Diacylglycerol kinase theta and zeta isoforms: regulation of activity, protein binding partners and physiological functions

(1)

Diacylglycerol kinase theta and zeta isoforms: regulation of activity,

protein binding partners and physiological functions

Los, Alrik Pieter

Citation

Los, A. P. (2007, January 25). Diacylglycerol kinase theta and zeta isoforms: regulation of

activity, protein binding partners and physiological functions. Retrieved from

https://hdl.handle.net/1887/9451

Version: Corrected Publisher’s Version

License: Licence agreement concerning inclusion of doctoral thesis in the

Institutional Repository of the University of Leiden

Downloaded from: https://hdl.handle.net/1887/9451

Note: To cite this publication please use the final published version (if applicable).

(2)

(3)

(4)

(5)

Design

Crispijn Los Joris Smidt Printed by

PrintPartners Ipskamp B.V., Amsterdam

(6)

kinase theta and

zeta isoforms:

Regulation of activity,

protein binding

partners and physio-

logical functions

Proefschrift

ter verkrijging van

de graad van Doctor aan de Universiteit Leiden, op gezag van de Rector Magnificus Dr. D.D. Breimer,

hoogleraar in de faculteit der Wiskunde en Natuurwetenschappen en die der Geneeskunde,

volgens besluit van het College voor Promoties te verdedigen op donderdag 25 januari 2007

klokke 15:00 uur

door

Alrik Pieter Los

Geboren te ’s-Hertogenbosch in 1975

(7)

Promotiecommissie

Promotor

Prof. Dr. W. H. Moolenaar Co-promotores

Dr. N. Divecha

Nederlands Kanker Instituut, Amsterdam Dr. W. J. van Blitterswijk

Nederlands Kanker Instituut, Amsterdam Referent

Prof. Dr. J. J. Neef jes Overige leden

Prof. Dr. P. ten Dijke

The research described in this thesis was performed at the

Division of Cellular Biochemistry of the Netherlands Cancer Institute, Amsterdam. This work was supported by the Dutch Cancer Society, grant NKI 2000-2209.

Publication of this thesis was financially supported by the Netherlands Cancer Institute and the Dutch Cancer Society.

(8)

Introduction to the signalling functions

of diacylglycerol kinase-ζ and -θ isoforms...9

Chapter 2

Structure–activity relationship of diacylglycerol

kinase θ.

(BBA. 2004 Mar 22;1636(2-3):169-174)

...45

Chapter 3

Introduction to the family of retinoblastoma

gene products...59

Chapter 4

The retinoblastoma family proteins bind to

and activate diacylglycerol kinase-ζ

(JBC. 2006 Jan 13;281(2):858-866)

...75

Chapter 5

Protein kinase

C

inhibits binding of diacylglycerol

kinase-ζ to the retino blastoma protein

(BBA Molecular Cell research, in press)

...99

Chapter 6

Is there a role for diacylglycerol kinase-ζ in cell

cycle regulation?...115

Chapter 7

Diacylglycerol kinase-ζ stimulates muscle

differentiation...129

Summary and general discussion...148

Nederlandse samenvatting...152

Curriculum Vitae...157

List of Publications...157

List of Abbreviations...158

Dankwoord...159

(9)

(10)

Introduction to the

signalling functions of

diacylglycerol kinase- ζ

and - θ ^isoforms

Chapter 1

(11)

(12)

Abstract

Diacylglycerol kinases (

DGK

s) phosphorylate the second

messenger diacylglycerol (

DAG

) yielding phosphatidic

acid (

PA

). Ten different mammalian

DGK

isoforms have

been described and in this review we will focus on the

DGK

ζ ^and

^DGK

θ isoforms.

DGK

ζ is ubiquitously ex-

pressed, whereas

DGK

θ expression is limited to certain

specific tissues. Both isoforms and their activities have

several subcellular locations, being regulated by lipids,

protein interactions and growth factors. This complex

regulation serves to temporally and spatially restrict

DGK

ζ ^and

^DGK

θ activity. Accumulating evidence suggest

that

DGK

ζ ^and

^DGK

θ have a dual role in signalling by reg-

ulating

DAG

-binding proteins as well as

PA

-binding pro-

teins. The former include protein kinase

C

(

PKC

),

Ras guanyl nucleotide-releasing protein (

R

as-

GRP

),

chimaerins, and Munc-13, whereas the

PA

-binding pro-

teins include phosphatidylinositol-5-kinases (

PIP5K

)

and mammalian target of rapamycin (m

TOR

). Therefore,

DGK

ζ ^and

^DGK

θ may play a crucial role in determining

the correct balance between

DAG

and

PA

signalling

pathways to regulate physiological processes.

Introduction to the signalling functions of diacylglycerol kinase-ζ and -θ isoforms

(13)

1212

1. Introduction

The lipid second messenger diacylglycerol (DAG) is produced upon stimulation with a variety of hormones that regulate many cellular processes including cell division, differentiation, cytoskeletal reorganisation and vesicular transport.

The diverse role of DAG in signal transduction is mediated by an array of DAG- binding proteins including protein kinase C (PKC), Ras guanyl nucleotide-releasing proteins (Ras-GRP), chimaerins, and Munc-13 (Brose et al., 2004). DAG is generated by G-protein coupled receptor-mediated activation of phospholipase C-β (^PLCβ) or tyrosine kinase-linked receptor-mediated activation of ^PLCγ. ^PLC hydrolyses phosphatidylinositol-4,5-bisphosphate (PI(4,5)P₂) yielding DAG and inositol-1,4,5-trisphosphate (IP₃). In addition, DAG is produced by phospholipase D-mediated hydrolysis of phosphatidylcholine (PC) followed by dephosphorylation of phosphatidic acid, and during de novo synthesis of phospholipids and triglyc- erides (van Blitterswijk et al., 1994).

DAG is involved in regulating many signalling pathways and its levels likely need to be tightly regulated. Uncontrolled synthesis of DAG may contribute to cellular transformation as tumours often show elevated levels of DAG (Kato et al., 1987; Kato et al., 1988). Attenuation of DAG is achieved by five possible mecha- nisms: phosphorylation by diacylglycerol kinases yielding phosphatidic acid (PA), hydrolysis by DAG lipase generating monoacylglycerol and fatty acids, conversion into PC by CDP-choline:DAG choline phosphotransferase or into phosphatidylethanolamine by CDP-ethanolamine:DAG ethanolamine phosphotransferase, or conversion by transphosphatidylation into bisphosphatidic acid (van Blitterswijk et al., 1994). It is thought that DGKs are involved in rapid attenuating of DAG produced after phosphatidylinositol hydrolysis.

The importance of DGKs in signalling is underlined by the size of the DGK family; recently, the tenth DGK isoform, DGKκ has been described (Imai et al., 2005). DGK isoforms can be divided into five classes, based on their structural domains. They all contain two or three cysteine-rich domains (CRD) that are thought to bind DAG, and a catalytic domain that is conserved among all DGKs.

In addition, each class of DGKs contains different signalling domains that link them to signal transduction pathways. In addition to the mammalian DGK isoforms, DGKs have been identified in Drosophila melanogaster, Caenorhabditis elegans, Dictyostelium disc oideum, and Arabidopsis thaliana (Topham, 2005).

