RESEARCH ARTICLE

Diacylglycerol kinase (DGKA) regulates the effect of theepilepsy and bipolar disorder treatment valproic acid inDictyostelium discoideumElizabeth Kelly, Devdutt Sharma, Christopher J. Wilkinson and Robin S. B. Williams*

ABSTRACTValproic acid (VPA) provides a common treatment for both epilepsyand bipolar disorder; however, common cellular mechanisms relatingto both disorders have yet to be proposed. Here, we explore thepossibility of a diacylglycerol kinase (DGK) playing a role in regulatingthe effect of VPA relating to the treatment of both disorders, using thebiomedical model Dictyostelium discoideum. DGK enzymes providethe first step in the phosphoinositide recycling pathway, implicated inseizure activity. They also regulate levels of diacylglycerol (DAG),thereby regulating the protein kinase C (PKC) activity that is linked tobipolar disorder-related signalling. Here, we show that ablation of thesingle Dictyostelium dgkA gene results in reduced sensitivity to theacute effects of VPA on cell behaviour. Loss of dgkA also providesreduced sensitivity to VPA in extended exposure during development.To differentiate a potential role for this DGKA-dependent mechanismin epilepsy and bipolar disorder treatment, we further show that thedgkA null mutant is resistant to the developmental effects of a range ofstructurally distinct branched medium-chain fatty acids with seizurecontrol activity and to the bipolar disorder treatment lithium. Finally, weshow that VPA, lithium and novel epilepsy treatments functionthrough DAG regulation, and the presence of DGKA is necessary forcompound-specific increases in DAG levels following treatment.Thus, these experiments suggest that, in Dictyostelium, loss ofDGKA attenuates a common cellular effect of VPA relating to bothepilepsy and bipolar disorder treatments, and that a range of newcompounds with this effect should be investigated as alternativetherapeutic agents.

This article has an associated First Person interview with the firstauthor of the paper.

KEY WORDS: Diacyclglycerol, Diacylglycerol kinase,Dictyostelium discoideum, Epilepsy, Lithium, Valproic acid

INTRODUCTIONDiacylglycerol kinases (DGKs) provide the first step in thephosphoinositide salvage pathway, functioning in thephosphorylation of diacylglycerol (DAG) to produce phosphatidic

acid (PA) (Whatmore et al., 1999). Through this activity, cellsregulate levels of lipid (Milne et al., 2008) and glucose metabolism(Banfic et al., 1993; Inoguchi et al., 1994), cell growth (Banfic et al.,1993) and cell signalling (Cerbón et al., 2005). In addition, DGKsregulate both inositol phosphate and phosphatidylinositol signallingthrough the recycling of DAG, and activation of protein kinase C(PKC; also known as PRKC) enzymes, which require DAG binding.Owing to the numerous roles of DAG, and the ten DGK isoforms inhumans, few studies have considered a role for DGK-dependentsignalling in disease states. The exceptions to this are severalknockout studies that have linked various DGK isoforms withepilepsy [DGKβ (Ishisaka et al., 2103), DGKδ (Leach et al., 2007)and DGKε (Rodriguez de Turco et al., 2001)] and bipolar disorder[DGKβ (Kakefuda et al., 2010; Squassina et al., 2009) and DGKη(Baum et al., 2007; Moya et al., 2010)]. In bipolar disorder, PKClevels are also elevated during manic episodes (Friedman et al.,1993; Wang and Friedman, 1996), and lithium, a common bipolardisorder treatment, is known to increase DAG levels (Brami et al.,1993; Drummond and Raeburn, 1984). Surprisingly, no studies, toour knowledge, have investigated DGK-dependent signalling asan overlapping process relating to both epilepsy and bipolardisorder treatment.

One widely used treatment for both epilepsy and bipolar disorderis the eight-carbon, branched-chain fatty acid, valproic acid (VPA).It was discovered accidentally in 1963 as a treatment for epilepsy(Meunier et al., 1963), and is also used in the treatment of migraine(Pryse-Phillips et al., 1997), with potential for the treatment ofcancer (Arce et al., 2006) and HIV (Lehrman et al., 2005). In aclinical context, it is used at a plasma concentration of 0.3-0.6 mM(DSMIV, 2000), but it is teratogenic, leading to enhanced likelihoodof birth defects if taken during pregnancy, thus limiting its clinicaluse (Jentink et al., 2010). To develop novel compounds lacking theside effects of VPA, but with a common therapeutic mechanism,many studies have sought to identify its cellular effects and potentialtargets. This research has been complicated by a broad range ofeffects caused by VPA, linked to varying therapeutic (or adverse)mechanisms (Terbach and Williams, 2009). One approach tosimplify this research is to use a tractable model system, such as thesimple biomedical model Dictyostelium discoideum.

Dictyostelium has been used to investigate complex cellularmechanisms of both bioactive natural products and drugs,including VPA. This pharmacogenetics research has employedDictyostelium, because its haploid genome enables the rapididentification and ablation of genes encoding proteins that controlsensitivity to a drug in growth or acute cell behaviour responses,thus implicating either the identified protein or the pathway thatit regulates in the action of a drug. In such an approach, the mutantcell line might also show reduced sensitivity to the effects of thecompound onmulticellular development, thus conferring resistance toReceived 10 May 2018; Accepted 3 July 2018

Centre for Biomedical Sciences, School of Biological Sciences, Royal HollowayUniversity of London, Egham TW20 0EX, UK.

*Author for correspondence (robin.williams@rhul.ac.uk).

R.S.B.W., 0000-0002-9826-6020

This is an Open Access article distributed under the terms of the Creative Commons AttributionLicense (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use,distribution and reproduction in any medium provided that the original work is properly attributed.

1

© 2018. Published by The Company of Biologists Ltd | Disease Models & Mechanisms (2018) 11, dmm035600. doi:10.1242/dmm.035600

Disea

seModels&Mechan

isms

the drug-dependent block in fruiting body formation. Dictyosteliumprovides an excellent system to investigate the developmentaleffects of VPA, because its well-defined process of cell migration,coalescence and differentiation to form multicellular fruiting bodiesis sensitive to VPA (Boeckeler et al., 2006; Williams et al., 2002).It is important to note that Dictyostelium provides the ability toassess these behaviours as mutually exclusive cell functions,controlled by groups of both independent and common proteins,yielding a considerable advantage over other model systems.Dictyostelium was the first model to show VPA-dependent effectson phospholipid signalling (Chang et al., 2012; Xu et al., 2007)relating to seizure control (Chang et al., 2013, 2014, 2015, 2016),and on inositol phosphate signalling (Eickholt et al., 2005; Frejet al., 2016; Shimshoni et al., 2007; Williams et al., 2002) relatingto bipolar disorder treatment. These studies have shown that VPAreduces phosphoinositide levels in a time- and dose-dependentmanner (Chang et al., 2012; Xu et al., 2007), independent ofphosphatidylinositol-3-kinase activity, inositol recycling andde novo inositol biosynthesis. In Dictyostelium, VPA has also beenfound to increase the levels of phospholipids, phosphatidylcholine(PC) and phosphatidylethanolamine (PE) (Elphick et al., 2012),implicating the phosphoinositide salvage pathway as a target pathwayof the drug. In both epilepsy and bipolar disorder, Dictyostelium hasbeen used to identify mechanisms underlying the therapeutic role ofVPA and also novel compounds for seizure control that have beenvalidated by confirmatory experiments in both in vitro and in vivomammalian models (Chang et al., 2012, 2014, 2015; Williams et al.,2002). These various studies suggest thatDictyostelium could providevaluable insights into the mechanism of VPA in therapeutic function,relating to both epilepsy and bipolar disorder treatment.Here, we investigate a role for DGKA in attenuating the effect of

VPA inDictyostelium. The use ofDictyostelium as a model providesthe opportunity to ablate the single dgkA gene, creating a stableisogenic cell line lacking all DGK activity, and enabling thesubsequent quantification of acute effects of VPA and congenerson cell behaviour and development in the absence of this enzyme(Chang et al., 2012; Cocorocchio et al., 2016, 2018; Xu et al.,2007). Previous studies show that ablation of the dgkA gene givesrise to altered development, where cells were able to form small,but relatively normal, fruiting bodies (Egelhoff et al., 1993), but thisstudy was complicated by the proposal that the enzyme functioned asa myosin II kinase, whereas it was later shown to provide DGKactivity (De La Roche et al., 2002; Ostroski et al., 2005). We showthat loss of DGKA, a proposed DGKθ orthologue, results in asignificant decrease in the potency of VPA in triggering acute cellbehaviour responses and in development. We further show that lossof DGKA reduces sensitivity to the inhibitory effects of a range ofother potential epilepsy treatments and a structurally dissimilarbipolar disorder treatment, lithium. These results suggest that DGKAmight regulate the cellular effects of VPA, relating to treatments forboth epilepsy and bipolar disorder, and newly identified alternativesto VPA could function through the same molecular mechanism.

