Cyclin‐dependent kinase activity enhances phosphatidylcholine biosynthesis in Arabidopsis by repressing phosphatidic acid phosphohydrolase activity

Summary Coordination of endomembrane biogenesis with cell cycle progression is considered to be important in maintaining cell function during growth and development. We previously showed that the disruption of PHOSPHATIDIC ACID PHOSPHOHYDROLASE (PAH) activity in Arabidopsis thaliana stimulates biosynthesis of the major phospholipid phosphatidylcholine (PC) and causes expansion of the endoplasmic reticulum. Here we show that PC biosynthesis is repressed by disruption of the core cell cycle regulator CYCLIN‐DEPENDENT KINASE A;1 (CDKA;1) and that this repression is reliant on PAH. Furthermore, we show that cyclin‐dependent kinases (CDKs) phosphorylate PAH1 at serine 162, which reduces both its activity and membrane association. Expression of a CDK‐insensitive version of PAH1 with a serine 162 to alanine substitution represses PC biosynthesis and also reduces the rate of cell division in early leaf development. Together our findings reveal a physiologically important mechanism that couples the rate of phospholipid biosynthesis and endomembrane biogenesis to cell cycle progression in Arabidopsis.


INTRODUCTION
Plant growth and development are primarily driven by the production of new cells (Polyn et al., 2015). In addition to DNA replication, cell proliferation requires an approximate doubling of core structural components such as the endomembrane system prior to each division (Jackowski, 1994;. Phospholipids are the main building blocks of the plant endomembrane system (Ohlrogge and Browse 1995) and hence their biosynthesis must occur in conjunction with cell cycle progression.
Temporal changes in phospholipid metabolism over the course of the cell cycle have 4 been well-documented in mammals, fungi, algae, dinoflagellates and bacteria (Jackowski, 1994;Saitoh et al., 1996;Janero and Barrnett, 1981;Kwok and Wong, 2005;Joseleau-Petit et al., 1984;Knacker et al., 1985). Similar changes must also take place in plants but surprisingly they have yet to be described, possibly because of difficulties in generating highly synchronised plant cell cultures (Kwok and Wong, 2005). Furthermore, there is currently no mechanistic understanding of how phospholipid biosynthesis is coupled to the cell cycle in plants.
The cell cycle is governed by conserved molecular machinery in which the core components are serine/threonine kinases, known as cyclin-dependent kinases (CDKs) (Inzé and De Veylder, 2006;Gutierrez, 2009;Polyn et al., 2015). The link between CDK activity and phospholipid biosynthesis is perhaps best understood in the budding yeast Saccharomyces cerevisiae where the enzyme Mg 2+ -dependent phosphatidic acid phosphohydrolase (Pah1p) is a direct target for multisite phosphorylation by both cell division cycle 28 (Cdc28p) and phosphate metabolism 85 (Pho85p) (Santos-Rosa et al., 2005;Choi et al., 2011;. Together Cdc28p and Pho85p modulate a range of cellular processes in response to environmental stimuli (Enserink and Kolodner 2010). Pah1p catalyses the conversion of phosphatidic acid (PA) to diacylglycerol (DAG), which is a key step controlling the partitioning of carbon flux between membrane and storage lipid biosynthesis in S. cerevisiae (Carman and Henry, 2007).
Pah1p is an amphitropic enzyme that switches between a phosphorylated soluble inactive form and a dephosphorylated membrane-bound active form (Choi et al., 2011). Pah1 is dephosphorylated by the nuclear envelope morphology 1sporulation 7 (Nem1p-Spo7p) protein phosphatase complex at the nuclear-endoplasmic reticulum (ER) membrane allowing the protein to associate with the membrane via a short Nterminal amphipathic helix (Karanasios et al., 2010;Choi et al., 2011). 5 Pah1p membrane disassociation and inactivation resulting from multisite phosphorylation by Cdc28p and Pho85p, leads to elevated levels of the enzymes' substrate (PA), which is the precursor for the cytidine diphosphate-diacylglycerol (CDP-DAG) pathway. This pathway is able to synthesise all phospholipid classes in S. cerevisiae (Carman and Henry, 2007). Accumulation of PA also enhances the expression of a suite of phospholipid biosynthetic genes and causes dramatic expansion of the nuclear-ER membrane (Santos-Rosa et al., 2005;Carman and Henry, 2007). The genes that are induced form part of the inositol-responsive circuit in S. cerevisiae (Carman and Henry, 2007). Expression of these genes is induced by the inositol requiring 2inositol requiring 4 (Ino2p-Ino4p) transcription factor complex, but is blocked by interaction of the repressor protein overproducer of inositol 1 (Opi1p) with Ino2p (Carman and Henry, 2007). Accumulation of PA relieves Opi1p-mediated repression of gene expression because it ties Opi1p to a protein called suppressor of choline sensitivity 2 at the nuclear-ER membrane (Loewen et al., 2004).
