mulk: a eukaryotic multi-substrate lipid kinase2004/07/13 · mulk reactions were incubated at 94 c...

1

MuLK: A eukaryotic multi-substrate lipid kinase

David W. Waggoner, Laura Beth Johnson, Philip C. Mann, Valerie Morris, John Guastella,Sandra M. Bajjalieh*

Department of Pharmacology, University of Washington, Seattle,

*To whom correspondence should be addressed at:Department of PharmacologyUniversity of WashingtonBox 357280Seattle, WA 98195

Phone: 206-616-2962FAX: [email protected]

Running title: Multi-substrate lipid kinase

JBC Papers in Press. Published on July 13, 2004 as Manuscript M405932200

Copyright 2004 by The American Society for Biochemistry and Molecular Biology, Inc.

by guest on April 29, 2020

http://ww

w.jbc.org/

Dow

nloaded from

http://www.jbc.org/

2

Summary

We report the identification and characterization of a novel lipid kinase that phosphorylates

multiple substrates. This enzyme, which we term MuLK for Multi Lipid Kinase, does not belong

to a previously described lipid kinase family. MuLK has orthologs in many organisms and is

broadly expressed in human tissues. Although predicted to be a soluble protein, MuLK co-

fractionates with membranes and localizes to an internal membrane compartment. Recombinant

MuLK phosphorylates diacylglycerol, ceramide and 1-acyl-glycerol but not sphingosine.

Although its affinity for diacylglycerol and ceramide are similar, MuLK exhibits a higher Vmax

toward diacylglycerol in vitro, consistent with it acting primarily as a diacylglycerol kinase.

MuLK activity was inhibited by sphingosine and enhanced by cardiolipin. It was stimulated by

calcium when magnesium concentrations were low and inhibited by calcium when magnesium

concentrations were high. The effects of charged lipids and cations on MuLK activity in vitro

suggest that its activity in vivo is tightly regulated by cellular conditions.


http://ww

w.jbc.org/

Dow

nloaded from

http://www.jbc.org/

3

Introduction

The membranes of eukaryotic organelles contain a shifting constellation of lipids that, in

addition to their structural role, also serve as modulators of protein function, scaffolding

molecules and ligands for G protein-coupled receptors. The generation, phosphorylation, and

dephosphorylation of mono- and diacylglycerols (DG), sphingosine, and ceramide produce

signaling lipids that modulate a vast array of cellular processes (1-5). In animal cells, the kinases

that act on DG (6), ceramide (7) and sphingosine (8) constitute three different families of

enzymes, each of which demonstrates substrate selectivity. In contrast, E. coli expresses a lipid

kinase that phosphorylates all three substrates (9). Although animal cell lipid kinases are specific

and clearly belong to distinct families based on their amino acid sequences, they share a similar

catalytic domain that was first identified in DG kinase a (10) and is therefore known as a DG

kinase domain.

Sequencing of the human and mouse genomes has revealed multiple putative lipid

kinases, several of which contain a DG kinase domain. In an attempt to identify cDNAs

encoding a calcium-activated ceramide kinase that co-purifies with neurotransmitter-containing

(synaptic) vesicles (11), we searched sequence databases for orphan lipid kinases that might

encode a ceramide kinase. We report here the characterization of a lipid kinase that

phosphorylates ceramide but also demonstrates significant activity toward DG and

monoacylglycerols. This enzyme, which we term MuLK for multi-substrate lipid kinase, is a

novel lipid kinase.


http://ww

w.jbc.org/

Dow

nloaded from

http://www.jbc.org/

4

Experimental procedures:

Materials -- Restriction endonucleases were purchased from Fermentas (Hanover, MD). The

cloning vector pFLAG-CTC, anti-FLAG agarose affinity gel, 3X-FLAG peptide and anti-FLAG

M2 monoclonal antibody were obtained from Sigma (Saint Louis, MO). Primers were purchased

from Sigma Genosys (The Woodlands, TX). EST clones of human (BE889885) and mouse

(AW321722) MuLK and mouse CERK (6400872) were obtained from ResGen (Invitrogen). The

normalized human cDNA library used for expression analysis and the pIRES-EGFP vector were

purchased from BD Bioscience (Clontech). Ceramide (bovine brain), 1, 2 –dioleoyl-sn-glycerol,

sn1-monooleoylglycerol, tetraoleoylcardiolipin and sphingosine were obtained from Avanti

Polar Lipids. 1-arachidonylglycerol (1-AG) and 2-arachidonylglycerol (2-AG) were purchased

from Cayman Chemical Company. EST clones of human MuLK (BE889885), mouse MuLK

(AW321722) and CERK (6400872) were obtained from ResGen (Invitrogen). Lipofectamine

was from Invitrogen. The monoclonal anti-GFP antibody (B-2) was obtained from Santa Cruz

Biotechnologies and horseradish peroxidase conjugated goat anti-mouse secondary antibody was

from Zymed. The chemiluminescent detection reagent was DuraWest from Pierce. Fluoromount

G and 16% paraformaldehyde were from Electron Microscopy Sciences. [g-32P]ATP (3000

Ci/mmol) was purchased from Perkin-Elmer. Triton X-100, b-octylglucoside, aprotinin,

pepstatin, leupeptin, and Complete-Mini protease inhibitors (without EGTA) were obtained from

Roche. Recombinant E. coli DG kinase, Hoechst stain and other reagents were from Sigma.

Genome Search for Potential Ceramide Kinases -- The human sphingosine kinase, sphingosine

kinase 2, was used as a query sequence in a TBLASTN search of the human genome database.