RESULTSDictyostelium DGKA represents the origins of the family ofmammalian DGK enzymesBecause DGKs catalyse the first step in the DAG salvage pathway(Fig. 1A), we initially investigated homology between the singleDictyosteliumDGKA protein and the tenmembers of the mammalianDGK family of enzymes (Fig. 1B-F). The Dictyostelium DGKAprotein shows a conserved domain structure generally found in theten human isoforms (Fig. S1), with three putative N-terminal

phorbol-ester/DAG-type 1 zinc finger domains and a DAG-kinasecatalytic domain (Fig. 1B). The catalytic site is highly conservedbetween the human and Dictyostelium proteins and broadlyconserved throughout other kingdoms (Fig. 1C). In addition, theDictyostelium enzyme retains a conserved glycine (G262) that,when mutated, abolishes enzymatic activity in COS-7 cells withouteffect on translocation to membrane (Los et al., 2004; van Baal et al.,2005) and two prolines (P245 and P246) necessary for full enzymaticactivity (Los et al., 2004). The Dictyostelium protein also containsthree cysteine residues necessary for membrane association ofthe mammalian protein (van Baal et al., 2005). Although theDictyostelium protein shows slightly stronger homology to thehuman DGKθ (De La Roche et al., 2002), the enzyme is likelyto function as a single, generic DGK owing to high sequenceconservation between Dictyostelium DGKA and all ten humanisoforms (Fig. S2). The cladistics tree supports evolutionaryconservation of the DGKs with tight clusters of DGK types 1-5 inorganisms with multiple isoforms (Fig. S3).

To analyse the role of the Dictyostelium DGKA enzyme, weablated the enzyme in an isogenic cell line. The dgkA− mutantwas produced by homologous recombination with a knockoutcassette, leading to the loss of 1072 bp from the central region ofthe encoding gene including the catalytic site (Fig. 1F; Fig. S4).Loss of dgkA gene expression was confirmed by reversetranscription polymerase chain reaction (RT-PCR) (Fig. 1E).The dgkA− mutant was then used to investigate sensitivity orresistance to VPA exposure in both acute cell behaviour assaysand in development.

Investigating a role for DGKA in regulating the acute effect ofVPA on Dictyostelium cell behaviourBecause we have previously shown that VPA acutely blocksDictyostelium cell behaviour (Boeckeler et al., 2006; Williamset al., 2002), we then assessed whether this effect was attenuatedfollowing ablation of DGKA. This was carried out by monitoringcell behaviour in wild-type and dgkA− cells prior to treatment, toidentify potential behavioural differences owing to loss of DGKA,and following exposure to VPA, to identify altered sensitivity to thebehavioural changes caused by VPA. The logic to this approach isthat loss of VPA sensitivity in the mutant would implicate a role forDGKA-related signalling or protein function in the effect of VPA onacute cell behaviour in this model. In these experiments, cells weretreated with pulsatile cyclic adenosine monophosphate (cAMP) tomimic the natural signalling pathways Dictyostelium uses to initiatethe process of changing from a unicellular lifestyle to a multicellularphase, beginning with aggregation. These cells, in early developmentwith rapid cell movement, were then exposed to VPA for 10 min tomonitor changes in cell behaviour (Fig. 2A). As an initial approach,we treated cells with VPA at 0.5 mM to show that wild-type cellshad halted in movement and formed a circular shape (Fig. 2B),whereas dgkA− cells continued to move and remained amoeboid inshape (Fig. 2C), after VPA treatment. We continued this analysisby employing a recently developed assay for monitoring acutechanges in cell behaviour following drug or compound exposure,enabling quantification of cell responses (Cocorocchio et al.,2016, 2018). Here, cell behaviour was recorded using time-lapsemicroscopy prior to and following the addition of VPA, andcomputer-generated cell outlines were analysed to describechanges in cell shape (circularity) (Fig. 2D-F), membraneprotrusion (Fig. 2G) and motility (Fig. 2H). In the absence ofVPA treatment, wild-type and dgkA– cells showed similarcircularity (0.71±0.02 and 0.65±0.04, respectively), protrusions

2

RESEARCH ARTICLE Disease Models & Mechanisms (2018) 11, dmm035600. doi:10.1242/dmm.035600

Disea

seModels&Mechan

isms

Fig. 1. See next page for legend.

3

RESEARCH ARTICLE Disease Models & Mechanisms (2018) 11, dmm035600. doi:10.1242/dmm.035600

Disea

seModels&Mechan

isms

(8.4±0.1 and 9.2±0.1, respectively) and motility (0.012±0.001and 0.017±0.001 µm/s, respectively, suggesting that thesecharacteristics and behaviours are not grossly affected by loss ofDGKA. Following treatment, in all cell behaviour characteristicsmonitored, dgkA− cells showed reduced sensitivity to the acuteeffects of VPA, with around a twofold reduction in IC50 valuesrelating to membrane protrusions and motility compared with thoseof wild-type cells (Fig. 2G,H). This was also seen when comparingthe average protrusion number in the absence and presence of VPA(0.5 mM), with wild-type cells showing a 5.6-fold reduction, anddgkA− cells showing a 1.5-fold reduction, following treatment.These results suggest that, following the deletion of dgkA, theeffect of VPA on acute cell behaviour is reduced, implicatingeither a role for DGKA-dependent signalling or the DGKA proteinin regulating the action of VPA in this model.

Rescuing the DGKA-dependent VPA sensitivity in acutecell behaviourWe next investigated whether the reduced VPA sensitivity in acutecell behaviour shown for dgkA− cells was through loss of DGKA, byexpression of a tagged DGKA-GFP in dgkA− cells (to producedgkA−/+). In these experiments, the expressed DGKA-GFP proteinwas of the expected size (130 kDa, Fig. 3A), and was localisedwithin the cytosol (Fig. 3B), consistent with the localisation of themammalian protein (van Baal et al., 2005). Acute cell behaviouranalysis of the resulting dgkA−/+ cells showed reinstated VPAsensitivity in relation to circularity, protrusion formation andmotility (Fig. 3C-E; Fig. S5). These results further support a rolefor DGKA-dependent signalling or the DGKA protein in regulatingthe sensitivity of cells to the effects of VPA.

Investigating a role for DGKA in regulating the effect of VPAon Dictyostelium developmentIn Dictyostelium, VPA has been shown to block development,where starving cells aggregate to form a mound and then developinto a mature fruiting body over a 24 h period (Boeckeler et al.,2006; Williams et al., 2002) (Fig. 4A). Because the development ofdgkA− cells produces morphologically normal fruiting bodies, withdistinct stalk and spore heads (Egelhoff et al., 1993), albeit ofreduced size, we can therefore analyse the sensitivity of this mutantto the VPA-dependent block in fruiting body formation. Loss ofVPA sensitivity in the mutant would suggest that VPA functions toblock development inDictyostelium through a mechanism dependentupon DGKA-related signalling or protein function. We thereforecompared fruiting bodymorphology, in wild-type, dgkA− and dgkA−/+

cells, starved on nitrocellulose filters (Fig. 4B-E), in the absence andpresence of a range of VPA concentrations (0.3 mM, 0.5 mM and1 mM) (Fig. 4C-E). The therapeutic (plasma) concentrations ofVPA in patients is ∼0.4 mM (DSMIV, 2000). In the absence ofVPA, wild-type, dgkA− and dgkA−/+ cells formed mature fruitingbodies, containing a stalk and a spore head, although the overallsize of the dgkA− fruiting body was reduced compared with that ofwild-type cells, consistent with previous reports (Abu-Elneel et al.,1996; De La Roche et al., 2002) (Fig. 4B). Fruiting body size wasrestored in the dgkA−/+ rescue cell line compared with that ofwild-type cells, consistent with complementing the loss of dgkA(Fig. 4B). In the presence of VPA at 0.3 mM (Fig. 4C) and 0.5 mM(Fig. 4D), wild-type fruiting body morphology was severelyaffected, reducing the aggregate size and stopping development atthe mound stage at 1 mM (Fig. 4E), consistent with earlier studies(Boeckeler et al., 2006; Williams et al., 2002). In contrast, dgkA−

cells were able to develop into mature fruiting bodies at 0.3 mM(Fig. 4C) and 0.5 mM (Fig. 4D) that were indistinguishable fromthose formed in the absence of VPA, although development wasalso arrested at the mound stage in the presence of 1 mM VPA(Fig. 4E). VPA sensitivity was restored in the dgkA−/+ cell line;fruiting body development was halted with a similar effect to thatshown for wild-type cells (Fig. 4C,D). These results suggest that thedevelopmental effect of VPA on Dictyostelium is at least partiallydependent upon the presence of the DGKA protein.

