Báo cáo khoa học: Annexin A2 binds to the localization signal in the 3¢ untranslated region of c-myc mRNA ppt

Here we show the specific binding of a trans-acting factor to the perinuclear localization ele-ment in the 3¢UTR of c-myc mRNA and identify this protein as annexin A2.. Gel retardation an

untranslated region of c-myc mRNA

Ian Mickleburgh1, Brian Burtle1, Hanne Holla˚s2, Gill Campbell3, Zofia Chrzanowska-Lightowlers4, Anni Vedeler2and John Hesketh1

1 School of Cell and Molecular Biosciences, University of Newcastle, UK

2 Department of Biomedicine, University of Bergen, Norway

3 Rowett Research Institute, Aberdeen, UK

4 School of Neurology, Neurobiology and Psychiatry, University of Newcastle, UK

The delivery of newly synthesized proteins to their site

of function is crucial for normal cell function There is

now evidence that in an increasing number of specific

cases this not only involves targeting signals within

proteins, but also signals in mRNAs resulting in their

localization and translation in different cytoplasmic

compartments [1–4] Messenger RNA localization is

observed during early development in Drosophila and

Xenopus, in highly polarized neurones and glial cells

[5–7], and in fibroblasts [8–10] Such mRNA localiza-tion is dependent on cis-acting sequences almost exclu-sively found in the 3¢ untranslated regions (3¢UTRs) of the mRNAs concerned [7–14] Using transfected cell lines expressing chimaeric gene constructs, 3¢UTR sequences have been found to be capable of targeting

a reporter sequence to different cytoplasmic sites and

to the cytoskeleton [8–13] Complementary experiments have shown that removal of the appropriate 3¢UTR

Keywords

cytoskeleton; mRNA localization;

RNA-binding protein; targeting; 3¢UTR

Correspondence

J Hesketh, School of Cell and Molecular

Biosciences, University of Newcastle,

Newcastle-upon-Tyne NE1 7RU, UK

Fax: +44 191 222 8684

Tel: +44 191 222 8744

E-mail: j.e.hesketh@ncl.ac.uk

(Received 1 September 2004, revised

20 October 2004, accepted 15 November

2004)

doi:10.1111/j.1742-4658.2004.04481.x

Messenger RNA trafficking, which provides a mechanism for local protein synthesis, is dependent on cis-acting sequences in the 3¢ untranslated regions (3¢UTRs) of the mRNAs concerned acting together with trans-act-ing proteins The C-MYC transcription factor is a proto-oncogene product involved in cell proliferation, differentiation and apoptosis Localization of c-myc mRNA to the perinuclear cytoplasm and its association with the cytoskeleton is determined by a signal in the 3¢UTR Here we show the specific binding of a trans-acting factor to the perinuclear localization ele-ment in the 3¢UTR of c-myc mRNA and identify this protein as annexin A2 Gel retardation and UV cross-linking experiments showed that pro-teins in fibroblast extracts formed complexes with the region of c-myc 3¢UTR implicated in localization; a protein of  36 kDa exhibited specific,

Ca2+-dependent binding Binding was reduced by introduction of a muta-tion that abrogates localizamuta-tion Using RNA-affinity columns followed by gel electrophoresis and mass spectrometry this protein was identified as annexin A2 The RNA–protein complex formed by cell extracts was further retarded by anti-(annexin A2) Purified annexin A2 bound to the same region of the c-myc 3¢UTR but binding was reduced by introduction of a mutation, as with cell extracts It is proposed that binding of annexin A2

to the localization signal in the c-myc mRNA leads to association with the cytoskeleton and perinuclear localization The data indicate a novel func-tional role for the RNA-binding properties of annexin A2 in perinuclear localization of mRNA and the association with the cytoskeleton

Abbreviations

DTT, dithiothreitol; MBP, myelin basic protein; MS, mass spectrometry; mRNP, messenger ribonucleoprotein; PVDF, poly(vinylidene fluoride); UTR, untranslated region.

results in loss of, or altered, localization For example,

the transport and localization of both myelin basic

protein (MBP) and b-actin mRNAs require a signal

within the 3¢UTR [8,14] RNA-containing particles are

found colocalized with cytoskeletal components [9,14]

and there is evidence that mRNAs are transported in

RNA granules [15,16] The detailed mechanisms of this

spatial organization of the protein synthetic apparatus

and mRNA localization by 3¢UTR signals are still

poorly understood, particularly the nature of the

pro-teins that bind to these localization signals

In fibroblasts, b-actin mRNA is transported to the

cell periphery, whereas mRNAs encoding the

tran-scription factors MYC and FOS are localized around

the nucleus and are associated with the cytoskeleton

[10,13] In c-myc mRNA the localization signal lies in

an 86-nucleotide region within the 3¢UTR and is

abro-gated by a mutation in a conserved AUUUA [11] The

RNA-binding protein(s) involved in this retention of

the c-myc mRNA on the cytoskeleton around the

nuc-leus is not known but it is likely to be distinct from

those with roles in the transport and peripheral

local-ization of mRNAs such as b-actin and MBP Here we

describe the specific binding of a trans-acting factor to

the region of the 3¢UTR of c-myc mRNA previously

shown to contain the localization element, and

iden-tify this protein as annexin A2 The multifunctional

annexin A2 has previously been reported to have

RNA-binding properties [17,18] and the data presented here indicate a functional role for such binding and provide evidence for a novel role of this protein in perinuclear localization of mRNA

