Báo cáo khoa học: Investigation of leucine-rich repeat kinase 2 Enzymological properties and novel assays pptx

In this study, highly purified, baculovirus-expressed proteins have been used, for the first time providing large amounts of protein that enable a thor-ough enzymatic characterization of t

Enzymological properties and novel assays

Vasanti S Anand1, Laurie J Reichling2, Kerri Lipinski1, Wayne Stochaj3, Weili Duan3, Kerry

Kelleher3, Pooja Pungaliya4, Eugene L Brown4, Peter H Reinhart1, Richard Somberg2, Warren

D Hirst1, Steven M Riddle2 and Steven P Braithwaite1

1 Wyeth Research, Discovery Neuroscience, Princeton, NJ, USA

2 Invitrogen Corporation, Discovery Sciences, Madison, WI, USA

3 Wyeth Research, Chemical and Screening Sciences, Cambridge, MA, USA

4 Wyeth Research, Biological Technologies, Cambridge, MA, USA

Parkinson’s disease (PD) is the second most prevalent

neurodegenerative disorder in humans, and has a

rela-tively poorly understood etiology Linkage analysis

studies in families with PD identified several mutations

in the leucine-rich repeat kinase 2 gene (LRRK2) [1,2]

Moreover, epidemiological studies have shown that

these mutations are the most prevalent cause of the

autosomal form of the disorder, with high penetrance

of certain mutations [3] The similarity in age of onset and clinical symptoms between familial and idiopathic forms may also provide insights into the pathways involved in sporadic cases of PD

LRRK2 is a large, 286 kDa, multidomain protein [4] consisting of a number of putative protein–protein

Keywords

LanthaScreen TM ; LRRK2; LRRKtide; moesin;

Parkinson’s disease

Correspondence

S P Braithwaite, Wyeth Research,

Discovery Neuroscience, Princeton,

CN8000, NJ 08543, USA

Fax: +1 732 274 4020

Tel: +1 732 274 4556

E-mail: braiths@wyeth.com

(Received 18 May 2008, revised 11

November 2008, accepted 12 November

2008)

doi:10.1111/j.1742-4658.2008.06789.x

Mutations in leucine-rich repeat kinase 2 (LRRK2) comprise the leading cause of autosomal dominant Parkinson’s disease, with age of onset and symptoms identical to those of idiopathic forms of the disorder Several of these pathogenic mutations are thought to affect its kinase activity, so understanding the roles of LRRK2, and modulation of its kinase activity, may lead to novel therapeutic strategies for treating Parkinson’s disease In this study, highly purified, baculovirus-expressed proteins have been used, for the first time providing large amounts of protein that enable a thor-ough enzymatic characterization of the kinase activity of LRRK2 Although LRRK2 undergoes weak autophosphorylation, it exhibits high activity towards the peptidic substrate LRRKtide, suggesting that it is a catalytically efficient kinase We have also utilized a time-resolved fluores-cence resonance energy transfer (TR-FRET) assay format (Lantha-ScreenTM) to characterize LRRK2 and test the effects of nonselective kinase inhibitors Finally, we have used both radiometric and TR-FRET assays to assess the role of clinical mutations affecting LRRK2’s kinase activity Our results suggest that only the most prevalent clinical mutation, G2019S, results in a robust enhancement of kinase activity with LRRKtide

as the substrate This mutation also affects binding of ATP to LRRK2, with wild-type binding being tighter (Km,app of 57 lm) than with the G2019S mutant (Km,appof 134 lm) Overall, these studies delineate the cat-alytic efficiency of LRRK2 as a kinase and provide strategies by which a therapeutic agent for Parkinson’s disease may be identified

Abbreviations

COR, C-terminus of Roc; FRET, fluorescence resonance energy transfer; GST, glutathione S-transferase; LRRK2, leucine-rich repeat kinase 2; LRRK2-FL, full-length leucine-rich repeat kinase 2; PD, Parkinson’s disease; Roc, Ras of complex; TR-FRET, time-resolved

fluorescence resonance energy transfer; 4E-BP, eukaryotic initiation factor 4E-binding protein.

interaction domains, including N-terminal ankyrin

repeats, a leucine-rich repeat region, and a C-terminal

WD40 domain It also contains a GTPase domain

composed of Ras of complex (Roc) and C-terminus of

Roc (COR) regions and a kinase domain Mutations

linked to PD are found throughout the protein,

includ-ing the kinase domain (G2019S and I2020T), the Roc–

COR domain (R1441C and Y1699C), the leucine-rich

repeats (I1122V), and the WD40 domain (R2385G) [4]

