the serum response factor nuclear localization signal: general

MOLECULAR AND CELLULAR BIOLOGY, Jan. 1995, p. 433–444 Vol. 15, No. 10270-7306/95/$04.0010Copyright q 1995, American Society for Microbiology

The Serum Response Factor Nuclear Localization Signal: GeneralImplications for Cyclic AMP-Dependent Protein Kinase

Activity in Control of Nuclear TranslocationCECILE GAUTHIER-ROUVIERE,1 MARIE VANDROMME,1 NICOLE LAUTREDOU,1 QIU QIONG CAI,2

FRANCK GIRARD,1 ANNE FERNANDEZ,1 AND NED LAMB1*

Cell Biology Unit, Centre de Recherche en Biochimie Macromoleculaire, Centre National de la Recherche Scientifique-Institut National de la Sante et de la Recherche Medicale, 34033 Montpellier Cedex,1 and Institut National de la Sante

et de la Recherche Medicale U148, 34090 Montpellier,2 France

Received 5 July 1994/Returned for modification 6 September 1994/Accepted 18 October 1994

We have identified a basic sequence in the N-terminal region of the 67-kDa serum response factor (p67SRF

or SRF) responsible for its nuclear localization. A peptide containing this nuclear localization signal (NLS)translocates rabbit immunoglobulin G (IgG) into the nucleus as efficiently as a peptide encoding the simianvirus 40 NLS. This effect is abolished by substituting any two of the four basic residues in this NLS.Overexpression of a modified form of SRF in which these basic residues have been mutated confirms theabsolute requirement for this sequence, and not the other basic amino acid sequences adjacent to it, in thenuclear localization of SRF. Since this NLS is in close proximity to potential phosphorylation sites for thecAMP-dependent protein kinase (A-kinase), we further investigated if A-kinase plays a role in the nuclearlocation of SRF. The nuclear transport of SRF proteins requires basal A-kinase activity, since inhibition ofA-kinase by using either the specific inhibitory peptide PKIm or type II regulatory subunits (RII) completelyprevents the nuclear localization of plasmid-expressed tagged SRF or an SRF-NLS-IgG conjugate. Directphosphorylation of SRF by A-kinase can be discounted in this effect, since mutation of the putative phosphor-ylation sites in either the NLS peptide or the encoded full-length SRF protein had no effect on nucleartransport of the mutants. Finally, in support of an implication of A-kinase-dependent phosphorylation in amore general mechanism affecting nuclear import, we show that the nuclear transport of a simian virus40-NLS-conjugated IgG or purified cyclin A protein is also blocked by inhibition of A-kinase, even thoughneither contains any potential sites for phosphorylation by A-kinase or can be phosphorylated by A-kinase invitro.

The 67-kDa serum response factor (p67SRF or SRF) is aubiquitous nuclear transcription factor that acts through bind-ing to a consensus DNA sequence, the serum response ele-ment (SRE) (38, 41, 53) present in the promoters of genesinvolved in the early mitogenic response (15, 52). All func-tional SREs so far examined contain the core sequence CC(A/T)6GG (termed the CArG sequence), which is also found inthe promoters of several muscle-specific genes (cardiac andskeletal actin genes and the dystrophin gene, for example) (22,34). Through its binding to these different promoters, SRF hasbeen implicated in control of both proliferation and muscledifferentiation, which could be confirmed through in vivo im-munodepletion with anti-SRF antibody injection (11, 13, 56).Correlating with its transcriptional activity, SRF is a nuclearphosphoprotein present throughout the cell cycle in all celllines tested to date (13).Studies of the requirements for nuclear translocation have

implicated basic sequences (referred to as nuclear localizationsignals [NLS]) sufficient to induce nuclear import (5, 9). NLSare generally short stretches of 5 to 10 amino acids containingseveral basic residues (arginine and/or lysine) (6). Some pro-teins (e.g., the simian virus 40 [SV40] large-T antigen) have anNLS comprising a contiguous run of basic amino acids (21, 25),while others (e.g., nucleoplasmin) have a bipartite motif (7,

10). It is believed that the basic region facilitates interactionwith NLS binding proteins (NBPs), which allow transportthrough nuclear pore complexes in an ATP-dependent manner(1, 47, 50, 59). These NBPs have been characterized for variousorganisms, and two of them, NBP70 and Nopp140, are phos-phorylated in vivo, an event required for their interaction withNLS (31, 51). Moreover, the possibility that nuclear localiza-tion might be regulated by variations in specific kinase activi-ties is illustrated by various examples (29, 37, 42). In particular,it has been shown that the SRE binding activity of the tran-scription factor NFIL-6 is inducible by a forskolin treatment,an event which correlates with an increase in both NFIL-6phosphorylation and its nuclear localization (33). Positive cor-relations have also been made between phosphorylation andnuclear translocation for c-rel and the cyclic AMP (cAMP)-dependent protein kinase (A-kinase) (35), and suggested forc-fos and A-kinase (45). Recently, we observed that the activenuclear import of the myogenic transcription factor MyoDrequires A-kinase activity in a manner that did not involvedirect phosphorylation of MyoD (55). This observation raisedthe following two possibilities for a potential role of A-kinasein this process: (i) phosphorylation by A-kinase of a compo-nent distinct from MyoD but specifically involved in MyoDnuclear import is required, and (ii) A-kinase activity is neededin a more general mechanism of control of nuclear transport.In the present study, we show that SRF contains a unique

and highly efficient NLS (SRF-NLS) located at the N-terminalpart of the protein. Since this NLS is located in close proximity

* Corresponding author. Mailing address: Cell Biology Unit,CRBM, CNRS-INSERM, Route de Mende, BP 5051, 34033 Montpel-lier Cedex, France. Phone: (33) 67 61 33 57. Fax: (33) 67 52 15 59.

433

to potential consensus phosphorylation sites for A-kinase, wequestioned whether direct phosphorylation by A-kinase couldmodulate the efficiency of the process of nuclear import.Whereas A-kinase activity was required for SRF nuclear im-port, site-specific mutations of SRF showed that A-kinase doesnot exert its control through direct phosphorylation of the sitespresent near SRF-NLS. Additionally, we show that the nucleartransport of cytoplasmically injected SV40-NLS-conjugatedIgGs or purified cyclin A is impaired through inhibition ofA-kinase activity. Since neither SV40-NLS nor cyclin A can bephosphorylated by A-kinase, these data show the general re-quirement for A-kinase in the mechanism of nuclear translo-cation control.