The complexity of the family is further increased by the existence of splicing variants of DGKs.

While the DGK family and characteristics and functions of each isoform were described in several reviews (van Blitterswijk and Houssa, 2000; Kanoh et al., 2002; Luo et al., 2004b; Goto and Kondo, 2004; Topham, 2005), this chapter is limited to recent progress on DGKζ and ^DGKθ isotypes only, focusing on their domain structure, (sub)cellular localisation and mode of regulation and activation and their role in signal transduction.

(15)

1414

Fig. 1.

Schematic representation of the domain organisation of DGKζ and DGKθ and their homologs.

Human DGKζ and DGKθ and their homologues contain a conserved catalytic domain, respectively two and three cysteine-rich domains (CRD) of which the most C-terminal CRD is extended with a stretch of 15 amino acids (Ext).

In addition, DGKζ contains a domain that is homologue to the phosphorylation site domain of the MARCKS protein (MARCKS-PSD) and four Ankyrin re- peats. Caenorabditis elegans DGKζ (ceDGKζ, gene 2G748, accession number NP_495301) lacks ankyrin repeats, whereas, Drosophila melanogaster DGK2 lacks a MARCKS-PSD. DGKθ contains a proline- and glycine-rich domain (Pro) and a Pleckstrin homology (PH) domain overlapping with a Ras association (RA) domain. Dictyostelium dis- coideum DGKA (ddDGKA) lacks a PH domain, but contains a large poly-asparagine stretch (Asp).

Domain Function Reference

CRDs

DAG binding?

Essential for DGK activity Localisation

(Bunting et al., 1996) (Santos et al., 2002) (Santos et al., 2002)

MARCKS-PSD Localisation

Phosphorylated by PKC

(Topham et al., 1998) (Luo et al., 2003b) pRB (pocket protein) binding (Los et al., 2006)

Catalytic domain Essential for DGK activity (Topham et al., 1998; Santos et al., 2002)

Ankyrin repeats Localisation

Leptin receptor (Ob-Rb) binding

(Goto and Kondo, 1996; Hozumi et al., 2003) (Liu et al., 2001)

PDZ-binding motif Syntrophin binding Localisation

(Hogan et al., 2001)

(Hogan et al., 2001; Abramovici et al., 2003) Table 1. DGKζ structural domains and their functions

(16)

2. Characteristics, cell/tissue distribution

and subcellular localisation of

DGK

ζ

2.1. Characteristics of DGK

ζ

In 1996, DGKζ was independently identified by two groups: Prescott and co-work- ers cloned DGKζ from human umbilical vein endothelial cells, whereas Goto and Kondo isolated DGKζ (^DGK-IV) cDNA from cDNA libraries of rat retina and brain (Bunting et al., 1996; Goto and Kondo, 1996). Two years later, Prescott and co-work- ers also cloned DGKζ from a mouse brain c^DNA library (Ding et al., 1998). Human and rat/mouse DGKζ c^DNAs encode a protein of, respectively, 928 and 929 amino acids and share 88% identity at the nucleic acid level and 95.5% at the amino acid level. DGKζ is classified in class ^IV^DGKs that also include DGKι, based on some characteristic domains (Fig. 1A; Table 1). The class IVDGKs contain a region that is similar to the phosphorylation-site domain (PSD) of myristoylated alanine-rich C-kinase substrate (MARCKS) (Bunting et al., 1996). The DGKζ-^MARCKS-PSD contains several putative PKC phosphorylation sites and overlaps with a nuclear localisation signal. In addition, DGKζ and ^DGKι are characterised by the presence of four tandem ankyrin repeats that are involved in protein-protein interactions.

Furthermore, DGKζ contains a functional ^PDZ-binding motif at its C-terminus (Fabre et al., 2000), and two PEST sequences that may regulate protein degradation (Bunting et al., 1996). The catalytic domain and the two CRDs are essential for DGKζ activity, since point-mutations in these domains inactivated the enzyme (Topham et al., 1998; Santos et al., 2002). In contrast, the ankyrin repeats and the PDZ-binding site are not essential for DGK activity.

The human DGKζ gene is localised on chromosome 11p11.2 and spans approx- imately 50 kb genomic sequence containing 32 exons (Ding et al., 1997). An alternative splice variant of DGKζ was isolated from a human skeletal muscle c^DNA library. It encodes a DGKζ variant of 1.117 amino acids (130 k^Da) that contains the same functional domains, but has a unique N-terminus in which no additional known domains could be identified (Ding et al., 1997). The alternative N-terminus of the DGKζ splice variant was encoded by an exon between the first two exons of the endothelial DGKζ form. The existence of tissue-specific ^DGKζ variants suggests that DGKζ has tissue-related functions and/or is specifically regulated in different tissues.

2.2. Conservation among species

Homologues of DGKζ have been found in at least two other species (Fig. 1A).

Drosophila melanogaster DGK2, which is encoded by the rdgA locus, is 42% identical and 61% similar to human DGKζ (Masai et al., 1993; Bunting et al., 1996). ^DGK2 has an elongated N-terminus with unknown function and does contain the four tandem ankyrin repeats but no MARCKS-PSD. A Caenorhabditis elegans DGK, encoded by gene 2G748 (also known as K06A1.6 or YK1670), is most similar to the DGKζ isoform (39% identity at the amino acid level). It also contains two ^CRDs, a catalytic domain and a domain that is similar to the MARCKS-PSD with four serines and a stretch of positively charged residues, but lacks ankyrin repeats.

(17)

1616

Tissue Cell type Subcellular localisation References

Embryogenesis: (Ding et al., 1998)

– Somites – Limb buds – Interdigital regions – dorsal root ganglia – vibrissa follicles

– rat lung (Katagiri et al., 2005)

Lung Various epithelial cells nucleus (Katagiri et al., 2005)

Alveolar type II cells nucleus

Heart Cardiomyocytes nucleus (Takeda et al., 2001)

Skeletal muscle Myoblasts sarcolemma (Abramovici et al., 2003)

Ovary follicles (theca cells) (Toya et al., 2005)

corpora lutea interstitial cells

Placenta labyrinthine zone (all cell types) (Toya et al., 2005)

Lymphoid system T cells cytosol (Santos et al., 2002)

Brain: (Goto and Kondo, 1996;

Hozumi et al., 2003)

– Cerebellum Purkinje cells nucleus+proximal

dendritic shaf t

granule cells nucleus

– Hippocampus Pyramidal cells nucleus

dentate granular cells nucleus scattered interneurons nucleus

– cerebral cortex most neurons nucleus

– olf actory bulb and tubercle

hypothalamic nuclei (Liu et al., 2001)

LβT2 gonadotrope cells nucleus/cytosol (Davidson et al., 2004) DDT1-MF2 ductus def erence

smooth muscle nucleus/cytosol (Fukunaga-Takenaka et al.,

2005)

IIC9 fibroblasts cytosol (Bregoli et al., 2001)

MEF, U2-OS, MCF7, MDCK,

N1E-115, N2A cytosol/nucleus our unpublished

observations

HL60 human leukemia cells (Batista, Jr. et al., 2005)

Table 2. Tissue distribution and subcellular localisation of DGKζ:

(18)

However, this protein has not been characterised. The presence of DGKζ in lower organisms indicates that DGKζ is evolutionally conserved, with essential functions that are already required for more primitive organisms.

