The N-terminal Membrane Occupation and Recognition Nexus Domain of Arabidopsis Phosphatidylinositol Phosphate Kinase 1 Regulates Enzyme Activity*

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.

AtPIPK1 is product-activated and the MORN domain binds PtdIns(4,5)P 2 . Adding back the MORN peptide to ⌬MORN or to the PtdOH-activated full-length protein increased activity ϳ2-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.
Phosphatidylinositol-4-phosphate (PtdInsP) 2 5-kinases catalyze the synthesis of phosphatidylinositol-(4,5)-bisphosphate (PtdIns(4,5)P 2 ), a key component in phosphoinositide (PI) signaling that regulates many cellular processes. In Arabidopsis, there are 11 isoforms of PtdInsP 5-kinases that have been divided into 2 subfamilies depending on the size of the proteins (1). Subfamily A consists of AtPIPK10 and AtPIPK11, which contain 401 and 421 amino acids, respectively. The subfamily B proteins (AtPIPK1-9) are larger, from 705 to 859 amino acids, and contain repeated 23-amino acid Membrane Occupation and Recognition Nexus (MORN) motifs in the N terminus. There are 7 MORN motifs in AtPIPK1-3 and 8 motifs in AtPIPK4 -9. The predicted B subfamily PIPKs from rice (2), tomato, and maize also have multiple MORN motifs in the N terminus based on data base analysis; however, these motifs are not found in PIPKs from other eukaryotes including mammals, Caenorhabditis elegans or yeast. In an attempt to understand what advantage there might be for a majority of the plant PIPKs to have this novel N-terminal extension we investigated the effect of the N-terminal MORN domain of AtPIPK1 on enzyme activity and subcellular distribution. MORN motifs were first described in junctophilin (3) and have since been reported to be present in several proteins involved in membrane fission (4,5). Junctophilin is an integral endoplasmic reticulum (ER) protein that is a component of junctional complex between the plasma membrane and ER of excitable cells. The N terminus of junctophilin contains 8 MORN motifs of 14 amino acids that are essential for plasma membrane binding. Takeshima et al. (3) proposed that the MORN motifs of junctophilin bind to membrane phospholipids. Toxoplasma gondii has 2 proteins, MORN1 and MORN2, which contain only multiple tandem MORN motifs and have no known enzyme activity (5). MORN1 consists of 14 MORN motifs of 23 amino acids each, and is thought to be part of a protein complex that forms a constrictive ring during nuclear division and daughter cell budding. In Arabidopsis, the only other protein reported to contain MORN motifs is ARC3 (4). ARC3 localizes to the chloroplast envelope at the site of division and is thought to be involved in building a scaffold of proteins on the chloroplast outer envelope, which are essential for organelle fission (4). Consistent with this function, arc3 mutant plants had fewer and larger chloroplasts compared with wild-type plants.
The function of the MORN-containing proteins described above is to bind membranes and/or facilitate the formation of a protein scaffold involved in tight membrane adhesion or organelle fission. The human protein ALS2 and the plant PIPKs are the only MORN motif containing proteins reported to have enzymatic function. ALS2 functions as a Rab5 guanine nucleotide exchange factor (GEF) and is involved in endosomal membrane trafficking (6). The C-terminal domain of ALS2 contains 8 MORN motifs within the Rab5GEF domain. At least 4 of the motifs were essential for Rab5GEF function. Mutations in ALS2 are associated with a rare juvenile form of amyotrophic lateral sclerosis (7).
In contrast to ALS2, the plant PIPK subfamily B proteins, exemplified by AtPIPK1 (At1g21980), have N-terminal MORN motifs and the motifs do not overlap with the catalytic domain. We show that the N-terminal MORN domain (amino acids 1-251) is essential for PtdOH activation and that expressing the MORN domain peptide in vivo or adding it to the PtdOH-activated enzyme increases PtdInsP kinase activity. Taken together, the data support a model where the MORN domain functions in the full-length protein to regulate the accessibility of lipids to the active site in a PtdOHsensitive manner.