Investigating a role for DGKA in regulating the effect ofother epilepsy and bipolar disorder treatments onDictyostelium developmentResearch into the molecular mechanisms of VPA has yet to identifya common mechanism for both epilepsy and bipolar disordertreatments. We therefore investigated the role of the DictyosteliumDGKA protein in regulating the effect of VPA relating to thetreatment of both epilepsy and bipolar disorders, using a range ofepilepsy treatments shown to be active in Dictyostelium and inmammalian epilepsy models (Chang et al., 2012, 2013, 2014, 2015,2016) and the commonly used bipolar disorder treatment, lithium(Williams et al., 1999, 2002). The logic to this approach was that ifthese compounds modulate the same signalling pathway or have thesame protein target, cells in which this pathway is blocked or has amissing target will show reduced responsiveness to each compound.This analysis was limited to assessing the developmental effectsof these compounds, because lithium does not show acute effectson cell behaviour in Dictyostelium (King et al., 2009). In theseexperiments, cells were again starved on nitrocellulose filter for 24 hto initiate development, in the presence of a range of compoundsat concentrations that inhibited development and fruiting bodyformation in wild-type cells, and the effects were comparedbetween wild-type and dgkA− cells (Fig. 5A). We first investigateddecanoic acid, a compound that attenuated phosphoinositidesignalling in Dictyostelium and provides a key component ofmedium-chain ketogenic diet for the treatment of drug-resistantepilepsy, with efficacy against seizure activity in an ex vivohippocampal slice model (Augustin et al., 2018; Chang et al., 2016).This compound arrested wild-type Dictyostelium development at themound stage (at 1.65 mM), whereas dgkA− cells were resistant to thecompounds and developed into mature fruiting bodies (Fig. 5B). In asimilar manner, the branched chain fatty acid 4-ethyloctanoic acidshows activity in both Dictyostelium phosphoinositide inhibition, inex vivo hippocampal seizure models and in neuroprotection (Changet al., 2012, 2013, 2015). This compound also blockedDictyosteliumdevelopment at the mound stage in wild-type cells (at 0.5 mM), but

Fig. 1. Diacylglycerol kinase (DGK), within the phosphoinositide salvagepathway, is highly conserved across kingdoms and ablated inDictyostelium. (A) Schematic of the phosphoinositide salvage pathway,showing the recycling of diacylglycerol (DAG) to phosphatidic acid (PA) fromthe extracellular membrane to the endoplasmic reticulum, via cytidinediphosphate (CDP)-DAG to phosphoinositide (PI), and then back to theextracellular membrane and phosphorylation to various phosphoinositides(PIP, PIP2). Phospholipase C (PLC) regenerates DAG and inositoltrisphosphate. (B) Schematic representation of Dictyostelium DGKA andhuman DGKθ, highlighting DAG binding domains and the catalytic domain.(C) Conservation amino acids within regions of the DGK catalytic domain(pink), with ‘.’ representing an intervening region. (D) Schematic ofhomologous recombination used to produce the dgkA knockout fragment,with arrows representing the primers used to create the two fragments whichwere inserted into the knockout vector, and the area in yellow indicating theregion deleted. (E,F) A DGKA null mutant was created, with the deletion of thecore catalytic domain of the protein (E). Arrows represent primers used todemonstrate loss of expression (RT-PCR) in the mutant (F).

4

RESEARCH ARTICLE Disease Models & Mechanisms (2018) 11, dmm035600. doi:10.1242/dmm.035600

Disea

seModels&Mechan

isms

did not block the development of dgkA− cells (Fig. 5C). As a relatedcontrol, a chemical with similar structure to these medium-chain fattyacids, 2-methylhexanoic acid, that showed no effect onDictyosteliumphosphoinositide regulation (Chang et al., 2012), and no effect inseizure control (Chang et al., 2013), was used. For this compound,both wild-type and dgkA− cells showed similar sensitivity duringdevelopment (shown for 0.5 mM), with both cell types showingdevelopment arrest at the mound stage (Fig. 5D). These resultssuggest that a range of compounds, identified in Dictyosteliumthrough the inhibition of phosphoinositide regulation and providingstrong seizure control inmammalianmodels, have reduced potency in

dgkA− cells, suggesting that the effect of these compounds isdependent upon the presence of DGKA.

We continued the analysis of dgkA− cells in response to the bipolardisorder treatment lithium. In Dictyostelium, lithium provides awell-described effect through inhibiting development related to ablock in inositol phosphate signalling (Williams et al., 1999,2002). To test for resistance to the developmental effect of lithiumfollowing loss of dgkA, development assays were repeated usinglithium chloride (at 8 mM). In wild-type cells, development wasblocked after aggregation, in early development, at first fingerstage (Fig. 5E), as previously reported (Williams et al., 1999, 2002).

Fig. 2. Loss of DGKA confers reducedsensitivity to the acute effects of VPA.(A) Schematic of the acute cell behaviourexperimental procedure. Cells wereinduced to early development by pulsatilecAMP before being visualised under amicroscope, and cell movement wasrecorded in the absence and presence ofVPA for 250 s and 750 s, respectively.(B,C) Brightfield images before and afteraddition of 0.5 mM VPA to WT (B) anddgkA− cells (C). Scale bars: 4 µm.(D,E) Quantification of cell circularity inthe absence and presence of a range ofVPA concentrations (0.01 mM to 0.7 mM) inWT (D) and dgkA− (E) cells, with datapresented as mean±s.e.m. (n=30 cells).(F-I) Schematic of the quantitativemeasurements taken during the assay (F).Black dashed line arrows represent themeasurement being taken and the resultingsecondary plots for data normalised tocontrol conditions of each cell type (100%)for circularity (G), membrane protrusions(H) andmotility (I) for WT (black circles) anddgkA− (blue squares) cells. A Kruskal–Wallis with Dunn’s post hoc test was used tocompare WT and dgkA− cell lines. Therewere significant differences in WT anddgkA− cell displacement, circularity,protrusions and motility, as indicated. IC50,half maximal inhibitory concentration; CI,95% confidence interval. **P≤0.01.

5

RESEARCH ARTICLE Disease Models & Mechanisms (2018) 11, dmm035600. doi:10.1242/dmm.035600

Disea

seModels&Mechan

isms

In contrast, dgkA− cells were able to develop into mature fruitingbodies, showing distinct stalks and spore heads, followinglithium treatment. These results suggest that the developmentaleffect of lithium chloride occurs through a mechanism dependentupon the presence of DGKA.

Investigating a role for DGKA as a regulator of the effect ofepilepsy and bipolar treatments on DAG levelsWe then went on to quantify the molecular changes inDictyostelium DAG levels in the presence and absence ofDGKA, and following treatment with VPA, lithium and othercompounds, using DAG-specific enzyme-linked immunosorbentassay (ELISA). In these experiments, cells were prepared in thesame way as for acute cell behaviour experiments, with pulsatilecAMP to induce early development. Results from this approachsurprisingly showed that, in the absence of treatment, dgkA− cellsshowed lower levels of DAG compared with wild-type cells, whichwere only partially restored on reintroduction of the enzyme withindgkA−/+ cells (Fig. 6A). We then treated cells with VPA atconcentrations in which dgkA− cells were resistant to treatment. Inthe presence of VPA (0.3 mM and 0.5 mM), DAG levelssignificantly increased in wild-type and dgkA−/+ cells comparedwith dgkA− cells (Fig. 6A). In a similar fashion, cells treated withlithium chloride (8 mM) also showed significantly increased DAGlevels. We also analysed the two unrelated structures withdemonstrated seizure control activity, again at concentrationsshown to block development in wild-type cells [decanoic acid(1.65 mM) and 4-ethyloctanoic acid (0.5 mM) (Chang et al., 2012,2013, 2015)], and found that these compounds elevated DAGlevels significantly above those in control (untreated wild-type)cells (Fig. 6B). These results suggest that, in Dictyostelium, loss ofDGKA reduces DAG levels, and treatment of wild-type cells withVPA, lithium and new compounds, gives rise to a common effect ofincreasing DAG levels.