2.3. Cell/tissue distribution of DGK

ζ

DGKζ is already expressed during embryogenesis in diverse developmental struc- tures as is summarised in Table 2. Intriguingly, DGKζ was not only expressed in tissues in which massive proliferation and differentiation occurs, including limb buds, but also in areas with extensive apoptosis. For example, DGKζ is expressed in interdigital regions of the limbs in which apoptosis is required to shape the digits. The broad distribution of DGKζ suggests a diverse role in embryogenesis (Ding et al., 1998). However, DGKζ knock-out mice did not show any defects during development as they do not have a phenotype (G. Koretzky, personal communi- cations), indicating that DGKζ is not essential for development.

DGKζ is broadly expressed throughout the adult body and in diverse cell types as is shown in Table 2. DGKζ m^RNAs could be detected in all human and rat tissues that were tested, but highest expression was found in brain (Bunting et al., 1996;

Liu et al., 2001). The alternatively spliced DGKζ variant identified in skeletal muscle was also expressed in heart, bladder and intestine muscle and in the lymphoid system (Ding et al., 1997; Zhong et al., 2002). Although not all of these tissues have been checked for DGKζ protein expression, the presence of ^DGKζ m^RNAs in all tested tissues suggests that DGKζ plays a role throughout the whole body.

The localisation of DGKζ in brain is extensively investigated. In brain, highest expression was found in Purkinje cells of the cerebellum and in pyramidal neurons of the hippocampus that integrate signals and are responsible for the output of the cerebellar cortex and hippocampus (Goto and Kondo, 1996; Hogan et al., 2001;

Hozumi et al., 2003). DGKζ is present in the cell bodies and dendrites of Purkinje cells but not in axons, suggesting that DGKζ is involved in signal receiving and integration. Since DGKζ is present in hippocampus, cerebral and cerebellar cortex, and olfactory bulb, which are brain areas known to be involved in synaptic plasticity, it is thought that DGKζ is involved in learning and memory function (Hozumi et al., 2003). Interestingly, DGKζ expression in brain is increased by 2-fold in adults compared to newborn mice (Ding et al., 1998), which may imply a role for DGKζ in neuronal development and learning. ^DGKζ is lower expressed in spinal cord and medulla that have autonomous functions, and in thalamus and corpus callosum that receive, process and distribute information and connect different areas (Ding et al., 1998), suggesting that it is less important for those brain functions.

In addition to its potential role in learning, DGKζ may also regulate energy balance as it is involved in leptin signalling (Liu et al., 2001). DGKζ is expressed in neurons of hypothalamic nuclei that regulate food uptake and body weight.

Interestingly, hypothalamic DGKζ expression was increased in mice that consumed a high-fat diet, whereas its levels then declined when the mice were maintained on the high-fat diet and became obese. The function of DGKζ in leptin signalling needs to be further investigated.

(19)

1818

The DGKζ homologue in Drosophila, ^DGK2, is localised in photoreceptors of the retina, but was not detectable in brain or in other parts of the body (Masai et al., 1993). DGKζ was also detected in the rat eye, but in different cell types (Goto and Kondo, 1996). DGKζ is localised in the inner granular layer of the retina in Müller’s supporting cells and in bipolar, horizontal and amacrine cells that are involved in transducing signals from the photoreceptors to ganglion cells, the output neurons of the retina. Since Drosophila DGK2 and mammalian DGKζ are differently localised they may have different functions in the retina.

2.4. Subcellular localisation of DGK

ζ

In addition to the broad tissue distribution, DGKζ can have several subcellular locations. Depending on cell-type and environmental conditions, it is localised within the nucleus (most often), cytosol and/or plasma membrane (Table 2).

DGKζ localisation is regulated by its signalling domains and by binding to other proteins.

2.4.1. Dependence on cell type, environmental conditions and development

In many, but not all cell types, DGKζ is found in the nucleus. In ^COS-7 cells, HEK293 cells and cultured hippocampal neurons, overexpressed full-length DGKζ is localised in the nucleus (Goto and Kondo, 1996; Hozumi et al., 2003; Avila-Flores et al., 2005). Alternatively spliced muscle-specific DGKζ is also localised in the nucleus (Ding et al., 1997). In skeletal muscle, on the other hand, DGKζ is localised at the sarcolemma (Abramovici et al., 2003). Fractionation analysis of IIC9 fibroblasts revealed that DGKζ is present in the cytosol, but not in the nucleus (Bregoli et al., 2001). In different cell types in lung and in cardiac myocytes, DGKζ is localised in the nucleus (Katagiri et al., 2005; Takeda et al., 2001). Overexpression of DGKζ-^GFP fusion proteins in LβT2 murine gonadotrope cells or ^DDT1-MF2 hamster ductus deferens smooth muscle cells also showed high expression in the nucleus and moderate cytosolic expression (Davidson et al., 2004; Fukunaga-Takenaka et al., 2005).

We also examined localisation of overexpressed DGKζ in ^COS-7 and HEK293 cells and also found DGKζ to be present in the nucleus, but only in a small percentage of cells. Other cells showed cytosolic and plasma membrane staining. We also examined DGKζ localisation in various types of cells that stably express ^DGKζ at low levels and observed that in (fixed) mouse embryonic fibroblasts, U2-OS osteosarcoma, MCF7 breast adeno carcinoma, N115 and N2A neuroblastoma, and Madin-Darby caning kidney cells, the majority of DGKζ is localised in the cytosol, whereas faint staining is present in the nucleus (our unpublished data). In a minority of cells, a stronger nuclear staining was visible, but that was especially in cells expressing high levels of DGKζ. Therefore, localisation of highly expressed DGKζ may be artificial to some extent.

Subcellular localisation of endogenous DGKζ is extensively studied in brain and appeared to be highly regulated under specific conditions. Endogenous DGKζ was shown to be nuclear localised in most neurons, including pyramidal cells in the cerebral cortex, all neurons in the hippocampus and in Purkinje and granule

(20)

cells in de cerebellum (Hozumi et al., 2003). Interestingly, DGKζ was translocated from the nucleus to the cytosol in hippocampal CA1 pyramidal neurons in ischemic rat brain, whereas such translocation was not observed in other hippocampal cells or other brain areas (Ali et al., 2004). In addition, in cerebral infarction, DGKζ levels rapidly decreased in ischemic cortical neurons, which occurred prior to neuronal degeneration (Nakano et al., 2006). The functional consequence of regulating DGKζ levels or translocation is not known. Finally, localisation of ^DGKζ in cerebellar Purkinje cells is dependent on the developmental stage of the cerebellum (Hozumi et al., 2003). During cerebellar development DGKζ relocalises from the perinuclear cytoplasm and axon hillock to the nucleus and dendrites, suggesting that it has a function during cerebral development.

2.4.2. Dependence on DGK

ζ

structural domains

The nuclear localisation signal (NLS) within the MARCKS-PSD is sufficient for nuclear localisation of DGKζ, as a ^MARCKS-GFP fusion protein was localised in the nucleus. In addition, a rat DGKζ ^NLS-GFP fusion protein was also localised in the nucleus of cultured hippocampal neurons (Hozumi et al., 2003). Deletion ofthe MARCKS-PSD or inactivation of the NLS by substitution of all basic into neutral amino acids dislocalised DGKζ from the nucleus to the cytosol.