EXPERIMENTAL PROCEDURES
Plasmid Constructs-Full-length AtPIPK1 (8), truncated versions of AtPIPK1, and for chimeric MORN fusions were cloned using primer sets designed as described in Table 1. To construct vectors for C-terminal-truncated proteins of AtPIPK1, a stop codon was added in the reverse primers. The P396A mutant was generated by PCR using the QuikChange site-directed mutagenesis kit (Stratagene, La Jolla, CA) with the oligonucleotide primers 5Ј-GGACAAGGTTTCCAGC-AGAAGGGACTAAG-3Ј and 5Ј-CTTAGTCCCTTCTGCT-GGAAACCTTGTCC-3Ј. PCR was carried out using Platinum Pfx DNA polymerase (Invitrogen, Carlsbad, CA). The PCR products were subcloned into the pENTR/SD/D-TOPO entry vector (Invitrogen), and the sequences were verified. The resulting entry clones were recombined with an E. coli expression vector, pDEST15 (Invitrogen), for production of N-terminal glutathione S-transferase (GST)-tagged fusion proteins or with the Gateway-compatible plant expression vector pK7WGF2 (Functional Genomics Division of the Department of Plant Systems Biology, Gent, Belgium) for production of green fluorescence protein (GFP) fusion proteins under the control of a CaMV 35 S promoter.
Protein Expression and Purification-The E. coli expression strain BL21(DE3)pLysS (Invitrogen) was transformed with the designed constructs and used to express the GST-tagged fusion proteins. Recombinant proteins were expressed and purified as described previously (8). Protein concentration was determined using the Bradford method (Bio-Rad) with BSA as a standard. Purified recombinant proteins bound to glutathione-Sepharose beads were stored at 4°C until use for immunoblots and lipid kinase assays. For some studies proteins were eluted with reduced glutathione elution buffer (50 mM Tris-HCl, pH 8.0, 20 mM reduced glutathione, 0.01% (w/v) Nonidet P-40, 100 mM NaCl). For gel electrophoresis, bound proteins were eluted by boiling in 2ϫ SDS sample buffer 5 min prior to 10% (w/v) SDS-PAGE.
Plant Transformation and Selection of Transgenic Lines-The recombinant binary plasmids (pK7WGF2-AtPIPK1, pK7WGF2-MORN, pK7WGF2-⌬MORN, and vector control pK7WGF2) were transformed into Agrobacterium tumefaciens EHA105 by the freeze-thaw method (9). For stable transformation, NT-1 cells were transformed using A. tumefaciens-mediated gene transfer following the protocol of Perera et al. (10). For each transformation, two independent kanamycin-resistant microcalli were selected. Cells were grown in NT-1 medium containing 50 g ml Ϫ1 kanamycin and were subcultured weekly into 25 ml of NT-1 culture medium containing 50 g ml Ϫ1 kanamycin as described by Perera et al. (10). For transient transfection, NT-1 protoplasts were transfected using polyethylene glycol-mediated gene transfer following the protocol of Jin et al. (11).
Preparation of Plasma Membranes-NT-1 cells were harvested at days 4 by filtration and immediately homogenized in three volumes of cold buffer (200 mM Suc, 30 mM Tris-HCl, pH 7.2, 3 mM EGTA, 1 mM MgCl 2 , 1 mM dithiothreitol, and 1 mM phenylmethylsulfonyl fluoride) in a ground-glass Dounce homogenizer with 1% (w/w) polyvinylpolypyrrolidone to facil- itate grinding. The crude extract was clarified by centrifugation at 5,000 ϫ g for 10 min at 4°C. The supernatant was used as total cell lysate or fractionated further (40,000 ϫ g, for 60 min, at 4°C) to yield microsomal and soluble protein fractions. The microsomal pellet was washed in buffer (30 mM Tris-HCl, pH 7.2, and 15 mM MgCl 2 ), centrifuged at 40,000 ϫ g for 30 min at 4°C, and the final pellet was resuspended in the same buffer. Plasma membrane-enriched fractions were prepared from microsomes by aqueous two-phase partitioning as described previously (12,13). For enzyme assays, membrane fractions were placed on ice and assayed immediately. Protein concentrations were estimated using the Bradford method (Bio-Rad) with BSA as a standard.
