Báo cáo khoa học: Phosphorylation-dependent binding of human transcription factor MOK2 to lamin A⁄C pot

A-type lamins have been shown to bind hsMOK2 in vitro and in vivo through the coil 2 domain common to lamin A and lamin C, whereas the lamin A⁄ C-binding site in hsMOK2 has been mapped t

transcription factor MOK2 to lamin A ⁄C

Maryannick Harper, Jeanne Tillit, Michel Kress and Miche`le Ernoult-Lange

CNRS-FRE2937, Institut Andre´ Lwoff, Villejuif, France

The zinc-finger transcription factor MOK2 recognizes

both DNA and RNA through its zinc-finger motifs [1]

This dual affinity suggests that MOK2 may play a role

in transcription, as well as in the post-transcriptional

regulation of specific genes We have shown that

MOK2 represses expression of the interphotoreceptor

retinoid-binding protein(IRBP) gene [2] IRBP contains

two MOK2-binding elements, a complete 18-bp

MOK2-binding site, located in intron 2, and the

essen-tial 8-bp core MOK2-binding site (corresponding to

the conserved 3¢-half site) which is in the IRBP

pro-moter MOK2 can bind to the 8-bp sequence in the

IRBP promoter and repress transcription from this

promoter In the IRBP promoter, the TAAAGGCT

MOK2-binding site overlaps with the

photoreceptor-specific Crx-binding element, suggesting that MOK2

represses transcription by competing with the cone–rod

homeobox protein for DNA binding and decreasing

transcriptional activation by the cone–rod homeobox

protein The particular arrangement of the two

MOK2-binding sites, observed in the human IRBP gene and also in a second human potential MOK2 tar-get gene, Pax3, suggests that MOK2 may repress tran-scription via a dual mechanism Previously, we identified lamin A⁄ C proteins as binding partners for hsMOK2 in a yeast two-hybrid screen [3] A-type lamins have been shown to bind hsMOK2 in vitro and

in vivo through the coil 2 domain common to lamin A and lamin C, whereas the lamin A⁄ C-binding site in hsMOK2 has been mapped to its N-terminal acidic domain Divergent evolution has been observed between human and mouse MOK2 genes which results

in the loss of this NH2-domain in the mouse gene [4]

An in silico search of MOK2 genes in different species has shown that the lamin-binding site is present only

in primate MOK2 proteins Furthermore, we have found that a fraction of human hsMOK2 protein is associated with the nuclear matrix We therefore suggested that hsMOK2 interactions with lamin A⁄ C and the nuclear matrix might be important for its

Keywords

Aurora A; JLP; JNK3; JSAP1; MOK2

Correspondence

M Ernoult-Lange, CNRS-FRE2937, Institut

Andre´ Lwoff, 7 rue Guy Moˆquet, 94801

Villejuif, France

Fax: +33 1 49 58 33 43

Tel: +33 1 49 58 33 46

E-mail: ernoult@vjf.cnrs.fr

(Received 15 December 2008, revised 4

March 2009, accepted 31 March 2009)

doi:10.1111/j.1742-4658.2009.07032.x

Human MOK2 is a DNA-binding transcriptional repressor Previously, we identified nuclear lamin A⁄ C proteins as protein partners of hsMOK2 Fur-thermore, we found that a fraction of hsMOK2 protein was associated with the nuclear matrix We therefore suggested that hsMOK2 interactions with lamin A⁄ C and the nuclear matrix may be important for its ability to repress transcription In this study, we identify JNK-associated leucine zip-per and JSAP1 scaffold proteins, two members of c-Jun N-terminal kinase (JNK)-interacting proteins family as partners of hsMOK2 Because these results suggested that hsMOK2 could be phosphorylated, we investigated the phosphorylation status of hsMOK2 We identified Ser38 and Ser129 of hsMOK2 as phosphorylation sites of JNK3 kinase, and Ser46 as a phos-phorylation site of Aurora A and protein kinase A These three serine resi-dues are located in the lamin A⁄ C-binding domain Interestingly, we were able to demonstrate that the phosphorylation of hsMOK2 interfered with its ability to bind lamin A⁄ C

Abbreviations

GST, glutathione S-transferase; IRBP, interphotoreceptor retinoid-binding protein; JIP, JNK-interacting proteins; JLP, JNK-associated leucine zipper; JNK, c-Jun N-terminal kinase; PKA, protein kinase A.

ability to repress transcription Lamins A and C are

the major products of the LMNA gene which is

expressed in most differentiated cells [5,6] Mutations

in the LMNA gene have been shown to cause a variety

of inherited human diseases (i.e laminopathies) We

investigated whether missense mutations located in the

coil 2 domain of lamin A⁄ C could affect the

interac-tion with hsMOK2 [7] Our results showed that none

of the tested mutations was able to disrupt binding to

hsMOK2 in vitro or in vivo However, we observed an

aberrant cellular localization of hsMOK2 into nuclear

aggregates induced by pathogenic lamin A and C

mutant proteins These results indicated that

patho-genic mutations in lamin A⁄ C lead to sequestration of

hsMOK2 into nuclear aggregates, which may

deregu-late MOK2 target genes

In this study, we identify two new partners of

hsMOK2, which belong to the c-Jun N-terminal kinase

(JNK)-interacting proteins (JIP) family The JIP family

regulates both the JNK and P38 kinase cascade [8–10]