The N-terminal part of DGKζ including the ^CRDs is also involved in regulating DGKζ localisation. In ^T-cells, DGKζ is localised in the cytosol (Santos et al., 2002).

However, when both CRDs of DGKζ are removed, it is mainly found in the nucleus.

In contrast, mutation of the first histidine of the CRD core structure that is sufficient to inactivate the enzyme, still results in localisation in the cytosol, suggesting that the N-terminus keeps DGKζ in the cytosol, but that functional ^CRDs are not required.

Removal of the C-terminus up to the catalytic domain (DGKζ-Δ^C) showed a nuclear localisation especially in intranuclear granules in COS-7 cells (Goto and Kondo, 1996). In contrast, in hippocampal neurons DGKζ-Δ^C is localised in the cytosol (Hozumi et al., 2003). Also localisation of the expressed C-terminus is different in COS-7 cells and hippocampal neurons, i.e. nucleus or cytosol respectively. The difference in localisation may reflect the difference in cell type or expression levels, which is extremely high in COS-7 cells. The cytosolic expression in hippocampal neurons of the DGKζ-Δ^C mutant, in which the NLS domain remained unaffected, suggests a role for the C-terminus in nuclear localisation. Indeed, the C-terminal PDZ-binding motif is involved in regulating subcellular localisation of DGKζ by binding to γ1-syntrophin (Hogan et al., 2001). The adapter protein γ1-syntrophin keeps ^DGKζ in the cytosol, whereas ^DGKζ recruits γ1-syntrophin to the nucleus in HeLa cervix carcinoma cells. Inhibiting the interaction by overexpressing DGKζ-^FLAG or a PDZ-binding motif mutant of DGKζ, disrupts the balanced localisation of both proteins and caused DGKζ to accumulate in the nucleus and γ1-syntrophin in the cytosol. Also in muscle, DGKζ is partly relocal- ised from the plasma membrane to the nucleus in cells with lower syntrophin levels or when the PDZ domain is masked by a FLAG tag (Abramovici et al., 2003).

Why DGKζ-Δ^C is localised in the cytosol of neurons remains an open question,

(21)

2020

since this mutant cannot be kept in the cytosol by syntrophin. Since this mutant also lacks ankyrin repeats, these particular domains may also be involved. Alter- natively, the NLS in the DGKζ-Δ^C mutant might be cryptic in the tertiary structure, and molecules interacting with the C-terminal region might determine whether or not the NLS is exposed to allow nuclear transport (Hozumi et al., 2003).

In conclusion, the NLS is not the only structural domain that regulates DGKζ subcellular localisation. Both C-terminal and N-terminal sequences may co-regulate DGKζ localisation, depending on interactions with proteins such as γ1-syntrophin and exposure to environmental signals, in a cell type-dependent fashion.

3. Characteristics, cell/tissue distribution

and subcellular localisation of

DGK

θ

3.1. Characteristics of DGK

θ

DGKθ was cloned in 1997 in our lab (Houssa et al., 1997). Human ^DGKθ c^DNA encodes a protein of 941 amino acids with a molecular mass of 110 kDa and is the only known isoform that belongs to the class VDGK category (Fig. 1B; Table 3).

DGKθ is the only ^DGK isoform that contains three instead of two CRDs. The third CRD of DGKθ is most homologous to the second ^CRD of other DGK isotypes and is extended with a stretch of 15 amino acids that is conserved among all DGKs (Houssa and van Blitterswijk, 1998). Mutational analysis showed that the CRD extension is essential for DGK activity (Los et al., 2004). DGKθ contains a catalytic domain and a proline- and glycine-rich domain with a putative SH3 domain binding site. Furthermore, it has a pleckstrin homology (PH) domain that overlaps with a Ras association domain. Truncation studies have shown that the entire primary sequence of the protein is required for full activity (Los et al., 2004). Even a minor 33 amino acid truncation at the C-terminus or removal of the N-terminal proline- and glycine-rich domain rendered the protein inactive. Only an N-terminal truncation that includes the first CRD yields some residual activity, 10-20% activity of wild-type DGKθ (Los et al., 2004).

3.2. Conservation among species

DGKθ homologues have been found in Dictyostelium discoideum and Caenorhab- ditis elegans (Fig. 1B). Dictyostelium DGK, which is encoded by the dgkA gene, was first thought to be a myosin II heavy chain kinase. However, the protein did not have a protein kinase domain, but a DGK catalytic domain (De la Roche et al., 2002). It consists of 887 amino acids and is most closely related to DGKθ sharing 40% identity and 56% similarity with human DGKθ. In addition to three ^CRDs, with the third being extended, and a catalytic domain, it has a short proline- and glycine-rich domain and an asparagine-rich domain, but lacks a PH domain.

DGK activity analysis showed that the protein indeed phosphorylates DAG (De la Roche et al., 2002; Ostroski et al., 2005). Southern blot analysis and BLAST searches revealed that dgkA is probably the only DGK in Dictyostelium (De la

(22)

Domain Function Reference

Pro/Gly-rich motif Essential for DGK activity (Los et al., 2004)

CRDs

DAG binding?

Essential for DGK activity Localisation

(Houssa and van Blitterswijk, 1998) (Los et al., 2004)

(van Baal et al., 2005)

PH/Ras unknown

Catalytic domain Essential for DGK activity RhoA binding

(Los et al., 2004)

(Los et al., 2004; McMullan et al., 2006)

Tissue Cell type Subcellular localisation References

Small intestine Duodenum Liver

(Houssa et al., 1997)

rat small arteries vascular smooth muscle,

endothelial cells nucleus (Walker et al., 2001)

IIC9, MDA-MB-453, MCF7, PC12,

A431, HeLa nucleus/cytosol

(Bregoli et al., 2001;

Tabellini et al., 2003;

van Baal et al., 2005) U2-OS, MDCK, COS, CHO, HEK293,

N1E-115, Jurkat-J6 (van Baal et al., 2005)

Brain: (Houssa et al., 1997)

– Cerebellum Purkinje cells granule cells – Hippocampus

– Cerebral cortex – olf actory bulb – brain stem nuclei

C. elegans neurons cell bodies (peri-nuclear)

and axons (Nurrish et al., 1999)

excretory canals

Roche et al., 2002). It would be interesting to know what function dgkA has in this organism, as it may help to understand the function of mammalian DGKθ.

The C. elegans dgk-1 gene encodes a protein of 950 amino acids that is 38%

identical to human DGKθ (Nurrish et al., 1999). ^DGK-1 also has three CRDs, a catalytic domain and a PH domain. Mutational analysis of dgk-1 revealed that all nine amino acid substitutions found to inhibit DGK-1 function were in the second or third CRD or in the catalytic domain, indicating the importance of these domains (Jose and Koelle, 2005).

Table 3. DGKθ structural domains and their functions

Table 4. Tissue distribution and subcellular localisation of DGKθ:

(23)

2222

3.3. Cell / Tissue distribution and (sub)cellular localisation of DGK

θ

DGKθ is less ubiquitously expressed in the body than ^DGKζ. Expression of ^DGKθ in tissues and cell lines is summarised in Table 4. In mammals and C. elegans, highest expression was found in neurons (Houssa et al., 1997; Nurrish et al., 1999).