Results

The perinuclear localization element in the c-myc mRNA has previously been mapped to between nucleo-tides 194 and 280 of the 3¢UTR: the b-globin reporter

is localized by nucleotides 194–440 (D3) and 194–280 (MW) from the wild-type c-myc 3¢UTR, but not by nucleotides 194–280 in which the AUUUA motif was mutated to AGGGA (MM) [11] Protein binding to this localization signal was investigated by gel retardation and UV cross-linking assays using RNA transcripts that corresponded to these regions of the 3¢UTR Gel retardation assays usingD3 transcripts showed complex formation with increasing amounts of a S100 cell extract from Ltk– fibroblasts (Fig 1A) Competitive experiments carried out using [32P]UTP[aP]-labelledD3 transcripts and unlabelled MW transcripts (Fig 1B) showed that the shorter 86-nucleotide transcripts competed effectively for protein binding to D3 tran-scripts There was almost total inhibition of complex formation at 80-fold molar excess In contrast, the presence of mutant MM transcripts, even at 80-fold molar excess, had little or no effect on protein

Fig 1 RNA–protein complex formation monitored by gel retardation assay Complex formation was studied using [ 32 P]UTP[aP]-labelled D3 RNA (nucleotides 194–440 of c-myc 3¢UTR) and S100 extract from Ltk – fibroblasts (A) RNA was incubated with increasing amounts of S100 protein extract (1–5 lg) Complex formation is observed with 2 and 5 lg (B) Complex formation was studied using [ 32 P]UTP[aP]-labelled D3 RNA (500 Bq) in the presence of 10–80-fold molar excess of unlabelled competitor RNA, either MW or MM transcripts as indicated.

binding to the D3 transcripts, showing that the MM

transcripts did not compete for protein binding These

experiments indicate that one or more proteins capable

of binding to nucleotides 194–440 of the c-myc 3¢UTR

are present in cytoplasmic extracts of Ltk– fibroblasts

and that this binding involves nucleotides 194–280

implicated in localization Furthermore, the data

indicate that the conserved AUUUA found necessary

for localization [11] is necessary for full binding

activity

The proteins binding to this region of the c-myc

3¢UTR were further investigated by UV cross-linking

of RNA–protein complexes followed by SDS⁄ PAGE

As shown in Fig 2A, the D3 RNA exhibited binding

to two major proteins and one other minor

compo-nent Comparison of the mobility of these proteins in

SDS⁄ PAGE with that of molecular mass standards

indicated that the major proteins were of approximate

molecular mass 36 and 50 kDa, with the minor protein

of  90 kDa The same pattern of binding was

observed with a cytoskeletal fraction (from fibroblasts)

that is known to be enriched in c-myc mRNA [9,19]

but no binding was observed with cytosolic or

endo-plasmic reticulum fractions (Fig 2B) Binding of the

proteins to nucleotides 194–280 was investigated by

carrying out competition experiments in which the cell

extract was incubated with both [32P]UTP[aP]-labelled

D3RNA and increasing amounts of unlabelled MW

prior to cross-linking As shown in Fig 2(A),

increas-ing amounts of unlabelled MW reduced the bindincreas-ing

of RNA to the 36 kDa protein such that binding was

reduced by 20-fold excess of competitor and was

almost undetectable in the presence of 80-fold molar

excess In contrast, the 80-fold molar excess of

unla-belled MW had comparatively little effect on binding

to the other proteins A similar excess of a nonspecific

RNA, the c-myc coding region, had no effect on

bind-ing (results not shown) These data indicate that the

36 kDa protein bound to the 194–280 region of the

3¢UTR

These observations were extended using shorter

transcripts (nucleotides 205–280 of the c-myc

3¢UTR), RNase T1 digestion and gel retardation As

shown in Fig 3, these shorter transcripts also formed

a complex with S100 extracts and complex formation

was unaffected by an excess of homoribopolymer A

and C (lanes 6, 7) but abolished by

homoribopoly-mer G (lane 8) and to a lesser extent by

homoribo-polymer U (lane 9); it was also Ca2+ sensitive

(compare lanes 2 and 4 with lanes 3 and 5) In

addi-tion, formation of radiolabelled complex disappeared

after incubation with unlabelled MW competitor

(lane 10), but not after incubation with MM (lane

11) Again these data support the view that complex formation is due to binding of a protein to the 3¢UTR implicated in localization

Biotinylated RNA transcripts linked to streptavidin-coated magnetic beads were used to isolate proteins binding to the region of c-myc 3¢UTR implicated in localization The beads were incubated with S100