Orthologs of S. cerevisiae Pah1 are also present in higher eukaryotes where they are often termed lipins (Carman and Henry, 2007). In both animals and plants, it has been shown that disruption of lipins can enhance PA content, trigger changes in nuclear and/or ER membrane morphology (Golden et al., 2009;Gorjánácz et al., 2009;Ugrankar et al., 2011;Mall et al., 2012;Bahmanyar et al., 2014;Eastmond et al., 2010) and alter gene expression (Finck et al., 2006;Donkor et al., 2009;Peterson et al., 2011;Eastmond et al., 2010;Craddock et al., 2015). In the case of Caenorhabditis elegans, although defects in nuclear and ER morphology are accompanied by a perturbation in phospholipid composition, total phospholipid content does not increase significantly (Bahmanyar et al., 2014). Arabidopsis thaliana contains two Pah1p orthologs called PAH1 and PAH2, which appear to be expressed in all 6 tissues (Nakamura et al., 2009;Eastmond et al., 2010). Analysis of a double mutant shows that extra-plastidial phospholipid content almost doubles and both the rate of phosphatidylcholine (PC) biosynthesis and turnover increases markedly (Eastmond et al., 2010;Wang et al., 2014;Craddock et al., 2015). In A. thaliana there are also gross changes in ER morphology, but unlike C. elegans there is no obvious defect in the nucleus (Eastmond et al., 2010;Bahmanyar et al., 2014). These data suggest that plants may use PAHs as a hub in the regulatory network governing phospholipid biosynthesis and endomembrane biogenesis (Eastmond et al., 2010;Craddock et al., 2015).
The mechanism by which PAH inactivation enhances phospholipid (and in particular PC) production in A. thaliana differs markedly from that of S. cerevisiae. This is because the inositol-responsive circuit found in S. cerevisiae (Carman and Henry, 2007) is not conserved in higher eukaryotes and A. thaliana also does not use PA to synthesise PC via the CDP-DAG pathway, because it cannot directly methylate phosphatidylethanolamine (PE) (Keogh et al., 2009). Instead it converts ethanolamine to choline by methylation mainly at the phospho-base level and uses the nucleotide (also known as Kennedy) pathway to assemble PC from CDP-choline and DAG (Keogh et al., 2009). Because DAG is the product of the reaction catalysed by PAH it is counterintuitive that disruption of this enzyme should lead to an increase in PC biosynthesis (Eastmond et al., 2010). However, characterisation of the pah1 pah2 mutant shows that A. thaliana possesses additional routes to synthesise DAG (Nakamura et al., 2009;Eastmond et al., 2010) and isotopic labelling experiments indicate that the rate of PC biosynthesis is more limited by the availability of CDPcholine, supplied by the enzyme cytidine 5'-triphosphate: phosphocholine cytidylyltransferase (CCT) (Kinney et al., 1987;Eastmond et al., 2010). Analysis of PC 7 biosynthesis in the pah1 pah2 mutant suggests that enhanced metabolic flux through the pathway is mainly a result of increased CCT activity (Eastmond et al., 2010;Craddock et al., 2015). In many eukaryotes CCT activity is determined by local membrane lipid composition with a relative enrichment of anionic lipids such as PA and its CDP-DAG pathway derivatives phosphatidylinositol (PI) and phosphatidylglycerol (PG) acting as a stimulant (Cornell and Northwood, 2000;Jackowski and Fagone, 2005). Arabidopsis CCT1 activity can also be stimulated by anionic lipids and expression of a lipid-insensitive version partially mimics the pah1 pah2 phenotype by stimulating both PC biosynthesis and ER proliferation (Craddock et al., 2015). These data suggest that in the pah1 pah2 mutant, PC biosynthesis increases to compensate for an imbalance in membrane lipid composition (Craddock et al., 2015). Interestingly, Szymanski et al., (2014) also reported that disruption of CDP-DAG SYNTHASE2 not only reduces PI and PG levels in A. thaliana, but also those of PC and PE.
Regardless of the precise mechanism of action, PAH activity has the potential to modulate phospholipid biosynthesis in A. thaliana (Eastmond et al., 2010;Wang et al., 2014;Craddock et al., 2015). This study aimed to investigate how PAH activity is governed, and particularly whether it might be coupled to cell cycle progression via the action of CDKs, as has been shown in S. cerevisiae (Santos-Rosa et al., 2005;Karanasios et al., 2010;Choi et al., 2011;. A. thaliana contains twelve CDKs that are classified into six groups, but only class A and class B CDKs have been implicated in core cell cycle regulation in plants (Inzé and De Veylder, 2006;Gutierrez, 2009). Among A. thaliana CDKs, S. cerevisiae Cdc28p and Pho85p share the greatest sequence similarity with CDKA;1, which is the only class A CDK present in A. thaliana and is distinguished by the conserved PSTAIRE cyclin-binding motif (Nowack et al.,8 2012) that is also found in both Cdc28p and Pho85p. CDKA;1 is expressed throughout the cell cycle (Segers et al., 1996), both in tissues with active cell division and also those with proliferative competence (Martinez et al., 1992). CDKA;1 can complement the S. cerevisiae cdc28Δ mutant suggesting that it functions in both G1 to S and G2 to M phase transitions (Ferreira et al., 1991). Mutant analysis in A. thaliana has confirmed that CDKA;1 is necessary for entry into S phase and thus for both mitotic cell division and endoreduplication (Nowack et al., 2012). We therefore focused our investigation on whether CDKA;1 might play a role in governing PAH activity in A. thaliana.