Ten hits with expect values of 7 ¥ 10-8 or less were obtained. As anticipated, the highest scoring

hits represented the known mammalian sphingosine kinase genes, as well as related DG kinase


http://ww

w.jbc.org/

Dow

nloaded from

http://www.jbc.org/

5

genes. However, the search also revealed a gene with a region that is 44% homologous to

hSPK2. The region of highest homology corresponded to a potential kinase catalytic domain that

is present in both sphingosine and DG kinases. This highest scoring sequence was used in a

BLASTP search of the non-redundant protein database, which revealed three unpublished

entries. One entry was identified as a putative lipid kinase, but did not describe any functional

data. The full-length protein sequence of the putative lipid kinase was then used to re-query the

human genome database and three homologous sequences were identified.

Sequence comparisons and domain predictions – The sequences of MuLK, CERK, sphingosine

and DG kinases were aligned using ClustalW with the BLOSUM protein weight matrix. A

phylogenetic tree was constructed and genetic distances calculated (percent divergence) using

the neighbor joining method of Saitou and Nei (12). The sequences included were H. sapiens

MuLK (CAB93536), M. musculus MuLK (CAC06108), H. sapiens sphingosine kinase 1

(Q9NYA1), M. musculus sphingosine kinase 1 (AAH37710), H. sapiens sphingosine kinase 2

(Q9NRAO), M. musculus sphingosine kinase 2 (Q9JIA7), H. sapiens ceramide kinase

(BAC01154), M. musculus ceramide kinase (NP_663450), H. sapiens DG kinases a

(AAH223523), d (Q16760), e (P52429), q (NP_001338), and z (Q13574) and E. coli DG Kinase

(P00556). Gene products of putative MuLKs from X. laevis (AAH43761), D. rerio (AAH45347)

were identified using a BLASTP search and compared to human and mouse MuLK sequences

using ClustalW. ClustalW identifies conservative changes by dividing amino acid residues into

5 groups as follows: small and hydrophobic; acidic; basic; hydroxyl amine and basic; others

(http://www/ebi.ac.uk/clustalw/). Phosphorylation consensus sites and protein motifs across

species were predicted using SwissProt Prosite.


http://ww

w.jbc.org/

Dow

nloaded from

http://www.jbc.org/

6

MuLK-EGFP expression and subcellular fractionation –A carboxy-terminal fusion of enhanced

green fluorescent protein to MuLK (MuLK-EGFP) was generated by ligating an EGFP PCR

product into Sal I and Not I sites of a plasmid already containing MuLK. This effectively

eliminated the intervening IRES and generated the desired fusion construct. To create the C-

terminal fusion of EGFP to CERK, CERK was amplified using PCR from IMAGE clone

6400872 (Invitrogen) and inserted into EcoR1/Sal1 sites of the pMuLK-EGFP construct,

replacing MuLK. To evaluate cellular localization, a 60 cm plate of 80% confluent HEK293T

cells was transfected with 1.6 ug of DNA of each construct (pIRES-EGFP, pMuLK-EGFP, and

pCERK-EGFP) using Lipofectamine 2000. A transfection efficiency of 60-70% was determined

by fluorescent microscopy. After 48 hours, cells were washed with ice-cold PBS and scraped

into PBS containing 2 mg/ml aprotinin, 0.5 mg/ml leupeptin, 0.7 mg/ml pepstatin and 0.2 mM

PMSF, and then lysed by sonication. Intact cells and nuclei were removed by centrifuging

2000xg for 5 min. Soluble and insoluble fractions of the cleared lysate were then generated by

ultracentifugation at 100,000xg for 1 hour at 4°C. Equal volumes of these samples were

separated on a 12.5% Tris-Glycine gel under reducing and denaturing conditions and then

transferred to PVDF. Blots were probed with monoclonal anti-GFP (B-2) antibody, followed by

HRP-conjugated goat-anti-mouse secondary antibody and detected using the Pierce DuraWest

enhanced chemiluminescence reagent. To visualize MuLK-EGFP expression in situ, transfected

cells were fixed 48 hours after transfection (4% paraformaldehyde for 20 min), washed 3 times

in 0.1 M glycine, incubated with Hoechst dye to label nuclei and coverslipped using

Fluoromount G as a mounting medium. EGFP fluorescence was visualized using an Applied

Precision Deltavision microscope and images were captured using Softworx software.


http://ww

w.jbc.org/

Dow

nloaded from

http://www.jbc.org/

7

mRNA expression analysis — To determine expression levels of MuLK in various human tissues

a 528 base pair product, encompassing the 3’ end of the coding sequence and 119 base pairs of

the 3’ untranslated region of human MuLK, was amplified from a BD Biosciences normalized

cDNA panel. Reactions were conducted according to the manufacturer’s directions. The

following two 29 base pair primers were used; 5’ – GGCACAACCACAGGATGCCCTTTCCCAAG; 3’ –

TGCCATGAAAATGCCCTGGGGACCCTCTG. MuLK reactions were incubated at 94°C for 30s and

followed by 28 cycles of 94° C for 5 sec and 68° C for 2 min. As a control for template loading,

a portion of glyceraldehyde 3-phosphate dehydrogenase (GAPDH) was also amplified using

manufacturer-supplied primers and cycle parameters. The relative amounts of GAPDH product

generated from each tissue were similar to the levels reported by the manufacturer. PCR products

were separated in a 2% agarose gel and stained with ethidium bromide. Product band net

intensity was quantified on a Kodak Image Station 440CF.

Expression of C-terminal FLAG-tagged MuLK -- A carboxy-terminal FLAG-epitope-tagged

mouse MuLK (MuLK-FLAG) was created by inserting MuLK into pFLAG-CTC. This construct

was expressed in BL21, a protease-deficient strain of E. coli. To generate protein, cultures of

transformed bacteria were grown under selection overnight, diluted 10-fold and then grown to an

A600 of 2 at which point protein expression was induced by adding 100 mM IPTG for four hours.