Immunoblots-Recombinant proteins were detected using monoclonal anti-GFP (BD Biosciences, San Jose, CA) or monoclonal anti-GST (BD Biosciences) antibodies at a dilution of 1:5,000 for 1 h followed by horseradish peroxidase-conjugated anti-mouse as the secondary antibody at a dilution of 1:20,000 for 1 h. Immunoreactivity was visualized by incubating the blot in SuperSignal West Pico Chemiluminescence substrate (Pierce) and exposure to x-ray film.
Liposome Binding Assay-Liposomes were prepared by a modification of published protocols (15,16). Phosphatidylcholine (65% PtdCho), phosphatidylserine (20% PtdSer), and the desired lipid, either 10% of PtdSer, phosphatidic acid (PtdOH), or phosphoinositides, were mixed as indicated. NBD-PtdCho (5%) was used to monitor lipid recovery. All lipids were obtained from Avanti Polar Lipids Inc. The lipid mixtures were dried under nitrogen gas, and resuspended to final concentration of 2-5 mg/ml in buffer containing 50 mM Hepes at pH 7.2, 100 mM NaCl, and 0.5 mM EDTA. Resuspended lipids were incubated at 37°C for 2 h and then sonicated in a bath sonicator until homogeneous liposomes formed (2 min). Liposomes were collected by centrifugation at 16,000 ϫ g for 10 min and resuspended in buffer containing 20 mM Hepes at pH 7.2, 100 mM NaCl, and 1 mM EGTA. Protein (2-5 g) was incubated with 0.64 M/50 l liposomes at 22°C and centrifuged at 16,000 ϫ g for 30 min. The supernatant (S) was removed, and the liposome pellet (P) resuspended to the same volume as supernatant. The supernatant and the pellet proteins were subjected to SDS-PAGE and immunoblotting with monoclonal anti-GST antibody as described above.
Membrane arrays (PIP-Strips) spotted with 100 pmol of phospholipids were purchased from Echelon Research Laboratories and used for protein-lipid overlay assays by following the manufacturer's instructions. Strips were incubated 2 h in TBST with 3% fat-free BSA at room temperature then transferred to the purified recombinant protein solution at 1 g ml Ϫ1 in TBST overnight at 4°C with gentle agitation. Each strip was washed three times in TBST buffer before incubating with peroxidaseconjugated anti-GST antibody (Santa Cruz Biotechnology, Santa Cruz, CA) to detect the bound GST fusion proteins. Antibody binding was visualized using ECL plus Western blotting detection reagents (Amersham Biosciences, GE Healthcare Life Sciences) and exposing the blot to x-ray film. PtdOH binding was confirmed using PtdOH bound to nitrocellulose membranes as described by Stevenson et al. (17).
In Vivo Imaging-Confocal fluorescence images and concurrent differential interference contrast (DIC) images were acquired with a Leica TCS SP confocal system using a Leica DM IRBE microscope and a 40ϫ N.A. 1.2 oil immersion objective (Leica, Wetzlar, Germany). Samples were excited with an argon laser at 488 nm and fluorescence emission was collected from 500 to 560 nm.

A Comparison of the AtPIPK1 MORN Motif Amino Acid
Sequences-An analysis of the AtPIPK1 (At1g21980) amino acid sequence using Pfam predicts 7 MORN motifs of 23 amino acids at the N terminus (amino acids 81-251) and a type I phosphatidylinositol-4-phosphate 5-kinase (PIPK) kinase homology domain at the C terminus that includes an activation loop AL (701-713) as indicated in Fig. 1A. In addition, there is a predicted linker region (amino acids 380 -409) according to Domain Linker Prediction.
Each of the 7 MORN motifs is slightly different; however, they all contain a consensus sequence of hydrophobic and glycine residues, YXGX(W/F)(X) 6 GXG(X) 6 G(X) 2 (Fig. 1B) interspersed with several basic amino acids. As shown in Fig.  1B, the putative MORN motifs are highly similar to those from junctophilins, an ER-plasma membrane junction protein (3), MORN1, a component of the T. gondii cell division apparatus (5), ALS2 (6), a Rab5GEF, and ARC3 (4), a chloroplast division factor. A multiple sequence alignment of the MORN sequences from the proteins above show that the MORN sequence of T. gondii MORN1 aligns most closely to MORN sequence of AtPIPK1 (Fig. 1C).