Disruption of CDKA;1 enhances PAH activity and supresses PC biosynthesis
In Arabidopsis thaliana it is possible to obtain null mutants in CDKA;1 (Nowack et al., 2012), but their growth is so severely retarded that biochemical analysis is problematic. However, weaker alleles of cdka;1 (referred to here as cdka;1D and cdka;1DE) have been created by transforming cdka;1 plants with construct expressing phosphomimicry versions of CDKA;1 that have greatly reduced kinase activity (Dissmeyer et al., 2007;2009). The cdka;1D line expresses a copy of CDKA;1 with a T161D substitution and the cdka;1DE line a copy with T14D and Y15E substitutions (Dissmeyer et al., 2007;2009). To determine whether CDKA;1 function effects the activity of PAH in A. thaliana we performed enzyme assays (Eastmond et al., 2010) on extracts from developing rosette leaves of wild type, cdka;1D and cdka;1DE ( Figure   1). PAH activity was increased by ~50% in both cdka;1D and cdka;1DE on a per unit fresh weight basis (Figure 1a). To investigate whether CDKA;1 function also effects 9 phospholipid metabolism we measured the rate of PC biosynthesis in the leaves by monitoring the incorporation of [methyl-14 C]choline into lipids (Eastmond et al., 2010).
The rate of biosynthesis was ~30% lower in cdka;1D and cdka;1DE plants than in wild type when expressed on a per unit fresh weight basis (Figure 1b). The total PC content of cdka;1D and cdka;1DE leaves was also lower (Table S1). Since a reduction in PAH activity leads to stimulation of PC biosynthesis and ER biogenesis in A. thaliana (Eastmond et al., 2010;Wang et al., 2014;Craddock et al., 2015), it is logical that the increase observed in cdka;1D and cdka;1DE plants may be responsible for the suppression of the pathway.

Suppression of PC biosynthesis by CDKA;1 is dependent on PAH activity
To investigate whether the elevated PAH activity in cdka;1D leaves is derived from PAH1 and/or PAH2 and whether this increase is responsible for the suppression of PC biosynthesis we created a pah1 pah2 cdka;1D triple mutant ( Figure 1). The increase in PAH activity detected in cdka;1D leaves was suppressed in pah1 pah2 cdka;1D (Figure 1a) and therefore is attributable to the PAH1 and/or PAH2 proteins, and not to other lipid phosphatases that are responsible for the remaining PAH activity present in pah1 pah2 (Nakamura et al., 2009;Eastmond et al., 2010). The rate of incorporation of [methyl-14 C]choline into lipids in cdka;1D was also recovered in the pah1 pah2 background to a rate that is substantially higher than wild type (Figure 1b).
Leaf area, cell number and cell size of pah1 pah2 cdka;1D plants were not significantly different (P > 0.05) to those of cdka;1D (Table S2) suggesting that the growth defects caused by CDKA;1 disruption (Dissmeyer et al., 2007) cannot be rescued by relieving the repression of phospholipid synthesis. Considered together these data suggest that 10 CDKA:1 supresses PAH activity and enhances PC biosynthesis, and that CDKA;1 exerts its effect on PC production in A. thaliana through PAH.

PAH1 is sufficient to mediate the effect of CDKA;1 on PC biosynthesis
A. thaliana contains two PAH genes with overlapping expression patterns (Eastmond et. al. 2010). Single pah1 and pah2 mutants have no obvious morphological phenotypes (Eastmond et. al. 2010). However, the pah1 mutant does exhibit a significant (P < 0.05) reduction in PAH activity and elevated rate of PC biosynthesis ( Figure 2). PAH1 is therefore the predominant isoform in this tissue. We have previously shown that constitutive expression of hemagglutinin (HA) tagged PAH1 under the control of the CaMV 35S promoter can complement the pah1 pah2 mutant (Craddock et al., 2015). To determine whether CDKA;1 can exert its effect on PC biosynthesis through PAH1 we introduced the same Pro35S:PAH1-HA T-DNA insertion event into the pah1 pah2 cdka;1D background via crossing ( Figure 1). The increase in PAH activity detected in cdka;1D leaves, but suppressed in pah1 pah2 cdka;1D, was rescued by Pro35S:PAH1-HA ( Figure 1a). Furthermore, the rate of incorporation of [methyl-14 C]choline into lipids that was supressed in cdka;1D, but recovered in pah1 pah2 cdka;1D, was again suppressed by Pro35S:PAH1-HA ( Figure   1b). These data suggest that the activity of PAH1 is supressed by CDKA;1 and that, in the absence of PAH2, this suppression is sufficient to stimulate PC biosynthesis.