In brain, DGKθ m^RNA was ubiquitously expressed in gray matter, but not in white matter. Highest expression was found in Purkinje and granule cells of the cerebellar cortex and in hippocampus, while moderate expression was observed in the olfactory bulb, cerebral cortex and brain stem nuclei (Houssa et al., 1997). The wider expression throughout the gray matter, suggest a more general function for DGKθ in brain (Houssa et al., 1997).

Subcellular fractionation and immunofluorescence of COS-7 cells overexpressing DGKθ showed ^DGKθ localisation in membranes, cytosol and in the perinuclear region (Houssa et al., 1999). In rat mesenteric small arteries, DGKθ was also found in nuclear extracts (Walker et al., 2001). In addition, DGKθ expression was observed in the nucleus and cytosol of several cell lines as is shown in Table 4. In MDA-MB-453, but also in some other cell lines, DGKθ was localised in nuclear speckles were it colocalised with PI(4,5)P₂, PLCβ1 and specific markers for speckles including the splicing component SC-35 and RNA polymerase II (van Blitter- swijk and Houssa, 2000; Tabellini et al., 2003). However, detection of DGKθ in nuclear speckles was observed with one specific DGKθ antibody and could not be reproduced with a different set of DGKθ-specific antibodies (Van Baal and Van Blitterswijk, unpublished observations), making it difficult to draw definite conclusions about this nuclear localisation.

4. Regulation of

DGK

ζ and

DGK

θ activity

Phosphorylation of DAG by DGKs can be regulated in two ways: by regulating catalytic activity of DGKs and by bringing DGKs into close proximity with their substrate (Table 5 and 6). Catalytic activity was shown to be regulated by lipids, protein interactions, and by protein modifications, including phosphorylation.

Translocation of DGKs to their substrate and/or catalytic activity is regulated upon stimulation with different growth factors.

4.1. Substrate specificity and regulation by lipids

Like other DGK isotypes, DGKζ and ^DGKθ prefer 1,2-diacyl-sn-glycerol as substrate over 1,3-diacyl-sn-glycerol and monoacylglycerol (Bunting et al., 1996; Ding et al., 1997; Houssa et al., 1997). An ether-linkage instead of ester-linkage at the sn-1 position, such as in 1-O-hexadecyl-2-arachidonoyl-sn-glycerol generally lowers the enzymatic rate (Epand et al., 2004). DGKζ activity is increased when negatively charged lipids, i.e. phosphatidylserine (PS), phosphatidylinositol (PI), PI(4,5)P₂ and PA, are included in mixed micelle assays (Thirugnanam et al., 2001). The posi- tive effect of PI(4,5)P₂ and other anionic lipids on DGKζ activity is thought to be

(24)

a consequence of strong interactions with the positively charged DGKζ-^MARCKS- PSD that may tighten DGKζ binding to membranes, like was shown for ^PI(4,5)P₂ binding to MARCKS (Wang et al., 2001). Our preliminary (unpublished) data from lipid blot assays suggest that the DGKζ-^MARCKS-PSD indeed binds to the anionic phosphoinositides.

In an assay with large unilamellar vesicles that more closely resemble cellular membranes, as they consist of a lipid bilayer, phosphatidylethanolamine (PE), which typically promotes negative curvature of the membrane, was shown to stimulate DGKζ binding to the membrane and to increase ^DGKζ activity (Fanani et al., 2004). Sphingomyelin inhibited DGKζ activity. Cholesterol, another lipid that promotes negative curvature, enhanced membrane binding of DGKζ but did not affect DGK activity. Cholesterol inhibited PE-mediated DGKζ activation but reverted DGKζ inhibition by sphingomyelin (Fanani et al., 2004), so that it seems to act as a moderating lipid. Since in biological membranes, sphingomyelin and cholesterol colocalise in caveolae/lipid raft microdomains, the above findings with artificial mixtures of these lipids could be of potential physiological relevance.

Category Cell type Factor Stimulatory/inhibitory Reference

Lipids in vitro PS, PI, PI(4,5)P2, PA + (Thirugnanam et al.,

2001)

PE + (Fanani et al., 2004)

SM -

Proteins PKCα - (Luo et al., 2003b)

in vitro pRB, p107, p130 + (Los et al., 2006)

Growth

f actors gonadotrope cells GnRH + / translocation (Davidson et al., 2004)

Jurkat+muscarine

receptor carbachol translocation (Santos et al., 2002)

Category Cell type Factor Stimulatory/inhibitory Reference

Lipids in vitro PS + (Bregoli et al., 2001)

in vitro PS, PA (ddDGKA) + (Ostroski et al., 2005)

Proteins RhoA - (Houssa et al., 1999)

Growth

f actors ^A431 ATP/ bradykinin / thrombin translocation (van Baal et al., 2005) IIC9 fibroblasts α-thrombin + / translocation (Bregoli et al., 2001)

PC12 NGF + (Tabellini et al., 2004)

arteries Noradrenaline + / translocation (Walker et al., 2001)

Table 5. Regulation of DGKζ activity:

Table 6. Regulation of DGKθ activity:

(25)

2424

However, no physical association of DGKζ to such lipid domains has been reported to date.

Similar to DGKζ, ^DGKθ is also stimulated by ^PS in mixed micelles assays, but other anionic lipids have not been tested (Bregoli et al., 2001). The Dictyostelium DGKθ homologue, ^DGKA was shown to be activated by PS as well as PA (Ostroski et al., 2005).

4.2. Regulation by proteins or protein modifications

PKCα was shown to bind to ^DGKζ and to phosphorylate the ^DGKζ-^MARCKS-PSD in HEK293 cells (Luo et al., 2003b). Phosphorylation of DGKζ inhibits ^DGKζ activity by 40%. In contrast, the retinoblastoma tumour suppressor protein (pRB) and its family members p107 and p130 also bind to DGKζ and stimulate ^DGKζ activity up to 5-fold (Los et al., 2006). These opposing effects of PKCα and p^RB-type proteins allow tight regulation of DGKζ activity.

DGKθ binds to the effector loop of ^GTP-bound RhoA (Houssa et al., 1999).

Associated active RhoA completely inactivates DGKθ, which may be explained by the binding of RhoA to the catalytic domain of DGKθ (Los et al., 2004).

4.3. Translocation and activation induced by growth factors DGKζ and ^DGKθ are activated and/or translocated upon stimulation with certain growth factors. In HEK293 cells, endogenous DGK activity diminished after serum starvation and increased upon serum addition, which was also observed in cells that overexpress DGKζ (Avila-Flores et al., 2005). This result indicated that ^DGKζ is activated by growth factors. A growth factor that stimulates DGKζ activity is gonadotrope releasing hormone (GnRH), which increased DGKζ activity by 2.5 or 6-fold in, respectively, SCL60 and Lβ^T2 gonadotrope cells (Davidson et al., 2004).

Upon GnRH stimulation, DGKζ was shown to bind to active c-^Src, which is required for GnRH-mediated activation of DGKζ and for translocation of cytosolic ^DGKζ to the plasma membrane. In Jurkat T cells expressing a muscarinic type I receptor, stimulation of the receptor with carbachol causes a rapid translocation of GFP- tagged DGKζ from the cytosol to the plasma membrane, which requires intact CRDs and phosphorylation of the MARCKS-PSD. Since CRDs are thought to bind to DAG, functional CRDs are probably required for DAG-mediated recruitment of DGKζ to the plasma membrane (Santos et al., 2002).