The N-terminal MORN Domain of AtPIPK1 Affects Enzyme Activity-To determine whether or not the N-terminal MORN domain affected enzyme activity and to characterize the catalytic domain of the enzyme, we made a series of deletion peptides. All of the constructs were designed as N-terminal GST fusions so that the recombinant, E. coliproduced peptides could be detected with antibodies to GST as shown in (Fig. 2A). Removing the N-terminal MORN domain (amino acids 1-251) produced ⌬MORN which had Ͼ4 times the specific activity of the full-length enzyme (Fig.  2B). Removing both the N-terminal MORN and C terminus (amino acids 697-752), which includes the predicted activation loop to generate ⌬MORN/⌬C decreased the activity compared with ⌬MORN but increased activity 1.5-2 times compared with the full-length enzyme. The predicted activation loop in the C-terminal region is similar to that found in the human PIPK1␣ where it defines the substrate specificity and is required for activity (18). We predicted that elimination of the predicted activation loop region (⌬C) in ⌬MORN would abolish the activity, but it did not. Notably, amino acids 252-435, which included a predicted linker region, were essential for activity. Deleting amino acids 252-435 in combination with the MORN domain (⌬MORN/⌬L or ⌬MORN/⌬L/⌬C), eliminated all enzyme activity (Fig.  2B). Together, the truncated proteins indicate that the MORN domain is not essential for activity and that removing the MORN increases activity.
Deleting the MORN Domain and the Predicted Activation Loop Increases the V max but Does Not Affect the Substrate Specificity or K m of AtPIPK1-To determine whether the truncated enzymes were promiscuous and would use any substrate, we characterized the substrate specificities from purified recombinant GST-tagged proteins using PtdIns3P, PtdIns4P, or PtdIns5P as substrates. As shown in Table 2, deleting the N-terminal MORN domain with or without the C-terminal activation loop (⌬MORN and ⌬MORN/⌬C) had no effect on substrate specificity. PtdIns4P was the preferred substrate for the full-length and the truncated proteins. PtdIns3P was phosphorylated to a lesser extent; however, when PtdIns5P was added, there was no detectable product formed. The results were similar whether the substrate was delivered in Triton X-100 or ␤-CD (8). As  shown in Table 3, the ⌬MORN and ⌬MORN/⌬C deletions did not significantly affect the K m ; however, the V max increased so that the V max /K m ratio increased 2-fold.
The MORN Domain Peptide Binds PtdOH and PtdIns(4,5)P 2 and Is Necessary for PtdOH Activation-As shown in supplemental data Fig. S1, the MORN domain, ⌬MORN and the fulllength enzyme all bound to PtdOH, PtdInsP, and PtdInsP 2 presented in lipid blots. To compare the lipid binding specificity, we used a liposome binding assay. In the liposome assay, the MORN domain preferentially bound to PtdOH and PtdIns(4,5)P 2 relative to PtdIns4P (Fig. 3, A and B). In contrast, ⌬MORN bound the substrate, PtdIns4P, more than PtdOH and PtdIns(4,5)P 2 .
PtdOH has been shown to activate both the HsPIPK1␣ and AtPIPK1 (8,19). Because the MORN domain bound PtdOH, we asked whether the MORN domain was necessary for PtdOH activation. The specific activity of ⌬MORN and ⌬MORN/⌬C did not change when assayed with PtdIns4P and PtdOH at a 1:1 molar ratio. Under these same conditions the purified, recombinant full-length AtPIPK1 was activated 2-fold (Table 3). These data indicated that the MORN domain was necessary for PtdOH activation.