DISCUSSIONMechanistic insights into the biochemical basis of therapeutictreatments provides opportunities to increase our understandingof signalling pathways underlying these disorders and forthe development of new treatments. In this study, we employed thesimple biomedical model system Dictyostelium to investigate thecellular mechanisms of the epilepsy and bipolar disorder treatment,VPA, through the key enzyme, DGKA, which is responsiblefor the phosphorylation of DAG to PA. In mammalian models,DAG is a second messenger derived from phosphatidylinositol4,5-bisphosphate (PIP2) through excitatory neurotransmitteractivation of metabotropic glutamate receptors to activate PLC(Rodriguez de Turco et al., 2001). DAG has numerous rolesincluding in the activation of PKCs (Newton, 1997), transientreceptors channels (Lucas et al., 2003) and the endocannabinoidsystem (Williams et al., 2003). The roles of DGK thus includeturning off DAG-dependent signalling, in addition to acting atthe start of the phosphoinositide salvage pathway for the recyclingof phosphoinositide signalling. By using Dictyostelium as a model,we were able to ablate all enzyme activity through loss of oneprotein, likely to represent the evolutionary origins of the family ofmammalian DGK enzymes, with conserved structure and keyamino acids. Consistent with the viability of yeast mutants, theDictyostelium enzyme is not vital (Han et al., 2008a,b), althoughthe encoded protein more closely resembles that found in highereukaryotes rather than that in other single-cell models, such asyeast and bacteria (Cai et al., 2009). We show that loss of theDictyostelium dgkA gene significantly reduced sensitivity to theacute effects of VPA on changes in cell behaviour and inmulticellular development, suggesting that the effect of VPA oncellular function is at least partially dependent upon DGKA inDictyostelium. It is therefore likely that VPA acts to change acellular function in this model to cause acute and developmentaleffects, where these changes are dependent on the presence ofDGKA, thus implicating either perturbedDGKA-related signalling orDGKA activity as a cellular mechanism for VPA action. We furtherused Dictyostelium development to analyse a range of compoundsassociated with both epilepsy and bipolar disorder treatments, toshow that loss of DGKA decreases sensitivity to both treatments. At abiochemical level, loss of DGKA appears to reduce DAG levels, andexposure to both epilepsy and bipolar disorder treatments increasesDAG levels. Removal of DGKA therefore counteracts the effect ofthese treatments in increasing DAG levels. These studies thus suggestthat, in Dictyostelium, DGKA might regulate a cellular mechanismcommon for drugs used to treat both epilepsy and bipolar disorder.

In this study, we used the dgkA− cell line to assess a role forDGKA in the biochemical activity of VPA. This approach isunusual, as we have removed all DGK activity in these cells. Ourearlier studies have shown that VPA acutely blocks cell behaviour,with a concomitant reduction in phosphoinositide levels (Changet al., 2012; Xu et al., 2007), and reduces inositol phosphate levels(Williams et al., 1999). Here, we extend these studies, usingidentical conditions, to show that loss of DGKA reduces the acuteeffect of VPA on cell behaviour, and blocks the VPA-dependentincrease in DAG levels in wild-type cells. The most likely rationalefor this resistant phenotype is that of molecular changes in thedgkA− mutant, either through loss of the enzyme as a direct targetfor VPA or by alterations in cell signalling regulated by DGKA toovercome the cellular effect of VPA in this model. Further studiesare necessary to investigate a direct role for VPA in inhibitingDGKA, and to examine the VPA-dependent regulation of thephosphoinositide salvage pathway (Fakas et al., 2011; Rodriguez

Fig. 3. VPA sensitivity is restored by expression of GFP-DGKA in thedgkA null mutant. (A) Western blot showing GFP-DGKA expressionwith loading control (streptavidin) for WT, dgkA− and dgkA−/+ cell lines.(B) Fluorescence image of GFP-DGKA localisation in dgkA−/+ cells.Scale bar: 10 µm. (C-E) In behavioural assays, cells were induced to earlydevelopment by pulsatile cAMP, before cell behaviour was recorded in theabsence and presence of VPA for WT (black circles), dgkA− (blue squares)and dgkA−/+ (pink triangles) cell response to 0.01-0.7 mM VPA for circularity(C), protrusions (D) and motility (E). Data are from n=30 cells, with secondaryplots using data normalised to control conditions of each cell type (100%).A Kruskal–Wallis with Dunn’s post hoc test was used to compare the three celllines, for which there were significant differences between WT and dgkA−

(black asterisks), and dgkA− and dgkA−/+ (blue asterisks), cells as indicated.*P≤0.05, **P≤0.01, ***P<0.001 and ****P≤0.0001.

6

RESEARCH ARTICLE Disease Models & Mechanisms (2018) 11, dmm035600. doi:10.1242/dmm.035600

Disea

seModels&Mechan

isms

de Turco et al., 2001; Walsh et al., 1995), phosphoinositide levels(Chang et al., 2012; Xu et al., 2007) and inositol phosphates(Williams et al., 1999, 2002).This study provides the first analysis of Dictyostelium DAG

levels. We show that nanomolar levels of DAG in unstimulatedDictyostelium cells (at ∼0.4 ng/107 cells) are similar to thosereported for human tissue (110 nmol/g), pigs (using porcine aorticendothelia cells) (13.7 nmol/107 cells) and mice fibroblast Swiss3T3 cells (1.4 nmol/107 cells) (Pettitt et al., 1997; Szendroedi et al.,2014). Although Dictyostelium DAG levels were shown to bevariable, consistent with those shown in in vivo mouse modelstudies (Rodriguez de Turco et al., 2001), the dgkA−mutant showeda significant and large reduction in DAG levels. This result wasunexpected, because elevated levels of DAG would be expectedfollowing removal of DGKA. Two potential mechanisms can besuggested for this result. First, the cellular response VPA treatment(or dgkA deletion) might be to upregulate the activity of theKennedy pathway, through which DAG is recycled to PA throughPE and PC to form CDP-DAG (Gibellini and Smith, 2010). Thistheory appears likely, because VPA treatment inDictyostelium leadsto the accumulation of nonpolar lipids in lipid droplets typicallyconsisting of DAG and triacylglycerol (TAG) species (Kalantariet al., 2010), and includes PE and PC (Elphick et al., 2012). Thus,enhanced activation of the Kennedy pathway might be induced byacute exposure to VPA or to a chronic block in DGKA activity(through gene ablation). A second potential mechanism could simply

involve the stabilisation or protection of DAG levels through bindingto DGKA, thus loss of the protein (as distinct to inhibition of theenzyme) leads to the reduction in DAG levels. It must also beacknowledged that VPA-dependent changes in DAG, althoughproviding a useful marker, could indeed arise through effects on avariety of other metabolites to give rise to this effect. However,through either direct or indirect processes, treated Dictyostelium cellsshowed a significantly altered DAG level, suggesting that DAG-dependent signalling might provide an important target pathway forVPA, lithium and other compounds.

DGK-related activity might play an important role in epilepsy.A range of studies have demonstrated a role for phosphoinositidesignalling in both seizure susceptibility and progression (Backmanet al., 2001; Chang et al., 2014; Shinoda et al., 2004; Vanhaesebroecket al., 2012), and this signalling is dependent upon DAG recyclingthrough DGK activity. In addition, mutations in several DGKisoforms have been linked with increased seizure activity (Ishisakaet al., 2103; Leach et al., 2007; Rodriguez de Turco et al., 2001).The data provided here suggest that, in Dictyostelium, one action ofVPA is related to the cellular function of DGK, because ablation ofDGKA attenuates the effect of VPA. Further studies will be necessaryto investigate a VPA-dependent effect on seizure control throughregulation of DGK-related signalling. It is also interesting to note thatprevious studies have shown a reduction in dendritic spines inepilepsy patients (Muller et al., 1993; Scheibel et al., 1974), thatDGKβ controls dendritic spine outgrowth and maturation (Hozumi

Fig. 4. Loss of DGKA confers reduced sensitivity tothe chronic effects of VPA on development.(A) Schematic of the development assay experimentalprocedure. Cells were shaken and placed on anitrocellulose filter in the presence and absence of VPAfor 24 h to enable development before being visualisedunder a dissection microscope. (B-E) Overhead and sideshots of developmental morphologies (B) under controlconditions, and in the presence of VPA at (C) 0.3 mM,(D) 0.5 mM and (E) 1 mM, in WT (left column), dgkA−

(middle column) and dgkA−/+ (right column) cells.Data are representative of triplicate independentexperiments. Scale bars: 0.5 mm (overhead view) and0.1 mm (side view).

7

RESEARCH ARTICLE Disease Models & Mechanisms (2018) 11, dmm035600. doi:10.1242/dmm.035600

Disea

seModels&Mechan

isms

et al., 2009), and that VPA treatment increases dendritic spineformation in mouse models (Yang et al., 2016). Together, these dataprovide a view consistent with a role for VPA in regulating cellfunction through alterations in DGK-dependent signalling pathwaysrelated to seizure activity and neurological effects.We also sought to investigate a role for Dictyostelium DGKA in

attenuating the effects of other potential epilepsy treatments. Here, weemployed a range of novel compounds, identified inDictyostelium toprovide a similar cellular function to VPA, that show seizure controleffects in mammalian models (Chang et al., 2016, 2012, 2013, 2014,2015). Importantly, these compounds lack the potential adverseeffect of histone deacetylase inhibition at therapeutically relevantconcentrations, and this effect has been linked to teratogenicityof VPA and resulting birth defects (Jentink et al., 2010). Usingthe Dictyostelium development assay to screen compounds, weshowed that loss of DGKA also provided reduced sensitivityto decanoic acid [the therapeutically active constituent of themedium-chain-triglyceride ketogenic diet (Augustin et al., 2018)]and to 4-ethyloctanoic acid. In contrast, 2-methyhexanoic acid, arelated compound with low activity in Dictyostelium withoutseizure control activity (Chang et al., 2012, 2013), did not showdifferential effects between the two cell lines. Furthermore,decanoic acid and 4-ethyloctanoic acid treatment also elevatedDAG levels at concentrations shown to block development in

wild-type cells, although the large increase in DAG levels shownhere should be investigated further to exclude potentialnonspecific effects on the assay. These studies therefore proposea role for DGKA in regulating the cellular effects of other potentialepilepsy treatments, beyond that of VPA.