The most prevalent of these mutations, G2019S [3,5–

7], within the Mg2+-binding region, has been shown to

increase the kinase activity of LRRK2 [8], leading to

neurodegeneration [9,10] and deficits in neurite

out-growth [11,12] The functional consequences and roles

of other mutations reported in the literature are

con-flicting, I2020T causing an increase in kinase activity

[13] or a decrease [14] Similarly, mutations in the

GTPase domain have been demonstrated to increase

kinase activity [8,15,16], whereas in other studies they

have had no effect [14] Characterization of these

mutations and understanding LRRK2’s pathogenic

function has proven to be challenging, due to technical

difficulties in expressing the protein The majority of

studies have used immunoprecipitated LRRK2 from

recombinant mammalian expression systems [8–10,13],

and there is one report of Escherichia coli-expressed

LRRK2 [17] These studies have investigated

auto-phosphorylation, or phosphorylation of the surrogate

substrate myelin basic protein, due to the lack of

knowledge of physiological substrate(s); however, both

of these are very weak events The recent identification

of moesin as a putative physiological substrate for

LRRK2 provided the first alternative for an in-depth

investigation of LRRK2’s enzymatic properties [14];

however, its physiological relevance remains to be

determined The only other proposed substrate of

LRRK2 is eukaryotic initiation factor 4E-binding

pro-tein (4E-BP), identified in Drosophila [18], which may

play a role in regulating protein translation, although

the precise residue that is phosphorylated remains to

be clarified In order to have a viable target for drug

development, it is essential to know whether LRRK2

has appreciable activity towards its substrates

In these studies, we have, for the first time, utilized

highly purified LRRK2 produced from

baculovirus-infected insect cells to generate significant quantities of

active proteins for thorough enzymatic

characteriza-tion Importantly, a truncated construct consisting of

all the conserved functional domains of LRRK2 was

found to behave similarly to the full-length protein,

proving that results obtained with such constructs are

valid We have investigated the detailed kinetics of

wild-type LRRK2 in terms of measuring the rate

con-stants of autophosphorylation and phosphorylation of LRRKtide, a short peptide substrate derived from moesin [14] This characterization significantly extends the results from previous studies, which have been lim-ited by protein supply [14], preventing the measure-ment of catalytic rate constants and other enzymatic parameters Furthermore, a time-resolved fluorescence resonance energy transfer (TR-FRET) methodology has been used to characterize LRRK2’s enzymological properties and assess the potency of small molecule, nonselective, kinase inhibitors Finally, we have assessed the effects of a number of common pathologi-cal mutations in LRRK2 on its enzymatic activity Overall, our studies provide a detailed enzymatic char-acterization of LRRK2’s kinase activity, and highlight its potential tractability as a drug target for PD

Results LRRK2 proteins expressed by baculovirus Previous studies have primarily used LRRK2 constructs expressed in mammalian expression systems such as HEK-293 cells [8–10] Owing to the low yields obtained from such recombinant overexpression, alternatives are preferable for larger-scale expression and enzymological characterization Expression in E coli has been previ-ously reported [17]; however, this study demonstrated difficulty in the production of large constructs consisting

of more than just the COR-kinase domains, and the kinase activity associated with these domains was found

to be relatively weak Therefore, in this study baculovi-rus-infected insect cells were used to express proteins Efficient expression of N-terminal glutathione S-trans-ferase (GST) fusion proteins of LRRK2 residues 970–

2527, consisting of the Roc, COR, kinase, WD40 and entire C-terminal domains, produced significant amounts of protein for in-depth characterization Mutant forms of LRRK2 were also used, namely the pathogenic mutants G2019S, I1122V, I2020T, R1441C and Y1699C, along with the predicted kinase-dead D1994A mutant [10,15,19], in which the critical aspar-tate residue in the catalytic loop of the kinase domain is mutated Separation of the proteins (5 lg of each prepa-ration) by SDS⁄ PAGE followed by Coomassie Blue staining demonstrated that they are all of similar, high, purity (> 85%) and have similar banding patterns of minor contaminants (Fig 1A) Western blotting showed that the major band at 204 kDa for the wild-type pro-tein is LRRK2 immunoreactive (Fig 1B) Samples were also characterized by MS, revealing the LRRK2 sequence as the dominant species; b-tubulin was also detected, but no other known kinases were detectable

(data not shown) As previous studies have shown that

the chaperone proteins Hsp90 and p50cdc37can interact

with LRRK2 in a recombinant expression system and

mammalian cells [13,20], western blots for these proteins

were performed, but immunoreactivity was not detected

(data not shown) Additionally, the presence of Hsp60

and Hsp70, which are often found to interact with

pro-teins expressed in insect cells, was investigated, but these

were also not detected (data not shown)