We therefore investigated the phosphorylation status of

hsMOK2 and identified two JNK3 phosphorylation

sites Furthermore, we also identified an Aurora A⁄

pro-tein kinase A (PKA) phosphorylation site on hsMOK2

Interestingly, using phosphomimetic substitution, we

determined that phosphorylation at this site interferes

with the ability of hsMOK2 to bind lamin A⁄ C

Results and Discussion

hsMOK2 interacts with JNK-associated leucine

zipper and JSAP1

To identify partners of hsMOK2 that might be

involved in regulating hsMOK2 functions, we

performed a two-hybrid yeast screen, as described

previously [3] One of the clones corresponded to the

N-terminal region of JNK-associated leucine zipper

(JLP) protein (amino acids 1–141), which is the most

recently identified member of the JIP group of scaffold

proteins [11] To determine which region of hsMOK2

interacts with JLP, we co-transformed the yeast strain

L40 with the library pGAD–JLP 1–141 vector and

pLex containing either the nonfinger acidic domain

(pLex–NH2) or the finger domain (pLex–finger) of

hsMOK2, and performed b-galactosidase assays The

JLP 1–141 domain interacted only with the finger

domain of hsMOK2 (Fig 1A) No interaction was

found with the NH2-acidic domain of hsMOK2

To corroborate the two-hybrid results and test for a

direct interaction between JLP and hsMOK2, the

JLP 1–141 domain was expressed as a glutathione

S-transferase (GST)–fusion protein in bacteria The

GST–JLP 1–141 protein was purified, immobilized on glutathione–agarose beads and incubated with nuclear extracts from HeLa cells transfected with full-length hsMOK2 Consistent with the results obtained in the yeast two-hybrid analysis, it was found that this N-ter-minal region of JLP protein (amino acids 1–141) bound to hsMOK2 (Fig 2A) To further define the region required for interaction with hsMOK2, we con-structed a deletion series by removing N- and C-termi-nal amino acids residues Similar amounts of different GST proteins were used in the binding assay As shown in Fig 2A and summarized in Fig 2B, hsMOK2 was bound by GST–JLP 1–101 and GST– JLP 21–101 deletion mutants at levels similar to those

LexA NH

2 Finger

–

pGAD–GH

pLex–hsMOK2

NH2 LexA Finger

–

pLexA

LexA

pGAD –JLP (1–141)

+/–

pGAD–GH

pLex–NH2

LexA NH

2

–

pGAD–GH

pLex–Finger

LexA Finger

+/–

pLex–NH2

LexA NH

2

pGAD –JLP (1–141)

+++

pLex–hsMOK2

pGAD –JLP (1–141)

+++

LexA Finger

pLex–Finger

pGAD –JLP (1–141)

A

B

Fig 1 Identification of JLP as a partner of hsMOK2 using the yeast two-hybrid screen and identification of the hsMOK2 interac-tion domain (A) Constructs expressing full-length or the indicated domains of hsMOK2 and the human polypeptide JLP 1–141 were co-transformed into yeast The specificity of the interaction between bait and prey was determined by estimating the degree

of color development after 90 min of incubation in the filter lift b-galactosidase assay, as described in Materials and methods (+++) High color blue development, (+ ⁄ )) very low color blue development, ( )) no color development (B) Amino acid sequence alignment of N-terminal of JLP 1–141 and JSAP1 1–146.

with the GST–JLP 1–141 protein Furthermore,

deletion of amino acids 71–101 strongly reduced

inter-action with hsMOK2, and GST–JLP 66–141 and

GST–JLP 66–101 proteins did not bind to hsMOK2

These results demonstrated that the minimal domain

of JLP to mediate hsMOK2 binding was located from

amino acids 21 to 101 and that the region surrounding

JLP residues 71–101 was required but was not

sufficient for this interaction

JLP exhibits sequence homology to JSAP1 (also

called JIP3) [12,13] In particular, they share 77.3%

homology in their N-terminal region (Fig 1B),

suggesting that hsMOK2 may also interact with JSAP1 We therefore tested its interaction with the N-terminal region of JSAP1 protein (amino acids 26–106), which corresponds to amino acids 21–101 of JLP hsMOK2 bound even more efficiently to GST– JSAP1 26–106 protein than to GST–JLP 21–101 (Fig 1A,D) We analyzed and compared the secondary structure of the JLP 1-141 and JSAP1 6–106 domains using the paircoil program [14] This program uses pairwise residue probabilities to detect coiled-coil motifs in protein sequence data, and the database of pairwise residue correlations suggests structural