11
To investigate whether PAH1 is a substrate for CDK-cyclin complexes, histidine x 6tagged PAH1 was expressed in Escherichia coli and affinity purified as previously described (Eastmond et al., 2010). In vitro phosphorylation assays were carried out on the recombinant protein using p13 Suc1 -bound CDK-cyclin complexes isolated from A. thaliana flower bud extracts (Dissmeyer et al., 2007) and [γ-32 P]ATP ( Figure 3).
Kinase activity was detected, which required the presence of both the protein substrate (His6-PAH1) and the CDKA;1-containing complex ( Figure 3a). CDKs commonly phosphorylate the consensus site S/T-P in target proteins (Nowack et al., 2012) and anti-pS/pT-P antibodies (α-MPM2) have previously been used to detect Cdc28p and Pho85p dependent phosphorylation of Pah1p (Santos-Rosa et al., 2005;Choi et al., 2011;. Using α-MPM2 antibodies we were also able to detect CDK-dependent phosphorylation of His6-PAH1 by immunoblotting ( Figure 3a). To screen directly for CDK-dependent phosphorylation sites the products of unlabelled reactions were separated by gel electrophoresis. His6-PAH1 was then subjected to tryptic digestion and the resulting peptides were separated via liquid chromatography and analysed by tandem mass spectrometry (LC-MS/MS). As shown previously (Eastmond et al., 2010), diagnostic mass fingerprints could be identified for 33 peptides, representing 43% coverage of the PAH1 protein. In addition to these peptides, a single phosphorylated peptide was also detected (FYDFQDDPP[pS]PTSEYGSAR) with a modification that corresponds to serine 162 in the intact native protein ( Figure S1).
Evidence from multiple proteomics studies contained in the PhosPhAt 4.0 database (Durek et al., 2010) suggests that this site is also phosphorylated in vivo (Jones et al., 2009;Ito et al., 2009;Nakagami et al., 2010;Reiland et al., 2011;Zhang et al., 2013;Wang et al., 2013). To investigate whether CDK-dependent phosphorylation of His6-PAH affects its activity, enzyme assays were also performed on the reaction products.

12
CDK treatment led to a ~65% reduction in PAH activity ( Figure 3b). These data suggest that PAH1 can be phosphorylated and inactivated by CDKs in vitro and that S162 is a target.

CDKA;1 is required for in vivo PAH1 phosphorylation
To investigate whether CDKA;1 activity effects the in vivo phosphorylation state of PAH1, and whether S162 plays a role in this, we created a dephosphorylated version of PAH1-HA containing a S162A substitution (Dissmeyer and Schnittger, 2011) and expressed it in pah1 pah2 under the control of the CaMV 35S promoter. In S.
cerevisiae, phosphorylation of Pah1p by Cdc28p and Pho85p reduces its electrophoretic mobility (Santos-Rosa et al., 2005;Choi et al., 2011;. Therefore, we immunoprecipitated PAH1-HA and PAH1 S162A -HA from pah1 pah2 leaf extracts and PAH1-HA from pah1 pah2 cdka;1D leaf extract and performed gel electrophoresis followed by immunoblotting ( Figure 4). From the pah1 pah2 genetic background α-HA antibodies detected PAH1-HA bands of differing mobility, whereas from pah1 pah2 cdka;1D a single dominant band was detected with the same mobility as the lowest band present in pah1 pah2 (Figure 4a). From pah1 pah2 expressing PAH1 S162A -HA the single lower band was also dominant (Figure 4a). Using α-MPM2 antibodies we were also able to detected a band in immunoprecipitates from pah1 pah2 expressing PAH1-HA, but were unable to from pah1 pah2 cdka;1D expressing PAH1-HA or pah1 pah2 expressing PAH1 S162A -HA (Figure 4a). To confirm that the shift in mobility and signal detected by α-MPM2 were both due to phosphorylation of PAH1-HA, immunoprecipitates were treated with Calf intestinal alkaline phosphatase (CIP) prior 13 to immunodetection (Figure 4b). Together these data suggest that PAH1 is phosphorylated at S162 in a CDKA;1-dependent manner.

