the n-terminal domain of g3bp enhances cell motility and ... · the regulationof mrna stability...

Angiogenesis, Metastasis, and the Cellular Microenvironment

The N-Terminal Domain of G3BP Enhances Cell Motilityand Invasion by Posttranscriptional Regulation of BART

Keisuke Taniuchi1,2, Isao Nishimori2, and Michael A. Hollingsworth1

AbstractThe regulation of mRNA stability plays an important role in the control of gene expression during cell motility

and invasion. We previously reported that GTPase-activating protein [Src homology 3 (SH3) domain] bindingprotein (G3BP), a marker of cytoplasmic stress granules that are formed in stressed cells and regulate mRNAstability, binds and degrades the mRNA of binder of Arl two (BART) that inhibits retroperitoneal invasion andhepatic metastasis of pancreatic cancer cells. Here, we report that overexpression of the amino (N)-terminal regionof G3BP, including the binding region for BARTmRNA, dominant-negatively inhibits formation of the complexbetween endogenous G3BP and BART mRNA, and increases the expression of BART. This, in turn, inhibits theinvasiveness of pancreatic cancer cells. On the other hand, the carboxy (C)-terminal region of G3BP is associatedwith phosphorylation of eukaryotic translation initiation factor 2a (eIF2a) that initiates stress granule assemblybut does not modulate the posttranscriptional regulation of BART mRNA. N-terminal G3BP also plays a role inregulating secreted matrix metalloproteinases, transcription factors, and a variety of genes involved in cell adhesionand motility. These results suggest that N-terminal G3BP contributes to posttranscriptional regulation of cellmotility and invasive capacity of pancreatic cancer. Mol Cancer Res; 9(7); 856–66. �2011 AACR.

Introduction

G3BP was originally identified on the basis of its ability tobind the Src homology 3 (SH3) domain of Ras GTPase-activating protein p120 (RasGAP), which functions as aphysiologic negative regulator of Ras signaling in exponen-tially growing cells (1–3). The phosphorylation of G3BPand its association with RasGAP are affected by extracellularstimuli (3); however, functional consequences of the bind-ing of G3BP to RasGAP have not been fully elucidated. TheN-terminus of G3BP is a nuclear transport factor-2 (NTF-2)-like domain homologous to NTF-2 and is followed byacidic and proline-rich regions, an RNA-binding domainwith an RNA recognition motif (RRM), and multiplearginine-glycine rich (RGG) motifs (1). The NTF-2–likedomain of G3BP influences the cellular localization ofproteins and its oligomerization with itself or with otherpartners (4). Indeed, in vitro binding assays suggest that the

NTF-2–like domain of G3BP is responsible for RasGAPbinding (5). Assembly of stress granules (SG) can bedominantly induced by G3BP overexpression (4). G3BPbinds to Caprin-1 through the NTF-2–like domain, andthis complex is localized in cytoplasmic granules that con-tain a major part of the cytoplasmic RNA (6). This complexinduces the formation of cytoplasmic SGs that form inresponse to a variety of cellular stresses (6). SGs containmRNAs, certain translation-initiating factors, and RNA-binding proteins that can be directed to either translationinitiation or decay pathways (4). G3BP is likely to regulatethe transport and translation of mRNAs of proteinsinvolved in cellular proliferation and migration in multiplecell types (6). This suggests that the NTF-2–like domain ofG3BP influences gene expression by affecting the steady-state levels of mRNA following posttranscriptional repres-sion or induction through binding to partner proteins andSG formation.Another interesting feature of G3BP is that the RRM

domain mediates the binding of G3BP to specific RNAsequences so that G3BP can exert its function as a dinu-cleotide-specific single-strand–specific endoribonuclease(7). G3BP can bind to the 30 untranslated region (30UTR)of human c-Myc mRNA through the RRM domain in aphosphorylation-dependent manner to increase its degrada-tion in vitro (3). It is the only known endoribonuclease thatrequires site-specific phosphorylation for its catalytic activity(3). A small number of general pathways seem to beresponsible for degrading most mRNAs; therefore, regula-tory mechanisms are thought to target the initial events thatdirect the mRNA into one of these pathways, to either

Authors' Affiliations: 1Eppley Institute for Research in Cancer and AlliedDiseases, and Department of Pathology and Microbiology, University ofNebraska Medical Center, Omaha, Nebraska; and 2Department of Gastro-enterology and Hepatology, Kochi University Medical School, Nankoku,Kochi, Japan

Note: Supplementary data for this article are available at Molecular CancerResearch Online (http://mcr.aacrjournals.org/).

Corresponding Author: Keisuke Taniuchi, Department of Gastroenterol-ogy and Hepatology, Kochi University Medical School, Nankoku, Kochi783-8505, Japan. Phone: 81-888-802-338; Fax: 81-888-802-338; E-mail:[email protected]

doi: 10.1158/1541-7786.MCR-10-0574

�2011 American Association for Cancer Research.

MolecularCancer

Research

Mol Cancer Res; 9(7) July 2011856

on November 19, 2020. © 2011 American Association for Cancer Research. mcr.aacrjournals.org Downloaded from

Published OnlineFirst June 10, 2011; DOI: 10.1158/1541-7786.MCR-10-0574

http://mcr.aacrjournals.org/

maintain it in a translationally inactive state, or permit itstranslation (8). The major decay-initiating events are dead-enylation, endonucleolytic cleavage, and decapping (9). c-Myc transcript levels, which are considered to be regulatedby G3BP, are subject to posttranscriptional regulation by asequential pathway involving deadenylation followed bydegradation of the mRNA body (10). In contrast, G3BPcan stabilize tau mRNA, a microtubule-associated proteinhighly regulated during neuronal cell differentiation (11). Inaddition to endoribonuclease activity, G3BP functions tostabilize mRNA. The studies that examine the intersectionof G3BP with RNA metabolism have proposed that theRRM domain of G3BP plays an important role in regulat-ing mRNA translation or decay, or both.We previously reported that binder of Arl two (BART)

inhibits retroperitoneal invasion and liver metastasis ofpancreatic ductal adenocarcinoma (PDAC) cells in anorthotopic xenograft model (12). Our findings providean explanation for the oncogenic function of G3BP toinduce tumor invasion and metastasis by mediating BARTmRNA decay. Here, we provide evidence that the N-terminal G3BP (G3BP1–180) functions in the posttranscrip-tional regulation of BART mRNA for the promotion of cellmotility and invasion in PDAC. The C-terminal G3BP(G3BP241–466) functions in SG assembly, but the expres-sion of BART and c-Myc mRNA is not regulated. More-over, G3BP1–180 plays a critical role in establishing a patternof gene expression that favors PDAC cell motility andinvasive behavior.