B

C

E

Fig 2 The N-terminal domains of JLP and JSAP1 bind to hsMOK2 in vitro (A) Mapping the interaction region of JLP using GST pull-down analysis Various JLP N-terminal regions and the homologous JSAP1 region were tested for their interaction with hsMOK2 Nuclear extracts (20 lg) from HeLa cells transfected with the expression vector hsMOK2 were incubated with an equal amount (10 lg) of recombinant GST fusion proteins bound to glutathione beads After washing the beads thoroughly, the bound proteins were eluted in SDS sample buffer, resolved by SDS⁄ PAGE and immunoblotted with an affinity purified anti-hsMOK2 serum The proteins were visualized by exposing the blots

to CL-Xposure film (Pierce) (B) Structure of the JLP deletion mutants The amino acid number of the encoded proteins is indicated for each construct Interactions observed in (A) are summarized on the right (C) Comparison of the predicted coiled-coil structure of JSAP1 6-146 domain (black) with wild-type (red) and mutant D68N (green) and F65L⁄ D68N (blue) JLP 1–141 domains The graphs of coiled-coil scores were determined using the PAIRCOIL program [14] (x-axis) Residue number (y-axis) Probability of a coiled-coil formation (D) GST pull-down

by JSAP1 21–101 domain and wild-type or mutant of JLP 26–106 domains was performed as described in (A) (E) Bound proteins were visu-alized with Fluor-S Max MultiImager and quantified with QUANTITY ONE software (Bio-Rad) Results were expressed as a percentage of binding

to JSAP1 (21–101) domain (mean ± SD of three different experiments).

features that stabilize or destabilize coiled-coils

Analy-sis showed that the predicted coiled-coil region in the

JSAP1 6–146 domain is more extended than in the

JLP 1–141 domain (Fig 2C) The coiled-coil region

begins at residue 69 in the JLP 1–141 domain, whereas

it begins 15 amino acids upstream in the JSAP1 6–146

domain Interestingly, a difference of three amino acids

was found between the two sequences of this region

(Fig 2C) The substitution in JLP sequence of F65

and D68 amino acids by the corresponding JSAP1

residues (Leu70 and Asn73) increases the probability

of residues 54–68 forming a coil-coiled region We

determined experimentally the effects of a single point

mutation D68N and double point mutation

F65L⁄ D68N in JLP (amino acids 21–101) on binding

to hsMOK2 Only the double substitution within the

JLP domain significantly enhanced the association of

JLP with hsMOK2 (Fig 2D,E) suggesting that the

single mutation D68N does not provide enough

stabi-lization of the coil-coiled region The association with

the double point mutation F65L⁄ D68N in JLP became

comparable with that observed with the JSAP1

domain These results confirmed that the region

between residues 59 and 73 in JSAP1 strongly

promotes binding to hsMOK2

To determine whether hsMOK2 interacts with

JSAP1 in mammalian cells, GST pull-down and

co-immunoprecipitation analyses were performed The

endogenous MOK2 protein is difficult to assess

because of its very low expression level, and so we

examined the in vivo interaction in transfected cells

Indeed, the endogenous MOK2 protein has been

detected only by electron microscopy [1] HEK293 cells

were transfected with constructs that expressed

hsMOK2 tagged with GST in the N-terminus (GST–

hsMOK2) and JSAP1 fused to a Flag epitope in the

N-terminus (Flag–JSAP1), either together or

sepa-rately As shown in Fig 3A, Flag–JSAP1 protein was

strongly detected in glutathione-bound proteins from

cells co-transfected with Flag–JSAP1 and GST–

hsMOK2 (lane 3) compared with those from cells

transfected with Flag–JSAP1 alone (lane 1) In reverse

experiments, GST–hsMOK2 protein was strongly

detected in anti-Flag immunoprecipitates from cells

co-transfected with Flag–JSAP1 and GST–hsMOK2

(Fig 3B, lane 3) compared with those from cells

trans-fected with GST–hsMOK2 alone (lane 2) To verify

equivalent recovery of GST–hsMOK2 (Fig 3A, lower)

or Flag–JSAP1 (Fig 3B, lower), the blots were

stripped and reprobed with anti-hsMOK2 or anti-Flag

serum, respectively These results demonstrated that

interaction between the full-length hsMOK2 and

JSAP1 proteins occurs in mammalian cells

hsMOK2 is phosphorylated in cells The interaction of hsMOK2 with JLP and JSAP1 scaf-fold proteins suggests that hsMOK2 activity could be modulated by phosphorylation by the JNK family of MAP kinases We examined the in vivo phosphoryla-tion status of transfected hsMOK2 HeLa cells trans-fected with GST–hsMOK2 were lyzed and incubated with glutathione–agarose to purify the GST–hsMOK2 fusion protein As negative and positive controls, we used GST alone and GST-tagged kinesin-12 (also called Kif 15), respectively [15] We used two commer-cially available antibodies to detect phosphorylation at serine or threonine residues in the hsMOK2 protein Purified GST fusion proteins were resolved in dupli-cate SDS⁄ PAGE gels and immunoblotted with either