DGKθ also translocates upon stimulation with growth factors. ^DGKθ translocated from cytosol to plasma membrane upon stimulation with the vasocon trictor noradrenaline, but not angiotensin II, in rat small mesenteric arteries (Walker et al., 2001). Furthermore, noradrenaline increased DGKθ activity in a ^PI 3-kinase dependent way. How PI 3-kinase regulates DGKθ activity needs to be further investigated, but Akt/PKB may be involved as increased DGK activity was physically associated with PKB/Akt upon noradrenaline stimulation. In A431 cells, extracellular ATP, bradykinin and thrombin that all signal via G-protein coupled receptors, induced DGKθ translocation from cytosol to plasma membrane (van Baal et al., 2005). In contrast, stimulation of cells with epidermal growth factor (EGF), insulin or insulin-like growth factor that signals via tyrosine kinase receptors

(26)

failed to translocate DGKθ. ^DGKθ also translocated upon stimulation with high concentrations (100 µM) of short-chain DiC8, indicating that high amounts of DAG are sufficient for translocation. Similar to DGKζ, mutation of residues in the DGKθ-^CRDs that are critical for DAG binding prevented growth factor and DiC8 induced translocation, suggesting that binding of DAG to CRDs causes DGKθ translocation.

Nuclear translocations of DGKζ and ^DGKθ are also observed. Phosphoryla- tion of DGKζ by activated ^PKCα translocated ^DGKζ from the nucleus to the cytosol (Topham et al., 1998). Stimulation of IIC9 fibroblasts with α-thrombin transiently increased nuclear DGKθ protein and ^DGKθ activity, suggesting that ^DGKθ translocated to the nucleus (Bregoli et al., 2001).

5. Signalling functions of

DGK

ζ and

DGK

θ

From the previous sections, we can conclude that DGKζ and ^DGKθ have specif - ic characteristics, tissue distribution and subcellular localisation, and that their activities are regulated via enzyme translocation and regulation of catalytic activity. These specificities link them to different signalling pathways and physiological functions. In this review, we provide evidence that most signalling functions are based on the regulation of the DAG-binding proteins PKC, Ras-GRP, chimaerins and Munc-13. Others are based on signalling properties of the DGK product PA. The data are summarised in Table 7. The (possible) implications of DGKζ activation for diverse signalling pathways are visualized in Figure 2.

5.1. DGK

ζ

as negative regulator of PKC signalling

Accumulating evidence suggests that DGKs attenuate PKC signalling. In fact, a dynamic interplay exists between DGKs and PKC, in which the two proteins regulate each other. Examples have been described especially for DGKζ, but also for DGKθ. The dynamic interplay between ^DGKs and PKC is best illustrated by four lines of evidence in which DGKζ and ^PKCα are shown to bind to each other and regulate activity and localisation of each other.

First, overexpression of DGKζ in ^HEK293 cells has been shown to inhibit PKCα activity, which was dependent on DGK activity (Luo et al., 2003a).

Second, PKCα regulates ^DGKζ localisation by phosphorylation of serines within the MARCKS-PSD of DGKζ (Topham et al., 1998). ^PKCα and ^PKCγ stimulated nuclear export of DGKζ in ^COS-7 cells. Nuclear export was regulated by phosphorylation of serines in the MARCKS-PSD of DGKζ as ^S/D-DGKζ mutants, in which these serines are replaced by aspartates that mimic phosphorylation, localised in the cytoplasm, whereas S/N-DGKζ mutants that cannot be phosphorylated localised in the nucleus. However, these studies were performed by confocal scanning laser microscopy and the percentage of cells with the specific localisation has not been described. Inhibiting nuclear export of DGKζ by ^PKC downregulation inhibited EGF-induced increases in nuclear DAG levels in A172 neuroblastoma

(27)

2626

cells, suggesting that DGKζ regulates nuclear ^DAG levels and, with that, nuclear PKC activity. Therefore, activated nuclear PKCα enhanced its own activity by stimulating DGKζ export. In contrast, in certain cell lines ^PKC-mediated phosphorylation translocated DGKζ to the plasma membrane, suggesting a negative feedback loop in which PKC facilitates DAG removal by DGKζ. For example, in ^T cells, PKC phosphorylation of the DGKζ-^MARCKS-PSD is required (but not sufficient) for DGKζ translocation to the plasma membrane upon receptor activation, as demon- strated by PKC downregulation and DGKζ-^MARCKS-PSD mutant studies (Santos et al., 2002). Furthermore, in C2C12 myoblasts, S/D-DGKζ showed increased plasma membrane localisation that was independent of receptor stimulation, suggesting that PKC-mediated phosphorylation is sufficient for membrane localisation in these cells (Hogan et al., 2001). However, PKC-regulated localisation of DGKζ in the latter study was only based on the S/D-DGKζ substitution mutant and no direct evidence involving PKC was shown. Thus, depending on the cell system, PKC stimulated DGKζ export from the nucleus or ^DGKζ translocation to the plasma membrane by phosphorylating the DGKζ-^MARCKS-PSD.

Third, PKC regulates DGKζ activity by phosphorylation. Part of the catalytic domain of DGKζ binds to ^PKCα and this interaction is inhibited upon ^TPA or PDGF- mediated activation of PKCα that phosphorylates serines of the ^DGKζ-^MARCKS- PSD (Luo et al., 2003a; Luo et al., 2003b). The S/D-DGKζ mutant in which ^PKC phosphorylation is mimicked was unable to bind PKCα, had reduced ^DGK activity, and was unable to inhibit PKCα activity. In contrast, the ^S/N-DGKζ mutant that cannot be phosphorylated, was insensitive for PKCα regulation. As expected, S/D-DGKζ-transfected ^HEK293 cells had higher DAG levels compared to S/N-DGKζ- transfected cells (Luo et al., 2003b), but whether DAG levels were increased compared to wild-type DGKζ was not shown. These results comply with a model in which DGKζ inhibits ^PKCα activity by keeping ^DAG levels low. Upon activation of PKCα, ^DGKζ is inhibited by phosphorylation which prolongs ^PKCα activation.

However, this model is based on overexpression studies, as a direct relationship between PKC activity, DAG levels and DGKζ activity has not been performed in cells that express only endogenous DGKζ and ^PKCα. In addition, these experiments were performed in whole cells rather than in nuclei, while PKCα was shown to regulate nuclear localisation of DGKζ, which may have an effect on ^DAG levels in whole cells. Therefore, it would be interesting to do these experiments also in isolated nuclei.

Fourth, we found an additional level of complexity by which PKCα regulates DGKζ activity. ^DGKζ was shown to bind to the retinoblastoma protein (p^RB) via its DGKζ-^MARCKS-PSD, which stimulated DGK activity (Los et al., 2006). Phos- phorylation of the DGKζ-^MARCKS-PSD by PKCα inhibited the interaction between DGKζ and p^RB (manuscript accepted). Therefore, in addition to the above men- tioned model, nuclear DGKζ bound to p^RB may keep DAG levels low, but upon activation of PKCα the interaction is disrupted and ^DGKζ stimulation by p^RB is replaced by PKCα-mediated inhibition of ^DGKζ activity and nuclear export of DGKζ.