Materials and Methods

Identification of the BART-binding region in G3BPWe produced 6 deletion constructs of G3BP (G3BP1–420,

G3BP1–330, G3BP1–240, G3BP1–180, G3BP1–135, andG3BP1–80 of accession number NM_005745) by cloningthe pcDNA3.1(þ)/Myc-HisA vector (Invitrogen) intoappropriate sites. COS7 cells were transiently transfectedwith each plasmid by using FuGENE6 (Roche). After 48hours of transfection, cell extracts were immunoprecipitatedwith agarose-conjugated anti-Myc antibody. Each Myc-tagged immunoprecipitate was incubated with biotiny-lated BART 30UTR (748–1703 of accession numberNM_012106) or glyceraldehyde 3-phosphate dehydrogen-ase (GAPDH) RNA immobilized on streptavidin magneticbeads for 2 hours at 4�C. Bound proteins in the pull-downmaterial were washed using strong permanent magnets(MPC-S; Dynal) and analyzed by Western blotting usinganti-Myc antibody.

Real-time quantitative reverse transcriptase-PCRTotal RNA was extracted using the RNeasy kit with

DNase I treatment, and reverse transcribed as describedearlier. The resultant cDNA was amplified by real-timePCRusing assays-on-demand primers and TaqMan probesfor BART (Hs00183708_m1), c-Myc (Hs00153408_m1),and GAPDH (Hs99999905_m1; Applied Biosystems).Each sample was run in triplicate for both BART and

GAPDH in a 20-mL reaction solution, using TaqManUniversal PCRMaster Mix according to the manufacturer'sinstructions (Applied Biosystems). The gene-specific reversetranscriptase-PCR (RT-PCR) products were measured con-tinuously on an ABI PRISM 7000 Sequence Detectionsystem (Applied Biosystems). Transcript levels of BARTand c-Myc were normalized to GAPDH in each sample byusing the standard curve method as described by themanufacturer.

Matrigel invasion assayThe 2-chamber invasion assay (24-well plates, 8-mm pore

size, membrane coated with a layer of Matrigel extracellularmatrix (ECM) proteins; Becton Dickinson) was used toassess cell invasion. A total of 5.0� 104 cells were seeded inserum-free medium into the top chamber and allowed toinvade toward 5% fetal calf serum (FCS) as a chemoat-tractant in the bottom chamber. After 20 hours of incuba-tion, the number of invading cells at the bottom of themembrane was estimated by counting 3 independent visualfields in a microscope.

Results

Overexpression of G3BP in PDACG3BP expression in cell lysates from PDAC cell lines and

from a line of immortalized normal pancreatic nestin-expressing epithelial (HPNE) cells (13) was determinedby Western blot analysis (Fig. 1A). All PDAC cell linesshowed high levels of G3BP expression compared withHPNE cells. G3BP expression in PDAC and in normaltissue sections was also investigated using an anti-G3BPmonoclonal antibody (Fig. 1B–E). In concordance with theresults of the Western blotting, immunohistochemical ana-lysis revealed strong signals of G3BP in the cytoplasm ofcancer cells, but its expression was hardly detectable innormal pancreatic tissues. Among the 5 PDACs examinedby immunohistochemistry, all specimens showed strongsignals of G3BP. These observations suggest that G3BPis overexpressed in PDAC compared with normal pancreas.

The acidic domain of G3BP is essential for binding toBART mRNAG3BP binds to the 30UTR of BART (748–1703 of

accession number NM_012106) and degrades BARTmRNA (12). To define the region of G3BP that is requiredfor its binding to BART-30UTR, the binding ability of 6Myc-tagged fragments of G3BP (Fig. 2A; 1–420, 1–330, 1–240, 1–180, 1–135, and 1–80) to BART-30UTR wereinvestigated. Each fragment was transiently expressed inCOS7 cells and high levels of expression were confirmedby Western blotting (Fig. 2B). Myc-tagged proteins in totalcell lysates were immunoprecipitated with anti-Myc anti-body and incubated with in vitro–transcribed biotinylatedBART-30UTR or control GAPDH RNA immobilized onstreptavidin magnetic beads. The immunoprecipitated com-plexes were then pulled down with magnets. G3BP1–420,G3BP1–330, G3BP1–240, and G3BP1–180 could interact with

G3BP Accelerates Cell Motility and Invasion

www.aacrjournals.org Mol Cancer Res; 9(7) July 2011 857




BART-30UTR, but G3BP1–135 and G3BP1–80 could notinteract (Fig. 2C). GAPDH mRNA failed to pull down theG3BP fragments (data not shown). These experimentssuggest that the region spanning amino acid residues 136to 180, located in the acidic domain, is sufficient for bindingBART-30UTR.

Inhibition of G3BP endoribonuclease activity forBART-30UTR transcript by synthesized polypeptides ofG3BP136–180To investigate the functional significance of the

G3BP136–180 fragment that includes the region that bindsto the BART-30UTR transcript, we used synthetic 45-amino acid polypeptides (G3BP136–180 peptide). BART-

30UTR was incubated with recombinant G3BP with orwithout the G3BP136–180 peptide, followed by pull-down ofthe BART transcript. When the G3BP136–180 peptide wasadded, it was able to inhibit formation of the complexbetween recombinant G3BP and BART-30UTR (Fig. 2D),thus indicating that this synthesized peptide functions in adominant-negative manner. To determine dominant-nega-tive effects of the G3BP36–180 peptide on the RNase activityof G3BP, we conducted an in vitro mRNA decay assay(Fig. 2E). In vitro–transcribed BART-30UTRwas efficientlycleaved after 30 minutes of incubation with recombinantG3BP in the absence of G3BP136–180 peptide, but mRNAsfor GAPDH were not cleaved. G3BP136–180 peptide treat-ment strongly decreased cleavage of the BART-30UTR byG3BP but had no effect on control GAPDH. These resultssuggest that the interaction of G3BP and BART-30UTR ismediated through the region of G3BP136–180 and thatG3BP136–180 is necessary for specific G3BP RNase activityon BART mRNA.