A

B

Fig 3 Interaction of hsMOK2 and JSAP1 in human cells Cultured HEK293 cells were transfected with expression vector for Flag– JSAP1 (lane 1), GST–hsMOK2 (lane 2) or co-transfected with both vectors (lane 3) (A) Whole-cell extracts (100 lg) were incubated with 30 lL of 50% slurry glutathione beads After washing the beads thoroughly, the bound proteins were eluted in SDS sample buffer, resolved by SDS ⁄ PAGE and immunoblotted with mouse (Flag M2) mAb The blot was stripped and reprobed with anti-hsMOK2 serum to verify equivalent recovery of the GST fusion pro-tein (lower) (B) Whole-cell extracts from transfected HEK293 cells (100 lg) were immunoprecipitated with 20 lL of anti-(Flag M2) aga-rose affinity gel After washing the beads thoroughly, the bound proteins were eluted in SDS sample buffer, resolved by SDS ⁄ PAGE and immunoblotted with an affinity purified anti-hsMOK2 serum The blot was stripped and reprobed with Flag M2 mAb to verify equivalent recovery of the Flag fusion protein (lower).

anti-(phosphoserine Q5) or anti-(phosphothreonine

Q7) serum None of the antibodies reacted with GST

alone (Fig 4, lane 1) Phosphorylation of GST–

hsMOK2 was detected only by the anti-phosphoserine

serum (Fig 4, lane 3), whereas phosphorylation of

GST–Kin-12–stalk2 was detected only by the

anti-phosphothreonine serum (Fig 4, lane 2) An anti-GST

serum confirmed that equivalent amounts of purified

GST fusion proteins were loaded and that the bands

revealed by the anti-phosphoserine or the

anti-phos-phothreonine serum corresponded to the migration of

GST–hsMOK2 and GST–Kin-12–stalk2 (Fig 4, left)

These results established that hsMOK2 is

phosphory-lated on serine residues in vivo

hsMOK2 is phosphorylated by JNK3, Aurora A

and PKA kinases in vitro

It is known that JNK kinases phosphorylate Ser⁄

Thr-Pro motifs in target proteins [16] hsMOK2 contains

four of these motifs: S28P, S38P, S129P and S191P

(Fig 5A) The sequence of human MOK2 is highly

conserved between primates, but only S129P motif is

conserved (Fig 5A) Because JNK3 is the JNK kinase

expressed primarily in the brain like MOK2 [16,17],

we examined hsMOK2 phosphorylation by JNK3

kinase in an in vitro kinase assay hsMOK2 was

expressed in Escherichia coli as GST fusion protein,

purified on glutathione–agarose and incubated with

activated recombinant JNK3 in the presence of

[32P]ATP[cP] The result showed that recombinant

hsMOK2 was a substrate for JNK3 in vitro (Fig 5B,

lane 1) To determine the possible contribution of the

four serine residues, we replaced individual serine

resi-dues with alanine and expressed the mutant proteins

as GST fusion in E coli Similar quantities of the pro-tein (as shown in the Coomassie Brilliant Blue-stained gel in Fig 5B, lower), were subjected to in vitro phos-phorylation with JNK3 kinase The results showed that replacement of Ser28 or Ser191 with Ala did not decrease the phosphorylation of hsMOK2 by JNK3, whereas the phosphorylation was markedly decreased when Ser38 or Ser129 were replaced with Ala The simultaneous replacement of Ser38 and Ser129 caused

Fig 4 hsMOK2 is a phosphoserine protein Whole-protein extracts (500 lg) from HeLa cells transfected with GST (lane 1), GST–Kin-12–stalk2 (lane 2) or GST–hsMOK2 (lane 3) were incubated with 50 lL of 50% slurry glutathione beads After thoroughly washing the beads, the bound proteins were eluted in SDS sample buffer, resolved in duplicate gels by SDS ⁄ PAGE and immunoblotted with either anti-(phosphoSerine Q5) serum (middle) or anti-(phosphoThreonine Q7) serum (right) The same blots were stripped and re-probed with anti-GST serum to confirm equivalent loading of GST fusion protein (left) The proteins were visualized by exposing the blots to CL-Xposure film (Pierce).