B A

Fig 2 UV cross-linking analysis of proteins binding to c-myc 3¢UTR (A) [ 32 P]UTP[aP]-labelled D3 (nucleotides 194–440 of c-myc 3¢UTR) RNA was incubated either with S100 extract alone or with a 2–80-fold molar excess of unlabelled competitor MW RNA (nucleo-tides 194–280 of c-myc 3¢UTR) After UV cross-linking, samples were subjected to SDS ⁄ PAGE and RNA–protein complexes were detected by the presence of radioactive bands In the absence of competitor the D3 RNA formed a complex with two major proteins, one of  36 kDa (indicated by arrowhead) and one of  50 kDa, and a minor protein The presence of competitor MW RNA reduced the complex formation between RNA and the 36 kDa protein but had no effect on complex formation with the larger protein (B) [ 32 P]UTP[aP]-labelled D3 RNA was incubated with protein from

a cytosolic (lane 1), cytoskeletal (lane 2) or membrane fraction (lane 3), or with S100 extract (lane 4), and after UV cross-linking, sam-ples were subjected to SDS ⁄ PAGE Note that complex formation occurred with proteins, including a 36 kDa protein (arrow), in the cytoskeletal fraction.

extract from mouse Ltk– fibroblasts and after removal

of excess extract and stringent washing, bound

mater-ial was released and subjected to SDS⁄ PAGE In the

absence of Ca2+the major protein bound to the RNA beads was of  50 kDa (Fig 4A, lane 1) Following incubation in the presence of 1 mm Ca2+ the major

Fig 3 RNA–protein complex formation with nucleotides 205–280 of the c-myc 3¢UTR Gel retardation using [ 32

P]UTP[aP]-labelled D205 RNA (12 fmoles; nucleotides 205–280 of 3¢UTR) and 1 lg of protein from an Ltk – fibroblast S100 extract RNase T1digestion was performed after the binding reaction Lane 1 contains free probe and lanes 2–5 show retardation complex formed (arrowhead) with extract in a buffer either containing 40 m M or 120 m M NaCl and in the absence and presence of 1 m M CaCl 2 The effects of competition with a 100-fold mass excess

of homoribopolymers poly(A), poly(C), poly(G) and poly(U) are shown in lanes 6–9, respectively Poly(G) and poly(U) dramatically reduce the calcium-dependent gel retardation complex In lanes 10 and 11 a 160-fold molar excess of MW (nucleotides 194–280 of c-myc 3¢UTR) and

MM (nucleotides 194–280 of c-myc 3¢UTR with AGGGA mutation), respectively, were used to compete with D205 RNA for protein binding Note that MW competes much more effectively than MM.

A

B

C

Fig 4 Isolation of proteins binding to nucleotides 205–280 of the c-myc 3¢UTR in the absence and presence of calcium Proteins from an Ltk–S100 extract (1 mg) were incubated with biotinylated D205 RNA anchored to SA-PMP beads (see Experimental procedures) or to control SA-PMP beads with no RNA attached Ten microlitres of unbound proteins (from binding solution after incubation) and half the volume of eluted proteins were separated by 12.5% (w ⁄ v) SDS ⁄ PAGE (A) and (C) show gels stained with Coomassie Brilliant Blue and in (B) western blotting analysis was performed with monoclonal anti-(annexin A2) IgG at a 1 : 5000 dilution In (A) and (B) the RNA-bound proteins and the unbound proteins (first wash) recovered in the absence of calcium are shown in lanes 1 and 2, respectively, and the RNA-bound and unbound proteins recovered in the presence of calcium are shown in lanes 3 and 4, respectively In (C) Lane 1 shows proteins eluted from SA-PMP alone (no biotinylated RNA) in the presence of calcium compared with the eluate from D205 RNA-bound SA-PMP (lane 2) Black arrowheads indicate 36 kDa protein and white arrowhead points to 50 kDa protein.

protein bound was of 36 kDa (Fig 4A, lane 3) and

the 50 kDa band was less intense Under these

condi-tions the 36 kDa protein did not bind to control beads

with no RNA (Fig 4C)