MATERIALS AND METHODS

Cells. Human foreskin fibroblasts (Hs-68) or rat embryo fibroblasts (REF-52)(28) were cultured in Dulbecco’s modified Eagle’s medium supplemented with8% fetal calf serum as described previously (13). Cells were plated 2 to 3 daysbefore use.Peptide synthesis and conjugation. Solid-phase synthesis of the different pep-

tides was performed by using a 9050 Milligen synthesizer with FMOC PAL-PEG-PS resin (Milligen, St-Quentin les Yvelines, France), 9-fluorenyl methyloxycarbonyl (FMOC) as temporary alpha-amino group protection, and benzo-triazol-1-yl-oxy-tris-(dimethylamino)-phosphonium-hexafluorophosphate (BOP)as the coupling reagent. Terminal cysteine was added to each peptide sequencefor the purpose of protein coupling.Peptides (see sequences below) were purified by reverse-phase high-pressure

liquid chromatography on Nucleosil 5C18 columns by using a linear gradient ofacetonitrile (0 to 30% gradient) with 0.1% aqueous trifluoroacetic acid. Purifiedpeptides or SRF-DB (13) was coupled to rabbit immunoglobulin G (IgG) (Sig-ma, St-Quentin Fallavier, France) by using the water-soluble cross-linking re-agent sulfo-succinimidyl-4-(N-maleimidomethyl-cyclohexane)-1-carboxylate (Sulfo-SMCC) (Pierce, Interchim, Montlucon, France) in a 50:1 molar ratio as describedby the manufacturer. Conjugates were gel filtered on PD10 (Pharmacia) equil-ibrated in 100 mM HEPES (N-2-hydroxyethylpiperazine-N9-2-ethanesulfonicacid) (KOH, pH 7.2) and resuspended at 2 mg/ml in 100 mM HEPES forsubsequent microinjection. The coupling ratio was evaluated to be ca. 2 to 5peptides per IgG molecule by the shift in molecular weight of the IgG heavychains by sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis. Se-quences of the peptides were as follows (mutated residues are underlined):SRF-NLS, 91LGAERRGLKRSLSE104; SRF-NLS-E95E96, 91LGAEEEGLKRSLSE104; SRF-NLS-E99E100, 91LGAERRGLEESLSE104; SRF-NLSAla,91LGAERRGLKRALAE104; SRFNLSAsp, 91LGAERRGLKRDLDE104.In vitro phosphorylation. SRF proteins obtained by in vitro translation in

reticulocyte lysate (the plasmid was kindly provided by R. Treisman, ImperialCancer Research Fund, London, England) were incubated with either caseinkinase II (CKII) (11) or A-kinase (Promega, Coger, Paris, France) in the pres-ence of 100 mM [g-32P]ATP (100 to 200 cpm/pmol) for 20 min at room temper-ature. The reaction was stopped by adding sample buffer to the reaction mixtureand then boiling for 2 min before loading the sample onto an SDS–12.5%polyacrylamide gel and performing autoradiography.Protein kinase A activation in living cells. Cells were made quiescent by

removal of serum for 36 h and then incubated with forskolin (15 mM) andmethyl-isobutyl-xanthine (MIX) (0.5 mM; Sigma) or 8-Br-cAMP (1 mM; Sigma)and MIX (0.5 mM), after which they were fixed at different times and analyzedfor SRF distribution by immunofluorescence using confocal laser scanning mi-croscopy.Measurements of intracellular cAMP levels. Two 60-mm-diameter dishes

(representing ca. 8 z 106 cells) of either quiescent cells, cells stimulated withserum for 30 min, or cells treated with forskolin (15 mM) and MIX (0.5 mM) for30 min were used for intracellular cAMP level measurements. Cells were scrapedinto phosphate-buffered saline (PBS), and the cAMP was extracted by an aque-ous ethanol solvent procedure and measured by using the cAMP assay kit fromAmersham (Les Ulis, France) according to the manufacturer’s instructions. Anaverage cell volume of 1.5 z 10213 liters was used for calculations of the intra-cellular concentrations of cAMP.Expression of tagged SRF by microinjection into mammalian cells and mu-

tagenesis. A BglI-EcoRI fragment from the full-length SRF cDNA (plasmidSRFpT7DATG obtained from R. Treisman, Imperial Cancer Research Fund)was ligated into the SmaI-EcoRI-digested BGF5 plasmid. The resulting plasmidwas used to transform HB101 bacteria. Cesium chloride-purified plasmid wasmicroinjected into the nuclei of growing cells, and expression of tagged HAP/SRF protein was detected by immunofluorescence 3 to 5 h later.Double-stranded SRF cDNA was mutagenized in the same vector by using the

transformer site-directed mutagenesis kit (Clontech, Ozyme, Montigny, France)as described by the manufacturer. The following purified oligonucleotides (Eu-rogentec, Seraing, Belgium) were used: selection primer, ACCGTCGACCACG

AGGGGGGG; mutagenic primer E95E96, CTGGGCGCCGAGGAGGAGGGCCTGAAGCGG; mutagenic primer Ala101-103, GGCCTGAAGCGGGCCCTGGCCGAGATGGAGATCGGT; mutagenic primer Asp101-103, GGCCTGAAGCGGGACCTGGACGAGATGGAGATCGGT.Microinjection and immunofluorescence. REF-52 or Hs-68 cells plated on

glass coverslips were treated with forskolin (15 mM) and MIX (0.5 mM) or8-Br-cAMP (1 mM) and MIX. At different times after treatment, cells were fixedin 3.7% formalin in PBS for 5 min. After extraction in acetone (2208C) for 30 sand rehydration in PBS containing 0.2% bovine serum albumin, cells werestained for SRF distribution as previously described (13) and for actin distribu-tion by using rhodamine-phalloidin (8). The stainings were analyzed with a Leicaconfocal laser scanning microscope.For nuclear transport studies, a solution of peptide-rabbit IgG conjugates (1