(28)

Three examples have been described in which DGKζ inhibits ^PKC signalling.

In serum-starved HEK293 cells in which 60% of endogenous DGKζ is downregulat ed by DGKζ si^RNA, different PKC isoforms translocated to the plasma membrane and become active. In contrast, control cells expressing normal DGKζ levels do not contain active PKC, suggesting that DGKζ keeps ^DAG levels low to prevent PKC activation under serum-starved conditions (Avila-Flores et al., 2005). Further- more, in gonadotrope cells, DGKζ is recruited and activated to inhibit ^PKC signalling. Activation of gonadotrope releasing hormone (GnRH) receptors continuously activates PLCβ1 to produce ^Ca²⁺ and DAG. To regulate DAG levels and PKC signalling, GnRH receptors activate c-Src, which recruits DGKζ to the plasma membrane and stimulate its activity (Davidson et al., 2004). Thereby, GnRH receptors also in-duce an off-switch for DAG production to limit PKC-mediated gonadotropin release to a short time interval. In addition, a physiological role of DGKζ in attenuating PKC signalling was found in cardiac hypertrophy, in which cardiomyocytes increase in size as a result of hypertrophic signals or increased functional demands.

Several PKC isoforms, including PKCε, are involved in cardiomyocyte hypertrophy and heart failure and therefore PKC activity need to be regulated. In cardiomyocytes overexpressing DGKζ or in hearts of transgenic mice with cardiac-specific DGKζ expression, hypertrophic agonist-mediated ^PKCε translocation to the plasma membrane is inhibited (Takahashi et al., 2005; Arimoto et al., 2006). Further- more, DGKζ suppressed transcription of the hypertrophic gene atrial natriuretic factor, protein synthesis, enlargement of cardiomyocytes and increase in heart weight; characteristics for cardiomyocyte hypertrophy. Since hypertrophic ago- nists increase myocardial DAG levels in hearts from wild-type mice, which was completely suppressed in hearts from transgenic DGKζ mice, ^DGKζ probably inhibits cardiomyo-cyte hypertrophy by removing DAG and, with that, by attenuating PKC signalling. However, no direct evidence is provided that DGKζ regu - lates PKC activity and therefore a role for other DAG-binding proteins on cardiac hypertrophy cannot be excluded.

5.2. DGK

θ

as negative regulator of PKC signalling

DGKθ also has been shown to be regulated by ^PKC and, conversely, may negatively regulate PKC signalling. In A431 cells, activation of PKCε by its pseudo-ε^RACK peptide resulted in translocation of PKCε to the plasma membrane within 3 minutes, followed by DGKθ translocation within 6 minutes of ^PKCε stimulation (van Baal et al., 2005). Consequently, PKCε activation is restricted to a short time interval. How PKCε facilitates ^DGKθ translocation is not known, but since ^PKCε, and also ^PKCη, bound to and phosphorylated DGKθ, it possibly increases the affinity of ^DGKθ-^CRDs for DAG, induces DGKθ binding to the plasma membrane via another domain, or stimulates binding of proteins that guide DGKθ to the plasma membrane.

5.3. Role of DGK

ζ

in the Ras -ERK pathway

Many receptors signal via the Ras-Mitogen-activated protein kinase (MAPK) signalling pathway that is implicated in the regulation of a variety of responses, including proliferation, senescence, differentiation and apoptosis (Dent et al.,

(29)

2828

2003). This signalling cascade is stimulated upon activation of the small G-protein Ras that leads to the sequential activation of Raf1, MEK and MAPK, also known as extracellular signal-regulated kinase (ERK). Activated ERK, in turn, regulates downstream kinases or transcription factors (Sebolt-Leopold and Herrera, 2004).

Two DAG-dependent proteins regulate Ras activation: PKC and Ras guanyl nucleotide-releasing protein (Ras-GRP). It is not known how PKC stimulates Ras activity. The involvement of PKC in Ras-ERK signalling is based on PKC inhibitors and prolonged TPA treatment, which inhibit, in some cases, growth factor-mediated ERK activation (Chiloeches et al., 1999; Bonfil et al., 2004; Clerk et al., 2006). Ras- GRP is an exchange factor that activates Ras by releasing GDP and thus facilitat- ing the binding of GTP (Kazanietz, 2000). DAG binds to the C1 domain of Ras-GRP, which is required for Ras-GRP activation (Ebinu et al., 1998; Tognon et al., 1998).

DGKζ was found to inhibit ^Ras-GRP activity and downstream signalling via the Ras-ERK pathway (Topham and Prescott, 2001). In HEK293 cells, co-expression of Ras and Ras-GRP with DGKζ, but not kinase-inactive ^DGKζ, decreased ^Ras-GTP levels, suggesting that removal of DAG prevents Ras-GRP activation. Since DGKζ is part of a signalling complex containing Ras-GRP and H-Ras, it may spatially regulate DAG levels to locally regulate Ras activation after receptor stimulation.

Indeed, in several cell systems, DGKζ negatively regulates ^Ras-ERK signalling upon G-protein coupled receptor stimulation. Whether either PKC or Ras-GRP is involved in DAG-dependent Ras activation in these systems is not entirely clear.

Overexpression of DGKζ in ^Lβ^T2 gonadotrope cells or cardiomyocytes shortened ERK activation upon stimulation with, respectively, GnRH or endothelin (Davidson et al., 2004; Takahashi et al., 2005). In T cells in which DGKζ is overexpressed, ^ERK activation was decreased upon T cell receptor ligation, whereas overexpression of kinase-inactive DGKζ prolonged ^ERK activation (Zhong et al., 2002; Topham and Prescott, 2001). Physiological evidence for DGKζ in regulating ^Ras-ERK signalling was obtained in T cells isolated from DGKζ knock-out mice. These ^DGKζ- deficient T cells have increased active Ras and have enhanced and prolonged MEK and ERK activation upon T cell receptor stimulation (Zhong et al., 2003).

Since the Ras-ERK pathway regulates large numbers of protein kinases and transcription factors, the functional consequence of DGKζ-mediated inhibition is dependent on the cell system. In Lβ^T2 gonadotrope cells, DGKζ prevents ^ERK- mediated activation of p90 ribosomal S6 kinase (Davidson et al., 2004), which regulates transcription factors and is involved in protein synthesis. Since gonadotrope releasing hormone stimulates transcription of the β-subunits of luteinizing hormone and follicle stimulating hormone in an ERK-dependent way (Kanasaki et al., 2005), DGKζ may regulate transcription of these gonadotropins. In other cell systems, DGKζ inhibits ^ERK-mediated activation of the transcription factor AP1, which stimulates in cardiomyocytes the expression of the hypertrophic gene atrial natriuretic factor, and, in T cells, the activation marker CD69 (Takahashi et al., 2005; Zhong et al., 2002). Thus, although the functional outcome is completely different, DGKζ regulates signalling in these cells in a similar way.