Exogenous overexpression of N-terminal G3BPdominant-negatively increases BART expression inPDAC cellsTo examine the global function of G3BP136–180 on

posttranscriptional regulation, we established PANC-1 deri-vative clones from a PDAC cell line. These clones consti-tutively expressed either exogenous N-terminal G3BP(G3BP-N, 1–180) with the NTF-2–like domain andG3BP136–180, or the C-terminal G3BP (G3BP-C, 241–466), including the RNA-binding domains (Fig. 3A). Wes-tern blot analysis shows the expression levels of the exo-genous G3BP fragments compared with mock-controlclones (Fig. 3B).BART and c-Myc mRNAs are specifically associated with

G3BP in PDAC cells (12). This was confirmed by immu-noprecipitation (IP) experiments in PANC-1 cells by usinganti-G3BP or control mouse IgG monoclonal antibodies(Fig. 3C). The immunoprecipitated transcripts were thenconverted to cDNA by reverse transcription, and RT-PCRwas conducted. We found that antibodies to G3BP speci-fically immunoprecipitated both BART and c-MycmRNAs. Nontarget GAPDH mRNA was also amplified,albeit inefficiently, and to the same extent in both IPgroups. These findings reveal the presence of low levelsof contaminating, nonspecific mRNA in all IP samples, yetverifies that equal amounts of input material was used, andshows that the G3BP-IP materials were enriched comparedwith the control IgG-IPs. Following this, quantitative RT-PCR analysis was conducted to identify the levels of BARTmRNA in the G3BP deletion mutant clones and to therebydetermine the functions of N-terminal G3BP on the post-transcriptional regulation of BART. We found that BARTmRNA was upregulated in N-terminal G3BP PANC-1clones but not in mock or C-terminal clones by real-timequantitative RT-PCR (Fig. 3D). The same result wasobserved by Western blotting (Fig. 3E). Interestingly,G3BP-N cells also increased c-Myc mRNA as comparedwith mock-control clones; however, G3BP-C failed to alter

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

Re

lative

sig

na

l(n

orm

aliz

ed

to

β-a

ctin)

IB: β-actin

IB: G3BP

1 2 3 4 65 NA

B

7 8

50

kDa

35

C

D E

Figure 1.Overexpression of G3BP in PDAC. A, G3BP expression in PDACcell lines in comparison with a normal pancreatic epithelial cell line (HPNE),examined by Western blot analysis (left). PDAC cells (1, PANC-1; 2,BxPC3; 3, S2-013; 4, SUIT-2; 5, COLO357; 6, HPAF; 7, MIA-Paca2; and 8,Capan2) and HPNE cells (N) were used. b-Actin was used as a quantitativecontrol. Scanning and densitometric analyses were conducted in arepresentative experiment (right). B–E, immunohistochemistry wasconducted on clinical samples of PDAC by using anti-G3BP antibody.PDAC specimens (B–D) and 1 normal pancreatic tissue specimen (E) areshown.

Taniuchi et al.

Mol Cancer Res; 9(7) July 2011 Molecular Cancer Research858




c-Myc mRNA levels (Fig. 3F). These results indicate thatG3BP-N functions in a dominant-negative manner bystabilizing BART mRNA and that G3BP-N also affectsthe mRNA levels of c-Myc that binds to the C-terminalRNA-binding domain of G3BP.

Exogenously overexpressed N- and C-terminal G3BPscolocalize with cytoplasmic RNA in PDAC cellsTo show the mechanism of overexpressed G3BP-N in

regulating the target transcripts via inhibition of endogenousG3BP activity, we investigated the subcellular colocalization

A

G3BP1–420

G3BP1–330

G3BP1–240

G3BP1–180

G3BP1–135

G3BP1–80

136–180

1

341 411 428 450240140120

NTF2 Acidic RRM RGG

466

Pull–down: BART–3′UTRB C

75

50

kDa

15

25

35

G3

BP

1–

420

G3

BP

1–

330

G3

BP

1–

240

G3

BP

1–

180

G3

BP

1–

135

G3

BP

1–

80

IB: Myc IB: Myc

75

50

kDa

15

25

35

D

kDa

50

50IB: G3BPInput

IB: G3BP

G3BP136–180 peptide : –

+Recombinant G3BP: +

Pull-down: BART–3′UTR

10 μmol/L 5 μmol/L

+

0.5

0.2

0.3

BART–3′UTR GAPDH

G3BP136–180 peptide

(10 μmol/L) Inp

ut R

NA

– +

Inp

ut

RN

A

– +

E

Recombinant G3BP

+ + + +

kb

1.0

1.5

G3

BP

1–42

0

G3

BP

1–33

0

G3

BP

1–24

0

G3

BP

1–18

0

G3

BP

1–13

5

G3

BP

1–80

Figure 2. Identification of the BART mRNA–binding region in G3BP. A, schematic representation of 6 COOH-terminal Myc-tagged G3BP deletionmutants. B, Western blotting with anti-Myc antibody on COS7 whole-cell lysates transfected with Myc-tagged G3BP deletion mutants. C, identificationof the region in G3BP that binds to BART mRNA. The biotinylated BART-30UTR transcript was incubated with each immunoprecipitated Myc-tagged G3BPdeletion protein from COS7 cells. Association of the biotinylated BART-30UTR with Myc-tagged G3BP was assessed by pull-down of the RNA usingstreptavidin-conjugated beads, followed by Western blotting with anti-Myc antibody. D, reduction of the complex formed between recombinant G3BP andbiotinylated BART-30UTR transcript by treatment with G3BP136–180 peptides. E, BART 30UTR or control GAPDH RNA was incubated with recombinantG3BP with or without pretreatment with G3BP136–180 peptides. The RNA was electrophoresed and visualized by SYBR Gold staining. Closed arrowhead,BART-30UTR; arrow, control GAPDH.






of Myc-tagged N- and C-terminal G3BPs, and total cellularRNA in G3BP-N and G3BP-C PANC-1 cells. Ethidiumbromide staining of total cellular RNA showed that the bulkof cytoplasmic RNA colocalized with the transfected Myc-tagged N- and C-terminal G3BP (Fig. 4A). In addition,RNase treatment eliminated all ethidium bromide staining

in the cytoplasm (Supplementary Fig. S1). These resultssuggest that not only G3BP-C, including the RRM andRGG domains, but also G3BP-N binds to RNA. Impor-tantly, G3BP-N was diffusely distributed in the cytoplasm,whereas the C-terminal form was recruited to the cytoplas-mic ribonucleoprotein (RNP) granules (Fig. 4A). In line