Phosphorylation of PAH1 is necessary for stimulation of PC biosynthesis
To investigate whether the phosphorylation state of PAH1 at S162 plays a role in governing PAH activity and PC biosynthesis in A. thaliana, we introduced 35S:PAH1 S162A -HA into the pah1 pah2 cdka;1D background ( Figure 1). PAH activity in pah1 pah2 leaves expressing PAH1 S162A -HA was significantly higher (P < 0.05) than in those expressing the wild type form of PAH1-HA ( Figure 1a). This was in spite of the fact that the abundance of protein is very similar (Figure 4a). By contrast, expression of PAH1 S162A -HA and PAH1-HA from the same transformation events in pah1 pah2 cdka;1D plants resulted in no significant difference (P > 0.05) in PAH activity ( Figure 1a). The rate of incorporation of [methyl-14 C]choline into PC in pah1 pah2 leaves expressing PAH1 S161A -HA was also significantly lower (P < 0.05) than in pah1 pah2 leaves expressing PAH1-HA, but there was no significant difference (P>0.05) in pah1 pah2 cdka;1D leaves ( Figure 1b). These data suggest that CDKA;1dependent phosphorylation of PAH1 at S162 supresses PAH activity and through this action enhances PC biosynthesis.

Phosphorylation of PAH1 effects membrane association as well as activity
The phosphorylation status of Pah1p determines the protein's localisation, as well as its activity in S. cerevisiae. The N-terminus of Pah1 contains an amphipathic helix responsible for membrane binding and phosphorylation by Cdc28p and Pho85p 14 causes a conformational change that disassociates the protein from the membrane (Karanasios et al., 2010;Choi et al., 2011;. In A. thaliana, both C-and Nterminally tagged PAH1 and PAH2 are predominantly soluble proteins when expressed under the CaMV 35S promoter (Nakamura et al., 2009;Eastmond et al., 2010). However, sequence analysis (Gautier et al., 2008) suggests that the amphipathic helix found at the N-terminus in S. cerevisiae Pah1p might also be a conserved feature in PAH1 and PAH2 ( Figure S2). To study whether phosphorylation of PAH1 affects the localisation of the protein we used immunoblotting to monitor PAH1-HA protein distribution in subcellular fractions of pah1 pah2 leaf extracts expressing either PAH1-HA or PAH1 S162A -HA ( Figure 5). In the case of PAH1-HA, the majority of protein was detected in the soluble (150,000g supernatant) fraction and only traces were detected in the microsomal membrane fraction (150,000g pellet) ( Figure 5a). This finding is consistent with those of Nakamura et al., (2009) who previously performed fractionation experiments on extracts from leaves expressing PAH1 tagged at the C-terminus with green fluorescent protein (GFP). However, in extracts from pah1 pah2 leaves expressing PAH1 S162A -HA, a more even distribution of protein was detected between the microsomal membrane fraction and the soluble fraction ( Figure 5a). The membrane fraction from pah1 pah2 leaves already contains PAH activity, which must be derived from other lipid phosphatases (Nakamura et al., 2009). However, enzyme assays performed on the different fractions from pah1 pah2 leaves expressing PAH1-HA or PAH1 S162A -HA show that the proportion of PAH activity in the membrane fraction is significantly (P < 0.05) enhanced in the case of PAH1 S162A -HA ( Figure 5b). Together these data support the hypothesis that phosphorylation of PAH1 by CDKA;1 reduces membrane association as well as activity.

Phosphorylation of PAH1 is required for normal rates of cell division
We have shown that expressing the dephosphorylated version of PAH1 in A. thaliana results in enhanced PAH activity and membrane residency in leaves and that consequently the rate of PC biosynthesis is repressed ( Figure 1&5). To investigate whether this has an impact on leaf growth and development, wild type (Col-0) plants and pah1 pah2 plants expressing either PAH1-HA or PAH1 S162A -HA were grown alongside each other and their phenotype compared ( Figure 6). Wild type plants and pah1 pah2 plants expressing PAH1-HA were indistinguishable at rosette stage.
However, pah1 pah2 plants expressing PAH1 S162A -HA exhibited a small but significant reduction (P < 0.05) in the area of mature rosette leaves (Figure 6a). The effect of PAH1 S162A -HA expression on leaf growth was investigated in more detail. Microscopic analysis of the abaxial surface of mature leaves showed that the number of epidermal cells was significantly reduced (P < 0.05) in pah1 pah2 plants expressing PAH1 S162A -HA, as compared to either wild type or pah1 pah2 plants expressing PAH1-HA ( Figure   6b). By contrast, the average area of each cell was significantly (P < 0.05) increased ( Figure 6c). These data suggest a defect in the rate (or duration) of cell division (De Veylder et al., 2001). To study this further, we performed kinematic analysis of cell division and expansion in the first true leaf pair of WT and pah1 pah2 expressing PAH1 S162A -HA (DMp1). In both genotypes, leaf area and abaxial epidermal cell number increased exponentially until around day ten to twelve and then the rates decreased (Figure 7a and b). Average cell area also remained stable in both genotypes until day ten and then began to increase exponentially (Figure 7c). In both genotypes the calculated rate of cell division was greatest at the earliest time point (day six) and declined (Figure 7d), as cell expansion accelerated (Figure 7c). These data suggest that the period of most rapid cell division is not substantially altered in 16 DMp1. However, within this phase (day six to ten) the rate was consistently lower in DMp1 than in WT (Figure 7d). These data suggest that phosphorylation of PAH1 at S162 is required for normal rates of cell division (De Veylder et al., 2001). Since we have shown that phosphorylation of S162 in PAH1 is reduced in leaves of the cdka;1D mutant, our data support the hypothesis that CDK-specific suppression of PAH1 catalytic function drives sufficient endomembrane production to allow normal rates of mitotic cell division during leaf growth and development in A. thaliana.