mg/ml) mixed 1/1 with unconjugated mouse IgGs (1 mg/ml; Sigma) was injectedinto the cytoplasm of cells. In some cases, either PKIm (5.1025 M), PKI(15-24)(5.1025 M) (8), the regulatory subunit RII of A-kinase (1.4 mg/ml), or thecatalytic subunit of A-kinase (0.1 to 1 mg/ml) was included in the injectionsolution. Cells were fixed at different times after injection and stained for thelocalization of the NLS-conjugated IgG by using biotinylated anti-rabbit anti-body (1:200; Amersham) for 30 min and streptavidin-Texas red (1:400; Amer-sham) for 30 min and for localization of the mouse IgGs, used as markers for theinjected cells, by using fluorescein-conjugated anti-mouse antibody (1:200; Cap-pel-Organon Teknika, Fresnes, France) for 30 min. The chromatin was stainedwith Hoechst stain (0.1 mg/ml; Sigma), applied just before cells were mountedand observed by confocal laser scanning microscopy.Plasmids were injected in the nuclei of proliferating cells by using a solution of

plasmid (0.1 mg/ml) containing rabbit marker IgGs (1 mg/ml; Sigma) eitheralone or together with PKIm. Five hours later, cells were fixed as described belowand stained for the expressed tagged SRF protein with the monoclonal antibody12CA5 (1:200; Dabco-USA) (58) for 30 min and then with fluorescein-conju-gated anti-mouse antibody (1:200; Cappel). Injected cells were stained withbiotinylated anti-rabbit antibody (1:200; Amersham) for 30 min and then withstreptavidin-Texas red (1:400; Amersham).Solutions of purified human cyclin A containing either mouse marker IgGs or

mouse IgGs and PKIm were microinjected in the cytoplasm of Hs-68 cells about5 to 8 h after serum refeeding (i.e., during the early G1 phase of the cell cycle).After 30 min or 1 h, cells were fixed and analyzed for cyclin A distribution asdescribed elsewhere (17).Confocal laser scanning microscopy. Dual-channel confocal laser scanning

microscopy was performed by using the Leica confocal laser scanning microscopeequipped with a krypton-argon ion laser with two major emission lines at 488 nmfor fluorescein isothiocyanate excitation and 568 nm for rhodamine or Texas redexcitation. Planapochromat lenses (403 or 633) were used, and the untreatedimages were directly transferred from the VME bus of a Leica Motorola 68040to a Silicon Graphics, Inc. (SGI, Mountain View, Calif.), IRIS Indigo worksta-tion (R3000). Images were deconvoluted, gamma mapped, and converted to SGIraster format by using Convert software (18). Figures were assembled completelywith SGI Showcase 2.01 and printed directly as postscript files by using a KodakColorease thermal sublimation printer.

RESULTS

A basic sequence in the N-terminal part of SRF functions asNLS. Analysis of the primary amino acid sequence of p67SRF

reveals four clusters of basic amino acids which could act aspotential NLS. From the N terminus of the protein, the firstsequence, termed NLS1, is located outside the predictedDNA-binding domain. The three other potential NLS (NLS2,NLS3, and NLS4) are located within the predicted DNA-bind-ing domain (Fig. 1a).The activity of these NLS was assessed by monitoring their

capacity to translocate nonnuclear proteins, for example, im-munoglobulins (rabbit IgG), into the nucleus following micro-injection into cultured mammalian cells. Rabbit IgGs werechosen because of their large size, which effectively prohibitspassive diffusion through the nuclear membrane, and theirinert nature once injected into cells. The efficiency of NLS1was tested with the peptide 91LGAERRGLKRSLSE104. Theefficiencies of NLS2, NLS3, and NLS4 were tested simulta-neously by using a portion of SRF spanning from residues 113to 265 and containing the DNA-binding domain of the protein(called SRF-DB) (12). This peptide includes the three otherpotential NLS (Fig. 1a). Both the NLS1-containing peptideand purified SRF-DB were chemically coupled to rabbit IgG(as detailed in Materials and Methods) and microinjected intothe cytoplasm of REF-52 or Hs-68 fibroblasts together with an

434 GAUTHIER-ROUVIERE ET AL. MOL. CELL. BIOL.

FIG. 1. Identification of potential NLS in the amino acid sequence of SRFprotein. (a) We have identified the presence of four potential NLS in the primaryamino acid sequence of SRF on the basis of the presence of clusters of basicresidues (arginine and lysine). NLS were numbered from the N terminus of theprotein, with NLS1 being outside the DNA-binding domain and the three others,NLS2, NLS3, and NLS4, being within the DNA-binding domain. For each ofthese basic sequences, a consensus A-kinase site (R/K-X-S/T or R/K-X-X-S/T) isfound associated. SRF-DB (delimited by the horizontal arrow) represents aportion of SRF we have expressed in bacteria that contains NLS2, -3, and -4 andcomprises the DNA-binding region of SRF. (b) Either NLS1-containing peptideor purified SRF-DB was coupled to rabbit IgG as described in Materials andMethods. Each of these conjugates was microinjected into the cytoplasm ofgrowing Hs-68 cells with an inert mouse marker antibody for subsequent iden-tification of injected cells. Cells were fixed at different times afterward andstained for the conjugated rabbit IgG and for the mouse marker antibody. PanelsA to D show injection of NLS1-IgG conjugates (A and C) fixed at 2 h (B) or 24h (D), and panels E to H show injection of SRF-DB-IgG conjugates (E and G)fixed at 2 h (F) or 24 h (H). Bar 5 10 mM. (c) Growing Hs-68 cells weremicroinjected with SRF-NLS-IgG conjugates together with mouse marker anti-body, fixed at different times afterward, and stained for mouse marker antibody(A, C, and E) and localization of SRF-NLS-IgG conjugates (B, D, and F). Cellswere fixed 15 min (A and B), 30 min (C and D), or 60 min (E and F) afterinjection. Bar 5 10 mM.