LRRK2 exhibits weak autophosphorylation

activity

We first assessed the kinase activity of

baculovirus-expressed wild-type LRRK2 by autophosphorylation

LRRK2 enzymes (50 nm) were incubated with

32P-labeled ATP (200 lm) for 30 min at 30C, and

sep-arated by SDS⁄ PAGE The resultant autoradiograms

showed that wild-type LRRK2 autophosphorylates,

whereas the predicted kinase-dead, D1994A, mutant

form [9] does not exhibit any autophosphorylation

(Fig 2A) The only band appearing on the

autoradio-grams was at the size of the LRRK2 protein (204 kDa),

indicating that none of the other minor contaminant

bands seen on Coomassie-stained gels (Fig 1A) are

other active kinases, or substrates for LRRK2 On

per-forming filter-binding assays, it was confirmed that

there is significant, although low, incorporation of 32P into LRRK2 as compared to reactions in the presence

of the D1994A kinase-dead LRRK2, or in the absence

of enzyme (Fig 2B) These findings are consistent with previous studies that have used proteins expressed in mammalian cells [8–10,14] To extend these findings and further understand the kinetics of LRRK2 autophosphorylation, a time-course experiment was

WT G2019S I2020T Y1

220-

160-

120-

50-

20-

250- 75-

1 50-

Fig 1 Characterization of LRRK2 proteins (A) Coomassie

Blue-stained gel of 5 lg of each GST–LRRK2 protein preparation

sepa-rated by SDS ⁄ PAGE (B) Western blot of 0.5 lg of wild-type (WT)

LRRK2 with antibody against LRRK2 indicates that the major

pro-tein is LRRK2 immunoreactive Data are representative of three

independent experiments.

WT D1994A -LRRK2

0 500 1000 1500

0.08 0.10 0.12 0.14 0.16 0.18 0.20

0.24 0.22

Time (min)

Time (min) 1 5 7 10 15 20 30

A

C

D

B

Fig 2 Autophosphorylation of wild-type (WT) LRRK2 (A) Autora-diogram of wild-type and kinase-dead (D1994A) LRRK2 proteins (50 n M ) that have been incubated with 200 l M ATP for 30 min at

30 C (B) Wild-type and kinase-dead (D1994A) LRRK2 proteins (50 n M ) were allowed to autophosphorylate in the presence of

200 l M ATP, and the incorporated 32 P was quantitated using filter-binding assays (data from three independent experiments) Signifi-cant autophosphorylation was observed in the presence of LRRK2

as compared to kinase-dead (D1994A) LRRK2 and without LRRK2 (**P < 0.01) (C) The autoradiograph shows a time-dependent increase in LRRK2 autophosphorylation (D) The rate of autophos-phorylation was determined using filter-binding assays and quench-ing reactions with 100 m M EDTA at varying reaction times Counts

of incorporation were then plotted with respect to time and fitted to

a linear equation to obtain the rate constant of autophosphorylation.

performed, demonstrating increased LRRK2

phosphor-ylation over time (Fig 2C) Expressing higher amounts

of purified protein gives the advantage of being able to

perform a detailed enzymatic characterization; hence,

the rate of32P incorporation into LRRK2 was

quanti-fied using filter-binding assays, and wild-type LRRK2

autophosphorylation was found to be very slow, with a

rate constant of 0.006 ± 0.0005 pmolÆmin)1(Fig 2D)

LRRK2 activity on LRRKtide

The identification of moesin as a potential substrate for

LRRK2 led to the identification of a 15 amino acid

peptide based on its sequence that is sufficient for LRRK2 activity, named LRRKtide [14] To test whether wild-type LRRK2 was able to phosphorylate LRRKtide, LRRK2 (100 nm) was mixed with LRRK-tide (300 lm), and reactions were initiated by adding ATP (200 lm) at 30C for 30 min, before loading reac-tions onto phosphocellulose filters There was significant incorporation of32P into LRRKtide in comparison with autophosphorylation as detected by the filter-binding assay (Fig 3A) As LRRKtide contains both a threo-nine residue and a tyrosine residue as potential phos-phorylation sites, we investigated which residue was targeted for phosphorylation by mutagenesis Mutation

2.5 2.0 1.5 1.0

0.0 0.5

0

5

10

15

20

25

30

***

LRRKtide

+ + – – + + –

*** ***

1 10 100 1000

10 000

1 10

100

1000

A

B

3 )

No peptide No peptide

+LRRK2 –LRRK2

32 P incorporation

–1 )

Fig 3 LRRKtide phosphorylation by wild-type LRRK2 (A) LRRK2 (100 n M ) was incubated with 200 l M ATP in kinase reaction buffer in the presence of 300 l M LRRKtide; filter-binding assays showed a significant incorporation of32P into the substrate as compared to autophospho-rylation (***P < 0.0001, column 1) or in the absence of enzyme (column 3) or with no LRRK2 or LRRKtide present (column 4) (data from three independent experiments) (B) LRRK2 phosphorylates LRRKtide at its threonine residue LRRK2 at 100 n M was incubated in the pres-ence of 200 l M ATP with a series of peptides (400 l M ); LRRKtide and a form of LRRKtide in which the tyrosine residue was mutated to phenylalanine (LRRKtide Y–F) showed significant and robust phosphorylation as compared to background (***P < 0.0001) A peptide in which the threonine of LRRKtide was mutated to alanine (LRRKtide T–A) showed no incorporation of 32 P as compared to the control of LRRK2 alone Data from three independent experiments (C) The rate of LRRKtide phosphorylation was determined by incubating 50 n M LRRK2 with 300 l M LRRKtide in kinase reaction buffer containing 200 l M ATP Reactions were quenched with 100 m M EDTA after 1, 5, 10,