BA

RT

mR

NA

no

rmal

ized

to

GA

PD

H

0

2

4

6

8

10

Mock# 5 G3BP-N# 9 G3BP-C# 7

P = 0.0004 P = 0.0002

1

341 411 428 450240140120

NTF2 Acidic RRM RGG

466

A

N-terminus-G3BP (1–180) C-terminus-G3BP (241–466)

G3BP136–180

B

30

35

kDa

IB: Myc

IB: β-actin

Moc

k# 5

G3B

P-N

# 9

Moc

k# 9

G3B

P-C

# 7

G3B

P-N

# 11

G3B

P-C

# 20

D

C

Con

trol

IP: G3B

P

c-Myc

BART

GAPDH

E

35

15

kDaIB: BART

IB: β-actin

Moc

k# 5

G3B

P-N

# 9

Moc

k# 9

G3B

P-C

# 7

G3B

P-N

# 11

G3B

P-C

# 20

c-M

ycm

RN

A

norm

aliz

ed t

o G

AP

DH

0

1

2

3

4

5


P = 0.039 P = 0.069

F

Figure 3. Stable overexpression of N- or C-terminal G3BP in PANC-1 cells. A, schematic representation of the N-terminal G3BP deletion mutantcontaining the NTF-2–like and acidic domains, and the C-terminal deletion mutant containing the RRM and RGGmotifs. B, exogenously overexpressed G3BPfragments were detected in 2 sets of transfected clones (G3BP-N and G3BP-C) by Western blot analysis with anti-Myc antibody. Mock-transfectedPANC-1 cells (mock) served as controls. C, the association of endogenousG3BPwith endogenous BART and c-MycmRNAs in PANC-1 cells was tested by IP.The transcripts in the IP materials were detected by RT-PCR. Mouse IgG monoclonal antibody was used as a control. D, quantitative RT-PCR analysisof BARTmRNA in stable control, G3BP-N, andG3BP-C cells of PANC-1. Quantification of mRNA levels was done in triplicates and repeated 3 times. Columns,mean; bars, SD. P value comparedwith mock-control and G3BP-C cells. E,Western blotting with anti-BART antibody showing 2 representative clones of eachof G3BP-N and G3BP-C cells of PANC-1. F, quantitative RT-PCR analysis of c-Myc mRNA in stable control, G3BP-N, and G3BP-C cells of PANC-1.Quantification of mRNA levels was done in triplicates and repeated 3 times. Columns, mean; bars, SD.P value comparedwith mock-control and G3BP-C cells.

Taniuchi et al.





with these results, overexpressed N-terminal G3BP1–141 isdiffusely distributed in the cytoplasm of HeLa cells (6), andC-terminal G3BP334–466 containing the RNA-bindingdomains is recruited to SGs in COS cells (4). G3BP is a

marker for SGs, which form in stressed cells (14), and whenoverexpressed, G3BP induces SG formation (4). The C-terminal RNA-binding domains become concentrated inSGs under oxidative stress with sodium arsenite (SA), and

A

B

G3B

P-N

# 9

G3B

P-C

# 7

DAPI

SA

–

Myc TIA-1 Merge

SA

+

G3B

P-N

# 9

G3B

P-C

# 7

a

b

Myc EB MergeG

3B

P-N

# 9

G3B

P-C

# 7

DAPI

C Z stack

a b

Figure 4. Overexpressed C-terminal G3BP is localized in cytoplasmic SGs in PDAC cells. A, immunocytochemical staining of exogenous Myc-taggedG3BP fragments and total cellular RNA in PANC-1 transfectants, as determined using anti-Myc antibody (green) and ethidium bromide (red), respectively, and40,6-diamidino-2-phenylindole (DAPI) staining (blue, nucleus). Bars, 10 mm. B, stable G3BP-N and G3BP-C cells of PANC-1 were exposed to500 mmol/L SA for 30 minutes. Immunocytochemical staining, as determined with anti-Myc antibody (green), anti-TIA-1 antibody (violet), and DAPI staining(blue). White arrows and arrowheads indicate Myc-tagged G3BP-C colocalizing with endogenous TIA-1 in SGs after treatment with or without SA,respectively. Yellow arrows indicate SGs in which TIA-1 is localized. Bars, 10 mm. C, confocal Z stack with white boxes (a and b in B) depicting the positionwhere SGs associate with Myc-tagged G3BP-C and endogenous TIA-1, as determined with anti-Myc antibody (green), anti-TIA-1 antibody (violet), and DAPIstaining (blue). Arrows and arrowheads indicate Myc-tagged G3BP-C colocalizing with endogenous TIA-1 in SGs after treatment with or without SA,respectively.






their ability to bind RNA is reportedly sufficient to con-centrate them in SGs (4). To verify that the C-terminal formis present in SGs, cells were immunostained under condi-tions of oxidative stress caused by treatment with SA(Fig. 4B). In G3BP-C cells, the C-terminal form colocalizedin SGs with the endogenous SG-specific protein TIA-1, withor without SA treatment (white arrows and arrowheads inFig. 4B). This colocalizationwas confirmed in the confocal Zstack (Fig. 4C). Thus, RNP granules induced by the over-expression of G3BP-C might consist of SGs. TIA-1 waslocalized at SA-induced cytoplasmic granules in G3BP-Ncells (yellow arrows in Fig. 4B). However, SA did not inducecolocalization of transfected G3BP-N with endogenousTIA-1 at SGs in G3BP-N cells (Fig. 4B). These resultssuggest that exogenously overexpressed G3BP-N is notassociated with SG formation and does not inhibit SA-induced stress granule formation.