Because UV cross-linking indicates the specific

bind-ing of a 36 kDa component to this region of the

3¢UTR and RNase T1 digest⁄ gel retardation shows

Ca2+ sensitivity of complex formation, our further

analysis focused on the 36 kDa component of the

pro-teins recovered in the eluates from the RNA beads

Western blotting (Fig 4B) suggested that this major

protein was annexin A2, an observation consistent

with the observed Ca2+ sensitivity of specific complex

formation, the previously observed RNA-binding

properties of annexin A2 [17,18] and the competition

by poly(G) (Fig 3; cf [18]) The band corresponding

to this major 36 kDa component was excised from the

gel, digested with trypsin and subjected to

MALDI-TOF⁄ MS Comparison of the digestion pattern with

the available database confirmed that this protein band

corresponds to mouse annexin A2

Taken together, the western blotting and MS data

show that the 36 kDa protein that is present in

mouse fibroblast extracts and which binds to

nucleo-tides 205–280 of c-myc 3¢UTR is annexin A2 Gel

retardation experiments with cell extracts in the

pres-ence of anti-(annexin A2) showed that such

antibod-ies caused increased retardation of the complex or

‘supershift’ (Fig 5), but that this did not occur with

a comparable concentration of a control IgG

Demonstration of this supershift with anti-annexin

provides further evidence that annexin A2 is present

in the complex formed by fibroblast extracts with

the c-myc 3¢UTR transcripts

In further experiments, purified annexin A2

demon-strated the ability to bind to c-myc transcripts in vitro

The c-myc transcripts corresponding to either exon 3

(which contains both coding and 3¢UTR regions), the

3¢UTR or the region of the 5¢UTR containing the

first 496 nucleotides, were labelled in vitro with

[32P]UTP[aP] and incubated with immobilized

ann-exin A2 heterotetramer on nitrocellulose membranes

As shown in Fig 6(A), transcripts of exon 3 or only

the 3¢UTR bound to annexin A2, whereas the

1–496-nucleotide transcript of exon 1 did not interact

Fur-thermore, purified annexin A2 bound to labelled MW

transcripts corresponding to the 194–280-mucleotide

3¢UTR region but markedly less (57%) to the mutant

MM transcripts (Fig 6B) There was essentially no

binding of annexin A2 to control antisense transcripts

(10% of binding to MW) These data indicate that,

in vitro, annexin A2 binds to myc transcripts that

con-tain 3¢UTR sequences

Discussion

Using two independent methods, namely gel retarda-tion assays and UV cross-linking, the experiments presented here provide evidence for the existence of

a protein of  36 kDa in fibroblast cell extracts that binds specifically to the region of the 3¢UTR previ-ously implicated in the localization of c-myc mRNA [11] RNA-affinity experiments followed by MS iden-tified this protein as annexin A2, supershift assays showed annexin A2 to be present in the complexes formed by the fibroblast extracts and c-myc 3¢UTR transcripts, and in vitro experiments indicated that purified annexin A2 binds to this region of the c-myc 3¢UTR In addition, assays with both cell extracts and purified annexin A2 indicated that the conserved

Fig 5 The effect of anti-(annexin A2) IgG on RNA–protein complex formation with nucleotides 205–280 of the c-myc 3¢UTR Binding reactions were carried out using [ 32 P]UTP[aP]-labelled D205 RNA (12 fmoles; nucleotides 205–280 of 3¢UTR) and 2 lg of protein from

an Ltk – fibroblast S100 extract in the presence of 120 m M NaCl and

1 m M CaCl 2 Following RNase T 1 digestion, 0.5 lg of antibodies was added and incubated with RNA-bound proteins where indica-ted Gel retardation was performed and complexes were separated for 4 h by native PAGE Lanes 1 and 2 contain labelled D205 RNA in the absence and presence of Ltk – protein, respectively, with the RNP complex indicated with the white arrowhead Anti-(annexin A2) IgG caused a supershift of the complex formed by Ltk– proteins (lane 3, black arrowhead) There was no apparent supershift by anti-biotin (lane 4) No complex is formed by anti-(annexin A2) IgG with D205 RNA in the absence of Ltk- proteins (lane 5).

AUUUA motif within this region of the 3¢UTR was

necessary for full binding of the protein These data

correlate closely with earlier in situ hybridization

data showing that not only is the 86-nucleotide

region spanning nucleotides 194–280 in the c-myc

3¢UTR sufficient to target b-globin to the perinuclear

cytoplasm and the cytoskeleton, but also that the

AUUUA element is required for this targeting ability

and for localization of c-myc mRNA [11] Thus, the

data suggest that annexin A2 is involved in the

association of c-myc mRNA with the cytoskeleton

and its localization

It has previously been shown that annexin A2 is

recovered in a fraction released from the cell matrix by

130 mm KCl [17,18] This fraction also contains

cyto-skeletal components such as actin, messenger

ribo-nucleoproteins (mRNPs) including polysomes, and

specific mRNAs such as c-myc [9,17,19] The

observa-tion that the 36 kDa protein which binds to the

local-ization signal in c-myc 3¢UTR is recovered in such a

cytoskeletal fraction but not in the cytosolic or

membrane fractions (Fig 2B) is consistent both with the binding protein being annexin A2 and with previ-ous observations that c-myc mRNA is recovered in this fraction

Annexins are multifunctional proteins that can inter-act with both membranes and the cytoskeleton [20,21]

It has been proposed that these interactions and the resulting localization of annexins, including A2, can be modulated by post-translational modifications [21] Annexin A2 interacts in a Ca2+-dependent manner with the two cytoskeletal proteins F-actin and non-erythroid spectrin [22] Both the monomeric and tetra-meric forms of annexin A2 are able to associate with F-actin in the presence of Ca2+[20] In addition, it has been suggested that annexin A2 is an integral member

of mRNP complexes [17,21] For example, UV cross-linking and immunoprecipitation of annexin A2 fol-lowed by phenol extraction revealed that annexin A2 was directly associated with small RNA sequences [21] that were most likely degraded mRNAs Further stud-ies have shown that annexin A2 is present only in mRNPs associated with the cytoskeleton, either in the form of actively translating mRNPs in cytoskeleton-bound polysomes or inactive mRNPs [17] Taken together with the ability of annexins to bind to F-actin, the observations that annexin A2 binds RNA and is an integral component of mRNP complexes [17,21] suggest that it may act as a linker between certain mRNAs and the actin filament system However, it is likely that only