435

inert mouse antibody for subsequent identification of injectedcells. The localizations of conjugated NLS1, conjugated SRF-DB, and unconjugated rabbit IgGs were analyzed by immuno-fluorescence at different times thereafter. As shown in Fig. 1b,when Hs-68 cells were injected with conjugated-NLS1-contain-ing peptide (panels A and C), a strong nuclear staining wasobserved 2 h after injection (panel B) and remained visibleafter 24 h (panel D). This shows that NLS1 (designated SRF-NLS) is competent to translocate coupled IgGs into the nu-cleus. As few as two SRF-NLS peptides conjugated per IgGmolecule would successfully result in IgG nuclear transloca-tion. In contrast, only cytoplasmic staining was observed in cellsinjected with IgG-conjugated SRF-DB (Fig. 1b, panels E and G)2 h (panel F) and 24 h (panel H) after injection, clearly showingthat SRF-DB does not contain a sequence efficient in inducingnuclear localization (even at a coupling ratio of two SRF-DBpeptides per IgG molecule). Similar results were obtained forREF-52 cells and with the injection of uncoupled IgGs whichremain in the cytoplasm for up to 48 h (data not shown).For a more detailed analysis of the kinetics of nuclear local-

ization, NLS1 peptides conjugated to rabbit IgGs were micro-injected together with mouse marker IgGs into the cytoplasm

of proliferative Hs-68 cells and cells were fixed at times be-tween 1 min and 2 h. As shown in Fig. 1c, as little as 15 minafter injection (panel A), the rabbit IgGs had already acquireda nuclear localization (panel B), which became more pro-nounced after 30 min (panels C and D). Sixty minutes afterinjection (Fig. 1c, panel E), the staining revealed a completenuclear accumulation of the IgG-conjugated NLS1 (panel F),with little or no residual cytoplasmic signal remaining. Similarkinetics of nuclear import were also observed with IgGs con-jugated to the NLS sequence of SV40 (CPKKRKV), a se-quence described as being highly efficient in promoting nuclearlocalization (20).Taken together, these results show that amongst the four

potential NLS present in SRF, only the first N-terminal se-quence, NLS1 (henceforth designated SRF-NLS), was able toconfer rapid nuclear localization on large and nonnuclear pro-teins such as IgGs.Mutation of the basic residues within the NLS sequence of

SRF completely prevents its nuclear localization. In order toassess if the NLS sequence we have identified in SRF proteinwas absolutely required for SRF nuclear transport and to dis-count the possibility that other sequences in the entire SRF

FIG. 2. Mutation of arginine 95 and arginine 96 of SRF prevents the nuclear import of the expressed tagged protein. Either wild-type (wt) HAP/SRF plasmid (Aand B) or mutated HAP/SRF-E95E96 plasmid (C and D), together with a rabbit marker antibody for subsequent identification of injected cells, was injected into thenuclei of growing Hs-68 cells. Five hours later, cells were fixed and stained for rabbit marker antibody (A and C) or for HAP/SRF expression and localization (B andD). Bar 5 10 mM.


protein could substitute for it, we examined the effect on SRFnuclear transport of mutating SRF-NLS. To distinguish theexpression of the cloned cDNA from the endogenous SRF, wefused a 9-residue influenza virus hemagglutinin peptide epi-tope (HAP) (YDVPDYASL) to the 59 end of the SRF cDNA.Expression of the fusion protein (HAP/SRF) was detected withmonoclonal antibody 12CA5 (58).Mutant SRF protein in which the two residues arginines 95

and 96 were substituted with glutamic acid was prepared asdescribed in Materials and Methods. Either wild-type or mu-tated SRF plasmids (0.1 mg/ml), in a eucaryote expressionvector under the regulation of an SV40 promoter, were in-jected into the nuclei (to effect maximal expression) of growingHs-68 or REF-52 cells. After 5 h (a time sufficient to allowprotein expression), cells were fixed and analyzed for HAP/SRF expression and localization by immunofluorescence. Asshown in Fig. 2, when cells were injected with plasmids ex-pressing wild-type SRF (panel A), the expressed epitope-tagged SRF protein was exclusively localized in the nucleus(panel B). No cytoplasmic staining was ever detected, even ifexpression was monitored for shorter periods of time (data notshown), showing that wild-type SRF is efficiently translocatedinto the nucleus at all times. In marked contrast, when cellswere injected with plasmids encoding SRF mutants in whichthe two basic residues (arginines 95 and 96) were exchangedfor glutamic acid (Fig. 2C), the expressed SRF was exclusivelycytoplasmic (panel D), showing that the mutated HAP/SRFfusion protein is no longer competent to translocate into thenucleus. Furthermore, using SRF-NLS peptides in which ei-ther arginines 95 and 96 and/or lysine 99 and arginine 100 weresubstituted with glutamic acid, we have observed that the nu-clear localization activity of SRF-NLS peptide required allthese basic amino acids (data not shown).These results confirm the efficiency of the NLS sequence we

have identified in SRF for the nuclear transport of the proteinand particularly show that this NLS is the essential and uniquemotif for SRF nuclear localization, with no other sequence inSRF protein substituting for its activity.Modulation of SRF distribution in a manner dependent on

A-kinase activity. The SRF-NLS sequence is located betweentwo phosphorylation sites that appear to be phosphorylated invivo (19): a potential site for CKII (83SESGEEEE90) and aputative consensus site for either A-kinase, calcium calmodulin(CaM)-dependent kinase II, or the ribosomal S6 kinase pp90rsk

(99KRSLS103). Whereas it has been shown already that SRF isphosphorylated in vitro by pp90rsk (43), we have also observedin vitro phosphorylation of SRF by A-kinase (data not shown).The close proximity of these phosphorylation sites to the NLSand our previous observations of MyoD nuclear localization(55) prompted us to investigate potential involvement of A-ki-nase-dependent phosphorylation of these sites in the subcellu-lar localization of SRF.We have previously described the nuclear localization of