After induction, bacteria were harvested by centrifugation, resuspended in ice cold 10 mM

MOPS, pH 7.2, containing 25 mM NaCl, 1 mM EGTA and protease inhibitors (1 pellet/10 ml

buffer) and probe sonicated. The lysate was centrifuged at 20,000 x g for 20 min and the

resulting supernatant was incubated for 1-14 hr with anti-FLAG IgG-coated agarose resin at 4o C

with continual turning. MuLK was eluted with 3-5 column volumes of MOPS buffer (see above)

containing 0.1 mg/ml 3X-FLAG. This produced a preparation enriched in MuLK. An unrelated


http://ww

w.jbc.org/

Dow

nloaded from

http://www.jbc.org/

8

pFLAG-CTC fusion protein, which served as a negative control, was generated and purified

using an identical procedure. Glycerol or ethylene glycol, and dithiothreitol were added to the

eluate to a final concentration of 20% vol/vol and 1 mM, respectively. For kinetic assays, the

recombinant protein was stored at 4oC and used within a week. Expression and purification of a

47 KDa protein was confirmed by Western analysis using the anti-FLAG M2 antibody.

Expression levels of MuLK in E. coli varied between preparations and protein yields were

generally low.

Lipid kinase activity assay – The activity of purified, recombinant mouse MuLK was assayed in

b-octylglucoside as previously described (13) or as follows. Briefly, in a final volume of 100 ml

purified recombinant MuLK was combined with lipid-detergent mixed micelles containing the

indicated substrate, detergent and cardiolipin at one-tenth the concentration of the lipid substrate

(unless stated otherwise). This mixture was buffered with 10 mM MOPS (pH 7.2) containing 100

mM NaCl, 1 mM EGTA, 3 mM total CaCl2 (2 mM free calcium, unless indicated), and was

initiated by the addition of ATP/MgCl2 (1 mM, 5 mCi [g-32P]ATP and 5 mM MgCl2 in the final

assay). The assay was incubated with constant agitation at 37o C for 20 min. The reaction was

stopped by addition of 1 ml chloroform:methanol (1:1). After brief vortexing, 350 ml of 2 M

KCl, 2 mM H3PO4 was added to generate two phases. The two phases were separated after

vigorous mixing and the lower (organic) phase removed to a second tube, the volume reduced

under a stream of nitrogen gas and the entire sample analyzed by thin layer chromatography

(TLC). TLC plates were developed for 80 minutes in chloroform, methanol, acetic acid

(65:15:5). Phosphorylated reaction products were identified by co-migration with standards

generated using E. coli DG kinase (Sigma, Saint Louis, MO) and quantified by scraping and

scintillation counting or by using STORM Image analysis (Molecular Dynamics Corp.) and


http://ww

w.jbc.org/

Dow

nloaded from

http://www.jbc.org/

9

Imagequant software. An unrelated purified pFLAG-CTC fusion protein containing the first 163

amino acids of synaptic vesicle protein 2A (SV2AN) was assayed as a control for contamination

by endogenous E. coli lipid kinases. Under these conditions, all reactions were linear with time,

proportional to added enzyme, and substrate conversion was less than two percent. Calcium

concentrations were calculated using the MaxChelator Program

(http://www.stanford/edu/~cpatton/webmaxcS.htm)). Kinetic parameters were calculated using

Prism Graph Pad software.

Results

Identification of two novel lipid kinases by homology searching

Based on the hypothesis that ceramide kinases would share structural and functional

domains with sphingosine kinase, we performed a database search using one of the human

sphingosine kinases, sphingosine kinase 2, as a query sequence as described under Experimental

Procedures. This search revealed a gene that contained both a region that is 44% homologous to

SPK2 and a potential lipid kinase catalytic domain that is present in sphingosine and DG kinases

(14). The predicted full-length protein encoded by this gene was obtained from a BLASTP

search. As described below, we have named this protein MuLK. The sequence of MuLK was

then used to re-query the human genome database. Three additional gene sequences were

identified. All three are distantly related to sphingosine and DG kinases, with sequence

homologies of 10-22%. One of these was recently described by Sugiura, et al., as a ceramide-

specific kinase which they have termed CERK (7). The other two shared 68-87% amino acid

sequence identity with MuLK. These three proteins therefore appeared to constitute a gene


http://ww

w.jbc.org/

Dow

nloaded from

http://www.jbc.org/

10

family. A BLASTP search using human MuLK as a query sequence revealed a mouse homolog

as well as putative homologs in multiple organisms.

The sequence of MuLK

Figure 1 shows the sequence of mouse MuLK. The most notable feature of the protein is

a DG kinase domain near the amino terminus. MuLK also contains a putative nuclear

localization sequence that overlaps with the amino terminal portion of the catalytic domain. The

nuclear localization domain is found in all MuLK orthologs except C. elegans. It is also found in

Type IV and V DG kinases (z, i, and q) (15) and has been shown to mediate the nuclear

localization of DG kinase z (16). This nuclear localization sequence occurs with high frequency

in the protein database however, so it remains to be determined whether it regulates the

subcellular localization of MuLK.

A comparison of putative MuLKs across selected vertebrate species revealed that the

amino acid sequence identity is well conserved among vertebrate MuLK proteins, especially

within the amino-terminal region that includes the DG kinase catalytic domain (Figure 1).

Additional MuLK orthologs were identified in D. melanogaster (AAF52895) and C. elegans

(T16422), A. thaliana (AAM91597) and O. sativa (BAC65388) indicating that MuLK-like

proteins are expressed in most eukaryotes. There was no MuLK-related protein sequence in S.

cerevisiae however, suggesting that, like DG and ceramide kinases, MuLK is expressed only in

higher eukaryotes.