A

B

Fig 5 Phosphorylation of hsMOK2 in vitro by JNK3 kinase (A) Alignment of primate MOK2 proteins highlighting the potential SP motifs for JNK kinases in bold The percentage identity with human

is indicated in parentheses (B) GST-tagged wild-type or mutant hsMOK2 bound to glutathione beads were incubated with recombi-nant JNK3 kinase in the presence of [ 32 P]ATP[cP] The proteins were then separated on SDS ⁄ PAGE The gel was subjected to Coomassie Brilliant Blue staining (lower) followed by autoradio-graphy (upper).

a larger decrease in phosphorylation The data

demon-strate that JNK3 phosphorylates hsMOK2 on Ser38

and Ser129 in vitro

Because JSAP1 is expressed along with JNK3 and

MOK2 in the brain, JSAP1 could bring together

MAPKs and hsMOK2 in this tissue Such a function

of JIP proteins has been previously proposed for two

other transcription factors, c-Myc and Max [11]

Inter-estingly, the region of JLP that binds Myc is similar to

the region of JSAP1 binding hsMOK2 It is tempting

to speculate that JSAP1 may enhance hsMOK2

phos-phorylation by JNK3 Unfortunately, our attempts to

demonstrate such an effect were unsuccessful The

JSAP1 protein had no effect when added to our in vitro

phosphorylation assay (data not shown), which may

be because the recombinant JNK3 is an activated form

of the kinase To address this issue in vivo, HEK293

cells, which do not express JSAP1, were transiently

transfected with tagged hsMOK2, with and without

JSAP1 JSAP1 did not stimulate hsMOK2

phosphory-lation, even after activation of endogeneous JNK

kinases following incubation of the cells with sorbitol

This does not preclude that JSAP1 may play a role in

hsMOK2 phosphorylation in brain tissue

A computer search for other potential

phosphoryla-tion sites indicated the existence of several putative

PKA, protein kinase C and caseine kinase II

phos-phorylation sites spread along the hsMOK2 sequence,

as well as two Aurora phosphorylation sites in the

lamin A⁄ C-binding N-terminal acidic domain of

hsMOK2 Because these two Aurora sites, centered on

amino acids Ser46 and Ser146, are strictly conserved

between primates [18] (Fig 6A), we tested whether

they could be phosphorylated by Aurora A kinase

In vitro phosphorylation experiments showed that

hsMOK2 was a substrate for recombinant Aurora A

kinase (Fig 6B) To determine which serine is

phosphorylated, we replaced Ser46 or Ser146, or both

Ser46 and Ser146, with alanine in GST–hsMOK2

constructs Incubation of these fusion proteins with

recombinant Aurora A revealed a minor reduction in

phosphorylation of the mutant containing only the

S146A mutation (Fig 6B, lane 3), compared with the

wild-type, and a remarkably reduced phosphorylation

of the two mutants containing the S46A mutation

(Fig 6B, upper, lanes 2, and 4) We concluded that

only Ser46 is a major Aurora A phosphorylation site

on hsMOK2 Recently, it has been reported that

human Aurora A and Aurora B kinases prefer

substrate sequences with an arginine residue at the

position )2 [19,20] Accordingly, only the sequence

surrounding Ser46 in hsMOK2 conforms to this

preference (RDSV) Lastly, because the consensus for

Aurora A is reminiscent of that of PKA, we examined the ability of PKA to phosphorylate hsMOK2 protein

in vitro We obtained the same phosphorylation pattern

of wild-type and mutant hsMOK2 as observed with Aurora A kinase (Fig 6C) We conclude that hsMOK2

is efficiently phosphorylated in vitro by Aurora A kinase and PKA at residue Ser46

Analysis of phosphomimetic mutations

on hsMOK2 capacity to bind DNA Binding of hsMOK2 to DNA may be affected, either positively or negatively, by phosphorylation There-fore, to determine whether serine phosphorylation would affect the ability of hsMOK2 to bind DNA, we introduced phosphomimetic mutations in hsMOK2 by replacing individual serine residues by aspartic acid

A

B

C

Fig 6 Phosphorylation of hsMOK2 in vitro by Aurora A and PKA (A) Alignment of primate MOK2 proteins highlighting the two conserved Aurora phosphorylation motifs in bold (B) GST-tagged wild-type or mutant hsMOK2 bound to glutathione beads were incubated with recombinant Aurora A in the presence of [32P]ATP[cP] The proteins were then separated on SDS ⁄ PAGE The gel was subjected to Coomassie Brilliant Blue staining (lower) followed by autoradiography (upper) (C) GST-tagged wild-type or mutant hsMOK2 bound to glutathione beads were incubated with recombinant PKA, in the presence of [ 32 P]ATP[cP] and visualized as in (B).