Exogenous overexpression of C-terminal G3BP induceseIF2a phosphorylationSG assembly is usually initiated by the phosphorylation of

eukaryotic translation initiation factor 2a (eIF2a), a trans-lational initiation factor that exists as a component of theeIF2-GTP-tRNAMet ternary complex (15). To furtheranalyze whether overexpressed G3BP-N and G3BP-C func-tion in SG formation, we investigated the phosphorylationof eIF2a in stable transfectants by immunoblotting(Fig. 5A). Overexpression of N-terminal G3BP did notinduce phosphorylation of eIF2a; however, C-terminal

G3BP did induce eIF2a phosphorylation. Subsequently,we decided to test the capacity of endogenous G3BP to bindto exogenously overexpressed N- or C-terminal G3BP.When SG formation is initiated, endogenous G3BP formsmultimers (16). Extracts prepared from mock-control,G3BP-N, and G3BP-C PANC-1 cells were subjected toIP with anti-Myc or anti-G3BP antibodies (Fig. 5B). Endo-genous G3BP was efficiently coimmunoprecipitated withthe Myc-tagged C-terminus G3BP but not with the Myc-tagged N-terminus G3BP. Combined with the result ofFigure 4B, these data suggest that overexpressed N-terminalG3BP did not induce self-association of G3BP and sub-sequent SG assembly; however, it did increase BARTmRNA in a dominant-negative manner (Fig. 3D and E).As reported previously (4), the C-terminus G3BP plays arole in inducing SG formation; however, it is not associatedwith the posttranscriptional regulation of BART or c-MycmRNAs (Fig. 3D–F).

Exogenous overexpression of N-terminal G3BP inhibitscell motility and invasion in PDACTo investigate the functional significance of N- and C-

terminal G3BP for cell motility and invasion in PDAC, weemployed in vitro tumor culture assays. G3BP-C PANC-1cells appeared more spindle-shaped and showed a mesench-ymal morphology as compared with mock-control andG3BP-N cells, both of which showed epithelial cell morphol-ogy (Fig. 6A). We next measured directed motility into anartificial wound, using a confluent monolayer culture ofG3BP-N or G3BP-C PANC-1 clones, and compared themwith mock-control cells (Fig. 6B and C). G3BP-N cellsshowed inhibition of motility activity into the wound areaas compared with mock and C-terminal PANC-1 cells.Despite the spindle-shaped morphologic change of theG3BP-C cells, no other significant changes were seen. Toverify these results, a motility assay on uncoated membraneswas conducted to quantifymotile cells. Figure6Dshowed thatG3BP-N PANC-1 cells were significantly less motile than themock-control or G3BP-C cells. In addition, in 2-chamberinvasion assays, G3BP-N clones were significantly less inva-sive than themock-control or G3BP-C cells (Fig. 6E). To ruleout bias due to differences in proliferation, we conductedMTT assays and the results showed that N- and C-terminalG3BP clones did not affect cell proliferation (SupplementaryFig. S2). These results suggest that the dominant-negativeexpression ofMyc-taggedN-terminalG3BP decreases cellularmotility and invasiveness of PDAC cells. Thus, in contrast tothe C-terminus G3BP, the N-terminus G3BP seems toaccelerate cellular motility and invasiveness.

Transcriptional effects of N-terminal G3BP in PDACTo determine the transcriptional program and pathways

regulated by the expression of exogenous G3BP-N, whichmight underlie the differences in cell motility and invasion,we conducted cDNA microarray gene expression profilingexperiments with 3 G3BP-N PANC-1 clones and 3 mock-control PANC-1 clones (Tables 1 and 2). Raw data from themicroarray experiments were uploaded onto the Gene

A

IB: Phospho-eIF2α

IB: β-actin

Mock# 5

G3B

P-N

# 9

Mock# 9

G3B

P-C

# 7

G3B

P-N

# 1

1

G3B

P-C

# 2

0

35

35

kDa

35IB: eIF2α

Mock# 5

(+ a

rsenite)

B

IB: G3BP*

IgG

Myc

G3B

P

IP:


IgG

Myc

G3B

P

IgG

Myc

G3B

P

kDa

50

Figure 5. Overexpressed C-terminal G3BP induces phosphorylation ofeIF2a and self-aggregation of endogenous G3BP in PDAC cells. A,Western blot analysis of endogenous phosphorylated eIF2a in G3BPdeletion mutants. Mock-control cells were exposed to 500 mmol/L arsenitefor 30 minutes under oxidative stress and used as a positive control. B,whole-cell extracts prepared from the mock-control and G3BP deletionmutants were immunoprecipitated with anti-Myc antibody or anti-G3BPantibody. Mouse IgG monoclonal antibody was used as a control.Immunoprecipitates were analyzed by Western blotting using anti-Mycand anti-G3BP antibodies. Arrows, Myc-tagged G3BP; asterisk,coimmunoprecipitated endogenous G3BP with Myc-tagged C-terminalG3BP.

Taniuchi et al.





Expression Omnibus Database (http:\\www.ncbi.nlm.nih.gov\geo\ GEO accession: GSE23342). Enrichment wasdefined as genes exhibiting a 2-fold increase in the averageratio in signal intensity. Five genes among the 14 upregulatedgenes were related to cell adhesion, motility, and invasion.Many of these genes have been previously described asmodulators of cell invasion in various tumor cell models.

The group of downregulated genes consists mainly of genesassociated with cell motility and invasion.We confirmed ourmicroarray data of the vast majority of genes tested at theRNA level by RT-PCR analysis (Supplementary Fig. S3).Our microarray results suggest that N-terminal G3BP pre-dominantly contributes to the regulation of genes involved incell motility, invasion, and adhesion in PDAC.

D

E

0

20

40

60

80

100

120

140

160

180

200

Mock# 5 Mock# 9 G3BP-N# 9 G3BP-N# 11 G3BP-C# 7 G3BP-C# 20

*

**

Num

ber of ce

lls

**

*

0

20

40

60

80

100

120

Mock# 5 Mock# 9 G3BP-N# 9 G3BP-N# 11 G3BP-C# 7 G3BP-C# 20

Num

ber of ce

lls

A

G3BP-N# 9Mock# 5 G3BP-C# 7

B

G3BP-N# 9

G3BP-C# 7

Mock# 5

0 h 20 h

C

0

20

40

60

80

100

120


Num

ber of ce

lls

***

*

*

*

*

Figure 6. Overexpression of N-terminal G3BP dominant-negatively suppresses cell motility and invasion in PDAC cells. A, morphology ofPANC-1 control, G3BP-N, and G3BP-C cells as analyzed by phase-contrast microscopy. B, wound healing assay with PANC-1 control, G3BP-N, andG3BP-C cells. The dashed lines indicate the border of the scratch (time 0) made by a plastic pipette tip. C, the number of cells that migrated into theinitially cell-free scratch was counted. Cells in 4 defined areas per group per experiment were quantified. Data are representative of 3 independent experiments.Columns, mean; bars, SD. *, P < 0.005; **, P < 0.004 compared with mock-control and G3BP-C cells, respectively. D, Transwell motility assay in control,G3BP-N, and G3BP-C cells. Migrated cells in 4 fields per group were counted. Data are representative of 3 independent experiments. Columns,mean; bars, SD. *, P < 0.005; **, P < 0.002 compared with mock-control cells. E, quantification of the 2-chamber invasion assay with PANC-1 control, G3BP-N,and G3BP-C cells. Invaded cells in 4 fields per group were counted. Data are representative of 3 independent experiments. Columns, mean; bars,SD. *, P < 0.002 compared with mock-control cells.