The human MuLK gene is located on chromosome 7q34 and contains 15 exons. The

mouse gene is located on chromosome 6B1 and also contains 15 exons. The other two human

MuLK genes found in our original query are on the X chromosome (Xq26) and the Y

chromosome (4q11.2). These genes lack many of the introns present in the chromosome 7


http://ww

w.jbc.org/

Dow

nloaded from

http://www.jbc.org/

11

MuLK gene and are riddled with nonsense, missense, and frameshift mutations suggesting that

they are non-functional retroposons. Similarly, the mouse genome has an intron-less MuLK

sequence on chromosome 13C3 that has two frameshift mutations.

Several protein kinase substrate consensus sites in MuLK were identified using

SwissProt-Prosite. Mouse MuLK contains nine casein kinase II sites, seven Protein Kinase C

sites and two Protein Kinase A sites. These sites are conserved between human and mouse, but

are not present in MuLK orthologs from zebrafish, fruit fly, nematode, Arabidopsis and rice. It

is not known if any are bona fide phosphorylation sites, though the regulation of lipid kinases by

protein kinase C isoforms reported for DG kinase Types I (17), IV (16) and V (15) and for

sphingosine kinase I (18) suggests that MuLK may be also regulated by phosphorylation.

To determine whether MuLK is a member of a previously identified lipid kinase family,

we compared MuLK sequences to those of other lipid kinases. Figure 2 shows a phylogram

comparison of MuLK to DG, ceramide and sphingosine kinases. Bootstrap analysis of the

aligned sequences revealed that human and mouse MuLK segregated to a unique branch. This

indicates that MuLK is a distinct lipid kinase that is not a member of a previously described

family.

MuLK is a ubiquitously expressed, membrane-associated protein.

Analysis of the amino acid sequence of MuLK predicts no signal sequence,

transmembrane domains or lipid modifications. This suggests that MuLK is a cytosolic, soluble

protein. To test this, we expressed MuLK fused to Enhanced Green Fluorescent Protein (EGFP)

in HEK293T cells and compared the proportion of MuLK in the cytosolic and membrane

fractions as described under Experimental Procedures. As a control for the effects of EGFP on


http://ww

w.jbc.org/

Dow

nloaded from

http://www.jbc.org/

12

protein localization, we also analyzed the localization of mouse CERK-EGFP, since untagged

CERK was previously shown to be largely membrane-associated (7). Western analyses of

cytosolic and membrane fractions revealed that the majority of MuLK was found in the

particulate fraction (100,000 x g pellet), a distribution similar to that of CERK. EGFP, expressed

alone, was largely cytosolic (Figure 3A). When expressed in Chinese Hamster Ovary cells, a

fibroblast cell line, MuLK-EGFP localized to an internal membrane compartment (Figure 3B).

This indicates that in unstimulated fibroblasts the enzyme is associated with endomembranes

rather than being cytoplasmic or nuclear as suggested by its sequence.

To determine the tissue distribution of MuLK, we surveyed RNA expression levels by

amplification of normalized cDNA pools from various tissues as described under Experimental

Procedures. Figure 4 shows that MuLK was expressed in all tissues surveyed. Expression was

highest in neuroendocrine organs, such as pancreas and brain, and relatively lowest in skeletal

muscle, kidney, and lung.

MuLK is a multi-substrate lipid kinase.

To determine whether MuLK is a bona fide lipid kinase and to test its substrate

specificity, we analyzed the ability of recombinant MuLK to phosphorylate various lipids in vitro

using an assay in which substrates were presented in detergent-lipid mixed micelles. Purified

recombinant MuLK-FLAG phosphorylated 1, 2-dioleoyl-sn-glycerol as well as

monoacylglycerols. It also demonstrated significant activity towards ceramide but not

sphingosine (Figure 5A). This indicates that MuLK can phosphorylate a number of neutral lipid

species but does exhibit some substrate selectivity. We have indirect evidence that MuLK does

not phosphorylate soluble polyols, since up to 20% (vol/vol) glycerol or ethylene glycol had no


http://ww

w.jbc.org/

Dow

nloaded from

http://www.jbc.org/

13

effect on its ceramide kinase activity (not shown). MuLK phosphorylated sn1-oleoylglycerol

and 1-arachidonoylglycerol (1-AG) to a similar extent. It showed little activity toward 2-

arachidonoylglycerol (2-AG) when the assay was conducted using Triton X-100 mixed micelles

(Figure 4A), however when kinase selectivity was assessed using b-octylglucoside mixed

micelles, a greater relative activity toward 2-AG was detected (not shown). Despite these

observations, we cannot be certain that 2-AG is a bona fide substrate since preparations of 2-AG

readily isomerize to 1-AG (19).

To confirm that the activity measured was not due to contaminating E. coli DG kinase, an

unrelated recombinant protein, SV2AN, purified in the same way as recombinant MuLK-FLAG

was assayed for lipid kinase activity. It exhibited minimal activity, indicating that MuLK was

responsible for the majority of the kinase activity measured (not shown).

MuLK DG kinase activity was inhibited 86-94% by the presence of 40-fold higher

concentrations of alternate substrates (Figure 5B). Assuming only one active site on MuLK,

this inhibition, together with the finding that MuLK phosphorylates ceramide, mono- and

diacylglycerols, suggests that MuLK’s substrates act as competitive inhibitors of one another.

Interestingly, sphingosine also inhibited DG kinase activity (Figure 5B), even though it was not a

substrate for the enzyme. Both ceramide and DG kinase activities were inhibited by sphingosine

in a dose-dependent manner (not shown).

MuLK demonstrates similar affinities for ceramide and DG but a higher Vmax toward DG.

