,5)P2 were present and measuring the fluorescence in a spectrofluorometer with an excitation wavelength of 470 nm and an emission wavelength of 530 nm. Four percent of the NBD-PtdIns4P added was recovered from glutathione-S-transferase-Arabidopsis thaliana phosphatidylinositol phosphate kinase 1 (GST-At PIPK1) with βCD. Open table in a new tab Fluorescence was quantified by extracting the lipids from beads after a 1 h incubation with the fluorescent lipid preparations. Extracted lipids were separated by TLC. The fluorescent lipids were quantified by eluting the lipids from the silica matrix in the regions where PtdIns4P and phosphatidylinositol-(4,5)-bisphosphate PtdIns(4,5)P2 were present and measuring the fluorescence in a spectrofluorometer with an excitation wavelength of 470 nm and an emission wavelength of 530 nm. Four percent of the NBD-PtdIns4P added was recovered from glutathione-S-transferase-Arabidopsis thaliana phosphatidylinositol phosphate kinase 1 (GST-At PIPK1) with βCD. After characterizing the effects of cyclodextrins on the activity of purified recombinant PtdInsP kinase, we examined the effect of cyclodextrins on the membrane-associated lipid kinase activity found in A. thaliana membrane fractions. A. thaliana membranes have endogenous PtdInsP kinases as well as PtdIns4P; therefore, PtdInsP kinase activity was assayed in the presence and absence of exogenous substrate. The results (Fig. 5)show that even when excess substrate was added, cyclodextrin decreased the specific activity of the membrane-associated PtdInsP kinase relative to adding substrate as Triton micelles or sonicated vesicles. αCD had the least effect on the specific activity, and γCD had the greatest effect. The decrease in specific activity when cyclodextrin is added to the membranes is the opposite of the effect when the purified recombinant protein is used, suggesting that cyclodextrin either preferentially binds to other lipids so that the cyclodextrin-PtdIns4P concentration is less than anticipated (i.e., there is not as much substrate available) or that the cyclodextrin removes a factor from the membranes that normally enhances PtdInsP kinase activity. Cyclodextrins are often added to membranes or cells to remove cholesterol (17Pike L.J. Lipid rafts: bringing order to chaos.J. Lipid Res. 2003; 44: 655-667Google Scholar); however, they will also bind to other lipids and can effectively remove phospholipids from membranes (13Tanhuanpaa K. Somerharju P. γ-Cyclodextrins greatly enhance translocation of hydrophobic fluorescent phospholipids from vesicles to cells in culture: importance of molecular hydrophobicity in phospholipid trafficking studies.J. Biol. Chem. 1999; 274: 35359-35366Google Scholar). The results of this study indicate that cyclodextrins are useful tools for delivering the negatively charged phospholipid, PtdIns4P, to the recombinant PtdInsP kinase. βCD was the most effective cyclodextrin for delivering PtdIns4P to purified, recombinant GST-AtPIPK1. Using βCD did not alter the substrate specificity of AtPIPK1 and did not alter the Km compared with assays using Triton X-100 or deoxycholate. We hypothesize that the cyclodextrin takes up the PtdInsP substrate and facilitates its delivery. Based on work by Fauvelle et al. (14Fauvelle F. Debouzy J.C. Crouzy S. Goschl M. Chapron Y. Mechanism of alpha-cyclodextrin-induced hemolysis. 1. The two-step extraction of phosphatidylinositol from the membrane.J. Pharm. Sci. 1997; 86: 935-943Google Scholar) and Anderson et al. (18Anderson T.G. Tan A. Ganz P. Seelig J. Calorimetric measurement of phospholipid interaction with methyl-β-cyclodextrin.Biochemistry. 2004; 43: 2251-2261Google Scholar), we propose that the fatty acid from PtdIns4P enters the cyclodextrin core and allows the enzyme access to the head group for phosphorylation. The affinity of cyclodextrin for lipids in general may explain, in part, the differences in the activities of recombinant GST-AtPIPK1 and lipid kinases present in A. thaliana microsomes. Based on the fact that cyclodextrins are used to deplete cholesterol from membranes, we hypothesize that when membranes are used, the lipid substrate prepared in cyclodextrin is competitively displaced by other lipids in the membrane fractions.}, number={9}, journal={JOURNAL OF LIPID RESEARCH}, author={Davis, AJ and Perera, IY and Boss, WF}, year={2004}, month={Sep}, pages={1783–1789} }