The MORN domain also bound PtdIns(4,5)P 2 . Therefore, we asked what affect the MORN domain peptide would have on enzyme activity if it was added back to the full-length protein or the truncated peptide constructs. When added to ⌬MORN and ⌬MORN/⌬C, the MORN domain peptide increased the specific activity (Fig. 4). The activity increased with the increasing molar ratio of the MORN domain peptide to ⌬MORN. In contrast, adding the MORN domain even at the highest molar ratio had no effect on the specific activity of the full-length enzyme (Fig. 4). GST was added as a control for these experiments and had no effect on the activity (data not shown). Because deleting the MORN domain increased the V max , we hypothesized that in the native, full-length AtPIPK1 the MORN domain would somehow affect the access of the substrate or product to the Liposomes were prepared by mixing phosphatidylcholine (65% PtdCho), phosphatidylserine (20% PtdSer), and an additional 10% of PtdSer, phosphatidic acid (PtdOH), PtdIns4P, or PtdIns(4,5)P 2 as indicated. NBD-PtdCho (5%) was used to monitor lipid recovery. The proteins (2-5 g) were incubated with 0.64 M/50 l liposomes at room temperature and centrifuged at 16,000 ϫ g as described under "Experimental Procedures." The supernatant (S) was removed, and the liposome pellet (P) was resuspended to the same volume as supernatant. A, supernatant and the pellet proteins were separated on SDS-PAGE, immunoblotted with a monoclonal anti-GST antibody, and visualized by autoradiography. B, liposome binding of the GSTtagged proteins was quantified by densitometry. The images were scanned, and the relative density was quantified using the Imager FX (Bio-Rad). The relative protein is reported as bound (pellet)/free (supernatant). The data are the averages from two independent experiments.

TABLE 2 Substrate preferences of recombinant GST-tagged, full-length AtPIPK1 and truncated proteins
The specific activity of purified recombinant GST-tagged AtPIPK1 full-length and truncated proteins was measured to determine the substrate preference. Equal amounts of purified protein were incubated with the indicated substrates, 125 M PtdIns4P, PtdIns3P, PtdIns5P, and ͓␥-32 P͔ATP as described under "Experimental Procedures." The 32 P-labeled lipids were separated by TLC and quantified using a Bioscan Imaging scanner. The product formed when using PtdIns4P as the substrate is significantly higher than with other substrates for all the recombinant proteins. Specific activity is shown as pmol/min⅐mg protein.

TABLE 3 Kinetic analysis and PtdOH activation analysis of recombinant GST-tagged, full-length AtPIPK1 and truncated proteins
The assays were done using Triton X-100 for substrate delivery as described under "Experimental Procedures." Equal amounts (125 M) of PtdIns4P and PtdOH were added for PtdOH activation analysis. Adding PtdOH alone gave no PtdInsP kinase activity (data not shown  (Fig. 5A). In both instances, the MORN domain decreased enzyme activity almost 50% when it was contiguous with the N terminus of the enzyme (Fig. 5B).
Mutations in the Predicted Linker Region Eliminate PtdOH Activation of AtPIPK1-The predicted linker region is between the MORN domain and the kinase homology. We proposed that if the MORN domain of AtPIPK1 bound PtdOH in the membrane, the resulting change in conformation in the linker region might expose the catalytic site and thereby increase the activity. Site-directed mutagenesis was used to test the function of the putative linker region (KQTDFDPKEKFWTRFPPEGT-KTTPPHQSVD). Based on the Domain Linker Predictor program we predicted that mutating Pro 396 (in bold) to Ala would result in a critical kink between two domains. The enzyme activity of the P396A mutant was the same as that of the wild type protein; however, P396A was not activated by PtdOH indicating that Pro 396 was essential for PtdOH activation (Fig. 6, A  and B). Of interest, P396A activity increased about 2-fold when the MORN peptide was added. This is a similar fold increase to that found with ⌬MORN ϩ MORN (Fig. 6B) even though the specific activity is not as high as that of ⌬MORN.
When the PtdOH-activated full-length enzyme was incubated with the MORN peptide, the specific activity increased 2-fold over that from PtdOH activation alone (Fig. 6B). The specific activity was comparable to that of ⌬MORN but less than ⌬MORN ϩ MORN. These data are consistent with the concept that in the full-length protein the N-terminal MORN domain restricts access to the active site until it binds PtdOH. From these data, however, it was not clear how the MORN peptide was enhancing activity.
AtPIPK1 Is Activated by PtdIns(4,5)P 2 -The lipid binding data (Fig. 3) indicated that the MORN peptide had a higher affinity for the PtdIns(4,5)P 2 than PtdIns4P. Based on the literature (20,21), we anticipated that PtdIns(4,5)P 2 would feedback regulate a membrane-associated PIPK. This was not the case with AtPIPK1. When PtdIns(4,5)P 2 was added in excess, to ⌬MORN or the full-length enzyme, activity increased 2-and 8-fold, respectively (Table 4). If the full-length enzyme was activated by PtdOH, the effect of PtdIns(4,5)P 2 was even greater (16-fold). Adding PtdIns(3,5)P 2 did not increase enzyme activity even when PtdOH was present. We concluded from these experiments and the lipid binding data that AtPIPK1 was prod- uct activated and the MORN domain, while not essential for PtdIns(4,5)P 2 activation, had the potential to enhance it.