DISCUSSION
In this study, we have provided evidence that the core cell cycle regulator CDKA;1 (Nowack et al., 2012) stimulates the rate of PC biosynthesis in Arabidopsis thaliana by phosphorylating the enzyme PAH. This post-translational modification reduces the catalytic activity of PAH and also impairs protein association with the membrane, where its substrate resides. We have previously shown that suppression of PAH activity enhances phospholipid content and causes ER proliferation (Eastmond et al., 2010). The mechanism of PAH action is yet to be fully understood, but the increase in the rate of PC biosynthesis would appear to be caused by activation of the rate-limiting enzyme CCT (Craddock et al., 2015). Taken together our data provide a mechanistic link to explain how phospholipid biosynthesis in plant cells is coordinated with cell cycle progression. We have also shown that expression of a de-phosphorylated form of PAH1 uncouples PC biosynthesis from CDKA;1 and reduces the rate of mitotic cell division in the leaf. This suggests that CDKA;1 phosphorylation of PAH1 plays a physiologically important role in plant growth and development. Defects in the biosynthesis of fatty acids and lipids have been shown to impair cell division in many eukaryotic systems (Cornell et al., 1977;Tercé et al., 1994;Saitoh et al., 1996;Al-Feel et al., 2003, Kwok andWong 2005;Bach et al., 2011;Atilla-Gokcumen et al., 2014).
At the most basic level coupling between the two processes is considered to be important because membranes are a basic building block of the cell and their constituent phospholipids must increase in mass prior to each division (Jackowski 1994;. There is evidence that specific molecular species of membrane lipids also play specialised roles in cell division (Al-Feel et al., 2003;Bach et al., 2011;Atilla-Gokcumen et al., 2014) and may even act as signals to govern cell cycle progression (Koeberle et al., 2013;Chauhan et al., 2015).
It is accepted that phospholipid metabolism changes dynamically over the course of the cell cycle with the pattern varying depending on species (Jackowski 1994;. For yeasts such as S. cerevisiae and Schizosaccharomyces pombe that undergo closed cell division there is a burst of phospholipid biosynthesis upon entry to mitosis in order to support the rapid expansion of the nuclear envelope (Santos- Rosa et al., 2005;Makarova et al., 2016). In accordance with this, CDK-dependent phosphorylation of both S. cerevisiae Pah1p and its S. pombe ortholog Ned1 (nuclear elongation and deformation protein 1) increases upon G2 to M transition (Santos-Rosa et al., 2005;Makarova et al., 2016). By contrast in mammalian cells, which undergo open cell division, net accumulation of phospholipids is generally associated with cell (and nuclear) expansion during S phase (Cornell et al., 1977;Jackowski 1994).
Interestingly in the yeast S. japonicas, which also uses open cell division, Ned1 is not hyperphosphorylated during G2 to M transition despite being a CDK substrate (Makarova et al., 2016). Unfortunately, it is not known how phospholipid metabolism varies over the course of the cell cycle in plants ( Kwok and Wong, 2005). However, given that plants employ open cell division, it is possible that phospholipid accumulation might also be associated with S phase.
Our data suggest that PAH1 phosphorylation by CDKA;1 at S162 is important for enzyme function, but it remains possible that additional sites are also phosphorylated in a CDK-dependent manner. PAH1 contains multiple putative target sites containing the minimal consensus sequence S/T-P and our proteomic analysis provided incomplete coverage ( Figure S1). Nevertheless, S162 is so far the only S/T-P site to have been identified by in vivo proteomic studies (Durek et al., 2010). In S. cerevisiae, Pah1p is phosphorylated at three sites by Cdc28p and at seven by Pho85p (Choi et al., 2011;. Not all sites are of equal importance. Disruption of some individual sites leads to a significant reduction in Pah1p activity and membrane association but the largest effect results from simultaneous disruption of all (Choi et al., 2012). Pah1p has been shown to be the target of several other kinases, including protein kinase A  and protein kinase C (Su et al., 2014). The mammalian Pah1p ortholog lipin-1 is also phosphorylated by mTORC1 (mechanistic target of rapamycin complex1) (Peterson et al., 2011). Data contained in the PhosPhAt 4.0 database (Durek et al., 2010) suggests that PAH1 is phosphorylated at non S/T-P sites and therefore is also likely to be targeted by other kinases. These kinases could regulate PAH activity and localisation in response to various stimuli, as has been shown in S. cerevisiae . PAH1 is mostly present in the cytosol, shown that in cells of S. cerevisiae cultured with oleic acid, PAH1 will also localise in the nucleus. The functional significance of this response is unknown but in mammals the nuclear localisation of lipin-1 is also controlled by phosphorylation (Peterson et al., 2011). Although mutant analysis suggests that PAH2 plays a less important role in A. thaliana leaves than PAH1 (Figure 2), the two genes do clearly overlap in function (Eastmond et al., 2010). Interestingly the PhosPhAt 4.0 database contains evidence to show that PAH2 is also phosphorylated at several sites including S524 (Durek et al., 2010), which corresponds to the consensus S/T-P. Hence PAH2 may also be phosphorylated by CDKA;1, and indeed by other kinases.
In S. cerevisiae dephosphorylation of Pah1p by the Nem1p-Spo7p protein phosphatase complex on the nuclear-ER membrane, plays a vital role in activating the enzyme and in allowing it to associate with the membrane (Choi et al., 2011;Karanasios et al., 2010). This complex (and its function) is also conserved in animals (Kim et al., 2007;, but it has yet to be identified or characterised in plants. The catalytic subunit Nem1p and its animal ortholog C-terminal domain nuclear envelope phosphatase 1 (CTDNEP1) are related to the small C-terminal domain phosphatase 1 (SCP1), which is a protein that dephosphorylates the highly conserved heptad repeats in the C-terminal domain of RNA polymerase II (Hsin and Manley, 2012). A. thaliana contains a family of SCP1-like proteins but those members that have been characterised to date do not localise to the nuclear envelope (Feng et al., 2010).
The regulatory subunit Spo7p and its animal ortholog, nuclear envelope phosphatase 1-regulatory subunit 1 (NEP1-R1), are very small transmembrane proteins that share little sequence similarity with one another (Han et al., 2012) and standard bioinformatics tools do not reveal any obvious homologues in A. thaliana.
Understanding how PAH1 is dephosphorylated and whether a functionally analogous phosphatase complex to Nem1p-Spo7p is involved will require further work.
In conclusion our data suggest that PAH1 is a direct target for phosphorylation by CDKA;1-cyclin complexes in A. thaliana and that this post-translational 20 modification plays an important physiological role in determining the rate of phospholipid biosynthesis. Given that CDKA;1 is a core cell cycle regulator and is instrumental in controlling both the G1 to S and G2 to M transitions (Nowack et al., 2012) it is likely that its phosphorylation of PAH1 helps to coordinate phospholipid biosynthesis with fluctuations in demand for membrane biogenesis that must take place at these key stages in cell cycle progression (Jackowski 1994;Santos-Rosa et al., 2005). In plants, a detailed description of how phospholipid metabolism changes over the course of the cell cycle is still lacking (Kwok and Wong, 2005). Further work will be required to address this knowledge gap and also to attain a more complete understanding of the underlying mechanisms that couple PAH1, and other key lipid metabolic enzymes, to the core cell cycle machinery.