Discussion

Wepreviously reported that G3BP plays a role in inducinginvasion and metastasis of PDAC cells by BART mRNAdecay (12). BART was originally identified as a bindingpartner of ARL2 (17), a small G-protein implicated in theregulation of microtubule dynamics or folding (18). Toinvestigate the mechanisms by which G3BP modulates cellmotility and invasion in PDAC cells, we overexpressed N-terminal G3BP1–180 and C-terminal G3BP241–466. Wefound that overexpressed G3BP-N, which contains theNTF-2–like domain and G3BP136–180, dominant-nega-tively increases BART expression and thereby inhibits cellmotility and invasion of PDAC cells. Thus, it is likely thatthe N-terminal domain of endogenous G3BP enhances theinvasive ability of PDAC cells by posttranscriptionally reg-ulating the level of specific transcripts. The association of aparticular mRNA with G3BP may be highly regulated andmight depend on interaction with other proteins (3, 6, 19).The interaction of G3BP with cyclin-dependent kinase 7(CDK7) mRNA is dependent on the interaction of RasGAPwith Filamin C (FLN-C; ref. 19). When RNA interference

(RNAi) interrupts the interaction between RasGAP andFLN-C, CDK7 mRNA is released from G3BP (19). Wealso reported that intracellular CD24 binds to G3BP andinhibits G3BP RNase activity in PDAC cells (12). Thus,binding proteins of G3BP may form a link between G3BPand the mRNAs and influence cellular mRNA stabilizationand protein synthesis. In addition, the NTF-2–like domainplays an important role both in recruiting G3BP to thevicinity of local mRNA concentrations and in facilitating itsinteraction with selected mRNA species. Our results suggestthat a fragment of the NTF-2–like domain as well as aminimum binding region of the acidic domain of G3BPcould contribute to the recruitment of G3BP to BARTmRNA for subsequent BART mRNA degradation.In mammalian cells, phosphorylation of eIF2a can trigger

the assembly of SGs and the self-aggregation of G3BP (15).In the present study, exogenously overexpressed C-terminalG3BP could upregulate the phosphorylation of eIF2a andinduced SG assembly. Despite the colocalization of over-expressed G3BP-C fragments with RNA observed by immu-nostaining of G3BP-C cells (Fig. 4A), no significant changesin the mRNA expression of BART or c-Myc were detected

Table 1. Genes upregulated by overexpression of N-terminal G3BP

GenBank accession number Ratio: G3BP-N/mock Gene full name (symbol)

Cell adhesion, invasion, and ECMNM_001831 2.6 Clusterin (CLU)NM_001621 2.3 Aryl hydrocarbon receptor (AHR)J03048 2.2 Hemopexin (HPX)NM_004559 2.2 Nuclease sensitive element binding protein 1 (NSEP1)M16006 2.1 Plasminogen activator inhibitor-1 (PAI1)

OthersNM_016606 2.4 Receptor accessory protein 2 (REEP2)NM_003465 2.4 Chitinase 1 (CHIT1)NM_006440 2.3 Thioredoxin reductase beta (TR)NM_012266 2.3 Heat shock cognate 40 (HSC40)AF289028 2.2 Inducible T-cell costimulator ligand (ICOSLG)X56692 2.2 C-reactive protein (CRP)NM_019618 2.1 Interleukin-1 homolog 1 (IL-1H1)NM_005953 2 Metallothionein 2A (MT2A)AF105067 2 Lipopolysaccharide-binding protein (LBP)

Table 2. Genes downregulated by overexpression of N-terminal G3BP

GenBank accession number Ratio: Mock/G3BP-N Gene full name (symbol)

Cell adhesion, invasion, and ECMNM_005928 3.1 Milk fat globule-EGF factor 8 (MFGE8)NM_005529 2.7 Heparan sulfate proteoglycan 2, or perlecan (HSPG2)NM_000041 2.4 Apolipoprotein E (APOE)NM_002305 2.1 Lectin, galactoside-binding, soluble, 1, or galectin 1 (LGALS1)

OthersNM_006026 2.3 H1 histone family, member X (H1FX)NM_001823 2.3 Creatine kinase, brain (CKB)

Taniuchi et al.





(Fig. 3D–F). SGsmay be sites ofmRNA sorting whereby thestructure and composition of individual mRNPs mightdetermine whether mRNAs are repacked into translationallycompetent mRNPs or degraded (14). G3BP has been shownto bind c-Myc (3) and tau mRNA (11), inducing degrada-tion through endoribonuclease activity and stabilization,respectively. Thus, the C-terminal G3BP may not alwaysfunction inmRNAdegradation in SGs, but it may instead besometimes associated with mRNA stabilization.SG assembly can be dominantly induced by full-length

G3BP overexpression (4). Moreover, G3BP binds toCaprin-1 through the NTF-2–like domain and this com-plex induces the formation of cytoplasmic SGs (6). Inaddition, G3BP1–141 containing the NTF-2–like domainis recruited to SGs in COS cells (4). In the present study,G3BP1–180 containing the NTF-2–like domain and theBART mRNA-binding region did not induce G3BP self-association in PDAC cells. Because G3BP self-association isrequired for SG formation, it is likely that the N-terminalG3BP1–180 does not play a role in SG formation in PDACcells. Nevertheless, G3BP1–180 is essential for degradingspecific mRNAs, because exogenously overexpressedG3BP1–180 dominant-negatively increased the expressionof mRNAs for BART and c-Myc. We conclude thatG3BP1–180 is involved in mRNA degradation in PDACcells, although the exact mechanism by which it modulatesmRNA degradation remains to be clarified. In this respect,further studies are needed to show the mechanism by whichG3BP binds and recruits BART mRNA to SGs.To identify the factors and pathways regulated by N-