To analyze further the substrate specificity of MuLK, we measured the apparent Km and

Vmax towards ceramide and DG in two different detergent-based assays. Using the b-

octylglucoside-based assay previously used to assay ceramide kinase activity (13) we found a


http://ww

w.jbc.org/

Dow

nloaded from

http://www.jbc.org/

14

trend towards a slightly lower Km for ceramide and a consistently higher Vmax for DG (not

shown). In a Triton X-100-based assay the apparent Km for ceramide (34 uM) again was slightly

lower but was similar to that for DG (45 uM) suggesting that the enzyme binds both lipids with

roughly equal affinity. Similar Km values were determined for dipalmitoylglycerol in TX-100

(not shown), suggesting that acyl chain length does not have significant effects on substrate

affinity. In this assay the Vmax for dioleoylglycerol (159 nmol/min/mg) was 4-5 times the rate

measured for ceramide (37 nmol/min/mg) (Figure 6A).

We also performed assays in which we varied the mol% of substrate in Triton X-100

micelles. Data derived in this manner can be compared to a surface dilution kinetic model (20).

We found that ceramide kinase activity conformed to such a model, yielding a Km of 3.78 mol%

and a Vmax of 68 nmol/min/mg protein (Figure 6B). By contrast, the Km for dioleoylglycerol was

independent of the surface concentrations tested, suggesting that it was too low to measure

(Figure 6C). These results were not an aberration of the assay system since surface dilution

kinetic parameters for a DG kinase activity from T-cells has been determined using similar

conditions (D. Waggoner, unpublished observations). These results highlight the sensitivity of

MuLK to substrate presentation and suggest that the affinities and kinetics measured in vitro

using recombinant enzyme may differ from those of the enzyme in vivo.

MuLK activity is modulated by surface charge and the ionic environment.

Kinases require a counter ion for ATP, typically Mg2. In some cases kinase activity is

also modulated by those same ions directly. DG kinases from E. coli (21) and pig brain (22) are

both stimulated by magnesium in excess of ATP concentrations. We examined the Mg2+


http://ww

w.jbc.org/

Dow

nloaded from

http://www.jbc.org/

15

requirement for MuLK’s DG and ceramide kinase activities and found that it stimulated MuLK

kinase activity maximally when included at 5-10 molar excess over ATP (not shown).

All reported endogenous ceramide kinase activities, as well as the recently identified

ceramide-specific kinase, CERK, are stimulated by Ca2+ (7,11,23,24). Type I DG kinases are

also activated by Ca2+ (6). To determine whether MuLK is regulated by Ca2+, we compared

MuLK activity in the presence of 1 mM EGTA (which buffers Ca2+at approximately 10-9 M) to

its activity in the presence of 1 mM free Ca2+. We found that the effects of Ca2+ varied with the

concentrations of Mg2+ in the assay (Figure 7). At low Mg2+/ATP ratios (0.5:1) Ca2+ stimulated

phosphorylation of both DG and ceramide two- to nine-fold. By contrast, when Mg2+ was in

excess (5:1 Mg/ATP), Ca2+ inhibited kinase activity two- to five-fold.

Cardiolipin has been reported to be an activator of E. coli DG kinase (9) and for that

reason was included in the assay. We found that MuLK required the presence of cardiolipin in

detergent micelles for ceramide kinase activity (not shown). Moreover, the stimulatory effect of

cardiolipin was concentration dependent for both ceramide and DG kinase activity and was most

dramatic at low Ca2+ concentrations (Figure 7). Together these findings indicate that MuLK

activity is influenced by the ionic environment and by the charge of the lipid surface.

Discussion

We have described a novel lipid kinase that differs from known lipid kinases not only in

sequence but also by demonstrating a broad substrate preference. Some DG kinases also

phosphorylate monacylglycerols (25) but little to no detectible ceramide kinase activity has been

reported for any of the isoforms ((26,27) and H. Kanoh, personal communication). Analogously,

the ceramide kinase CERK displays no activity toward DG (7). The only other lipid kinase with


http://ww

w.jbc.org/

Dow

nloaded from

http://www.jbc.org/

16

the broad substrate specificity of MuLK is E. coli DG kinase (9) (for review see (28)).

Therefore, MuLK’s broad substrate selectivity appears to be unique among eukaryotic lipid

kinases. MuLK orthologs were found in many eukaryotes ranging from plants to human, and it

is expressed in all human tissues surveyed. This suggests that MuLK provides an important

function that is required by all cells. We found no MuLK homologs in yeast, however,

consistent with its function being subsumed by another enzyme in lower eukaryotes.

MuLK activity demonstrates complex regulation.

The effects of lipids, detergents and ions on MuLK activity suggest that its action in vivo

is likely to be under complex regulation, a feature MuLK shares with other lipid kinases. The

requirement for, and stimulatory effect of cardiolipin parallels the effects of anionic lipids on E.

coli DG kinase (21,29), and eukaryotic DG kinases a (30,31) and z (32), and contrasts with

inhibitory effects on other isoforms (10,27,32,33). The cationic lipid sphingosine, which exhibits

both stimulatory and inhibitory effects on different DG kinase isoforms (34-37) potently inhibits

MuLK activity. We originally hypothesized that sphingosine acts as a competitive inhibitor of

MuLK, however inhibition was not stereospecific and was sensitive to the concentration of

cardiolipin in the assay (not shown) suggesting sphingosine’s inhibitory action is more complex

and may include charge neutralization at the lipid (substrate) surface.

We found that Ca2+ could either stimulate or inhibit MuLK activity, depending on the

concentrations of Mg2+ and cardiolipin in the reaction. Similar reciprocal magnesium-dependent

effects have been reported for PI3K (37), and calcium effects on DG kinase a vary with reaction

conditions and are phosphatidylserine-dependent (33). The inhibitory effect of Ca2+ could reflect

surface charge neutralization of cardiolipin, which in turn decreases enzyme-membrane


http://ww

w.jbc.org/

Dow

nloaded from

http://www.jbc.org/

17

association. Consistent with the conditional effects of Ca2+, MuLK lacks a consensus calcium-

binding domain like the EF-hand domains present in Type I DG kinases. On the other hand, an

alignment of MuLK with DG kinases a, e and z (the latter two lack EF hand domains and are

regulated by Ca2+ in a complex fashion (32)), reveals that all four proteins contain a conserved

aspartate in their carboxy terminal domains which is important to the stimulatory effect of Ca2+

on DG kinase a (30). Therefore, MuLK may share part of a novel motif that mediates the effects

of Ca2+ on these lipid kinases.