15, 20 and 30 min, before loading onto phosphocellulose filters and subsequent washing and counting Data from three independent experi-ments (D) The apparent Km of wild-type LRRK2 for LRRKtide was determined by incubating 50 n M LRRK2 with varying concentrations of LRRKtide in the presence of 200 l M ATP Data were fitted to a hyperbola to yield an apparent Kmof 186 ± 77 l M Data from three indepen-dent experiments.

of the tyrosine to a phenylalanine (LRRKtide Y–F) did

not significantly affect the phosphorylation of the

pep-tide, whereas mutation of threonine to alanine

(LRRK-tide T–A) completely blocked the ability of the pep(LRRK-tide

to be phosphorylated (Fig 3B) Therefore, LRRK2

appears to act on LRRKtide as a serine⁄ threonine

kinase with no tyrosine kinase activity

To assess the rate of phosphorylation of this

pep-tide, LRRKtide was incubated with LRRK2 for

vari-ous times The rate of phosphorylation was determined

to be 0.7 ± 0.02 pmolÆmin)1, approximately 100-fold

faster than the measured rate constant of

autophos-phorylation (Fig 3C) Furthermore, the apparent Km

of LRRKtide was determined by performing reactions

with LRRK2 at varying concentrations of LRRKtide,

and was determined to be 186 ± 70 lm (Fig 3D); this

is consistent with data obtained using proteins

expressed in mammalian cells [14]

Activity of wild-type and G2019S mutant forms

of LRRK2

Having identified that LRRKtide is an efficient

sub-strate for LRRK2, we assessed the catalytic efficiency

of wild-type LRRK2 on this peptide Varying

concen-trations of LRRK2 were incubated with 300 lm

LRRKtide, and the incorporation of 32P over the

course of the reaction was measured, yielding a specific

activity of 42 ± 1.5 pmolÆmin)1Ælg)1 (Fig 4A) In

addition, to investigate the effect of the G2019S

muta-tion on the catalytic activity of LRRK2, varying

con-centrations of G2019S LRRK2 were incubated with

LRRKtide and ATP, and yielded a specific activity of

138 ± 7 pmolÆmin)1Ælg)1 (Fig 4A), about three-fold

greater than that determined for wild-type LRRK2

As G2019S LRRK2 showed greater activity than the

wild-type, and as the mutation is located within the

activation segment of the kinase domain, we

investi-gated its influence on the affinity of ATP for LRRK2

Proteins were incubated with 400 lm LRRKtide in the

presence of varying concentrations of ATP, and the

incorporation of32P into LRRKtide was assessed The

apparent Kmfor ATP of wild-type LRRK2 was found

to be 57 ± 4 lm (Fig 4B), approximately three-fold

lower than that of G2019S LRRK2, which had an

apparent Kmof 134 ± 2 lm (P < 0.01; Fig 4C)

A time-resolved fluorescent based assay for

measuring LRRK2 activity

Because LRRK2 showed high activity with LRRKtide,

we were able to convert the radioactive assay into a

TR-FRET-based LanthaScreenTM format A

fluores-cein-labeled LRRKtide is used as the substrate, and after a kinase reaction has occurred, a terbium-labeled antibody against phospho-LRRKtide is added for

0 5 10 15 20 25 30 35

0 20 40 60 80 100 120 140

0 100 200 300 400 500 600

G2019S WT

WT

G2019S

A

B

C

Fig 4 Activity of wild-type (WT) and G2019S LRRK2 on LRRKtide (A) The specific activity of G2019S LRRK2 ( ) is greater than that

of wild-type LRRK2 (d) Proteins were incubated at varying concen-trations with 300 l M LRRKtide and 200 l M ATP for 30 min, and the amount of 32 P incorporated per minute was calculated Data from three independent experiments (B) Wild-type LRRK2 or G2019S LRRK2 at 100 n M was incubated with 400 l M LRRKtide in the pres-ence of varying concentrations of ATP The apparent Km for ATP for wild-type LRRK2 is 57 ± 4 l M and that for G2019S LRRK2 is

134 ± 2 l M

detection Fluorescence resonance energy transfer

(FRET) occurs from the terbium-labeled antibody to

the fluorescein dye on the phosphorylated peptide

Reactions were performed with varying concentrations

of wild-type LRRK2 in the presence of 400 nm

fluores-cein–LRRKtide and 1 mm ATP for 1 h at room

temperature The reaction was stopped by addition of

10 mm EDTA, and phosphorylation was detected by

the terbium-labeled antibody against

phospho-LRRK-tide FRET was measured by the emission ratio at

520⁄ 495 nm The EC50 for wild-type LRRK2 was

found to be 2728 ± 884 ngÆmL)1, whereas G2019S

LRRK2 showed approximately two-fold greater

activ-ity, with an EC50 of 1276 ± 505 ngÆmL)1 (Fig 5),

comparable to the differences seen in specific activity

in radiometric assays (Fig 4A)