Gel-shift experiments were performed with nuclear

extracts of HeLa cells expressing wild-type or mutant

hsMOK2 proteins and a [32P]-labeled double-stranded

oligonucleotide, corresponding to the 18-bp

MOK2-binding site present in hsIRBP intron 2 Immunoblot

analysis attested that similar amounts of the wild-type

and mutant proteins were used in the gel-shift assay

(data not shown) As shown in Fig 7, phosphomimetic

substitutions at positions Ser46, Ser38 and Ser129 had

no effect on protein–DNA complex abundance at any

o the NaCl concentrations tested Hence, the two

phosphomimetic mutants retained the ability to bind

the specific DNA sequence with high affinity,

indicat-ing that this bindindicat-ing is not regulated by

phosphoryla-tion at Aurora A⁄ PKA or JNK sites

Effect of hsMOK2 phosphorylation on its capacity

to bind lamin A⁄ C

The two JNK phosphorylation sites of hsMOK2 are

located in the lamin A⁄ C-binding N-terminal acidic

domain We therefore examined whether

phosphoryla-tion of hsMOK2 at JNK3 or Aurora A⁄ PKA sites had

an impact on the interaction with lamin A⁄ C Nuclear

extracts from HeLa cells expressing wild-type or

mutant hsMOK2 proteins were prepared and

incu-bated with an equal amount of GST–DlaminC bound

to glutathione beads Phosphomimetic substitutions at Ser38, Ser129 or the double Ser38Ser129 mutation did not markedly decrease binding to DlaminC (Fig 8A, upper) By contrast, phosphomimetic substitution at Ser46 markedly decreased the binding, although ala-nine substitution at Ser46 had no effect (Fig 8A, lower) The data indicated that the phosphorylation of hsMOK2 at the Aurora A⁄ PKA site interfered with its ability to bind lamin A⁄ C in vitro We then sought evi-dence that a similar effect occurs in vivo We used the characteristic that mutations in lamin A⁄ C lead to sequestration of hsMOK2 in nuclear aggregates (Fig 8B) [7] HeLa cells were co-transfected with the expression vector for the Q294P lamin C mutant and wild-type hsMOK2 or hsMOK2 mutated at position S46 The nonphosphorylatable hsMOK2–S46A protein was found in the nuclear aggregates induced by the Q294P lamin C mutant like hsMOK2–WT protein (Fig 8B), whereas the phosphomimetic hsMOK2– S46D protein exhibited a homogeneous nuclear pattern (Fig 8B) The hsMOK2–S46D protein was therefore not displaced in nuclear aggregates, demonstrating that phosphomimetic substitution at Ser46 also prevents the interaction with lamin A⁄ C in vivo

To confirm that phosphorylation in vivo can disrupt the interaction between hsMOK2 and lamin A⁄ C, we examined the localization of hsMOK2–WT in cells treated with the phosphatase inhibitor orthovanadate

to enhance cellular phosphorylation HeLa cells co-transfected with expression vector for lamin C–Q294P and hsMOK2–WT were incubated with

1 mm sodium orthovanadate for 8 h In this condition,

no sequestration of hsMOK2 by mutant lamin A⁄ C was observed, confirming the importance of phosphor-ylation for hsMOK2 and lamin A⁄ C interaction (Fig 8C, upper) However, the same observation was made using hsMOK2–S46A (Fig 8C, lower), indicat-ing that the effect of orthovanadate treatment can be mediated by phosphorylation at another position Human MOK2 is a DNA-binding transcriptional repressor and its interaction with lamin A⁄ C and the nuclear matrix may be important for its ability to repress transcription Such involvement of lamins A⁄ C has been proposed previously for the transcriptional activator pRb [21–23] pRb controls cell-cycle progres-sion by negatively regulating the E2F transcription fac-tor in a phosphorylation-dependent manner [24] The active (hypophosphorylated) form of pRb co-localizes with lamins A⁄ C at the nuclear periphery in vivo and binds to lamins in vitro [21] Thus, transcriptional repression by pRb correlates with its lamin-binding activity Similarly, transcriptional repression by hsMOK2 might be correlated with its lamin-binding

Fig 7 Effects of hsMOK2 phosphomimetic mutations on DNA

binding activity EMSA was performed on whole-cell extracts

derived from HeLa cells transfected with various hsMOK2

expres-sion plasmids as indicated The amount of extracts was adjusted to

obtain equal level of the various hsMOK2 proteins The 32 P-labeled

double-stranded oligonucleotide corresponds to the 18-bp MOK2

binding site of human IRBP gene The binding reaction was

performed in buffer containing various concentrations of NaCl.

activity The simplest scenario is one in which the

lamin A⁄ C–hsMOK2 complex stabilizes a repressive

complex on DNA, preventing gene activation To

allow gene activation, hsMOK2 would be

phosphory-lated and then released from lamin A⁄ C This

regula-tion may take place following activaregula-tion of kinases

such as PKA in response to various signaling

pathways In addition, Aurora A kinase is specifically

activated before mitosis [25,26] Mitotic nuclear

envelope breakdown requires disassembly of the

nuclear lamina Lamins A and C are rapidly released

throughout the nucleoplasm in early prophase [27,28]

hsMOK2 dissociation from lamin A⁄ C at

hsMOK2-regulated loci in early mitosis may contribute to the

dispersion of lamins A⁄ C into the cytoplasm

Materials and methods

Plasmid constructs

pCMV–hsMOK2, pCMV–laminC(Q294P), pGEX–hsMOK2, pGEX–DlaminC and pGEX–Kin-12-stalk2 vectors have been described previously [2,3,7,15] Point mutations were introduced into hsMOK2 constructs using the