Is MuLK a chameleon kinase?

The activity of MuLK in vitro was greatest toward DG, suggesting that it may act

primarily as a DG kinase in vivo. On the other hand, given that detergents, lipids and ionic

conditions critically affect MuLK activity, it is likely that both substrate preference and the

amount of lipid phosphate generated in vivo may vary from those measured in vitro. This feature

is not unique to MuLK; for example, E. coli DG kinase expressed in mammalian cells causes

large increases in ceramide 1-phosphate levels, rather than phosphatidic acid (38). The more

pronounced effect of cardiolipin on ceramide (vs. DG) kinase activity suggests that MuLK’s

product(s) could change with alterations in its lipid environment.

In addition to phosphorylation of DG and ceramide, our results indicate that MuLK also

demonstrates measurable phosphorylation of sn-1 monoacylglycerols. Given that we did not

optimize the assay for that substrate, it raises the possibility that lysophosphatidic acids, which

are ligands for G-protein coupled receptors (2,3), may have a source other than phospholipases

A1/A2 or lyso- phospholipase D (39). Further, MuLK phosphorylation of 2-AG, an endogenous

cannabinoid, raises the possibility that MuLK could act in that signaling pathway as well.

Finally, the finding that MuLK activity is inhibited by sphingosine indicates that MuLK


http://ww

w.jbc.org/

Dow

nloaded from

http://www.jbc.org/

18

represents another point of cross-talk between glycerol- and sphingosine-based lipid signaling

pathways (40).

Given its broad substrate selectivity and modulation by membrane components and ionic

environment, MuLK could function as a “chameleon” kinase, whose action depends on cellular

location and conditions. Future studies of MuLK’s action in vivo will provide important clues to

the role of this unique enzyme in cellular function.


http://ww

w.jbc.org/

Dow

nloaded from

http://www.jbc.org/

19

References

1. Ghosh, S., Strum, J. C., and Bell, R. M. (1997) Faseb J 11, 45-50

2. Luquain, C., Sciorra, V. A., and Morris, A. J. (2003) Trends Biochem Sci 28, 377-383

3. Mills, G. B., and Moolenaar, W. H. (2003) Nat Rev Cancer 3, 582-591

4. Ye, X., Fukushima, N., Kingsbury, M. A., and Chun, J. (2002) Neuroreport 13, 2169-2175

5. Waggoner, D. W., Xu, J., Singh, I., Jasinska, R., Zhang, Q. X., and Brindley, D. N.(1999) Biochim Biophys Acta 1439, 299-316

6. Topham, M. K., and Prescott, S. M. (1999) J Biol Chem 274, 11447-11450

7. Sugiura, M., Kono, K., Liu, H., Shimizugawa, T., Minekura, H., Spiegel, S., andKohama, T. (2002) J Biol Chem 277, 23294-23300

8. Liu, H., Chakravarty, D., Maceyka, M., Milstien, S., and Spiegel, S. (2002) Prog NucleicAcid Res Mol Biol 71, 493-511

9. Schneider, E. G., and Kennedy, E. P. (1973) J Biol Chem 248, 3739-3741

10. Sakane, F., Imai, S., Kai, M., Wada, I., and Kanoh, H. (1996) J Biol Chem 271, 8394-8401

11. Bajjalieh, S. M., Martin, T. F., and Floor, E. (1989) J Biol Chem 264, 14354-14360

12. Saitou, N., and Nei, M. (1987) Mol Biol Evol 4, 406-425

13. Bajjalieh, S. M., and Batchelor, R. H. (2000) Methods in Enzymology 311, 207-213

14. Topham, M. K., and Prescott, S. M. (1999) J Biol Chem 274, 11447-11450

15. van Blitterswijk, W. J., and Houssa, B. (2000) Cell Signal 12, 595-605

16. Topham, M. K., Bunting, M., Zimmerman, G. A., McIntyre, T. M., Blackshear, P. J., andPrescott, S. M. (1998) Nature 394, 697-700

17. Kanoh, H., Yamada, K., Sakane, F., and Imaizumi, T. (1989) Biochem J 258, 455-462

18. Johnson, K. R., Becker, K. P., Facchinetti, M. M., Hannun, Y. A., and Obeid, L. M.(2002) J Biol Chem 277, 35257-35262

19. Stella, N., Schweitzer, P., and Piomelli, D. (1997) Nature 388, 773-778


http://ww

w.jbc.org/

Dow

nloaded from

http://www.jbc.org/

20

20. Carman, G. M., Deems, R. A., and Dennis, E. A. (1995) J Biol Chem 270, 18711-18714

21. Walsh, J. P., and Bell, R. M. (1986) J Biol Chem 261, 6239-6247

22. Kanoh, H., Kondoh, H., and Ono, T. (1983) J Biol Chem 258, 1767-1774

23. Dressler, K. A., and Kolesnick, R. N. (1990) J Biol Chem J Biol Chem 265, 14927-14921

24. Hinkovska-Galcheva, V. T., Boxer, L. A., Mansfield, P. J., Harsh, D., Blackwood, A.,and Shayman, J. A. (1998) J Biol Chem 273, 33203-33209

25. Kanoh, H., Iwata, T., Ono, T., and Suzuki, T. (1986) J Biol Chem 261, 5597-5602

26. Younes, A., Kahn, D. W., Besterman, J. M., Bittman, R., Byun, H. S., and Kolesnick, R.N. (1992) J Biol Chem 267, 842-847

27. Houssa, B., Schaap, D., van der Wal, J., Goto, K., Kondo, H., Yamakawa, A., Shibata,M., Takenawa, T., and van Blitterswijk, W. J. (1997) J Biol Chem 272, 10422-10428