Potency of broad-spectrum kinase inhibitors on

LRRK2

The LanthaScreenTMformat allows the rapid and

effi-cient screening of kinase inhibitors, and therefore allows

further characterization of LRRK2 properties A set of

approximately 120 known kinase inhibitors was

screened against wild-type and G2019S LRRK2 (data

not shown) A small panel of kinase inhibitors was then

selected for further study (Table 1), and dose–response

relationships were obtained using wild-type (Fig 6A)

and G2019S (Fig 6B) LRRK2 with 200 nm LRRKtide

and ATP at apparent Km values The inhibitors tested

showed dose-dependent inhibition of the kinase activity

(Fig 6A,B), with the most potent compound being

staurosporine, with an IC50 of 1.8 ± 0.09 nm for

G2019S LRRK2 (Table 1) There was no significant

dif-ference in the inhibitory efficacy of compounds between

wild-type and G2019S LRRK2 (Table 1)

Effect of additional clinical mutations on LRRK2

kinase activity

As we had seen significant differences in activity

between wild-type and G2019S LRRK2 using the

LRRKtide peptide, further analysis of other mutant

forms of LRRK2 was performed Previously,

conflict-ing results have been reported for the effects of these

mutations [8,13–15], although the majority of studies

have investigated autophosphorylation, or other

surro-gate substrates that are weakly phosphorylated We

therefore tested the effects of each mutation on both

autophosphorylation and LRRKtide phosphorylation,

to determine whether there are any differences

Wild-type and mutant LRRK2 proteins, G2019S, D1994A,

R1441C, Y1699C, I1122V, and I2020T, were incubated

with 32P-labeled ATP for 30 min at 30C, the reac-tions were stopped, proteins were separated by SDS⁄ PAGE, and the resultant gel was exposed to a phosphoimaging screen (Fig 7A, top) G2019S LRRK2 showed significantly greater autophosphoryla-tion than wild-type LRRK2 The other mutant LRRK2 proteins were not significantly different in activity with respect to the wild-type (Fig 7A) As the G2019S mutation provides a serine residue, and hence

a potential extra phosphorylation site, we sought to confirm that the increased autophosphorylation observed was not due to phosphorylation at this site Analysis by MS failed to detect signal at this residue;

0.0 0.2 0.4 0.6 0.8 1.0

1.2

[WT]

[G2019S]

Log concentration WT/G2019S LRRK2 (ng·mL –1 )

Fig 5 A time-resolved fluorescence-based LanthaScreenTMassay effectively measures LRRK2 kinase activity Titration of wild-type and G2019S LRRK2 demonstrates concentration-dependent phos-phorylation of LRRKtide, with the mutant protein being more active Varying concentrations of each protein were incubated with 400 n M fluorescein–LRRKtide and 1 m M ATP in kinase reaction buffer for

1 h at room temperature The reaction was stopped, and phosphor-ylation of LRRKtide was detected by addition of terbium-labeled antibody against LRRKtide FRET was measured by excitation at

495 nm, and the emission ratio between 525 nm and 495 nm was calculated to measure the phosphorylation of the substrate Data are illustrated as a single experiment representative of at least three independent runs.

Table 1 IC50 values of nonspecific kinase inhibitors on wild-type and G2019S LRRK2 Inhibitors were tested using the Lantha-Screen TM format, mean IC50± SD, data from three independent experiments for each compound.

Compound Wild-type IC50(n M ) G2019S IC50(n M ) JAK3 inhibitor VI 22 ± 2.5 40 ± 4

Staurosporine 2.0 ± 0.1 1.8 ± 0.09

however, as G2019S LRRK2 also displays increased

activity on LRRKtide peptide (Fig 4A), it is evident

that this mutation leads to a protein with greater

activity than the wild-type

The specific activities of mutant LRRK2 enzymes

with respect to 32P incorporation into LRRKtide were

measured by incubating varying concentrations of each

enzyme in the presence of 300 lm LRRKtide for

vari-ous times G2019S LRRK2 demonstrated the greatest

activity of all the mutants, although R1441C LRRK2

also showed significantly greater specific activity than

the wild-type Interestingly I2020T LRRK2 showed

significantly lower activity As would be predicted, the

D1994A kinase-dead mutant showed significantly less

activity on LRRKtide than the wild-type (Fig 7B)

Furthermore, the mutant forms of LRRK2 were tested in the LanthaScreenTM assay G2019S LRRK2 showed the greatest activity as compared to the wild-type, and I2020T LRRK2 the least activity (Fig 7C), comparable to the data obtained from the radiometric assay (Fig 7B)