Plant material and growth conditions
The construction of the Arabidopsis thaliana transgenic lines pah1 pah2 (DM), pah1 pah2 Pro35S:PAH1-HA (DMP), cdka;1D and cdka;1DE have been described previously (Eastmond et al., 2010;Craddock et al., 2015;Dissmeyer et al., 2007;2009). Seeds of A. thaliana were surface-sterilised, applied to agar plates containing half-strength MS salts (Sigma-Aldrich) and imbibed in the dark for 4 d at 4°C after which the plates were then placed vertically in a growth chamber set to 22°C 16 h light/18°C 8 h dark; PPFD = 250 μmol m −2 s −1 . After 7 days the seedlings were transplanted to soil and grown under the same conditions as described above. PCR performed on genomic DNA was used to genotype T-DNA mutants (Eastmond et al., 2010;Dissmeyer et al., 2007).

PC radiolabelling and PAH enzyme assays
Radiolabel feeding experiments were carried out using rosette leaves that were cut into 2 mm diameter strips with a razor blade. They were then placed in vials containing 0.5 mL of half-strength MS salts supplemented with 1 mM [methyl-14 C]choline chloride.
The leaf tissue was vacuum infiltrated for 5 min and, following incubation at 22°C with gentle agitation for up to 3 h, the tissue was blotted dry and frozen in liquid nitrogen.
Lipid extraction and analysis of the labelled PC were performed using the methods described by Tasseva et al. (2004). PAH assays were performed on total homogenates and immunoprecipitates obtained from rosette leaves using the method described by Craddock et al. (2015). The PC content was also directly quantified from total tissue extracts using the enzyme-coupled fluorometric procedure described previously (Eastmond et al., 2010).