The MORN Domain Localizes to the Plasma Membrane and Increases Endogenous Enzyme Activity-For these studies we used tobacco cells grown in suspension culture, a model system that has low endogenous PtdInsP 2 (10). We generated transgenic tobacco cell lines expressing the MORN and ⌬MORN fused to GFP. GFP-AtPIPK1 or AtPIPK1 without GFP could not be detected in transgenic tobacco cells or Arabidopsis plants either after stable transformation of the cells or plants using A. tumefaciens-mediated transformation or after transient transformation of protoplasts isolated from tobacco cells. The lack of AtPIPK1 expression was not because increasing PtdIns(4,5)P 2 was toxic. We could increase PtdIns(4,5)P 2 production in both tobacco cells and Arabidopsis plants severalfold by expressing HsPIPK1␣, and the cells and plants were viable. 3 When we stably or transiently expressed GFP alone, GFP-MORN, and GFP-⌬MORN in tobacco cells, we could detect significant GFP fluorescence in the cytosol (supplemental data, Fig. S2). Furthermore, when we isolated membranes from the stably transformed GFP-MORN and GFP-⌬MORN cells, we detected a significant amount of GFP by immunoblotting in the lower phase membrane proteins (endomembranes) isolated by two-phase partitioning (Fig. 7, A and C). These observations raised the possibility that some of the in vivo cellular fluorescence was attributable to GFP produced by proteolysis of the fusion peptide. Similar proteolysis of GFP peptides has been reported (22) and complicates the interpretation of the data. For these reasons, we used cell fractionation, immunoblotting, and lipid kinase assays rather than confocal imaging to analyze the distribution of the GFP fusion peptides. GFP-MORN was detected primarily in plasma membrane enriched fraction isolated by aqueous two phase partitioning (Fig. 7A). We could not detect GFP-⌬MORN in the isolated plasma membrane fraction using GFP antibodies (Fig. 7A); therefore, to compare the distribution of GFP-⌬MORN and GFP-MORN we measured enzyme activity using membranes from transgenic and wild-type cells. The plasma membrane fraction from the GFP-⌬MORN cell lines had higher PtdInsP 5-kinase specific activity indicating increased plasma membrane localization of the ⌬MORN enzyme. The lower phase by aqueous two-phase partitioning also had a slight increase in activity (Fig. 7B) which is consistent with the increased intracellular fluorescence. With the MORN domain-expressing cells, the specific activity of the plasma membrane PtdInsP 5-kinase was also higher than that of the wild-type cells even though the MORN domain peptide alone has no enzyme activity (Fig. 7B). Plasma membranes from the tobacco cells contain from 5-10 mol % PtdOH, 3 and our in vitro data that indicate the MORN peptide will activate the full-length protein in the presence of PtdOH. Therefore, we concluded that the MORN peptide enhanced the activity of the endogenous plasma membrane enzyme.
The MORN Domain and PtdInsP Kinase Activity Increase in the Lower Phase Membrane Fraction in Response to Hyperosmotic Stress-Several groups have shown that total cellular PtdInsP 2 increases in plants after several min of hyperosmotic stress (23-25) but the subcellular localization of PIPKs has not been characterized. When tobacco cells were exposed to hyperosmotic stress (0.8 osmolal sorbitol for 15 min) prior to isolating plasma membranes by aqueous twophase partitioning, the GFP-MORN peptide redistributed from the plasma membrane and was recovered primarily in the endomembrane or lower phase fraction (Fig. 7C). Consistent with the loss of the MORN peptide, the PtdInsP kinase activity in the plasma membrane fraction decreased (Fig. 7D). The decrease in plasma membrane enzyme activity also was evident in the ⌬MORN cells.