Full-length LRRK2 displays comparable properties

to the truncated form Our previous studies were all performed using a trun-cated construct consisting of all conserved structural

970–2527, as this could be obtained in reasonable quantities for performing enzymological characteriza-tion To further validate our findings, we wanted to ensure that full-length LRRK2 (LRRK2-FL) behaved similarly We were able to purify very low quantities

of LRRK2-FL, as a FLAG-tagged construct, with activity in a dimer fraction being separated using size exclusion chromatography Autophosphorylation experiments showed that the full-length and truncated forms of LRRK2 had comparable activities (Fig 8A) Using the TR-FRET LanthaScreenTM format, the

2375 ± 536 ngÆmL)1 (Fig 8B), which is very similar

to that obtained for the truncated wild-type LRRK2 construct (Fig 5) Furthermore, staurosporine, the most potent kinase inhibitor of the truncated variant

of LRRK2, was determined to have an IC50 of 8.2 ± 0.8 nm (Fig 8C), which is similar to that obtained for truncated wild-type LRRK2 (Fig 6) The low yields of LRRK2-FL obtained precluded a more thorough characterization of its enzymological proper-ties; however, these data suggest that the truncated construct behaves very similarly to the full-length protein in terms of its kinase activity

Discussion

In the present study, we have investigated the kinase activity of a new protein source of LRRK2, with respect to autophosphorylation and LRRKtide phos-phorylation These studies demonstrate LRRK2 to be

an effective kinase whose activity is dependent upon its substrate, and how mutations in LRRK2 that have been clinically linked to PD may affect its func-tion Finally, a novel fluorescence-based assay system using LanthaScreenTM technology, which robustly measures LRRK2 kinase activity, and is amenable for testing the efficacy of small molecule kinase inhibi-tors, has been evaluated Overall, this adds to the enzymological characterization of LRRK2 and

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

0.1

0.2

0.3

0.4

0.5

0.6

0.7

JAK3 inhibitorVI

K252A

Staurosporine

Su-11248

Ro-31-8220

JAK3 inhibitorVI

K252A

Staurosporine

Su-11248

Ro-31-8220

Log concentration inhibitor (n M )

Log concentration inhibitor (n M )

A

B

Fig 6 Inhibition of wild-type and G2019S LRRK2 by nonspecific

kinase inhibitors (A) The indicated inhibitors demonstrate

dose-dependent inhibition of wild-type LRRK2 activity Reactions were

performed in the presence of 3.4 lgÆmL)1LRRK2, 400 n M

fluores-cein–LRRKtide, 57 l M ATP, and varying doses of inhibitors (B) The

same inhibitors were also tested against 1.0 lgÆmL)1 G2019S

LRRK2 in a similar manner, except with 134 l M ATP.

provides a protein and assay system that can be

utilized for significantly higher throughput than has

previously been possible

Our studies have, for the first time, used baculovirus-expressed proteins The majority of previous studies on LRRK2 have used proteins expressed in mammalian cells [8,9,13,14] that can only be produced with low yields and low purity One study has reported the use

of E coli-expressed proteins [17], but only short con-structs, consisting of either the kinase or COR-kinase domains of LRRK2 Baculovirus-mediated production

of proteins gives the advantage of being able to pro-duce large amounts of post-translationally modified LRRK2 protein with all of its enzymatic domains In line with other studies, the generation of full-length LRRK2 has been difficult; however, we have been able

to express and purify the full-length protein in Sf21 cells on a small scale These studies allowed us to demonstrate that a shorter construct lacking the N-terminal region behaves the same as full-length LRRK2 with respect to its kinase activity This region contains no conserved structural features, and no clini-cally relevant mutations have been characterized within

it Therefore, characterization of the kinase activity of

a construct lacking this domain gives valuable data about LRRK2 Our findings significantly extend the characterization of LRRK2’s enzymological properties, our results being consistent with the only previously published parameter, Kmof LRRKtide, obtained from proteins expressed in mammalian cells, and adding the characterization of specific activity and Km for ATP Unlike studies using other expression systems [13,17],

we have shown that LRRK2, when expressed in insect cells, does not copurify with chaperone proteins

G2019S D1994A R1441C 1699C I1122V I2020T

0

100

200

300

400

***

32 P incorporation (% WT)

0.0

0.2

0.4

0.6

0.8

1.0

1.2

[D1994A]

[WT]

[G2019S]

[I2020T]

[I1122V]

[R1441C]

[Y1699C]

WT G2019S D1994A R1441C Y1699C I1122V I2020T

0

20

40

60

80

100

120

140

160

***

***

***

*

***

A

C

B

Fig 7 Kinase activity of pathogenic LRRK2 mutants (A) Autophos-phorylation of LRRK2 mutants G2019S, D1994A, I1122V, R1441C, Y1699C and I2020T was assessed by autoradiography Kinase reac-tions were performed with 50 n M each mutant in the presence of