DGK-regulated signalling also provides an attractive therapeutictarget for bipolar disorder. Data provided here demonstrate that thepreviously reported effect of the bipolar disorder treatment lithiumon Dictyostelium development is overcome by loss of DGKA(Boeckeler et al., 2006; Williams et al., 2002). Earlier studies inDictyostelium have shown that both VPA and lithium attenuateinositol phosphate signalling as a mechanism for bipolar disordertreatments (Eickholt et al., 2005; Shimshoni et al., 2007; Williamset al., 2002, 1999). The reduction in phosphoinositide signallingcaused by VPA, through attenuation of DGKA, might thereforeunderlie this change in inositol phosphate levels as a potentialmechanism for bipolar disorder treatment. This association is furthersupported by the role of DAG in bipolar disorder-associatedsignalling. In mammalian models, DGK competes with PKC forbinding to DAG, thereby regulating PKC activity, and dysregulationof the PKC signalling pathway has been widely demonstrated inbipolar disorder patient populations and following therapeutictreatment (Sakai and Sakane, 2012; Saxena et al., 2017). Numerousstudies have identified upregulated PKC activity in bipolar disorder

Fig. 5. Loss of DGKA confers reduced sensitivityto the chronic developmental effects of arange of novel antiepileptic treatments and thebipolar disorder treatment, lithium chloride.Overhead and side view images of development inWT (left column), dgkA− (middle column) anddgkA−/+ (right column) cells (A) under controlconditions and in the presence of (B) decanoic acid(1.65 mM), (C) 4-ethyloctanoic acid (0.5 mM),(D) 2-methylhexanoic acid (0.5 mM) and (E) lithiumchloride (8 mM). Data are representative of triplicateindependent experiments. Scale bars: 0.5 mm(overhead view) and 0.1 mm (side view).

8

RESEARCH ARTICLE Disease Models & Mechanisms (2018) 11, dmm035600. doi:10.1242/dmm.035600

Disea

seModels&Mechan

isms

patients (Friedman et al., 1993; Wang and Friedman, 1996), withthis activity being reduced by both lithium (Friedman et al., 1993;Wang and Friedman, 1989) and VPA (Ramadan et al., 2011;Watson et al., 1998) in animal models and patient studies (Katselet al., 2008; Soares and Mallinger, 1997; Wang et al., 2001).Associations with bipolar disorder have been made with DGKβand DGKη (Baum et al., 2007; Kakefuda et al., 2010; Moya et al.,2010; Squassina et al., 2009), with lithium treatment resulting inaccumulation of DAG in a range of models (Brami et al., 1993;Drummond and Raeburn, 1984), independent of phospholipase D(Nilssen et al., 2005). DGK also interacts with NR5A1, which in turnbinds to phosphatidylinositol headgroups, in particular PI(3,4)P2 andPI(3,4,5)P3, providing a mechanism of DGK-dependent regulationof phosphoinositides beyond that of DAG recycling (Blind et al.,2014). Thus a VPA- and lithium-dependent effect on signallingcomponents controlled by, or involving, DGK activity could providea biochemical mechanism underlying the treatment of patients withbipolar disorder.Developing new treatments for bipolar disorder has remained a

difficult prospect owing to a lack of clarity regarding a therapeuticmechanism of current drugs. One approach to overcome this isbased upon the observation that a range of antiepileptic drugs arealso effective in the treatment of bipolar disorder (Rogawski andLoscher, 2004). In earlier studies, we have successfully translated acommon molecular mechanism relating to both epilepsy and bipolardisorder treatments from Dictyostelium to mammalian neurons,demonstrating that it is possible to model bipolar disorder drugmechanisms in Dictyostelium (Williams et al., 2002, 1999). In ourcurrent study, we show that novel compounds (decanoic acid and4-ethyloctanoic acid) function through modulating a DGKA-relatedpathway commonly affected by lithium and VPA, suggesting thatthese compounds should be investigated as new treatments forbipolar disorder as well as epilepsy.In summary, VPA is a widely used treatment for both epilepsy

and bipolar disorder, without a clear mechanism of action and withsignificant side effects. Here, we propose that signalling regulatedby DGK could provide a key role in the mechanisms of actionfor VPA (Fig. 7), in addition to other proposed therapeutic

compounds that could provide treatments for both epilepsy andbipolar disorder.

MATERIALS AND METHODSReagentsAll compounds were purchased from Sigma-Aldrich (Dorset, UK) unlessotherwise stated. Axenic medium, SM agar, phosphate buffer (KK2) andblasticidin were obtained from ForMedium (Norfolk, UK). Penicillin-streptomycin was purchased from Gibco (Paisley, UK), decanoic acid wasfrom Alfa Aesar (Massachusetts, USA) and all enzymes were purchasedfrom Thermo Fisher Scientific (Hemel Hempstead, UK).

Cell culture, strains and plasmidsDictyostelium cell lines were grown at 22°C in Axenic medium whichcontained 100 µg/ml penicillin-streptomycin. dgkA− cell lines weregenerated from wild-type (Ax2) cells. Creation of the dgkA knockout

Fig. 6. Dictyostelium DGKA-dependent DAG regulation following treatment with VPA, lithium and novel antiepileptic compounds. WT, dgkA−

and dgkA−/+ cells were induced into early development by pulsatile cAMP and exposed to VPA, lithium or novel antiepileptic treatments, as indicated,for 10 min prior to ELISA quantification. (A) DAG levels in WT (black), dgkA− (blue) and dgkA−/+ (pink) cells under control conditions and in the presenceof 0.3 mM and 0.5 mM VPA and 8 mM LiCl. (B) DAG levels in WT cells in the absence (control) and presence of decanoic acid (1.65 mM) and 4-ethyloctanoicacid (0.5 mM). Data are mean±s.e.m. A Mann–Whitney test was used to compare the different cell lines and conditions, with a significant difference shownbetweenWT and dgkA− cells (**P<0.01) and between dgkA− and dgkA−/+ cells (+P<0.05). Data are representative of quadruplicate independent experimentswith duplicate technical repeats.

Fig. 7. Proposing a mechanism for VPA, linking epilepsy and bipolardisorder with the phosphoinositide salvage pathway. Phosphoinositidesignalling (pink oval) has been widely shown to be deregulated in seizuresand epilepsy, is regulated by VPA, and is dependent upon the salvagepathway involving DAG and DGKs. In addition, protein kinase C (PKC)activity is regulated by DAG, and numerous studies have described changesin PKC pathway activation in bipolar disorder studies (blue oval), in whichpatients are treated with VPA or lithium. Our data, based on themodel systemDictyostelium, suggest that loss of the DGK enzyme reduces the effect ofboth VPA and lithium, and reduces DAG levels, and that both compounds(and other potential new epilepsy treatments) function to commonly elevateDAG levels. Further pre-clinical and clinical studies are therefore needed toinvestigate DAG and related signalling as a target pathway in the treatment ofboth disorders.

9

RESEARCH ARTICLE Disease Models & Mechanisms (2018) 11, dmm035600. doi:10.1242/dmm.035600

Disea

seModels&Mechan

isms

construct was as previously described (Pakes et al., 2012). DNA wasamplified from the 5′ and 3′ regions of the dgkA gene (DDB_G0277223)(Fey et al., 2013) and cloned into the pLPBLP vector on either side of ablasticidin resistance cassette using the restriction enzymes BamHI, PstI,NcoI andHindIII. The knockout cassettewas excised from the vector throughrestriction digestion and electroporated into wild-type cells. Transformantswere selected in medium containing 10 µg/ml blasticidin, and survivingcolonies were screened by PCR for homologous integration using a genomicand vector control bands and a diagnostic band. RT-PCR (First StrandcDNA Synthesis Kit, Thermo Fisher Scientific) was performed to confirmloss of gene transcription. The dgkA−/+ rescue cell line was produced usingthe pTX-FLAG-DGK-GFP plasmid vector, which was kindly provided byT. Egelhoff (De La Roche et al., 2002).