terminal G3BP and to obtain leads into the mechanismsbehind the inhibition of cell invasion by dominant-nega-tively overexpressed N-terminal G3BP, we conducted glo-bal expression analyses (Tables 1 and 2). Fourteen geneswere upregulated (Table 1) and 6 genes were downregulated(Table 2) after N-terminal G3BP was overexpressed inPANC-1 cells. Strikingly, 5 of the upregulated transcriptscoded for proteins involved in cellular adhesion and/orinvasion. Of these, CLU, a secreted glycoprotein withstress-induced expression in various diseased and agedtissues, decreases cell proliferation and metastatic behaviorof prostate cancer cells (20). AHR participates in signalingpathways regulating cell–cell and cell–substratum adhesion(21). HPX and hemopexin domains of human proteinsfulfill functions in the activation of matrix metalloproteinase(MMP), the inhibition of MMPs, dimerization, binding ofsubstrates or ligands, cleavage of substrates, and endocytosis(22). NSEP1, a member of the Y-box binding factor familyof evolutionarily conserved DNA/RNA-binding factors,decreases collagen type I mRNA and protein levels (23).PAI-1, a member of the serine protease inhibitor (serpins)family that blocks the activity of urokinase-type plasmino-

gen activator (uPA), is downregulated in PDAC (24) anddecreases tumor invasiveness and metastatic potential in celllines derived from colon cancer, fibrosarcoma (25), andmurine melanoma (26). Concerning the downregulatedgenes, 4 of the identified transcripts code for proteinsinvolved in cellular adhesion and/or invasion. These includeMFGE8 identified as a marker of breast cancer (27) withangiogenesis-promoting activity (28) and anticoagulantactivity (29). HSPG2 is a major component of blood vesselbasement membranes and positively regulates angiogenesis(30). Interestingly, APOE induces expression of HSPG2(31). LGALS1 increases motility of glioma cells and thereorganization of the actin cytoskeleton associated withincreased expression of RhoA, a protein that modulatesactin polymerization and depolymerization (32). Accord-ingly, the N-terminal region of G3BP could play a role inregulating secreted MMPs, transcription factors, and avariety of genes involved in cell adhesion and motility,which, in turn, would enhance cell motility and invasioncapacity of PDAC cells. Future studies are essential todetermine the endoribonuclease activity of G3BP towardspecific mRNAs identified in this cDNA microarray ana-lysis. Furthermore, the invasive activity of these genes inPDAC cells should be determined by the use of RNAiknockdown techniques.The data presented here show that G3BP1–180 containing

the novel RNA-binding residue G3BP136–180 contributes toposttranscriptional regulation of specific mRNAs for thepromotion of cell motility and invasion in PDAC. More-over, the N-terminus of G3BP plays a critical role inestablishing a pattern of gene expression that favors PDACcell motility and invasive behavior. Thus, G3BP1–180,especially G3BP136–180, could represent a novel target forblocking tumor invasion and metastasis in PDAC.

Disclosure of Potential Conflicts of Interest

No potential conflicts of interest were disclosed.

Acknowledgements

The authors thank Michel Ouellette for providing the HPNE cells, David Kellyfor cDNAmicroarray analysis, and Thomas Caffrey and TomokoMinakuchi for theirtechnical help.

Grant Support

This work was funded by the NIH with a grant to M.A. Hollingsworth(R01CA057362).

The costs of publication of this article were defrayed in part by the payment of pagecharges. This article must therefore be hereby marked advertisement in accordancewith 18 U.S.C. Section 1734 solely to indicate this fact.

Received December 17, 2010; revised May 12, 2011; accepted May 13, 2011;published OnlineFirst June 10, 2011.

References1. Parker F, Maurier F, Delumeau I, Duchesne M, Faucher D, Debussche

L, et al. A Ras-GTPase-activating protein SH3-domain-binding pro-tein. Mol Cell Biol 1996;16:2561–9.

2. Tocque B, Delumeau I, Parker F,Maurier F, MultonMC, SchweighofferF. Ras-GTPase activating protein (GAP): a putative effector for Ras.Cell Signal 1997;9:153–8.






3. Gallouzi IE, Parker F, Chebli K, Maurier F, Labourier E, Barlat I, et al. Anovel phosphorylation-dependent RNase activity of GAP-SH3 bind-ing protein: a potential link between signal transduction and RNAstability. Mol Cell Biol 1998;18:3956–65.

4. Tourri�ere H, Chebli K, Zekri L, Courselaud B, Blanchard JM, BertrandE, et al. The RasGAP-associated endoribonuclease G3BP assemblesstress granules. J Cell Biol 2003;160:823–31.

5. Kennedy D, French J, Guitard E, Ru K, Tocque B, Mattick J. Char-acterization of G3BPs: tissue specific expression, chromosomallocalisation and rasGAP(120) binding studies. J Cell Biochem 2001;84:173–87.

6. Solomon S, Xu Y, Wang B, David MD, Schubert P, Kennedy D, et al.Distinct structural features of caprin-1 mediate its interaction withG3BP-1 and its induction of phosphorylation of eukaryotic translationinitiation factor 2alpha, entry to cytoplasmic stress granules, andselective interaction with a subset of mRNAs. Mol Cell Biol 2007;27:2324–42.

7. Tourri�ere H, Gallouzi IE, Chebli K, Capony JP, Mouaikel J, van derGeer P, et al. RasGAP-associated endoribonuclease G3BP: selectiveRNA degradation and phosphorylation-dependent localization. MolCell Biol 2001;21:7747–60.

8. Jacobson A, Peltz SW. Interrelationships of the pathways of mRNAdecay and translation in eukaryotic cells. Annu Rev Biochem 1996;65:693–739.

9. Irvine K, Stirling R, Hume D, Kennedy D. Rasputin, more promiscuousthan ever: a review of G3BP. Int J Dev Biol 2004;48:1065–77.

10. Brewer G. Regulation of c-myc mRNA decay in vitro by a phorbolester-inducible, ribosome-associated component in differentiatingmegakaryoblasts. J Biol Chem 2000;275:33336–45.

11. Atlas R, Behar L, Elliott E, Ginzburg I. The insulin-like growth factormRNA binding-protein IMP-1 and the Ras-regulatory protein G3BPassociate with tau mRNA and HuD protein in differentiated P19neuronal cells. J Neurochem 2004;89:613–26.