SRF throughout the cell cycle in mammalian cells (13). Inorder to monitor the effects of A-kinase activation, we treatedcells with forskolin and MIX and subsequently examined thedistribution of SRF by confocal microscopy using two fibro-blast lines (REF-52 and Hs-68). Since we have previously ob-served that the maximum effects of A-kinase activation occurin quiescent cells (14), SRF staining was performed with qui-escent cells. In order to control for the level of A-kinase acti-vation, we monitored the extent of actin filament reorganiza-tion (24), as shown in Fig. 3A and C. In addition, we havecarried out measurements of intracellular cAMP concentra-tion, to compare the cAMP levels in quiescent, serum-stimu-lated, and forskolin-treated cells. The results of these measure-

ments are summarized in Table 1. Quiescent cells, where thelevel of cAMP and therefore of active A-kinase is minimal(Table 1), present an intact actin microfilament network (Fig.3A) and SRF is essentially distributed in the nucleus (Fig. 3B).A low level of cytoplasmic staining is also detectable in thesecells because of the high degree of sensitivity of the confocallaser scanning microscope technique (Fig. 3B). When cellswere treated for 20 min with forskolin (15 mM) and MIX (0.5mM), which elevated intracellular cAMP levels by three- tofourfold (Table 1), we observed the complete disappearance ofthis diffuse cytoplasmic distribution of SRF, concomitant witha more granulated nuclear staining (Fig. 3D). No effects wereobserved when MIX alone was used (data not shown), whichcorrelates with an absence of detectable changes in cAMPlevel with MIX alone. The changes shown in Fig. 3A to D arefully representative of the staining observed in all cells treatedto activate A-kinase. Similar results for SRF distribution wereobtained when 8-Br-cAMP (1 mM) (a cell-soluble analogue ofcAMP) was used to activate A-kinase or following microinjec-tion of the highly purified catalytic subunit of A-kinase (datanot shown). A 20-min drug treatment was sufficient to allow astrong activation of A-kinase (as shown in Fig. 3C) withoutinducing extensive morphological modifications that couldhave contributed to a modification in the SRF distribution.Previous articles have reported that a reduction of the cAMPlevel occurs before induction of oocyte meiotic division (2, 27,32) and mitotic entry in mammalian cells (23). We stained cellsfor the distribution of SRF at the onset of mitosis (i.e.,prophase). Figure 3E and F shows an area of growing cells witha cell in prophase in the center (marked by an arrow) discern-ible by the condensed chromatin in the Hoechst-stained DNA(panel E). In this prophase cell, where the level of A-kinase islow, cytoplasmic SRF staining is clearly visible, whereas noneof the surrounding interphase cells show cytoplasmic stainingfor SRF (Fig. 3F). The appearance of cytoplasmic staining forSRF in prophase cells occurs prior to any sign of nuclearenvelope disassembly, as checked by staining cells for nuclearlamin organization (data not shown), and therefore is unlikelyto be due to an increased porosity of the nucleus. In addition,we have shown before that throughout the different phases ofmitosis, the majority of SRF remains preferentially associatedwith the condensed chromatin (13). These data illustrate thatthe subcellular localization of SRF is apparently modulatedwith respect to variations in intracellular A-kinase activity. Thepresence of cytoplasmic SRF is detected only when A-kinaselevels are reduced during the cell cycle.We have previously described the effective inhibition of A-

kinase activity in vivo, using a modified stable version of thespecific inhibitor peptide for A-kinase, PKIm (8). We thereforeexamined if complete inhibition of intracellular A-kinase ac-tivity through microinjection of PKIm modifies the nuclearlocalization of overexpressed tagged SRF proteins. PKIm wasmicroinjected into cells 5 h after microinjection of a plasmidencoding HAP/SRF protein. When the localization of the ex-pressed protein was examined 15 min later, we observed acomplete loss of nuclear localization of the expressed taggedSRF protein (Fig. 3G and H). In the two double-injected cells(identified by the presence of both mouse antibodies againstthe HAP tag and the rabbit marker antibodies injected withPKIm, marked by arrows in Fig. 3G and H), only a cytoplasmicdistribution of the expressed HAP/SRF protein is detectedafter 15 min. In contrast, the two adjacent cells, which havebeen injected only with the HAP/SRF wild-type plasmid, showa strong nuclear staining resulting from the nuclear accumu-lation of the expressed tagged SRF. As a control, cells injectedwith only rabbit marker antibody show no staining, effectively

VOL. 15, 1995 SRF NUCLEAR LOCALIZATION SIGNAL AND ROLE OF A-KINASE 437

FIG. 3. SRF distribution after modulation of A-kinase activity. In order to examine SRF distribution after A-kinase activation, quiescent Hs-68 fibroblasts platedon glass coverslips were treated with forskolin (30 mM) and MIX (0.5 mM). After a 20-min exposure to the drugs, cells were fixed with formalin and analyzed for thedistribution of SRF and actin as described in Materials and Methods. Shown are immunofluorescence micrographs of quiescent untreated cells (A and B) or quiescentcells treated with drugs for 20 min (C and D) showing actin (A and C) or SRF (B and D) distribution. An area of growing cells containing a prophase cell (markedby an arrow) is shown (E and F); panel E shows DNA staining with Hoechst, and panel F shows endogenous SRF distribution. Growing Hs-68 cells were injected withwild-type (wt) HAP/SRF plasmid. Five hours later, cells were reinjected with a solution of PKIm containing rabbit marker antibody and fixed after 1 h, after which theywere stained for rabbit marker antibody (G) and HAP/SRF localization (H). Arrows represent the double-injected cells. Bars 5 10 mM.


excluding the possibility that cross-reactivity during the immu-nofluorescence staining technique could account for these ob-servations (data not shown). These data, in addition to showingthat inhibiting A-kinase prevents nuclear translocation of neo-synthesized SRF, also demonstrate that inhibition of A-kinaseresults in the exit from the nucleus of the SRF already accu-mulated in the nuclear compartment during the previous 5 h ofexpression. Indeed, 15 min after PKIm injection, little or noneof the expressed HAP/SRF proteins remained in the nucleus.Moreover, no SRF proteins are detected in the nucleus for aslong as PKIm continues to be active (i.e., up to 4 to 5 h afterits injection), confirming that nuclear import of neosynthesizedSRF was also prevented by inhibiting A-kinase. Our observa-tion of the effect of A-kinase inhibition on SRF nuclear reten-tion and/or exit differs from our previous observations withMyoD protein. In the latter case, inhibition of A-kinase had noeffect on the nuclear localization of the MyoD proteins alreadypresent in the nucleus (55).The rate of import by SRF-NLS is dependent upon A-kinase

activity. In light of these effects of A-kinase activation or in-hibition on SRF distribution, we questioned whether the nu-clear import of SRF-NLS-conjugated IgGs was directly mod-ulated by A-kinase. To address this, solutions containing SRF-NLS-conjugated IgGs, inert mouse marker antibodies, andeither the catalytic subunit of A-kinase (0.1 mg/ml) or thespecific inhibitory peptide PKIm (5 z 1025 M) were injectedinto the cytoplasm of cells.In order to have the minimum basal A-kinase activity, qui-