a subfraction of annexin A2 has this function [21] While this study was in progress it was observed that annexin A2 binds to c-myc mRNA [18] Our data extend this observation by showing that the binding is

to a specific region within the 3¢UTR implicated in mRNA localization and association with the cytoskele-ton, consistent with the finding that c-myc mRNA is translated on cytoskeleton-bound polysomes [19] Few trans-acting factors involved in association of mRNAs with the cytoskeleton or in mRNA localiza-tion have been identified in mammalian cells ZBP1 and hnRNPA2 have been implicated in b-actin and MBP mRNA localization [14,23], whereas HAX1 and eEF1c bind the region of the 3¢UTR of vimentin mRNA [24] implicated in localization [25] The involve-ment of a different protein, annexin A2, in c-myc mRNA localization may reflect the different location

of c-myc mRNA (perinuclear cytoplasm and cytoskele-ton) and⁄ or different interactions with the cytoskeleton – with actin microfilaments in the case of c-myc and with intermediate filaments for vimentin

In conclusion, the data presented here indicate that annexin A2 binds to the 3¢UTR of c-myc mRNA and that the binding is to the defined section of the 3¢UTR

B A

Fig 6 The binding of different c-myc transcripts to purified annexin

A2 (A) Annexin A22p112heterotetramer (0.75, 1.5 and 3.0 lg) was

immobilized on nitrocellulose membranes and the binding of 2

fmo-lÆmL)1 (5000 Bq) of uniformly [32P]UTP[aP]-labelled c-myc

tran-scripts was performed as described in Experimental procedures.

Transcripts corresponded to nucleotides 1–496 of the 5¢UTR (lane

1), exon 3 (lane 2), and the 3¢UTR (lane 3) (B) 2 lg of annexin

A22p112 heterotetramer was incubated with 2 fmoles of MW

(nucleotides 194–280 of c-myc 3¢UTR; lane 1), MM (nucleotides

194–280 of c-myc 3¢UTR with AGGA mutation; lane 2) or antisense

MW (lane 3) transcripts Incubation of MW with BSA was included

as a negative control (lane 4) The binding was performed in

solu-tion, 1 lgÆlL)1 yeast tRNA being present to prevent nonspecific

RNA binding, and this was followed by UV cross-linking, RNase

treatment and 10% (w ⁄ v) SDS ⁄ PAGE Binding was visualized using

a Canberra Packard Instant Imager Migration position of annexin

A2 is indicated by an arrowhead.

previously implicated in localization [11] Our

hypothe-sis is that the RNA-binding properties of annexin A2

have a novel functional role in perinuclear localization

of mRNA and the association with the cytoskeleton

Further studies are in progress to investigate the role

of other proteins in the perinuclear localization RNP

complex

Experimental procedures

Subcloning of fragments of the c-myc 3¢UTR and

in vitro transcription

Three sequences from the mouse c-myc 3¢UTR, namely

bases 194–440 (D3), 194–280 containing a conserved

AUUUA (MW), and 194–280 with a three-base change

within the AUUUA sequence (MM) were transferred from

vectors PM13 delta3, pSVc-myc1 and pSVc-mycSK⁄ CL

[11] into pBluescriptII SK (Stratagene, Amsterdam, the

Netherlands) so as to maintain the RNA polymerase sites

Vector sequences 3¢ of the T7 promoter (including a tract

of seven C residues) and bases 194–205 of the 3¢UTR

sequence were removed from the MW construct by

diges-tion with XhoI and KpnI to generateD205 which contains

bases 205–280 of the 3¢UTR Radiolabelled D3 transcripts

were synthesized from linearized vectors using RNA

Tran-scription Kit (Stratagene) and [32P]UTP[aP] (800 CiÆ

mmol)1) Templates for the transcription of MW, MM and

D205 were generated by PCR with forward (5¢-TGAGCGC

GCGTAATACG-3¢) and reverse (5¢-GCCCTATTTACAT

GGAAAATTGG-3¢) primers and products purified using

QIAquick columns (Qiagen, Crawley, Sussex, UK)

Tran-scripts were labelled with [32P]UTP[aP] (800 CiÆmmol)1)

using a MAXIscript kit (Ambion, Austin, TX, USA),

extracted with phenol⁄ chloroform and precipitated with

ethanol Incorporation of radionucleotide into RNA was

assessed by scintillation counting and unlabelled RNA was

quantified spectrophotometrically; integrity was verified by

denaturing gel electrophoresis Transcripts corresponding

to nucleotides 1–496 at the 5¢-end of c-myc mRNA were

generated by in vitro transcription from the cDNA

contain-ing exons 1 + 2 of the mouse c-myc gene in pBluescript

SK A 362 bp fragment encoding the 3¢UTR of mouse

c-myc mRNA was synthesized by PCR from pBluescript

containing the complete genomic mouse c-myc gene (a gift

from T McDonnell, University of Texas, Houston, TX,

USA) as template with forward (5¢-TACTGCAGACT

GACCTAACTCGAGGAGG-3¢) and reverse (5¢-GCGGA

ATTCTATGGTACATGTCTTAAAATC-3¢) primers

con-taining a PstI site and an EcoRI site, respectively This

PCR product was inserted between the corresponding sites

of the pGEM 3Zf + vector and used for in vitro

transcrip-tion All constructs were sequenced in both directions to

confirm the orientation and sequences of the inserts

Cell extracts Ltk– fibroblasts were grown to  90% confluence in Dul-becco’s modified Eagle’s medium supplemented with 10% fetal calf serum and in a humidified atmosphere of 5% CO2

at 37
C S100 protein extracts were prepared following the method of Behar et al [7] with modifications Cells were resuspended in lysis buffer (130 mm NaCl, 5 mm MgCl2,