Conversely, DGKζ was found to be a substrate for ^ERK. In Cos-7 and HEK293 cells stimulated with serum, DGKζ is phosphorylated on serine residues near

(30)

the ankyrin repeats that conform to the minimal MAPK phosphorylation consensus sequence (Abramovici et al., 2003; Avila-Flores et al., 2005). Activated ERK indeed phosphorylated DGKζ in ^COS-7 cells. This phosphorylation was found to block DGKζ association to the cytoskeleton (Abramovici et al., 2003). The functional consequence is not known. It would be interesting to investigate if phosphorylation of DGKζ by ^ERK reflects a feedback mechanism that allows tight regulation of the Ras-ERK pathway.

5.4. Role of DGK

ζ

in actin cytoskeleton rearrangement

DGKζ is involved in actin rearrangement by a mechanism that is not entirely clear.

At least, DGKζ associates with the cytoskeleton in different cell systems and is involved in Rac signalling, as will be discussed in more detail.

DGKζ binds to the cytoskeleton, which is subject to regulation. Phosphor- ylation of DGKζ by ^ERK inhibited binding to the cytoskeleton in C2C12 myoblasts, whereas GnRH stimulation of gonadotrope Lβ^T2 cells increased DGK activity (probably DGKζ) bound to the cytoskeleton (Abramovici et al., 2003; Davidson et al., 2004). In skeletal muscle and neurons, DGKζ binds indirectly to the actin cytoskeleton via syntrophins (Abramovici et al., 2003). Syntrophin is a component of the dystrophin glycoprotein complex that provides a strong mechanical link between the actin cytoskeleton and the extracellular matrix, which protects muscle cells for contraction-induced membrane damage (Lapidos et al., 2004). DGKζ binds in a complex to syntrophin and dystrophin, which localises DGKζ at the sarcolemma (Hogan et al., 2001; Abramovici et al., 2003). Since DGKζ is present at sites were active actin remodelling occurs, it is thought to regulate actin rearrangement (see below). Interestingly, in mdx (X-chromosome-linked muscular dystrophy) mice that contain a point-mutation in dystrophin causing a Duchenne muscular dystrophy- like phenotype, DGKζ was mislocalised to the nucleus, which may play a role in abnormal cytoskeletal changes that contribute to the pathogenesis.

Actin cytoskeletal remodelling is important for changes in cell morphology, adhesion and motility. Several signalling pathways regulate actin remodelling in which the Rho family of small GTPases, including RhoA, Rac and Cdc42, play a central role. In Swiss 3T3 cells, RhoA induces the formation of stress fibers and the assembly of focal adhesions, Rac regulates the formation of lamellipodia and membrane ruffles, whereas Cdc42 controls the formation of filopodia and focal adhesions (Etienne-Manneville and Hall, 2002). DGKζ binds to ^Rac1, both in its active (Rac1^V12) and inactive (Rac1^N17) form (Yakubchyk et al., 2005). Although the Rac1 binding region of DGKζ is localised in or near the cysteine-rich domains, point-mutations that mimic (PKC-mediated) phosphorylation of the DGKζ-^MARCKS- PSD (S/D-DGKζ) prevented the binding to ^Rac1^V12 as well as to Rac1^N17, suggesting that PKC regulates DGKζ binding to ^Rac. Indeed, TPA stimulation inhibited DGKζ binding to Rac1, which was abolished by a PKC inhibitor (Yakubchyk et al., 2005).

In several cell systems, DGKζ and ^Rac1 colocalised, which provides further evidence that the two proteins associate: In C2C12 myoblasts and in N1E-115 neuroblastoma cells, DGKζ colocalised with ^Rac and F-actin in membrane ruffles and at the leading edge of lamellipodia (Abramovici et al., 2003; Yakubchyk et al., 2005).

(31)

3030

Signalling

pathway/effect DGK function DGK isoform (possible) physiological

function References

PKC inhibiting PKC (DAG↓) DGKζ /DGKθ Downregulating PKC signalling

(Luo et al., 2003a; van Baal et al., 2005; Avila- Flores et al., 2005)

DGKζ prevention of cardiac

hypertrophy (Arimoto et al., 2006)

DGKζ inhibition of

gonadotropin release? (Davidson et al., 2004)

stimulating PKC (PA↑)

(Jose Lopez-Andreo et al., 2003; Limatola et al., 1994)

Ras /ERK inhibiting Ras-GRP/PKC

(DAG↓) ^DGKζ Inhibiting Ras-ERK

signalling

(Topham and Prescott, 2001; Zhong et al., 2003; Zhong et al., 2002; Takahashi et al., 2005; Davidson et al., 2004)

DGKζ Inhibition of

transcription?

(Davidson et al., 2004;

Takahashi et al., 2005)

DGKζ inhibition of T cell

activation (Zhong et al., 2003) stimulating Raf1

translocation or a Ras-GTPase inhibitor, or inhibiting Ras-GAP by PA?

(Rizzo et al., 2000; Tsai et al., 1990; Tsai et al., 1989)

Actin

rearrangement small GTPase signalling DGKζ

neurite formation / actin rearrangement in lamellipodia /

acetylcholine clustering at NMJ?

(Abramovici et al., 2003; Yakubchyk et al., 2005)

small GTPase signalling DGKθ (Houssa et al., 1999;

McMullan et al., 2006) inhibiting chimaerins

(DAG↓)?

stimulating PIP5K (PA↑) ^DGKζ (Luo et al., 2004a)

stimulating PAK1 (PA↑) (Bokoch et al., 1998)

Neuronal functioning

inhibiting Munc f amily proteins (DAG↓)

ceDGK1 DGKζ?

DGKθ?

regulating synaptic plasticity / involved in learning and memory?

(Nurrish et al., 1999) (Chase et al., 2004) (Matsuki et al., 2006) (McMullan et al., 2006) (Hozumi et al., 2003)

stimulating NSF (PA↑) (Manif ava et al., 2001)

Table 7. Signalling functions of DGKζ and DGKθ:

(32)

Signalling

pathway/effect DGK function DGK isoform (possible) physiological

function References

Cell cycle regulating DAG and/or PA

levels? ^DGKζ cell cycle regulation (Topham et al., 1998;

Los et al., 2006)

stimulating PP1cγ (^PA↑)

(Jones et al., 2005;

Jones and Hannun, 2002)

PA signalling stimulation of mTOR

(PA↑) ^DGKζ protein synthesis

regulation

(Avila-Flores et al., 2005)

stimulating Arf, NSF, kinesin (PA↑)

vesicular trafficking

regulation (Manif ava et al., 2001)

Fig. 2.

Signal transduction pathways and physiological functions in which DGKζ is involved.

DAG-binding proteins are in bold and PA-binding proteins are in bold and italics. Solid arrows represents signalling steps in which DGKζ is known to be involved. Dashed arrows indicate signalling steps that may be affected by DGKζ. Dashed arrows with 2 arrow heads indicate proteins that bind to

DGKζ. Physiological functions in which pRB was shown to be involved are boxed with a solid line, whereas possible physiological functions are boxed by a dashed line. Abbreviations: ABP, actin binding proteins; ANF, atrial natriuretic factor;

PAK1, p21-activated kinase; p90RSK, p90 ribosomal S6 kinase; NSF, N-ethylmale- imide-sensitive fusion protein; Arf, ADP- ribosylation factor; PP1γ, protein phos- phatase-1 catalytic subunit.

Table 7. (continued)

Diacylglycerol kinase theta and zeta isoforms: regulation of activity, protein binding partners and physiological functions