escent cells were used for injection of the catalytic subunit ofA-kinase and the NLS-conjugated IgGs. As shown in Table 1,these cells have approximately twofold less cAMP than doserum-stimulated cells. Cells were fixed after 30 min, a timewhen the distribution of injected peptide-IgG complexes wastotally nuclear in serum-stimulated growing cells. In correla-tion with the observation that the intracellular levels of A-ki-nase are lower in quiescent cells, we observed that the kineticsof nuclear translocation were slower in quiescent than in pro-liferating cells. Indeed, when quiescent cells injected only withthe SRF-NLS-IgG conjugate were fixed after 30 min (Fig. 4a,panel A), we observed approximately equivalent cytoplasmicand nuclear distributions of the IgG-conjugated peptide (panelB). Comparison of this distribution in quiescent cells with thatobserved 30 min after injection in growing cells (compare Fig.4a, panels A and B, with Fig. 1c, panels C and D) shows thatin the latter case, the nuclear accumulation of injected SRF-NLS-IgG is almost complete by 30 min. In a similar manner,when the quiescent cells were coinjected with the catalyticsubunit of A-kinase and NLS-IgG conjugate (Fig. 4a, panel C),a predominant nuclear location of the IgGs was observed by 30min (panel D). The residual cytoplasmic staining in panel D iscompletely lost when cells are injected with a higher concen-tration of A-kinase (0.5 mg/ml) (data not shown). However, asreported previously (24), high levels of A-kinase result in mor-phological changes in cells which may interfere with our as-sessment of cytolocalization. A similar result was also obtainedby treating the injected cells with forskolin to activate A-kinase(data not shown). This result shows that elevation of intracel-lular A-kinase levels significantly increases the rate of nuclearimport of SRF-NLS-conjugated IgGs.To examine the effect of inhibiting A-kinase activity on the

nuclear import by SRF-NLS, growing cells were coinjectedwith SRF-NLS conjugate and PKIm and fixed 1 h after micro-injection, a time sufficient for the conjugated IgG to localizeinto the nucleus. As shown in Fig. 4b, when injected in thepresence of the inhibitory peptide PKIm (panel A), the SRF-NLS-conjugated IgGs are found exclusively in the cytoplasm

(panel B), showing the requirement for A-kinase activity in thenuclear transport by SRF-NLS. This effect was specific forA-kinase, since injection of a noninhibitory peptide, PKIm(15-24) (8) (Fig. 4b, panel C), had no effect on the nuclear local-ization of SRF-NLS (panel D). Moreover, inactivation of A-kinase through coinjection of the regulatory subunit ofA-kinase inhibited nuclear transport of SRF-NLS-conjugatedIgGs (Fig. 4b, panels E and F) in a similar manner. In contrast,specific inhibition of Ca21-phospholipid-dependent protein ki-nase through microinjection of another specific inhibitory pep-tide, C-PKI (19), did not result in any alteration in the nuclearimport of SRF-NLS-conjugated IgGs (data not shown).These results clearly show that A-kinase activity is required

for the nuclear import of IgG conjugated to SRF-NLS and thata reduced rate of import by the SRF-NLS sequence is observedwhen the endogenous level of A-kinase is low, i.e., in quiescentcells.Mutation of the SRF-NLS A-kinase phosphorylation site

does not affect its nuclear import activity. Since A-kinase ac-tivity seems to be acutely involved in the active nuclear importof both SRF-NLS-conjugated IgGs and SRF protein, we ex-amined the implication of direct phosphorylation of the puta-tive A-kinase site previously identified in close proximity to thebasic residues in the SRF-NLS sequence (20). To examine thisprocess, synthetic peptides in which the two phosphate accep-tor amino acids (serines 101 and 103) were substituted foreither a nonphosphorylatable residue, alanine (SRF-NLSAla),or a residue mimicking a phosphorylation charge, aspartic acid(SRF-NLSAsp), were synthesized. In vitro phosphorylation as-says using A-kinase confirmed that these peptides were notphosphorylatable (data not shown).Peptides were chemically coupled to rabbit IgGs, and the

efficiency of nuclear import of the conjugate was tested bymicroinjection. Both SRF-NLSAla and SRF-NLSAsp peptidestranslocate the conjugated rabbit IgGs into the nucleus asefficiently as does the wild-type SRF-NLS peptide. In addition,no significant differences in the nuclear import kinetics couldbe detected between these three peptides. However, coinjec-tion of the A-kinase inhibitor peptide PKIm with any of thesepeptides effectively inhibited nuclear import and resulted inthe cytoplasmic retention of the peptide-IgG conjugate, asobserved with the wild-type SRF-NLS peptide (data notshown). To confirm these observations, we constructed plas-mids coding for mutant forms of the whole SRF protein inwhich these key phosphoacceptor amino acids were substitutedfor nonphosphorylatable alanine or aspartic acid. Mutant plas-mids in which the two serines 101 and 103 have been substi-tuted with either alanine (HAP/SRF-Ala101-103) or asparticacid (HAP/SRF-Asp101-103) were injected in the nuclei ofgrowing cells. After 5 h, cells were fixed and stained for HAP/SRF localization by immunofluorescence. Both mutant formsof the protein localize into the nuclei, as observed with thewild-type plasmid (data not shown).

TABLE 1. Concentration of intracellular cAMP in quiescent cellsand after 30 min of serum stimulation or forskolin treatment

Cell line

cAMPconcn inquiescentcells (mM)

Serum-stimulated cells Forskolin-treated cells

cAMP concn(mM)

Foldincreasea

cAMP concn(mM)

Foldincreasea

Hs-68 1.32 2.2 1.7 4.92 3.8REF-52 1 1.8 1.8 3.3 3.3

a Expressed as fold increase in cAMP levels over the levels measured inquiescent cells and averaged from three independent series of experimentsconducted as described in Materials and Methods.