30 mm Tris⁄ HCl pH 7.6, 2 mm dithiothreitol [DTT]) con-taining 0.5% (v⁄ v) Nonidet P-40 and EDTA-free protease inhibitor cocktail (Roche, Lewes, East Sussex, UK), and lysed by passing them through a 21-gauge needle seven times Large debris were removed by centrifugation at

5000 g for 10 min and the supernatant fluid was diluted with 3 vol of 40 mm NaCl lysis buffer and centrifuged at

100 000 g for 1 h, 10% (v⁄ v) glycerol was added to the supernatant fluid before freezing in aliquots in liquid nitro-gen Cytosolic, cytoskeletal and membrane fractions were prepared using a sequential detergent⁄ salt extraction proce-dure as described previously [9,26] Cell pellets were resus-pended in 1 mL of buffer F (10 mm Tris, pH 7.6, 0.25 m sucrose, 25 mm KCl, 5 mm MgCl2, 0.5 mm CaCl2) contain-ing 0.05% (v⁄ v) Nonidet P-40, and after 10 min at 4
C, the suspension was centrifuged at 1000 g for 5 min The supernatant fluid (cytosolic fraction) was removed and after one wash in buffer F the pellet was resuspended in 1 mL of buffer F containing 130 mm KCl and 0.05% (v⁄ v) Noni-det P-40, incubated for 10 min and centrifuged at 2000 g for 10 min The supernatant fluid (cytoskeletal fraction) was removed, membrane components of the pellet solubi-lized by incubation in buffer F containing 130 mm KCl, 0.5% (v⁄ v) Nonidet P-40 and 0.5% (w ⁄ v) deoxycholate for

10 min and the membrane fraction collected by centrifuga-tion at 3000 g for 10 min

Gel retardation and UV cross-linking assays Gel retardation reactions were carried out with 1–5 lg S100 extract and 500 Bq of 32P-labelled RNA in binding buffer (10 mm Hepes, pH 7.6, 3 mm MgCl, 40 mm NaCl, 5% (w⁄ v) glycerol, 1 mm DTT and 10 lg tRNA) in a total volume of 10 lL at 22
C for 30 min For competition experiments, labelled and unlabelled RNA were added simultaneously Products were separated on 5% (w⁄ v) non-denaturing polyacrylamide gels (60 : 1, 1· TBE at

20 VÆcm)1 for 3 h) For gel retardation analysis combined with T1 treatment, digestion was carried out by adding

40 units of RNase T1 to the binding reaction and incuba-tion continued for 5 min before the addiincuba-tion of 2 lL of 20% (w⁄ v) Ficoll For supershift assays, 0.5 lg mouse anti-(annexin A2) IgG or a control IgG (antibiotin) was added

to binding reactions after RNase T1digestion and incuba-ted for 30 min at 4
C prior to the addition of Ficoll Complexes were separated by electrophoresis for 2 h at

20 VÆcm)1through 5% (w⁄ v) nondenaturing polyacrylamide

(79 : 1, 0.5· TBE) gels Gels were dried and analysed by

autoradiography UV cross-linking reactions were carried

out with 10–15 lg protein and 5000 Bq of 32P-labelled

RNA in 10 lL of binding buffer containing 100 mm NaCl

and 125 ng tRNA Heparin (5 mgÆmL)1) was then added

and the incubation continued for 5 min The reaction tubes

were then placed on ice and irradiated in a Spectrolinker

with 2· 960 mJ to cross-link protein–RNA complexes

Unprotected RNA was then removed by incubation with

4 lg RNase A and 12 units RNase T1for 15 min at 37
C,

and the samples subjected to SDS⁄ PAGE [27]