Amsterdam, The Netherlands) The prokaryotic expression vector pET29, containing Aurora A cDNA, was a kind gift from C Prigent (Faculte´ de Me´decine, Rennes, France)

generated from pGAD–JLP(1–141) by PCR using 5¢

C

Fig 8 Effect of hsMOK2 phosphorylation on its capacity to bind lamin A ⁄ C (A) For in vitro interaction, nuclear extracts from HeLa cells overexpressing wild-type or mutant hsMOK2 were incubated with 10 lg of GST–DlaminC or GST alone bound to glutathione beads The amount of nuclear extracts was adjusted to obtain equal levels of the various hsMOK2 proteins Input lanes correspond to 5% (upper) and 10% (lower) of the extracts used for binding reaction After thoroughly washing the beads, the bound proteins were eluted in SDS sample buffer, resolved by SDS ⁄ PAGE and immunoblotted with affinity purified anti-hsMOK2 serum The proteins were visualized by exposing the blots to CL-Xposure film (Pierce) (B) For in vivo interaction, HeLa cells were co-transfected with expression vector for lamin C–WT or lamin C–Q294P mutant and hsMOK2–WT, hsMOK2–S46A or hsMOK2–S46D After 36 h, cells were fixed and double stained sequentially with lamin A ⁄ C mAb and anti-hsMOK2 serum Cells were observed with a Leica DMR microscope and an Apochromat 63 · 1.32 oil immersion objective (C) HeLa cells were co-transfected with expression vector for lamin C–Q294P and hsMOK2–WT or hsMOK2–S46A Sixteen hours after transfection, the cells were treated with 1 m M sodium orthovanadate for 8 h, fixed and double-stained sequentially with lamin A ⁄ C mAb and anti-hsMOK2 serum.

primers and 3¢ primers containing an EcoRI site and a

SalI site, respectively After digestion with EcoRI and

SalI, the products were cloned into the corresponding

sites of the pGEX–6P1 vector (Amersham Pharmacia

pcDNA3–Flag–JSAP1 plasmid containing the entire coding

K Yoshioka [29]

Cell culture, transfections and protein extracts

Human HeLa or HEK293 cells were routinely maintained

in Dulbecco’s modified Eagle’s medium supplemented

with 10% fetal calf serum For transient transfections,

glass coverslip, 24 h prior to transfection with 2 lg of

transfected HeLa cells from three 100-mm Petri dishes

buffer (25 mm Hepes, pH 7.5, 150 mm NaCl, 10%

with a protease inhibitor cocktail without EDTA (Roche

Diagnostics, Meylan, France) and disrupted by sonication

After centrifugation at 15 000 g for 10 min, the total

from one 100-mm Petri dish plated with transfected

described for transfected HeLa cells Nuclear extracts

were prepared according to the Dignam method [31] and

dialysis and centrifugation, the nuclear extract was frozen

Brebieres, France)

Antibodies

Affinity-purified rabbit polyclonal anti-hsMOK2 serum

was obtained as described previously [2] Mouse

Biotechnology, Heidelberg, Germany) Mouse monoclonal

anti-(Flag M2) and anti-(Flag M2) agarose affinity gel

France) Mouse (phosphoThreonine Q7) and

anti-(phosphoSerine Q5) sera were purchased from Qiagen

(Courtaboeuf, France) Rhodamine (TRITC)-conjugated

anti-(rab-bit IgG) and peroxidase-conjugated rabanti-(rab-bit anti-(mouse

IgG) sera were purchased from Jackson Immunoresearch

Laboratories (Bar Harbor, ME, USA)

Yeast two-hybrid screen

Yeast two-hybrid screening using human hsMOK2 as bait was described previously [3] A human HeLa S3 Match-maker cDNA library (BD Clontech, St-Germain-en-Laye, France), constructed in the pGAD–GH vector expressing the GAL4 activation domain fusion protein, was trans-formed into L40 containing the pLex–hsMOK2 construct The cDNA inserts of positive clones were isolated by direct PCR of yeast colonies The cDNA inserts were further

sequence similarity in the GenBank database with the program blast

Immunofluorescence microscopy

followed by incubation with the secondary antibody The incubations were for 1 h each and were carried out

4¢-6-diamidino-2-phenylindole for 1 min The slides were mounted in

Ltd, London, UK) Immunofluorescence microscopy was

immersion objective Photographs were taken using a Micromax (Princeton Instruments, Evry, France) CCD camera and metaview (Universal Imaging Corp.) software

Purification of GST fusion proteins and kinase assay

protein extracts were prepared and purified as described previously [3] The purity and amount of the recombinant

staining with Coomassie Brilliant Blue The GST fusion proteins bound to glutathione beads were used as 50%

full-length human Aurora A protein was expressed in

agarose as described by Cremet et al [32] The Aurora A protein solution was concentrated using a centricon 10

purchased from Upstate (St-Quentin-en Yvelines, France) and the N-terminal His tagged human catalytic subunit