Taken together, these results show that the requirement forA-kinase activity in the SRF-NLS-mediated nuclear transportof either NLS-IgG conjugate or overexpressed tagged SRFprotein does not operate through the direct phosphorylation ofthe A-kinase sites present in the SRF-NLS sequence.Inhibition of A-kinase activity ubiquitously affects the pro-

cess of translocation. Inhibition of A-kinase seems to affect thenuclear translocation of SRF by a mechanism which does notinvolve direct phosphorylation of SRF itself. This observation,together with our previous finding of an indirect effect ofA-kinase on the nuclear transport of MyoD (55), led us toexamine if a requirement for A-kinase activity operates in ageneral mechanism in the nuclear import of all proteins. Wetherefore investigated the effect of inhibiting A-kinase activityon the import of SV40-NLS, since this NLS has served as amodel sequence in the analysis of the process of nuclear im-port. Growing Hs-68 cells were coinjected with SV40-NLS-conjugated IgGs and either the inhibitory peptide PKIm (Fig.5A and B), a noninhibitory peptide, PKIm(15-24) (panels Cand D), or the A-kinase regulatory subunit RII (panels E andF). One hour after injection, in cells injected with either PKIm(Fig. 5A) or the RII subunit (panel E), the SV40-NLS-conju-gated IgGs were exclusively localized in the cytoplasm (panelsB and F), clearly showing the requirement for A-kinase activityin the nuclear transport by SV40-NLS. Controls using theinactive form of protein kinase, PKI(15-24) (Fig. 5C), showedno modification of the nuclear localization of SV40-NLS (pan-el D). Since SV40-NLS (PKKKRK) does not contain any plau-sible phosphorylation sites, these observations further illus-trate that the modulation of nuclear import by A-kinase doesnot involve direct phosphorylation of the imported nuclearprotein. As such, A-kinase activity may be required to phos-phorylate other components of the nuclear import machinery.To extend this observation, we have examined the effect of

inhibiting A-kinase activity on the nuclear accumulation ofcyclin A, a nuclear protein that does not contain any potentialphosphorylation sites for A-kinase. Purified human cyclin Aprotein (17) was injected in the cytoplasm of Hs-68 cells duringthe G1 phase together with either inert mouse marker antibodyalone, PKIm, or inactive PKIm(15-24). Cells were fixed 30 minthereafter and stained for the presence and localization ofinjected cyclin A and marker antibody. As described before(17), 30 min after injection, the injected cyclin A protein isfound in the nuclei. In contrast, inhibition of A-kinase activitywith PKIm restricted the cyclin A protein staining to the cyto-plasm, showing that A-kinase activity is also required for cyclinA nuclear localization. The microinjection of PKIm(15-24),however, had no effect on the nuclear localization of cyclin A(data not shown). Taken together, these results support therequirement of A-kinase in a general mechanism of nucleartranslocation control.

DISCUSSION

In this report, we show that the active nuclear import of SRFis directed by a unique NLS located within the N-terminalregion of the protein, outside the DNA-binding domain. Theefficiency of nuclear translocation of SRF is modulated by thelevel of A-kinase activity without involving direct phosphory-

lation of SRF by A-kinase. Rather, A-kinase forms part of amore general mechanism of nuclear translocation control,since the nuclear transport of both the SV40-NLS-conjugatedIgGs (an NLS sequence without any phosphorylation sites) andcyclin A protein (which contains no potential A-kinase phos-phorylation sites) is also prevented by inhibition of A-kinase.Taken together, our data show that active nuclear targeting ofproteins through NLS involves a mechanism which requiresthe activity of A-kinase.SRF contains an active NLS outside its DNA-binding do-

main. SRF is a constitutive nuclear protein throughout the cellcycle in various mammalian cell lines (13), having a locationconsistent with its function as a transcriptional regulator. Sinceits apparent molecular mass (67 kDa) is too great to permitrandom diffusion through the nuclear pore, we would antici-pate there to be one or more sequences responsible for nuclearlocalization within SRF. Through microinjection, we have as-certained that only one of the four clusters of basic amino acidspresent along the SRF sequence acts in the nuclear localization

FIG. 5. Nuclear localization driven by SV40-NLS depends upon A-kinaseactivity. Growing Hs-68 cells were injected with SV40-NLS-IgG conjugates andmouse marker IgGs together with the A-kinase inhibitory peptide (PKIm) (Aand B); a noninhibitory peptide PKIm(15-24) (C and D); or the regulatorysubunit of A-kinase, RII (E and F). Cells fixed after 1 h were stained as describedin Materials and Methods. Panels A, C, and E show detection of the mousemarker antibody, and panels B, D, and F show detection of the carrier-conju-gated rabbit IgG.

FIG. 4. Nuclear localization driven by SRF-NLS depends upon A-kinase activity. (a) Quiescent Hs-68 cells were injected either with SRF-NLS-IgG conjugates alone(A) or with the catalytic subunit of A-kinase (C) and mouse marker antibody. After 30 min, cells were fixed and stained for mouse marker antibody (A and C) andIgG localization (B and D). (b) Growing Hs-68 cells were injected with SRF-NLS-IgG conjugates and the A-kinase inhibitory peptide (PKIm) (A and B); noninhibitorypeptide, PKIm(15-24) (C and D); or the regulatory subunit of A-kinase, RII (1.4 mg/ml) (E and F). One hour after injection, cells were fixed and stained for detectionof the mouse marker antibody (A, C, and E) or the NLS-peptide-conjugated rabbit IgG (B, D, and F).


of SRF. This NLS spans from amino acids 95 to 100 and ishighly efficient in promoting nuclear transport. It functions aseffectively as the SV40 large-T antigen NLS sequence withrespect to the rate of nuclear import when coupled to IgG,even when as few as one or two NLS peptides are coupled toan IgG molecule. This high level of efficiency is consistent withthe fact that SRF contains only one functional NLS and pre-vious reports that there is a direct link between the number ofNLS present in a given protein, their relative efficiencies, andthe extent of nuclear import (44).The nuclear localization domain of SRF we have identified

(amino acids 95 to 100) is present in a portion of the proteinexclusive to SRF, since it is not found in the various membersof the SRF-related family (human and Xenopus MEF-2,RSRFC4, RSRFC9, and RSRFC2) (3, 40) or the yeast homo-logues (MCM1 and ARG80) (54). As such, the nuclear trans-port mechanism of SRF may be implicated in specific aspectsof SRF protein function, particularly those processes in whichrapid nuclear localization would be highly important. For ex-ample, SRF has been implicated in the rapid down regulationof c-fos, an event thought to involve newly synthesized SRF(36). Clearly, under such circumstances, when the rate of nu-clear import is critical to the temporal regulation of cellular foslevels, the rapid nuclear targeting of SRF would be of impor-tance. Finally, it is interesting that SRF-NLS does not matchwith either the single-motif consensus sequence (16) or thebipartite consensus drawn previously from the identification ofother NLS (5). This raises the possibility that such a hybridstructure may represent another, third class of NLS, the effi-ciency and fidelity of which are similar to those of the other twoclasses.The process of nuclear translocation requires A-kinase.