In both the wild type (NT-1) and GFP-⌬MORN transgenic lines there was an increase in PtdInsP kinase activity in the lower phase fraction after hyperosmotic stress (Fig. 8A). To determine whether lipid binding was sufficient to cause the change in enzyme distribution, we extracted lipids from membranes isolated from the stressed and non-stressed wild-type cells, made liposomes and monitored the binding of the E. coli-expressed, full-length AtPIPK1. As shown in Fig. 8B, AtPIPK1 was recovered in the liposome pellet with all the extracts; however, if cells were osmotically stressed, there was 30% less binding to lipid vesicles isolated from plasma membranes and 35% increased binding to lipid vesicles from lower phase membranes.
Our results indicate that the increase in whole cell PtdInsP 2 in response to hyperosmotic stress in plants reflects an increase in intracellular PtdInsP 5-kinase activity. These observations along with the observed redistribution of the GFP-MORN peptide from the plasma membrane to the lower phase make a compelling argument that the N-terminal MORN domain can regulate the membrane association as well as the activity of AtPIPK1. 3 Y. J. Im and W. F. Boss, unpublished results. B, MORN peptide increased the activity of the full-length enzyme if PtdOH was present. ⌬MORN activity was increased by adding the MORN peptide and as predicted PtdOH had no effect. P396A was not activated by PtdOH but the activity increased if the MORN peptide was added. For these experiments the MORN peptide was added at a 2:1 ratio with each enzyme and 125 M PtdOH (a 1:1 ratio with PtdIns4P) was used as in Table 2

DISCUSSION
MORN motifs are found in several different proteins most of which are involved in tight membrane adhesion or organelle fission but have no enzymatic activity (3-6, 26, 27). In plants, these motifs are associated with the largest family of plant PIPKs, enzymes essential for the synthesis of PtdIns(4,5)P 2 . We have used AtPIPK1 as an example of the MORN-containing plant PIPKs to investigate the potential functions of this N-terminal domain. The data we present show that the AtPIPK1 MORN domain provides an additional, plant-specific mechanism for regulating PIPK activity.
AtPIPK10, which does not contain an N-terminal MORN domain, has the same K m but a 10-fold lower V max than AtPIPK1 and is not activated by PtdOH (8). This was our first clue that the MORN domain might be involved in PtdOH activation. This was substantiated by the fact that ⌬MORN was not PtdOH activated. PtdOH activation also can be eliminated by mutating Pro 396 to Ala in the linker region between the MORN domain and the kinase homology domain. The P396A mutant is as active as the PtdOH-activated enzyme suggesting that PtdOH binding normally causes a change in conformation in the linker region. The conformational change likely increases access to the active site because deleting the MORN domain from AtPIPK1 resulted in a more active, truncated enzyme, ⌬MORN.
PtdOH increases in plants in response to drought, abscisic acid, and cold and genetically altering PtdOH production by altering expression of selective PLDs affects plant responses to environmental stress (28 -33). Because some PLD isoforms require PtdInsP 2 for activity (34), redistribution of AtPIPK1 with its MORN-membrane binding motifs could also selectively target PLD isoforms to propagate a lipid-mediated signal.