Dictyostelium random cell movement assayAs previously described (Cocorocchio et al., 2016, 2018), wild-type andtransformant cells (1×107) were harvested from mid-log-phase-growingshaking cultures and washed in phosphate buffer, resuspended at 1×107

cells in 6 ml phosphate buffer, and pulsed with 30 nM cAMP at 6 minintervals for 5 h at 120 rpm. Cells were further washed in phosphatebuffer, resuspended in 4 ml phosphate buffer and diluted 1:10. Cells(250 µl) were placed in a Nunc Lab-Tek coverglass chamber (Thermo FisherScientific) and incubated for 15 min in order for the cells to adhere prior totime-lapse microscopy for 900 s. After a control period of 225 s, 250 µl ofdouble concentrate compound was added, and change in cell behaviourwas monitored at 30 cells per condition from at least three independentexperiments. To analyse random cell movement, the ImageJ Quimp 11bsoftware package and accompanying scripts for MATLAB analysis wereused. Individual cells were segmented and cell behaviour before and afterdrug addition were quantified using measures of circularity (where theratio of perpendicular axes through each cell provides a ratio, with ‘1’indicating a perfect circle), protrusion formation (defined within motilitymaps as regional peaks exceeding an average speed of 0.1 µm/s, andcounted automatically over a running window of 10 frames) to representprotrusive activity within short time periods, centred around discrete timepoints) and motility between each frame (local membrane velocitiesbetween frame motility maps), counted automatically over a ten-framerunning window.

Dictyostelium development assayDictyostelium wild-type and mutant cells (1×107 cells) were washed inphosphate buffer and developed on a nitrocellulose membrane filter above ahydrophobic membrane filter (both from Millipore, Watford, UK) in theabsence or presence of compounds. Filters were incubated for 24 h at 22°Cprior to imaging. Experiments were repeated over at least three independentexperiments.

DAG analysisLevels of DAG were determined using ELISA (General ELISA Kit forDiacylglycerol, E2038Ge, Amsbio, Abingdon, UK). Cells were prepared bypulsing with cAMP (as described for cell behaviour assays). DAG levelswere then measured from 5×106 cells, treated with VPA (0.3 mM–0.5 mM)or lithium chloride (8 mM) for 10 min in shaking suspension. Cells werewashed in phosphate buffer and resuspended in 100 µl phosphate buffer.The cell wall fraction was obtained from six rounds of freeze-thaw (3 mineach), centrifuged (10,000 g, 10 min, 4°C) and the pellet was resuspendedin 100 µl sample diluent from the kit. ELISA was conducted following thesupplier’s instructions. Data are derived from at least triplicate independentexperiments.

Data analysis and statisticsIn the analysis of cell behaviour, a Kruskal–Wallis analysis with Dunn’spost hoc test was employed to test for statistically significant changesbetween three independent groups (wild type, dgkA− and dgkA−/+) that werenot normally distributed. Differences were considered statisticallysignificant if the P-value was less than 0.05. In the analysis of DAG

levels, a Mann–Whitney analysis was employed to test for statisticallysignificant changes between groups, again with data that were notnormally distributed but enabling direct paired groups.

AcknowledgementsWe thank M. Walker for comments on the manuscript, T. Egelhoff for provision of theGFP-DGK-expressing plasmid, and Dictybase.org and the Dictybase Stock Centrefor provision of Dictyostelium strains and materials.

Competing interestsR.S.B.W. has a range of patents submitted regarding new treatments for epilepsyand bipolar disorder, including compounds cited in this study.

Author contributionsConceptualization: C.J.W., R.S.B.W.; Methodology: E.K., D.S., R.S.B.W.;Validation: D.S.; Formal analysis: E.K., D.S., R.S.B.W.; Investigation: E.K., D.S.,R.S.B.W.; Data curation: E.K., R.S.B.W.; Writing - original draft: E.K., R.S.B.W.;Writing - review & editing: E.K., D.S., C.J.W., R.S.B.W.; Visualization: E.K.,R.S.B.W.; Supervision: C.J.W., R.S.B.W.; Project administration: C.J.W., R.S.B.W.;Funding acquisition: R.S.B.W.

FundingThis work was supported by the National Centre for the Replacement, Refinementand Reduction of Animals in Research (NC/L001500/1).

Supplementary informationSupplementary information available online athttp://dmm.biologists.org/lookup/doi/10.1242/dmm.035600.supplemental

ReferencesAbu-Elneel, K., Karchi, M. and Ravid, S. (1996). Dictyostelium myosin II is

regulated during chemotaxis by a novel protein kinase C. J. Biol. Chem.271, 977-984.

Arce, C., Perez-Plasencia, C., Gonzalez-Fierro, A., de la Cruz-Hernandez, E.,Revilla-Vazquez, A., Chavez-Blanco, A., Trejo-Becerril, C., Perez-Cardenas, E.,Taja-Chayeb, L., Bargallo, E. et al. (2006). A proof-of-principle study ofepigenetic therapy added to neoadjuvant doxorubicin cyclophosphamide forlocally advanced breast cancer. PLoS ONE 1, e98.

Augustin, K., Khabbush, A., Williams, S., Eaton, S., Orford, M., Cross, J. H.,Heales, S. J. R., Walker, M. C. and Williams, R. S. B. (2018). Mechanisms ofaction for the medium-chain triglyceride ketogenic diet in neurological andmetabolic disorders. Lancet Neurol. 17, 84-93.

Backman, S. A., Stambolic, V., Suzuki, A., Haight, J., Elia, A., Pretorius, J.,Tsao, M.-S., Shannon, P., Bolon, B., Ivy, G. O. et al. (2001). Deletion of Pten inmouse brain causes seizures, ataxia and defects in soma size resemblingLhermitte-Duclos disease. Nat. Genet. 29, 396-403.

Banfic, H., Žizak, M., Divecha, N. and Irvine, R. F. (1993). Nucleardiacylglycerol is increased during cell proliferation in vivo. Biochem. J.290, 633-636.

Baum, A. E., Akula, N., Cabanero, M., Cardona, I., Corona, W., Klemens, B.,Schulze, T. G., Cichon, S., Rietschel, M., Nothen, M. M. et al. (2007).A genome-wide association study implicates diacylglycerol kinase eta (DGKH)and several other genes in the etiology of bipolar disorder. Mol. Psychiatry13, 197-207.

Blind, R. D., Sablin, E. P., Kuchenbecker, K. M., Chiu, H.-J., Deacon, A. M.,Das, D., Fletterick, R. J. and Ingraham, H. A. (2014). The signaling phospholipidPIP3 creates a new interaction surface on the nuclear receptor SF-1. Proc. Natl.Acad. Sci. USA 111, 15054-15059.

Boeckeler, K., Adley, K., Xu, X., Jenkins, A., Jin, T. andWilliams, R. S. B. (2006).The neuroprotective agent, valproic acid, regulates the mitogen-activated proteinkinase pathway through modulation of protein kinase A signalling in Dictyosteliumdiscoideum. Eur. J. Cell Biol. 85, 1047-1057.

Brami, B. A., Leli, U. and Hauser, G. (1993). Elevated phosphatidyl-CMP is not thesource of diacylglycerol accumulation induced by lithium in NG108-15 cells.J. Neurochem. 60, 1137-1142.

Cai, J., Abramovici, H., Gee, S. H. and Topham, M. K. (2009).Diacylglycerol kinases as sources of phosphatidic acid. Biochim. Biophys. Acta1791, 942-948.

Cerbon, J., Falcon, A., Hernandez-Luna, C. and Segura-Cobos, D. (2005).Inositol phosphoceramide synthase is a regulator of intracellular levels ofdiacylglycerol and ceramide during the G1 to S transition in Saccharomycescerevisiae. Biochem. J. 388, 169-176.

Chang, P., Orabi, B., Deranieh, R. M., Dham, M., Hoeller, O., Shimshoni, J. A.,Yagen, B., Bialer, M., Greenberg, M. L., Walker, M. C. et al. (2012).The antiepileptic drug valproic acid and other medium-chain fatty acids acutelyreduce phosphoinositide levels independently of inositol in Dictyostelium. Dis.Model. Mech. 5, 115-124.

10

RESEARCH ARTICLE Disease Models & Mechanisms (2018) 11, dmm035600. doi:10.1242/dmm.035600

Disea

seModels&Mechan

isms

Chang, P., Terbach, N., Plant, N., Chen, P. E., Walker, M. C. andWilliams, R. S. B. (2013). Seizure control by ketogenic diet-associated mediumchain fatty acids. Neuropharmacology 69, 105-114.

Chang, P., Walker, M. C. andWilliams, R. S. B. (2014). Seizure-induced reductionin PIP3 levels contributes to seizure-activity and is rescued by valproic acid.Neurobiol. Dis. 62, 296-306.

Chang, P., Zuckermann, A. M. E., Williams, S., Close, A. J., Cano-Jaimez, M.,McEvoy, J. P., Spencer, J., Walker, M. C. andWilliams, R. S. B. (2015). Seizurecontrol by derivatives of medium chain fatty acids associated with the ketogenicdiet show novel branching-point structure for enhanced potency. J. Pharmacol.Exp. Ther. 352, 43-52.