Creation of DNA constructs and transformation
PAH1 S162A was created from a PAH1 template in entry vector pDONR207 as previously described in (Craddock et al., 2015) using the Quikchange Lightning Site-
The gene cassette was then cloned into the destination vector pEG101 using the Gateway LR clonase enzyme mix (Invitrogen Ltd.). Transformation of the plasmids into Agrobacterium tumefaciens strain GV3101 was achieved by heat shock and into

22
Arabidopsis wild type and pah1 pah2 by the floral-dip method (Clough and Bent, 1998). Herbicide resistance was used to select transformants containing T-DNA insertions and homozygous lines were crossed into pah1 pah2 cdka;1D (TM).

PAH protein extraction and analysis
Subcellular fractionation, protein extraction, quantification, sodium dodecyl sulfatepolyacrylamide gel electrophoresis (SDS-PAGE) and immunoblotting were performed as described previously (Craddock et al., 2015) but with the following minor modifications. Phosphoenolpyruvate carboxylase activity was measured in subcellular fractions using the method described by Gregory et al. (2009). Detection of phosphorylated PAH1-HA by immunoblotting was carried out with leaf extracts prepared according to the method of Dissmeyer et al., (2007), using an ice cold extraction buffer containing phosphatase and protease inhibitors.
Immunoprecipitations were then performed by incubating 1 mg of total leaf protein with 40 μg of α-HA antibodies and 100 μl of protein A-Sepharose CL-4B beads (10% slurry, w/v) in a total volume of 0.5 ml. Immune complexes were collected by centrifugation at 1,500 g for 20 s and were washed repeatedly before being used directly for SDS-PAGE or pre-treated with calf intestine alkaline phosphatase (CIP) solution at 37 ºC.

Recombinant PAH1 expression and in vitro phosphorylation assays
Heterologous expression of His6-PAH1 in E. coli and affinity purification were performed essentially as previously described (Eastmond et al., 2010) CDK-cyclin complexes were purified from A. thaliana flower buds using p13 Suc1 -Sepharose beads and used for kinase assays following the protocols described by Dissmeyer and Schnittger (2011). Each reaction contained: 30 µl of washed beads together with 8.5 μl of water, 3.5 μl each of 500 mM Tris-Cl (pH 7.8), 150 mM MgCl2, 50 mM EGTA, and 10 mM DTT and 5 μl of His6-PAH1 (~0.1 mg ml -1 ). The addition of 4.1 μl of water, 0.3 μl of 1 mM Li-ATP, and 6 μCi of [γ-32 P]ATP (220 TBq/mmol) started the reactions, which were then incubated for 60 min at room temperature. To terminate 24 the reactions, the mixtures were spotted onto P81 phosphocellulose paper: the papers were then washed five times with 75 mM phosphoric acid and subjected to scintillation counting. Proteomic analysis of unlabelled reaction products was performed following the procedures previously described in Rajangam et al. (2013). In brief, the protein was digested in gel with trypsin, the peptides were resolved using ultra-performance liquid chromatography (UPLC) and analysed by a Quadrupole Time-of-Flight mass spectrometer (Q-ToF MS). The data were used to interrogate the National Center for Biotechnology Information non-redundant database using a MASCOT tandem MS/MS search (http://www.matrixscience.com) to identify PAH1 peptides.

Analysis of leaf growth
A kinematic analysis (De Veylder et al., 2001) of the abaxial epidermal cells of the first true leaf leave pair between 6 to 22 days after sowing (DAS) was performed following the method described by (Nelissen et al., 2013), except that leaves between 6 and 10 DAS were stained with propidium iodide and images were obtained using confocal microscopy. Older leaves were cleared with 70% (v/v) ethanol and mounted in 100% (v/v) lactic acid. The leaves were imaged using a binocular microscope and abaxial epidermal cells from two positions (Nelissen et al., 2013) were imaged with an Axiophot microscope (Zeiss) using differential interference contrast optics. Cell outlines were generated by hand and analysed using ImageJ (http://rsbweb.nih.gov/ij/) and the average cell area and estimated number of cells per leaf were calculated by dividing leaf blade area by cell area. The rate of cell division was calculated as the relative rate of increase in cell number over the course of time. To achieve this the logarithmic values of the average cell number were locally fitted with a five-point 25 quadratic function, of which the first derivative was the cell division rate (Nelissen et al., 2013).

Statistical analysis
The GenStat (2011, 14th edition; ©VSN International Ltd, Hemel Hempstead, UK) statistical system was used to carry out Student's t-tests or one-way analysis of variance (ANOVA). For ANOVA, following significant (P < 0.05) F-test results, means were compared using the appropriate least significant difference value at the 5% (P = 0.05) level of significance, on the corresponding degrees of freedom.

Accession Numbers
Sequence data from this article can be found in the GenBank/EMBL data libraries under the following accession numbers: PAH1, At3g09560; PAH2, At5g42870; CDKA;1, At3g48750. Table S1. Leaf PC content of all genotypes Table S2. Leaf morphology of selected genotypes   Asterisks denote a statistically significant difference from PAH1 alone (P < 0.05). The asterisk denotes a statistically significant difference from DMP (P < 0.05).