Our experiments provide evidence that the MORN domain is involved in more than just PtdOH activation and have uncov- . GFP-MORN was present in the plasma membrane fraction of non-stressed cells and moved to endomembranes (the lower phase fraction) after cells were exposed to hyperosmotic stress. A, immunoblot of plasma membrane and lower phase membrane proteins isolated from wild-type tobacco cells (NT-1) and transgenic cells containing each of the GFP constructs. Proteins were visualized using antisera raised against the GFP tag. The antiserum recognized a nonspecific band of ϳ70 kDa in plasma membranes of all cells and a 56-kDa band of the predicted mass of GFP-MORN in the plasma membranes from MORN transgenic lines (indicated by the arrow). A 28-kDa band corresponding to GFP was evident in the lower phase fractions. B, PtdInsP 5-kinase activity was monitored in the isolated membrane fractions from each cell line (solid bars, plasma membrane; gray bars, lower phase; n.d., not detectable). C, immunoblot of plasma membrane and lower phase membrane proteins isolated from transgenic cells expressing GFP-MORN and GFP-⌬MORN, which were treated with hyperosmotic stress (0.8 osmolal sorbitol in conditioned medium for 15 min) or conditioned medium alone (Ϫ). Membrane proteins were separated and visualized with antisera raised against GFP as in A. The migration of GFP-MORN is indicated with an arrow. D, PtdInsP kinase activity is lost from the plasma membranes of osmotically stressed cells. Cells were treated with hyperosmotic stress and enzyme activity was measured as is B. The values are the average of duplicates from two independent experiments. The MORN domain peptide not only increased enzyme activity of ⌬MORN, it also increased the specific activity of the activated, full-length enzyme. Notably, even though the MORN peptide itself has no enzyme activity, when the peptide was produced in tobacco cells the plasma membrane PtdInsP kinase activity increased 2-fold. Ma et al., (27) have shown that the MORN domain will associate with yeast plasma membranes in vivo. Our data demonstrate that the peptide will localize to a plant plasma membrane and provide in vivo evidence that the MORN peptide binds both PtdOH and PtdInsP 2 and increases enzyme activity. One mechanism for the MORN peptide to enhance the activity of the endogenous membrane would be by clustering PtdInsP 2 and facilitating product activation. We do not know whether product activation is specific for AtPIPK1 or is typical of all plant PIPKs. To our knowledge, product activation has only been reported for the non-membrane-associated PIPK activity of rat pituitary cells (20). Proteins that contain hydrophobic/acidic regions which target and penetrate the membrane usually undergo a conformational change as the protein is recruited and activated at the membrane surface (35). Such a dynamic process is envisioned for AtPIPK1 as the MORN domain first binds PtdOH and undergoes a conformational change in the linker region. The series of hydrophobic and basic amino acids in the MORN motifs might bind and cluster PtdInsP 2 by a combination of hydrophobic and electrostatic interactions in a manner similar to the N terminus of MARKS (36).
The impact of lipids on the subcellular distribution of AtPIPK1 was revealed in lipid binding studies using the recombinant, full-length enzyme and lipids extracted from hyperosmotic stressed and non-stressed cells. Based on our results, it is likely that when stimuli such as hyperosmotic stress increases PtdOH, the MORN domain of AtPIPK1 binds PtdOH and increases PtdInsP 2 biosynthesis. The biological relevance of these observations is that they reveal a means of generating new microdomains of PtdInsP 2 within the cell in response to stress.
The differential lipid binding while consistent with the changes in the enzyme activity we observed, is likely not the only factor affecting enzyme activity or distribution. Protein (de)phosphorylation or other interacting proteins can also result in either an electrostatic switch or conformational change that would recruit the PIPKs to membranes in response to stress stimuli. PtdInsP 5-kinase activity can be down regulated by phosphorylation (37)(38)(39) in both plants and animals. Recently, Yamamoto et al. (37) demonstrated that after treating HeLa cells with hyperosmotic stress, the human type I PIPK␤ redistributed from the soluble fraction to the plasma membrane as a consequence of dephosphorylation, and the plasma membrane PtdInsP kinase activity increased. In contrast to HeLa cells, the increased PtdInsP kinase activity in the hyperosmotically stressed tobacco cells was associated with the lower phase fraction and not the plasma membrane. It remains to be seen whether PIPK1 is also dephosphorylated in response to hyperosmotic stress. While this dephosphorylation would favor membrane binding it is not the determining factor as E. coli expressed AtPIPK1 selectively bound isolated membrane lipids. ⌬MORN might also be involved in targeting because transgenic tobacco cell lines expressing ⌬MORN showed a greater than 2-fold increase in PtdInsP kinase activity in the lower phase membranes after hyperosmotic stress.
Presumably, in planta, specific AtPIPK isoforms will be targeted in a tissue-specific and cell specific manner to provide tight regulation of PtdInsP 2 -mediated processes. The PI pathway is essential for membrane biogenesis (40 -43) and enzymes in the pathway regulate chromatin structure and RNA transport in the nucleus (44,45). Multiple mechanisms for regulating the subcellular distribution of the plant PIPKs would provide additional means for plants to "time" a signaling event. Such a timing device would contribute positively to the multifaceted signaling pathways of these sessile organisms.
In summary, MORN motifs have previously been characterized for their roles in membrane binding and organelle fission. This work describes a new function for MORN motif-containing peptides and a new mechanism for regulating PIPKs and targeting them to selective membranes.