Chang, P., Augustin, K., Boddum, K., Williams, S., Sun, M., Terschak, J. A.,Hardege, J. D., Chen, P. E., Walker, M. C. and Williams, R. S. B. (2016).Seizure control by decanoic acid through direct AMPA receptor inhibition. Brain139, 431-443.

Cocorocchio, M., Ives, R., Clapham, D., Andrews, P. L. and Williams, R. S.(2016). Bitter tastant responses in the amoeba Dictyostelium correlatewith rat andhuman taste assays. ALTEX 33, 225-236.

Cocorocchio, M., Baldwin, A. J., Stewart, B., Kim, L., Harwood, A. J.,Thompson, C. R. L., Andrews, P. L. R. and Williams, R. S. B. (2018).Curcumin and derivatives function through protein phosphatase 2A and presenilinorthologues in Dictyostelium discoideum. Dis. Model. Mech. 11.

De La Roche, M. A., Smith, J. L., Rico, M., Carrasco, S., Merida, I., Licate, L.,Cote, G. P. and Egelhoff, T. T. (2002). Dictyostelium discoideum has a singlediacylglycerol kinase gene with similarity to mammalian theta isoforms. Biochem.J. 368, 809-815.

Drummond, A. H. and Raeburn, C. A. (1984). The interaction of lithium withthyrotropin-releasing hormone-stimulated lipid metabolism inGH3 pituitary tumourcells. Enhancement of stimulated 1,2-diacylglycerol formation. Biochem. J.224, 129-136.

DSMIV (2000). American Psychiatric Association: Diagnostic and Statistical Manualof Mental Disorders. Washington: American Psychiatric Association.

Egelhoff, T. T., Lee, R. J. and Spudich, J. A. (1993). Dictyostelium myosin heavychain phosphorylation sites regulate myosin filament assembly and localization invivo. Cell 75, 363-371.

Eickholt, B. J., Towers, G., Ryves, W. J., Eikel, D., Adley, K., Ylinen, L.,Chadborn, N., Harwood, A., Nau, H. and Williams, R. S. (2005). Effects ofvalproic acid derivatives on inositol trisphosphate depletion, teratogenicity, GSK-3β inhibition and viral replication: a screening approach for new bipolar disorderdrugs based on the valproic acid core structure.Mol. Pharmacol. 67, 1426–1433.

Elphick, L. M., Pawolleck, N., Guschina, I. A., Chaieb, L., Eikel, D., Nau, H.,Harwood, J. L., Plant, N. J. and Williams, R. S. B. (2012). Conserved valproic-acid-induced lipid droplet formation in Dictyostelium and human hepatocytesidentifies structurally active compounds. Dis. Model. Mech. 5, 231-240.

Fakas, S., Konstantinou, C. and Carman, G. M. (2011). DGK1-encodeddiacylglycerol kinase activity is required for phospholipid synthesis duringgrowth resumption from stationary phase in Saccharomyces cerevisiae. J. Biol.Chem. 286, 1464-1474.

Fey, P., Dodson, R. J., Basu, S. and Chisholm, R. L. (2013). One stop shop foreverything Dictyostelium: dictyBase and the Dicty Stock Center in 2012.MethodsMol. Biol. 983, 59-92.

Frej, A. D., Clark, J., Le Roy, C. I., Lilla, S., Thomason, P., Otto, G. P.,Churchill, G., Insall, R., Claus, S. P., Hawkins, P. et al. (2016). The inositol-3-phosphate synthase biosynthetic enzyme has distinct catalytic and metabolicroles. Mol. Cell Biol. 36, 1464-1479.

Friedman, E., Hoau-Yan-Wang, Levinson, D., Connell, T. A. and Singh, H.(1993). Altered platelet protein kinase C activity in bipolar affective disorder, manicepisode. Biol. Psychiatry 33, 520-525.

Gibellini, F. and Smith, T. K. (2010). The Kennedy pathway–de novo synthesisof phosphatidylethanolamine and phosphatidylcholine. IUBMB Life 62,414-428.

Han, G.-S., O’Hara, L., Carman, G. M. and Siniossoglou, S. (2008a). Anunconventional diacylglycerol kinase that regulates phospholipid synthesis andnuclear membrane growth. J. Biol. Chem. 283, 20433-20442.

Han, G.-S., O’Hara, L., Siniossoglou, S. and Carman, G. M. (2008b).Characterization of the yeast DGK1-encoded CTP-dependent diacylglycerolkinase. J. Biol. Chem. 283, 20443-20453.

Hozumi, Y., Watanabe, M., Otani, K. and Goto, K. (2009). Diacylglycerol kinasebeta promotes dendritic outgrowth and spine maturation in developinghippocampal neurons. BMC Neurosci. 10, 99.

Inoguchi, T., Xia, P., Kunisaki, M., Higashi, S., Feener, E. P. and King, G. L.(1994). Insulin’s effect on protein kinase C and diacylglycerol induced by diabetesand glucose in vascular tissues. Am. J. Physiol. 267, E369-E379.

Ishisaka, M., Tsuruma, K., Shimazawa, M., Shirai, Y., Saito, N. and Hara, H.(2013). Increased seizure susceptibility in a mouse with diacylglycerol kinase βdeficiency. Neurosci. Med. 4, 177-122.

Jentink, J., Loane, M. A., Dolk, H., Barisic, I., Garne, E., Morris, J. K. andde Jong-van den Berg, L. T.; EUROCATAntiepileptic Study Working Group.(2010). Valproic acid monotherapy in pregnancy and major congenitalmalformations. N. Engl. J. Med. 362, 2185-2193.

Kakefuda, K., Oyagi, A., Ishisaka, M., Tsuruma, K., Shimazawa, M., Yokota, K.,Shirai, Y., Horie, K., Saito, N., Takeda, J. et al. (2010). Diacylglycerol kinasebeta knockout mice exhibit lithium-sensitive behavioral abnormalities. PLoS ONE5, e13447.

Kalantari, F., Bergeron, J. J. M. andNilsson, T. (2010). Biogenesis of lipid droplets– how cells get fatter. Mol. Membr. Biol. 27, 462-468.

Katsel, P., Davis, K. L., Li, C., Tan,W., Greenstein, E., Kleiner Hoffman, L. B. andHaroutunian, V. (2008). Abnormal indices of cell cycle activity in schizophreniaand their potential association with oligodendrocytes.Neuropsychopharmacology33, 2993-3009.

King, J. S., Teo, R., Ryves, J., Reddy, J. V., Peters, O., Orabi, B., Hoeller, O.,Williams, R. S. B. and Harwood, A. J. (2009). The mood stabiliser lithiumsuppresses PIP3 signalling in Dictyostelium and human cells. Dis. Model Mech.2, 306-312.

Leach, N. T., Sun, Y., Michaud, S., Zheng, Y., Ligon, K. L., Ligon, A. H.,Sander, T., Korf, B. R., Lu, W., Harris, D. J. et al. (2007). Disruption ofdiacylglycerol kinase delta (DGKD) associated with seizures in humans and mice.Am. J. Hum. Genet. 80, 792-799.

Lehrman, G., Hogue, I. B., Palmer, S., Jennings, C., Spina, C. A., Wiegand, A.,Landay, A. L., Coombs, R. W., Richman, D. D., Mellors, J. W. et al. (2005).Depletion of latent HIV-1 infection in vivo: a proof-of-concept study. Lancet366, 549-555.

Los, A. P., van Baal, J., de Widt, J., Divecha, N. and van Blitterswijk, W. J.(2004). Structure-activity relationship of diacylglycerol kinase theta. Biochim.Biophys. Acta 1636, 169-174.

Lucas, P., Ukhanov, K., Leinders-Zufall, T. and Zufall, F. (2003). Adiacylglycerol-gated cation channel in vomeronasal neuron dendrites isimpaired in TRPC2 mutant mice: mechanism of pheromone transduction.Neuron 40, 551-561.

Meunier, H., Carraz, G., Neunier, Y., Eymard, P. and Aimard, M. (1963).Pharmacodynamic properties of N-dipropylacetic acid. Therapie 18, 435-438.

Milne, S. B., Ivanova, P. T., Armstrong, M. D., Myers, D. S., Lubarda, J.,Shulga, Y. V., Topham, M. K., Brown, H. A. and Epand, R. M. (2008). Dramaticdifferences in the roles in lipid metabolism of two isoforms of diacylglycerol kinase.Biochemistry 47, 9372-9379.

Moya, P. R., Murphy, D. L., McMahon, F. J. and Wendland, J. R. (2010).Increased gene expression of diacylglycerol kinase eta in bipolar disorder.Int. J. Neuropsychopharmacol. 13, 1127-1128.

Muller, M., Gahwiler, B. H., Rietschin, L. and Thompson, S. M. (1993).Reversible loss of dendritic spines and altered excitability after chronic epilepsyin hippocampal slice cultures. Proc. Natl. Acad. Sci. USA 90, 257-261.