200 l M ATP, and reactions were allowed to proceed for 30 min at

30 C, before being stopped by separation using SDS ⁄ PAGE The gels were exposed to a Phosphoimager screen and autoradiograms were developed; the results are representative of experiments per-formed on three independent occasions The autophosphorylated variant LRRK2 bands were quantitated and normalized with respect

to wild-type (WT) LRRK2 Autophosphorylation of G2019S LRRK2 was significantly higher than that of wild-type LRRK2 (***P < 0.0001) (B) Specific activities of mutant LRRK2 proteins with respect to 32 P incorporation into LRRKtide were assessed using filter-binding assays G2019S LRRK2 (***P < 0.0001), and R1441C LRRK2 (*P < 0.05) activities were significantly higher than that of wild-type LRRK2 D1994A LRRK2 and I2020T LRRK2 activi-ties were found to be significantly lower than that of wild-type LRRK2 (***P < 0.0001) Data from three independent experiments (C) The activities of mutants were additionally assessed using the LanthaScreen TM format Experiments were performed as described previously; data are representative of at least three independent experiments.

However, these proteins are highly active, indicating that, although chaperone-mediated folding may be important for LRRK2, its maintained interaction is not a prerequisite for kinase activity Interestingly, by

MS, we found that b-tubulin copurified with LRRK2;

a recent study has also identified an interaction between the proteins [21]

Recent data indicate that LRRK2 predominantly exists as a dimer and undergoes cis-mediated intramo-lecular autophosphorylation [22] Many protein kinases require phosphorylation of residues within their kinase domains to be activated [23], and therefore can act as substrates of kinases, including themselves The S6⁄ H4 serine⁄ threonine kinase, for example, has been found to autophosphorylate at an exponential rate constant of 0.91 min)1, with the reaction going to completion in

 8 min [24] Both the fibroblast growth factor and RET receptor tyrosine kinases have been found to auto-phosphorylate, with rate constants of 0.05 s)1(complete reaction in  3 min) and 4.2 min)1 (complete reaction

in 3 min), respectively [25,26] LRRK2 is also able to undergo autophosphorylation, which is indicative of such a process, and the majority of studies on LRRK2

to date have used this property to assay its activity In our studies, we have demonstrated that LRRK2 can autophosphorylate, but this process is inefficient, with data displaying linear kinetics, indicating that the reac-tion does not go to complereac-tion within 30 min This sug-gests that LRRK2 is a relatively poor substrate for itself, in comparison to other well-characterized kinases, and that this may not be its major functional role in a physiological environment

The search for more relevant substrates of LRRK2 has led to the identification of members of the ezrin– radixin–moesin family of proteins [14] as potential substrates Although the physiological relevance of these proteins as substrates has not been proven, they clearly provide an in vitro substrate that LRRK2 can phos-phorylate with higher efficiency [14] Their functional roles, with respect to involvement with the cytoskeleton,

fit with observations of modulation of LRRK2 expr-ession affecting neuronal morphology [11] In these studies, we have used the short peptide LRRKtide, based around the putative phosphorylation site of moesin [14], to further characterize the enzymological activity of LRRK2 The homology of LRRK2, although low, has placed it in the TKL family of protein kinases [27], whose members have both serine⁄ threonine and tyrosine kinase activity With respect to LRRKtide, we have demonstrated that LRRK2 acts purely as a serine⁄ threonine kinase, and phosphorylates LRRKtide significantly more efficiently than it phosphorylates itself Potentially, LRRK2 could autophosphorylate in

0.2

0.4

0.6

0.8

1.0

1.2

1.4

0.0

0.1

0.2

0.3

0.4

0.5

A

B

C

Fig 8 Full-length LRRK2 is active (A) Autoradiogram of

autophos-phorylation of full-length and truncated wild-type LRRK2 constructs.

Full-length and truncated wild-type LRRK2 (100 n M ) were incubated

with 200 l M32P-labeled ATP for 30 min at 30 C to allow

autophos-phorylation to proceed Reactions were subjected to SDS ⁄ PAGE,

gels were exposed to a Phosphoimager screen, and an

autoradio-gram was developed The result is representative of three

indepen-dent experiments (B) Varying concentrations of full-length and

truncated LRRK2 proteins were incubated with 100 n M fluorescein–

LRRKtide and 50 l M ATP in kinase reaction buffer for 1 h at room

temperature The reaction was stopped, and phosphorylation of

LRRKtide was detected by addition of 1 n M terbium-labeled

anti-body against LRRKtide FRET was measured by excitation at

495 nm, and the emission ratio between 525 nm and 495 nm was

calculated to measure the phosphorylation of the substrate Data

are illustrated as a single experiment representative of three

inde-pendent runs (C) Dose-deinde-pendent staurosporine inhibition of

full-length LRRK2 LanthaScreen TM assays were performed with

3.6 lgÆmL)1 full-length LRRK2, 100 n M fluorescein–LRRKtide,

50 l M ATP and varying concentrations of staurosporine in kinase

reaction buffer for 1 h at room temperature.