28. Bielawska, A., Perry, D. K., and Hannun, Y. A. (2001) Anal Biochem 298, 141-150

29. Walsh, J. P., and Bell, R. M. (1986) J Biol Chem 261, 15062-15069

30. Abe, T., Lu, X., Jiang, Y., Boccone, C. E., Qian, S., Vattem, K. M., Wek, R. C., andWalsh, J. P. (2003) Biochem J 375, 673-680

31. Sakane, F., Yamada, K., Imai, S., and Kanoh, H. (1991) J Biol Chem 266, 7096-7100

32. Thirugnanam, S., Topham, M. K., and Epand, R. M. (2001) Biochemistry 40, 10607-10613

33. Walsh, J. P., Suen, R., Lemaitre, R. N., and Glomset, J. A. (1994) J Biol Chem 269,21155-21164

34. Yamada, K., Sakane, F., Imai, S., and Takemura, H. (1993) Biochim Biophys Acta 1169,217-224

35. Yamada, K., and Sakane, F. (1993) Biochim Biophys Acta 1169, 211-216

36. Sakane, F., Yamada, K., and Kanoh, H. (1989) FEBS Lett 255, 409-413

37. Sakane, F., Yamada, K., Kanoh, H., Yokoyama, C., and Tanabe, T. (1990) Nature 344,345-348

38. Ramer, J. K., and Bell, R. M. (1990) J Biol Chem 265, 16478-16483


http://ww

w.jbc.org/

Dow

nloaded from

http://www.jbc.org/

21

39. Umezu-Goto, M., Kishi, Y., Taira, A., Hama, K., Dohmae, N., Takio, K., Yamori, T.,Mills, G. B., Inoue, K., Aoki, J., and Arai, H. (2002) J Cell Biol 158, 227-233

40. Brindley, D. N., Abousalham, A., Kikuchi, Y., Wang, C. N., and Waggoner, D. W.(1996) Biochem Cell Biol 74, 469-476

Acknowledgements: The authors thank Ken Custer for generating the MuLK-FLAG construct

and for help throughout the project, the Biochemistry Department at the University of

Washington for use of their Phosphorimager and Ken Custer and Dr. Joe Beavo for reviewing

the manuscript. L.BJ. was supported by a training grant from the NIH (T32 GM07750). This

work was supported by a grant from NIDA (R21 DA14954) to S.M.B.


http://ww

w.jbc.org/

Dow

nloaded from

http://www.jbc.org/

22

Figure Legends

Figure 1: The sequence of MuLK

The predicted amino acid sequence of mouse MuLK is shown with the DG kinase catalytic

domain indicated by a black line and a putative nuclear localization sequence indicated by a grey

line. ClustalW was used to determine the similarity among representative MuLK orthologs,

including mouse (M. musculus, CAC06108) human (H. sapiens, CAB93536), frog (X. lavius

(AAH43761) and zebrafish (D. rerio, AAH45347). Stars indicate identical amino acids across the

four species and double dots indicate conservative changes as determined by ClustalW and

described under Experimental Procedures. Additional orthologs were also found in D.

melanogaster (AAF52895), C. elegans (T16422), A. thaliana (AAM91597) and O. sativa

(BAC65388).

Figure 2: Phylogram analysis reveals MuLK to be a distinct lipid kinase.

The BLOSUM weight matrix of ClustalW (origin 2) was used to align H. sapiens MuLK

(CAB93536), M. musculus MuLK (CAC06108), H. sapiens sphingosine kinase 1 (Q9NYA1), M.

musculus sphingosine kinase 1 (AAH37710), H. sapiens sphingosine kinase 2 (Q9NRAO), M.

musculus sphingosine kinase 2 (Q9JIA7), H. sapiens ceramide kinase (BAC01154), M.

musculus ceramide kinase (NP_663450), H. sapiens DG kinases a (AAH223523), d (Q16760),

e (P52429), q (NP_001338), and z (Q13574) and E. coli DG Kinase (P00556). The neighbor

joining algorithm was used to evaluate genetic distances (12) and construct the dendogram

shown. Bootstrap analysis by PAUP confirmed that MuLK segregated from other lipid kinases

to its own branch in 96/100 iterations.


http://ww

w.jbc.org/

Dow

nloaded from

http://www.jbc.org/

23

Figure 3: MuLK is a membrane-associated protein.

A) MuLK co-fractionates with the membrane fraction. Shown is an immunoblot probed with an

anti-GFP antibody demonstrating the presence of EGFP-immunoreactive protein in nuclei-free

cell homogenates (H), cytosol (C) and membrane (M) fractions from cells expressing either

EGFP alone, MuLK-EGFP or CERK-EGFP. Both MuLK-EGFP and CERK-EGFP fractionated

primarily with the membrane fractions, suggesting that both are membrane-associated.

B) MuLK localizes to an internal membrane compartment. Shown are fluorescent images of

Chinese Hamster Ovary cells expressing either enhanced green fluorescent protein (EGFP) or

EGFP fused to the carboxy terminus of MuLK (MuLK-EGFP). MuLK-EGFP localized to an

internal membrane compartment.

Figure 4: MuLK is broadly expressed.

Top) PCR amplification products of a 528 bp 3’ fragment of the human MuLK cDNA amplified

from a normalized panel of human cDNAs. Lanes correspond to the tissue indicated in the graph

below. MuLK was expressed in all tissues surveyed.

Bottom) Quantification of PCR products reveals that MuLK expression was highest in the

neuroendocrine tissues pancreas and brain. Net band intensities from three experiments were

averaged. Error bars represent the standard deviation.


http://ww

w.jbc.org/

Dow

nloaded from

http://www.jbc.org/

24

Figure 5: MuLK phosphorylates ceramide, diacylglycerol, 1-AG and 2AG, but not

sphingosine.