12. Taniuchi K, Nishimori I, Hollingsworth MA. Intracellular CD24 inhibitscell invasion by post-transcriptional regulation of BART throughinteraction with G3BP. Cancer Res 2011;71:895–905.

13. Lee KM, Yasuda H, Hollingsworth MA, Ouellette MM. Notch2-positiveprogenitors with the intrinsic ability to give rise to pancreatic ductalcells. Lab Invest 2005;85:1003–12.

14. Kedersha NL, Anderson P. Stress granules: sites of mRNA triage thatregulate mRNA stability and translatability. Biochem Soc Trans2002;30:963–9.

15. Kedersha NL, Gupta M, Li W, Miller I, Anderson P. RNA-bindingproteins TIA-1 and TIAR link the phosphorylation of eIF-2 alpha tothe assembly of mammalian stress granules. J Cell Biol 1999;147:1431–42.

16. Anderson P, Kedersha NL. RNA granules. J Cell Biol 2006;172:803–8.17. Sharer JD, Kahn RA. The ARF-like 2 (ARL2)-binding protein, BART.

Purification, cloning, and initial characterization. J Biol Chem1999;274:27553–61.

18. Zhou C, Cunningham L, Marcus AI, Li Y, Kahn RA. Arl2 and Arl3regulate different microtubule-dependent processes. Mol Biol Cell2006;17:2476–87.

19. Lypowy J, Chen IY, Abdellatif M. An alliance between Ras GTPase-activating protein, Filamin C, and Ras GTPase-activating protein SH3domain-binding protein regulates myocyte growth. J Biol Chem2005;280:25717–28.

20. Moretti RM,Marelli MM,Mai S, Cariboni A, Scaltriti M, Bettuzzi S, et al.Clusterin isoforms differentially affect growth and motility of prostatecells: possible implications in prostate tumorigenesis. Cancer Res2007;67:10325–33.

21. Barouki R, Coumoul X, Fernandez-Salguero PM. The aryl hydrocar-bon receptor, more than a xenobiotic-interacting protein. FEBS Lett2007;581:3608–15.

22. Piccard H, Van den Steen PE, Opdenakker G. Hemopexin domains asmultifunctional liganding modules in matrix metalloproteinases andother proteins. J Leukoc Biol 2006;81:870–92.

23. Norman JT, Lindahl GE, Shakib K, En-Nia A, Yilmaz E, Mertens PR.The Y-box binding protein YB-1 suppresses collagen alpha 1(I) genetranscription via an evolutionarily conserved regulatory element in theproximal promoter. J Biol Chem 2001;276:29880–90.

24. Warnecke-Eberz U, Prenzel KL, Baldus SE, Metzger R, Dienes HP,Bollschweiler E, et al. Significant down-regulation of the plasminogenactivator inhibitor 1 mRNA in pancreatic cancer. Pancreas 2008;36:173–7.

25. Cajot JF, Bamat J, Bergonzelli GE, Kruithof EK, Medcalf RL, TestuzJ, et al. Plasminogen-activator inhibitor type 1 is a potent naturalinhibitor of extracellular matrix degradation by fibrosarcomaand colon carcinoma cells. Proc Natl Acad Sci U S A 1990;87:6939–43.

26. Alizadeh H, Ma D, Berman M, Bellingham D, Comerford SA, GethingMJ, et al. Tissue-type plasminogen activator-induced invasionand metastasis of murine melanomas. Curr Eye Res 1995;14:449–58.

27. Larocca D, Peterson JA, Urrea R, Kuniyoshi J, Bistrain AM, Ceriani RL.A Mr 46,000 human milk fat globule protein that is highly expressed inhuman breast tumors contains factor VIII-like domains. Cancer Res1991;51:4994–8.

28. Silvestre JS, Th�ery C, Hamard G, Boddaert J, Aguilar B, Delcayre A.Lactadherin promotes VEGF-dependent neovascularization. Nat Med2005;11:499–506.

29. Shi J, Gibert GE. Lactadherin inhibits enzyme complexes of bloodcoagulation by completing for phospholipid-binding sites. Blood2003;101:2628–36.

30. Zhou Z, Wang J, Cao R, Morita H, Soininen R, Chan KM, et al.Impaired angiogenesis, delayed wound healing and retarded tumorgrowth in perlecan heparan sulfate-deficient mice. Cancer Res2004;64:4699–702.

31. Pillarisetti S. Lipoprotein modulation of subendothelial heparan sulfateproteoglycans (perlecan) and atherogenicity. Trends Cardiovasc2000;10:60–5.

32. Camby I, Belot N, Lefranc F, Sadeghi N, de Launoit Y, Kaltner H,et al. Galectin-1 modulates human glioblastoma cell migration intothe brain through modifications to the actin cytoskeleton and levelsof expression of small GTPases. J Neuropathol Exp Neurol 2002;61:585–96.

Taniuchi et al.





2011;9:856-866. Published OnlineFirst June 10, 2011.Mol Cancer Res Keisuke Taniuchi, Isao Nishimori and Michael A. Hollingsworth Invasion by Posttranscriptional Regulation of BARTThe N-Terminal Domain of G3BP Enhances Cell Motility and

Updated version

10.1158/1541-7786.MCR-10-0574doi:

Access the most recent version of this article at:

Cited articles

http://mcr.aacrjournals.org/content/9/7/856.full#ref-list-1

This article cites 32 articles, 18 of which you can access for free at:

Citing articles

http://mcr.aacrjournals.org/content/9/7/856.full#related-urls

This article has been cited by 4 HighWire-hosted articles. Access the articles at:

E-mail alerts related to this article or journal.Sign up to receive free email-alerts

Subscriptions

Reprints and

[email protected]

To order reprints of this article or to subscribe to the journal, contact the AACR Publications Department at

Permissions

Rightslink site. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC)

.http://mcr.aacrjournals.org/content/9/7/856To request permission to re-use all or part of this article, use this link



http://mcr.aacrjournals.org/lookup/doi/10.1158/1541-7786.MCR-10-0574

http://mcr.aacrjournals.org/content/9/7/856.full#ref-list-1

http://mcr.aacrjournals.org/content/9/7/856.full#related-urls

http://mcr.aacrjournals.org/cgi/alerts

mailto:[email protected]

http://mcr.aacrjournals.org/content/9/7/856


the n-terminal domain of g3bp enhances cell motility and ... · the regulationof mrna stability...

Documents