the cells that it is being prepared in, and hence be

already highly phosphorylated at this site, therefore

pre-venting much additional phosphorylation from taking

place Our results indicate that additional

phosphoryla-tion can still take place, showing linear kinetics over the

course of the reactions performed; therefore,

phosphor-ylation at this site is not saturated Nonetheless, these

findings give rise to the notion that caution must be used

when interpreting results based purely on

autophospho-rylation, as this is a relatively weak activity Even

though the physiological relevance of moesin in relation

to LRRK2 is still unclear, the findings indicate that

LRRK2 has the ability to be a kinase with significant

activity and high substrate turnover With specific

activ-ities of 42 pmolÆmin)1Ælg)1 for wild-type LRRK2 and

138 pmolÆmin)1Ælg)1 for G2019S LRRK2 for such a

large, 204 kDa protein, this activity is respectable, in

line with the findings of others [14] We have shown that

4E-BP [18] is also phosphorylated by our LRRK2

pro-teins (data not shown), but the precise site of

phosphor-ylation and the physiological relevance are unclear It

will be interesting to assess the activity on other

physio-logical substrates as they are identified

The linkage of mutations in LRRK2 to the

develop-ment of PD has led to the interest in this protein In

this study, we initially investigated the role of the most

prevalent mutation found in humans, G2019S This

mutation significantly increases the specific activity of

LRRK2, and also alters its apparent Kmfor ATP The

G2019S residue lies within the activation loop of the

kinase, and potentially leads to the introduction of an

extra residue that can be phosphorylated, placing

LRRK2 in a more active conformation [17] Our

find-ings confirm that the G2019S mutation increases

kinase activity, with respect not only to

autophospho-rylation, but also to the peptidic substrate LRRKtide

The increased activity seen with respect to LRRKtide

implies that differences seen are due to increased

activ-ity of the protein and not just to the presence of an

extra phosphorylation site The mutation also affects

the apparent Km of ATP, indicating that it modifies

the active site of the enzyme to alter the affinity for

ATP The R1441C and Y1699C mutations within the

Roc and COR regions, despite having been linked to

PD, did not, in our hands, increase kinase activity with

respect to autophosphorylation, but resulted in small

increases in activity with respect to LRRKtide

phos-phorylation Mutations within the Roc and COR

regions have previously been demonstrated to increase,

decrease or not affect kinase activity [14,15,17] The

differences in the results may be due to different

expression systems, construct lengths, or levels of

GTP, as the Roc region forms a GTPase domain that

has been shown to modulate kinase activity [15,16,28,29] The I2020T mutation has previously been shown to increase [13,15], decrease [14] or have

no effect on kinase activity [17] Our studies indicate that the I2020T mutation causes a decrease in the kinase activity of LRRK2 in the context of LRRKtide; this is possibly due to the critical role of this residue in the activation loop of the kinase domain, causing it to

be in a less active state, but this is not observed with respect to autophosphorylation, which is a much weaker event It remains to be determined whether other substrates will be found to be affected differently

by these mutant forms of LRRK2 Nonetheless, the differences in the results seen in multiple studies between different mutant forms of LRRK2 suggest that LRRK2 may either have multiple roles or act at multiple points in pathways relevant for PD Further-more, different mutations in LRRK2 lead to PD with pleiomorphic pathology and symptoms For example, patients with mutations in the GTPase domains have been shown to have different combinations of tauopa-thies and synucleinopatauopa-thies, in addition to the hall-mark neuronal degeneration in the substantia nigra [2] Additionally, the occurrence of the other mutations in LRRK2 is not as common as that of G2019S [30–32] With our findings that different mutations differen-tially affect the kinase activity of LRRK2, yet all lead

to PD, albeit with somewhat different symptoms, it appears that LRRK2 is a central protein in processes underlying the disease The mutations that do not affect kinase activity may affect the localization of LRRK2, or other properties that modulate its roles in

a critical pathway that underlies the disorder

The prevalence, penetrance and functional signifi-cance of the G2019S mutation make the kinase activity

of LRRK2 of major interest in developing therapeutic strategies for PD We have therefore taken advantage

of a time-resolved fluorescence based assay, Lantha-ScreenTM, to assess the activity of LRRK2; this can be used as a high-throughput assay to screen for inhibi-tory compounds The LanthaScreenTMformat with the LRRKtide peptide is comparable to radiometric assays, and has been effectively used to demonstrate that a number of nonselective kinase inhibitors display inhibitory activity on LRRK2

These studies have demonstrated that LRRK2 acts

as a serine⁄ threonine kinase with appreciable activity

in relation to a peptidic substrate, as compared to its autophosphorylation, which is too weak and inefficient

a process for thorough and high-throughput assays This study has been enabled by the generation of a baculovirus-expressed protein that contains all of the conserved structural domains of LRRK2 and that