Protein isolation using paramagnetic beads

A 0.6 mL aliquot suspension of prewashed MagneSphere

streptavidin-coated paramagnetic particles (SA-PMP,

Promega, Southampton, UK) was resuspended in 0.1 mL

0.5· NaCl ⁄ Cit containing 100 lg bovine serum albumin

(BSA) and 100 lg yeast tRNA and incubated at  20
C

for 1 h with shaking The suspension was washed twice

with 0.3 mL 0.5· NaCl ⁄ Cit and then incubated with 20 lg

biotinylated D205 transcripts (labelled with biotin-16-UTP

[Roche] and produced by in vitro transcription) in 0.3 mL

0.5· NaCl ⁄ Cit for 10 min at room temperature Unbound

RNA was removed by washing twice with 0.3 mL 0.5·

NaCl⁄ Cit SA-PMP with RNA bound were then incubated

at 4
C for 1 h with 1 mg cell extract in 40 mm NaCl lysis

buffer containing 0.5 mgÆmL)1 yeast tRNA, 0.2 mgÆmL)1

BSA, and 800 unitsÆmL)1RNasin (Promega), with or

with-out 1 mm CaCl2, in a total volume of 0.5 mL After

5· 1 mL washes with 40 mm NaCl lysis buffer, proteins

bound to the RNA were eluted by boiling in 45 mm

Tris⁄ HCl, pH 6.8, 10% (w ⁄ v) glycerol, 1% (w ⁄ v) SDS, 1%

(v⁄ v) 2-mercaptoethanol and 0.01% (w ⁄ v) bromophenol

blue

Western blotting and mass spectrometry

Proteins were separated by SDS⁄ PAGE using a 12.5%

(w⁄ v) acrylamide separating gel These were either stained

with Coomassie Brilliant Blue or the proteins were

trans-ferred to a poly(vinylidene fluoride) (PVDF) membrane by

semidry electroblotting Bands of interest from

Coomassie-stained gels were excised for in-gel trypsin digestion

followed by MALDI-TOF⁄ MS (carried out by J Gray,

Institute for Cell and Molecular Biosciences, University of

Newcastle, UK) Proteins transferred to PVDF membranes

were incubated with a monoclonal antibody to

annex-in A2 (BD Transduction Laboratories, 1 : 5000 dilution)

following the manufacturer’s instructions After

incuba-ting with anti-(mouse horseradish peroxidase-conjugated)

serum (Sigma-Aldrich UK, Poole, Dorset, UK), the blot

was developed using the POD chemiluminescence kit

(Roche) Chemiluminescence was detected by exposure to

Kodak X-Omat A2-5 film

RNA–annexin A2 binding assays The heterotetrameric annexin A22p112 complex was puri-fied from pig intestinal epithelium [28] mRNA filter bind-ing assays, usbind-ing in vitro transcribed RNA and immobilized native annexin A2 tetramer, were carried out as described previously [17], except that 1· Denhardt’s solution [0.02% (w⁄ v) Ficoll, 0.02% (w ⁄ v) polyvinylpyrrolidone, 0.02% (w⁄ v) BSA] was added to the RNA binding solution (10 mm triethanolamine, pH 7.4, 50 mm KCl, 1 mm DTT,

2 mm MgSO4, 1 mm CaCl2and 1 lgÆlL)1of yeast tRNA), supplemented with 20 UÆmL)1 RNasin (Promega), to reduce the background Membranes were incubated with

2 fmoleÆmL)1of transcript for 20 min at room temperature, washed rapidly three times and then a further four times for 15 min in binding solution lacking yeast tRNA and Denhardt’s solution The RNA binding was quantified and visualized using a Canberra Packard Instant Imager (Perkin Elmer, Pangbourne, UK) mRNA–annexin A2 binding assays in solution were performed as described by Kwon and Hecht [29] Purified in vitro transcribed mouse c-myc RNA probes were heated at 72
C for 3 min and cooled slowly to room temperature mRNA (2 fmol; 5000 Bq) was incubated with the annexin A22p112 complex in RNA binding solution, supplemented with 20 UÆmL)1 RNasin (Promega) containing 2.5% (w⁄ v) Ficoll for 20 min in a final volume of 20 lL After incubation, the RNA probes were covalently cross-linked to the proteins by exposure to

UV light and RNase T1and A were then added for 30 min

at 37
C to digest unprotected RNA (see above) Nucleo-tide–protein complexes were separated from degraded mRNA probe by SDS⁄ PAGE [27], the gels dried and mRNA–protein binding visualized using a Canberra Packard Instant Imager

Acknowledgements

The work was supported by BBSRC (grant 13⁄ C13737

to JEH), the Scottish Office Agriculture, Environment and Fisheries Department and the Norwegian Cancer Society (AV) ZMACL thanks The Royal Society for support We thank Sandra Fulton for advice on UV cross-linking assays, and Luc Veyrune and Jean-Marie Blanchard for providing vectors