Newton, A. C. (1997). Regulation of protein kinase C. Curr. Opin. Cell Biol.9, 161-167.

Nilssen, L. S., Dajani, O., Christoffersen, T. and Sandnes, D. (2005).Sustained diacylglycerol accumulation resulting from prolonged G protein-coupled receptor agonist-induced phosphoinositide breakdown in hepatocytes.J. Cell. Biochem. 94, 389-402.

Ostroski, M., Tu-Sekine, B. and Raben, D. M. (2005). Analysis of a noveldiacylglycerol kinase from Dictyostelium discoideum: DGKA. Biochemistry44, 10199-10207.

Pakes, N. K., Veltman, D. M., Rivero, F., Nasir, J., Insall, R. andWilliams, R. S. B.(2012). The Rac GEF ZizB regulates development, cell motility and cytokinesis inDictyostelium. J. Cell Sci. 125, 2457-2465.

Pettitt, T. R., Martin, A., Horton, T., Liossis, C., Lord, J. M. andWakelam, M. J. O.(1997). Diacylglrcol and phosphatidate generated by phospholipase C and D,respetively, have distinct fatty acid compositions and functions; phospholipaseD-derived diacylglyceol does not activate protein kinase C in porcine aorticendothelial cells. J. Biol. Chem. 28, 17354-17359.

Pryse-Phillips, W. E., Dodick, D. W., Edmeads, J. G., Gawel, M. J., Nelson, R. F.,Purdy, R. A., Robinson, G., Stirling, D. and Worthington, I. (1997). Guidelinesfor the diagnosis and management of migraine in clinical practice. CanadianHeadache Society. CMAJ 156, 1273-1287.

Ramadan, E., Basselin, M., Taha, A. Y., Cheon, Y., Chang, L., Chen, M. andRapoport, S. I. (2011). Chronic valproate treatment blocks D2-like receptor-mediated brain signaling via arachidonic acid in rats. Neuropharmacology61, 1256-1264.

Rodriguez de Turco, E. B., Tang, W., Topham, M. K., Sakane, F.,Marcheselli, V. L., Chen, C., Taketomi, A., Prescott, S. M. and Bazan, N. G.(2001). Diacylglycerol kinase epsilon regulates seizure susceptibility andlong-term potentiation through arachidonoyl- inositol lipid signaling. Proc. Natl.Acad. Sci. USA 98, 4740-4745.

Rogawski, M. A. and Loscher, W. (2004). The neurobiology of antiepileptic drugsfor the treatment of nonepileptic conditions. Nat. Med. 10, 685-692.

Sakai, H. and Sakane, F. (2012). Recent progress on type II diacylglycerol kinases:the physiological functions of diacylglycerol kinase delta, eta and kappa and theirinvolvement in disease. J. Biochem. 152, 397-406.

Saxena, A., Scaini, G., Bavaresco, D. V., Leite, C., Valvassoria, S. S.,Carvalho, A. F. and Quevedo, J. (2017). Role of protein kinase C in bipolardisorder: a review of the current literature. Mol. Neuropsychiatry 3, 108-124.

11

RESEARCH ARTICLE Disease Models & Mechanisms (2018) 11, dmm035600. doi:10.1242/dmm.035600

Disea

seModels&Mechan

isms

Scheibel, M. E., Crandall, P. H. and Scheibel, A. B. (1974). The hippocampal-dentate complex in temporal lobe epilepsy. A Golgi study. Epilepsia 15, 55-80.

Shimshoni, J. A., Dalton, E. C., Jenkins, A., Eyal, S., Kwan, K.,Williams, R. S. B., Pessah, N., Yagen, B., Harwood, A. J. and Bialer, M.(2007). The effects of CNS-active valproic acid constitutional isomers, cyclopropylanalogues and amide derivatives on neuronal growth cone behaviour. Mol.Pharmacol. 71, 884-892.

Shinoda, S., Schindler, C. K., Meller, R., So, N. K., Araki, T., Yamamoto, A.,Lan, J.-Q., Taki, W., Simon, R. P. and Henshall, D. C. (2004). Bim regulationmay determine hippocampal vulnerability after injurious seizures and in temporallobe epilepsy. J. Clin. Invest. 113, 1059-1068.

Soares, J. C. and Mallinger, A. G. (1997). Intracellular phosphatidylinositolpathway abnormalities in bipolar disorder patients. Psychopharmacol. Bull.33, 685-691.

Squassina, A., Manchia, M., Congiu, D., Severino, G., Chillotti, C., Ardau, R.,Piccardi, M. and Zompo, M. D. (2009). The diacylglycerol kinase eta gene andbipolar disorder: a replication study in a Sardinian sample. Mol. Psychiatry14, 350-351.

Szendroedi, J., Yoshimura, T., Phielix, E., Koliaki, C., Marcucci, M., Zhang, D.,Jelenik, T., Muller, J., Herder, C., Nowotny, P., Shulmsn, G. I. and Roden, M.(2014) Role of diacylglycerol activation of PKCθ in lipid-induced muscle insulinresistance in humans. Proc. Natl. Acad. Sci. USA 26, 9597-9602.

Terbach, N. and Williams, R. S. B. (2009). Structure-function studies for thepanacea, valproic acid. Biochem. Soc. Trans. 37, 1126-1132.

van Baal, J., de Widt, J., Divecha, N. and van Blitterswijk, W. J. (2005).Translocation of diacylglycerol kinase theta from cytosol to plasma membrane inresponse to activation of G protein-coupled receptors and protein kinaseC. J. Biol.Chem. 280, 9870-9878.

Vanhaesebroeck, B., Stephens, L. and Hawkins, P. (2012). PI3Ksignalling: the path to discovery and understanding. Nat. Rev. Mol. Cell Biol.13, 195-203.

Walsh, J. P., Suen, R. and Glomset, J. A. (1995). Arachidonoyl-diacylglycerolkinase. Specific in vitro inhibition by polyphosphoinositides suggests a

mechanism for regulation of phosphatidylinositol biosynthesis. J. Biol. Chem.270, 28647-28653.

Wang, H.-Y. and Friedman, E. (1989). Lithium inhibition of protein kinase Cactivation-induced serotonin release. Psychopharmacology (Berl) 99, 213-218.

Wang, H.-Y. and Friedman, E. (1996). Enhanced protein kinase C activity andtranslocation in bipolar affective disorder brains. Biol. Psychiatry 40, 568-575.

Wang, H.-Y., Johnson, G. P. and Friedman, E. (2001). Lithium treatment inhibitsprotein kinase C translocation in rat brain cortex. Psychopharmacology (Berl)158, 80-86.

Watson, D. G., Watterson, J. M. and Lenox, R. H. (1998). Sodium valproate down-regulates the myristoylated alanine-rich C kinase substrate (MARCKS) inimmortalized hippocampal cells: a property of protein kinase C-mediated moodstabilizers. J. Pharmacol. Exp. Ther. 285, 307-316.

Whatmore, J., Wiedemann, C., Somerharju, P., Swigart, P. and Cockcroft, S.(1999). Resynthesis of phosphatidylinositol in permeabilized neutrophils followingphospholipase Cbeta activation: transport of the intermediate, phosphatidic acid,from the plasma membrane to the endoplasmic reticulum for phosphatidylinositolresynthesis is not dependent on soluble lipid carriers or vesicular transport.Biochem. J. 341, 435-444.

Williams, R. S. B., Eames, M., Ryves, W. J., Viggars, J. and Harwood, A. J.(1999). Loss of a prolyl oligopeptidase confers resistance to lithium by elevation ofinositol (1,4,5) trisphosphate. EMBO J. 18, 2734-2745.

Williams, R. S. B., Cheng, L., Mudge, A.W. andHarwood, A. J. (2002). A commonmechanism of action for three mood-stabilizing drugs. Nature 417, 292-295.

Williams, E.-J., Walsh, F. S. and Doherty, P. (2003). The FGF receptor uses theendocannabinoid signaling system to couple to an axonal growth response. J. CellBiol. 160, 481-486.

Xu, X., Muller-Taubenberger, A., Adley, K. E., Pawolleck, N., Lee, V. W. Y.,Wiedemann, C., Sihra, T. S., Maniak, M., Jin, T. and Williams, R. S. B. (2007).Attenuation of phospholipid signaling provides a novelmechanism for the action ofvalproic acid. Eukaryot. Cell 6, 899-906.

Yang, E. J., Ahn, S., Lee, K., Mahmood, U. and Kim, H. S. (2016). Early behavioralabnormalities and perinatal alterations of PTEN/AKT pathway in valproic acidautism model mice. PLoS ONE 11, e0153298.

12

RESEARCH ARTICLE Disease Models & Mechanisms (2018) 11, dmm035600. doi:10.1242/dmm.035600

Disea

seModels&Mechan

isms