UK)

The kinase assays were performed in 25 lL of 50 mm

0.2 mm dithiothreitol, 0.2 mm sodium orthovanadate,

and 10–15 lL of a 50% slurry of GST fusion proteins

bound to the beads The reactions were incubated at

for 5 min The proteins were then separated on a Nupage

4–12% Bis-Tris gels (Invitrogen, Illkirch, France) The gel

was subjected to Coomassie Brilliant Blue staining, dried

and analysed using a Phosphoimager apparatus

(Mole-cular Dynamics, Orsay, France)

GST pull-down assay and

co-immunoprecipitation

protein, immobilized on glutathione–agarose beads was

added to 20 lg of nuclear proteins from transfected HeLa

cells, in a total volume of 400 lL Hepes buffer For in vivo

GST pull-down assay or co-immunoprecipitation,

whole-cell extracts (100 lg) from transfected HEK293 were

incu-bated either with 30 lL of 50% slurry glutathione beads or

with 20 lL of anti-(Flag M2) agarose affinity gel After 2 h

boiling in Laemmli buffer Bound proteins were separated

indicated antibodies using the Supersignal West Pico

Chemiluminescent Signal kit (Pierce) The proteins were

visualized by exposing the blots to CL-XPosure film

France)

Electrophoretic mobility shift assay

A 25-bp oligonucleotide corresponding to the sequence of

human IRBP intron 2 containing the 18-bp MOK2-binding

site (5¢-CTGCAGGACTTGTCAGGGCCTTTAA-3¢) was

used as a probe The double-strand oligonucleotide was

labeled with T4 polynucleotide kinase (Biolabs, Ipswich,

a 15% polyacrylamide gel End-labeled oligonucleotides

(0.2 ng) were incubated for 20 min at room temperature in

20 lL Hepes buffer containing 2 lg poly(dI–dC), various

concentrations of NaCl and whole-protein extracts from

HeLa cells transfected with wild-type or mutant hsMOK2

The amount of extract was adjusted to obtain an equivalent

level of the transfected hsMOK2 proteins Complexes were

analyzed by electrophoresis on a nondenaturing

0.5· TB buffer (45 mm Tris borate, pH 8.3) at 4 C at

200 V EDTA was omitted in all binding and

electrophore-sis buffers to avoid denaturing hsMOK2

Acknowledgements

We thank Katsuji Yoshioka for providing the pcDNA3–Flag–JSAP1 expression vector and Claude Prigent for pET29–Aurora A vector We also thank Vanessa Philipot for technical assistance and Domi-nique Weil for critical reading of the manuscript This research was supported by grants from the Centre National de la Recherche Scientifique, the Fondation Raymonde et Guy Strittmatter and the association Retina France

References

1 Arranz V, Harper F, Florentin Y, Puvion E, Kress M

& Ernoult-Lange M (1997) Human & mouse MOK2 proteins are associated with nuclear ribonucleoprotein components & bind specifically to RNA & DNA through their zinc finger domains Mol Cell Biol 17, 2116–2126

2 Arranz V, Dreuillet C, Crisanti P, Tillit J, Kress M & Ernoult-Lange M (2001) The zinc finger transcription factor, MOK2, negatively modulates expression of the interphotoreceptor retinoid-binding protein (IRBP) gene J Biol Chem 276, 11963–11969

3 Dreuillet C, Tillit J, Kress M & Ernoult-Lange M (2002) In vivo & in vitro interaction between human

Nucleic Acids Res 30, 4634–4642

4 Ernoult-Lange M, Arranz V, Leconiat M, Berger R & Kress M (1995) Human & mouse Kruppel-like (MOK2) orthologue genes encode two different zinc finger pro-teins J Mol Evol 41, 784–794

5 Fisher DZ, Chaudhary N & Blobel G (1986) cDNA sequencing of nuclear lamins A & C reveals primary & secondary structural homology to intermediate filament proteins Proc Natl Acad Sci USA 83, 6450–6454

6 McKeon FD, Kirschner MW & Caput D (1986) Homologies in both primary & secondary structure between nuclear envelope & intermediate filament proteins Nature 319, 463–468

7 Dreuillet C, Harper M, Tillit J, Kress M & Ernoult-Lange M (2008) Mislocalization of human transcription factor MOK2 in the presence of pathogenic mutations

8 Dhanasekaran DN, Kashef K, Lee CM, Xu H & Reddy

EP (2007) Scaffold proteins of MAP-kinase modules Oncogene 26, 3185–3202

9 Morrison DK & Davis RJ (2003) Regulation of MAP kinase signaling modules by scaffold proteins in mam-mals Annu Rev Cell Dev Biol 19, 91–118

10 Yoshioka K (2004) Scaffold proteins in mammalian MAP kinase cascades J Biochem 135, 657–661

11 Lee CM, Onesime D, Reddy CD, Dhanasekaran N & Reddy EP (2002) JLP: a scaffolding protein that tethers