References

1 Bassell G & Singer RH (1997) mRNA and cytoskeletal filaments Curr Opin Cell Biol 9, 109–115

2 Hesketh JE (1994) Translation and the cytoskeleton: a mechanism for targeted protein synthesis Mol Biol Rep

19, 233–243

3 Janssen R-P (2001) mRNA localization: message on the move Nat Rev Mol Cell Biol 2, 247–256

4 Tekotte H & Davis I (2002) Intracellular mRNA

localiza-tion: motors move messages Trends Genet 18, 636–642

5 Ainger K, Avossa D, Diana AS, Barry C, Barbarese E

& Carson JH (1997) Transport and localization of

exo-genous myelin basic protein mRNA microinjected into

oligodendrocytes J Cell Biol 138, 1077–1087

6 Barbarese E, Koppel DE, Deutscher MP, Smith CL,

Ainger K, Morgan F & Carson JH (1995) Protein

trans-lation components are colocalised in granules in

olgo-dendrocytes J Cell Sci 108, 2781–2790

7 Behar L, Marx R, Sadot E, Barg J & Ginzburg I (1995)

cis-Acting signals and trans-acting proteins are involved

in tau mRNA targeting in neurites of differentiating

neuronal cells Int J Devl Neurosci 13, 113–127

8 Kislauskis EH, Li Z, Taneja KL & Singer RH (1993)

Isoform-specific 3¢-untranslated sequences sort

alpha-cardiac and beta-cytoplasmic actin messenger RNAs to

different cytoplasmic compartments J Cell Biol 123,

165–172

9 Hesketh J, Campbell G, Piechaczyk M & Blanchard JM

(1994) Targeting of c-myc and b-globin coding

sequences to cytoskeletal-bound polysomes by c-myc 3¢

untranslated region Biochem J 298, 143–148

10 Wilson IA, Brindle KM & Fulton AM (1995)

Differen-tial localisation of the mRNA of the M and B forms of

creatine kinase in myoblasts Biochem J 308, 599–605

11 Veyrune J-L, Campbell GP, Wiseman J, Blanchard J-M

& Hesketh JE (1996) A localisation signal in the 3¢

untranslated region of c-myc mRNA targets c-myc

mRNA and b-globin reporter sequences to the

peri-nuclear cytoplasm and cytoskeletal-bound polysomes

J Cell Sci 109, 1185–1194

12 Oleynikov Y & Singer RH (1998) RNA localization:

different zipcodes, same postman? Trends Cell Biol 8,

381–383

13 Dalgleish G, Veyrune JL, Blanchard JM & Hesketh J

(2001) mRNA localisation by a 145-nucleotide region of

the c-fos 3¢ untranslated region J Biol Chem 276,

13593–13599

14 Munro TP, Magee RJ, Kidd GJ, Carson JH, Barbarese

E, Smith LM & Smith R (1999) Mutational analysis of

a heterogeneous nuclear ribonucleoprotein A2 response

element for RNA trafficking J Biol Chem 274, 34389–

34395

15 Smart FM, Edelman GM & Vanderklish PW (2003)

BDNF induces translocation of initiation factor 4E to

mRNA granules: evidence for a role of synaptic

micro-filaments and integrins Proc Natl Acad Sci USA 100,

14403–14408

16 Knowles RB, Sabry JH, Martone ME, Deerinck TJ,

Ellisman MH, Bassell GJ & Kosik KS (1996)

Translocation of RNA granules in living neurons

J Neurosci 16, 7812–7820

17 Vedeler A & Holla˚s H (2000) Annexin II is associated with mRNAs which may constitute a distinct subpopu-lation Biochem J 348, 565–572

18 Filipenko NR, MacLeod TJ, Yoon CS & Waisman DM (2003) Annexin A2 is a novel RNA-binding protein

J Biol Chem 279, 8723–8731

19 Hesketh JE, Campbell GP & Whitelaw PF (1991) c-myc mRNA in cytoskeletal-bound polysomes in fibroblasts Biochem J 274, 607–609

20 Gerke V & Moss SE (1997) Annexins and membrane dynamics Biochim Biophys Acta 1357, 129–154

21 Arrigo AP, Darlix JL & Spahr PF (1983) A cellular protein phosphorylated by the avian sarcoma virus transforming gene product is associated with ribonu-cleoprotein particles EMBO J 2, 309–315

22 Regnouf F, Rendon A & Pradel LA (1991) Biochemical characterization of annexins I and II isolated from pig nervous tissue J Neurochem 56, 1985–1996

23 Ross AF, Oleynikov Y, Kislauskis EH, Taneja KL & Singer RH (1997) Characterization of a beta-actin mRNA zipcode-binding protein Mol Cell Biol 17, 2158–2165

24 Al-Maghrebi M, Brule H, Padkina M, Allen C, Holmes

WM & Zehner Z (2002) The 3¢ untranslated region of human vimentin mRNA interacts with protein com-plexes containing eEF1 and HAX1 Nucleic Acids Res

320, 5017–5028

25 Bermano G, Shepherd RK, Zehner ZE & Hesketh JE (2001) Perinuclear mRNA localisation by vimentin 3¢ untranslated region requires a 100 nucleotide sequence and intermediate filaments FEBS Lett 497, 77–81

26 Vedeler A, Pryme IF & Hesketh JE (1991) The charac-terization of free, cytoskeletal-bound and membrane-bound polysomes in Krebs II ascites and 3T3 cells Mol Cell Biochem 100, 183–193

27 Laemmli U (1990) Cleavage of structural proteins dur-ing assembly of the head of the bacteriophage T4 Nature 227, 680–685

28 Gerke V & Weber K (1984) Identity of p36K phos-phorylated upon Rous sarcoma virus transformation with a protein purified from brush borders: calcium-dependent binding to non-erythroid spectrin and F-actin EMBO J 3, 227–233

29 Kwon YK & Hecht NB (1991) Cytoplasmic protein binding to highly conserved sequences in the 3¢ unlated region of mouse protamine 2 mRNA, a trans-lationally regulated transcript of male germ cells Proc Natl Acad Sci USA 88, 3584–3588