A) MuLK activity toward the indicated substrate (DOG, dioleoylglycerol; MOG,

monooeoylglycerol; 1-AG, 1-arachidonylglycerol; 2-AG, 2-arachindonylglycerol; Cer, ceramide;

Sph, sphingosine) and 0.5mM (clear bars) or 5 mM (hatched bars) was assayed in Triton X-100

micelles as described under Experimental procedures. Data shown are the average of duplicates

+/- the range and are representative of two experiments. MuLK phosphorylated all substrates

tested with the exception of sphingosine.

B) MuLK activity toward 50 mM DG (dioleoylgylcerol) in Triton X-100 micelles was evaluated

in the presence or absence of 2 mM of the indicated additional lipid. Data shown are the average

of duplicates +/- the range and are representative of two experiments.

Figure 6: MuLK demonstrates similar affinity for ceramide and DG and a higher Vmax

toward DG.

A) MuLK activity toward various concentrations of DG (dioleoylgylcerol) (filled circles) or long

chain ceramides (filled squares) in TX-100 micelles was determined as described under

Experimental Procedures. Data shown are the average of duplicates +/- the range and are

representative of three experiments. Kinetic parameters were generated using Prism Graph Pad.

B and C) Surface dilution analysis of MuLK activity toward 0.8 mM long chain ceramides (B)

or 2 mM dioleoylgylcerol (C) was determined in Triton X-100 micelles containing cardiolipin at

one tenth the total concentration of substrate, as described under Experimental Procedures.

Kinetic parameters were generated using Prism Graph Pad. Data shown are the average of

duplicates +/- the range and are representative of two experiments.


http://ww

w.jbc.org/

Dow

nloaded from

http://www.jbc.org/

25

Figure 7: MuLK activity is modulated by cations and anionic lipids.

Shown is MuLK activity assayed in varying concentrations of cardiolipin, Ca2+ and Mg2+.

MuLK DG and ceramide kinase activities were evaluated in solutions containing 1 mM (2

mol%) dioleoylgylcerol (A and B) or 1 mM (2 mol%) long chain ceramides (C and D) and either

0.1 or 1 mM cardiolipin in the presence or absence of 1 mM free Ca2+. The assays were initiated

with the addition of 1 mM ATP and either 0.5 mM (A and C) or 5 mM (B and D) Mg2+. Data

shown are the average of duplicate determinations +/- the range and are representative of two

experiments.


http://ww

w.jbc.org/

Dow

nloaded from

http://www.jbc.org/

Figure 1


http://ww

w.jbc.org/

Dow

nloaded from

http://www.jbc.org/

Figure 2


http://ww

w.jbc.org/

Dow

nloaded from

http://www.jbc.org/

A

B

KDa200

100

75

50

37

2520

Figure 3

EGFP MuLK-EGFP

H M C H M C H M C


http://ww

w.jbc.org/

Dow

nloaded from

http://www.jbc.org/

Brain Heart Kidney Liver Lung Muscle Pancreas Placenta

Rel

ativ

e in

tens

ity (

pixe

ls X

100

0)B

and

inte

nsity

(pi

xels

X 1

000)

Figure 4

564

831947

13751584

bp


http://ww

w.jbc.org/

Dow

nloaded from

http://www.jbc.org/

0

10

20

30

40

50

DOG MOG 1-AG 2-AG Cer Sph

Lipid Substrate

Kin

ase

Act

ivity

(n

mol

/min

/mg)

0

40

80

120

160

DOG MOG 1-AG 2-AG Cer Sph

Lipid Substrate

Kin

ase

Act

ivity

(n

mol

/min

/mg)

0.5mM

5mM

0

10

20

30

40

None MOG 1-AG 2-AG Cer Sph

Added Lipid (2 mM)

DG

K A

ctiv

ity

(nm

ol/m

in/m

g)A

B

Figure 5


http://ww

w.jbc.org/

Dow

nloaded from

http://www.jbc.org/

0 250 500 750 10000

50

100

150

200

Lipid substrate (mM)

0 2 4 6 8 100

10

20

30

40

50

Cer (mol%) DOG (mol%)

0

50

100

150

200

250

0 2 4 6 8 10

DG

K A

ctiv

ity (

nmol

/min

/mg)

A

B C

Figure 6


http://ww

w.jbc.org/

Dow

nloaded from

http://www.jbc.org/

Figure 7


http://ww

w.jbc.org/

Dow

nloaded from

http://www.jbc.org/

and Sandra M. BajjaliehDavid W. Waggoner, Laura Beth Johnson, Philip C. Mann, Valerie Morris, John Guastella

MuLK: A novel eukaryotic multi-substrate lipid kinase

published online July 13, 2004J. Biol. Chem.

10.1074/jbc.M405932200Access the most updated version of this article at doi:

Alerts:

When a correction for this article is posted•

When this article is cited•

to choose from all of JBC's e-mail alertsClick here


http://ww

w.jbc.org/

Dow

nloaded from

http://www.jbc.org/lookup/doi/10.1074/jbc.M405932200

http://www.jbc.org/cgi/alerts?alertType=citedby&addAlert=cited_by&cited_by_criteria_resid=jbc;M405932200v1&saveAlert=no&return-type=article&return_url=http://www.jbc.org/content/early/2004/07/13/jbc.M405932200.citation

http://www.jbc.org/cgi/alerts?alertType=correction&addAlert=correction&correction_criteria_value=early/2004/07/13/jbc&saveAlert=no&return-type=article&return_url=http://www.jbc.org/content/early/2004/07/13/jbc.M405932200.citation

http://www.jbc.org/cgi/alerts/etoc

http://www.jbc.org/

mulk: a eukaryotic multi-substrate lipid kinase2004/07/13 · mulk reactions were incubated at 94 c...

Documents