Various reports have implied that nuclear localization may bemodulated by specific variations in kinase activity, with phos-phorylation sites often found adjacent to NLS and in somecases modulating their efficiency. In particular, A-kinase hasbeen implicated in modulation of the subcellular localizationfor a variety of transcription factors. Indeed, a consensus sitefor A-kinase is found near the NLS in several members of therel family. Mutation of this site to a nonphosphorylatable res-idue demonstrated that phosphorylation by A-kinase was im-plicit in the translocation of c-rel from the cytoplasm to thenucleus (36). The nuclear localization of at least four otherproteins appears to be positively regulated by A-kinase. Thetranscription factor NFIL-6 binds to the SRE of the c-fospromoter in response to forskolin treatment. This induciblebinding correlates with direct increases in phosphorylation andnuclear localization (33). The nuclear accumulation of thec-fos protein requires continuous stimulation by growth factorsin a cAMP-dependent manner (45). In addition, the activationand nuclear translocation of the Drosophila Dorsal gene prod-uct requires phosphorylation by A-kinase (39). Recently, wehave shown that the nuclear import of MyoD is positivelymodulated by A-kinase through a mechanism independent ofMyoD phosphorylation (55). Here, we have shown that A-ki-nase-dependent phosphorylation is a necessary event in theprocess of nuclear localization of SRF. In addition, two newand important features concerning this requirement for A-ki-nase activity have been demonstrated in this study. Firstly,phosphorylation of SRF by A-kinase is not involved in deter-mining the cytolocale of SRF, even though the site KRSLSpresent in SRF-NLS is a potential phosphorylation site forA-kinase. Secondly, the role of A-kinase in the mechanism ofnuclear translocation described here is part of a more generalprocess and not exclusive to SRF. This last point is clearlyillustrated by the consequences of A-kinase inhibition on the

nuclear translocation of both cyclin A and the SV40-NLS-conjugated IgGs.That the blocking of nuclear transport is truly the conse-

quence of A-kinase inhibition is demonstrated by the repro-ducibility of these results by using two different approaches.Nuclear transport was impeded not only by microinjection ofthe inhibitor peptide PKIm, which might have interfered withsome events in the nuclear transport process other than inhi-bition of A-kinase, but also by the specific regulatory subunit ofA-kinase, RII, effectively ruling out possible extraneous effectsof PKIm. In addition, we have shown that nuclear transloca-tion occurs at a lower rate under two different physiologicalconditions when the level of cAMP (and consequently of A-kinase activity) was reduced during the cell cycle. We haveshown this level to be at its lowest in quiescent cells, beingactivated 1.5- to 2-fold upon serum stimulation (Table 1) andreduced again at the onset of mitotic induction (23). This latterreport implied that a basal level of A-kinase is maintainedthroughout the cell cycle and required to drop to enable cellsto transit from the interphase state to mitosis. We have shownthat the rate of nuclear import of either SRF-NLS- or SV40-NLS-conjugated IgGs is notably reduced in quiescent cells(Fig. 1 and 4) and that an increase in cytoplasmic staining forendogenous SRF is clearly detectable in cells enteringprophase (Fig. 3). Moreover, a partially cytoplasmic distribu-tion for normally nuclear proteins (such as proliferating cellnuclear antigen, c-myc, and c-fos) has been previously ob-served in quiescent cells (57). Since it has been reported beforethat cAMP levels are reduced in quiescent cells (60) and weshow here the same result for two different cell lines (Table 1),these data further support the link between a low level ofA-kinase activity and a reduced efficiency in nuclear import.In addition to a requirement for A-kinase activity in the

active nuclear translocation of proteins, our data show thatinhibition of A-kinase results in the rapid exit of SRF from thenucleus. This effect differs from what we observed previouslywith MyoD protein, which remained in the nucleus upon mi-croinjection of A-kinase inhibitors (55), suggesting that the twoprocesses of nuclear import and nuclear retention or exit maybe functionally distinguished. However, a more detailed anal-ysis of the implication of A-kinase in the mechanism of nuclearretention or exit of MyoD and SRF would be required to drawa conclusion about this process.A-kinase-dependent process is a more general mechanism

modulating nuclear import. SV40-NLS was one of the firstNLS to be identified and has subsequently been used to modelthe process of nuclear translocation. Since SV40-NLS does notcontain any phosphorylation sites (serine, threonine, or ty-rosine), our observations using RII or PKIm to prevent nucleartranslocation driven by SV40-NLS definitely establish that A-kinase activity is required in the general mechanism whichbrings about active nuclear import.The present study provides the first evidence for the general

implication of a protein kinase, namely, A-kinase, in the con-trol of cytonuclear translocation of proteins. Whether this im-plication of A-kinase control reflects a direct effect on theimport process remains to be shown, since we cannot excludethe possibility that A-kinase may act through another kinase orphosphatase, itself directly implicated in the control of theimport machinery proteins. Nuclear protein import is a multi-step process that results in transport of the protein from itscytoplasmic site of synthesis into the nucleus (48). An earlystep in nuclear protein import involves the specific recognitionof the NLS in the protein by NBPs, followed either by thetransport of this complex through nuclear pores into the nu-cleus in an ATP-dependent manner or by the interaction of


this complex with the pores, with subsequent translocation ofthe nuclear protein into the nucleus and recycling of the cyto-solic receptor. NBPs have been characterized for various or-ganisms (26, 30, 46, 47, 49, 50, 51). At least two of them,NBP70 and Nopp140, are phosphorylated in vivo, an eventrequired for their interaction with NLS (31, 51). In addition, byexamining NBP sequences, one can easily identify severalphosphorylation sites for different kinases, including A-kinase,again supporting the possibility that A-kinase-dependent phos-phorylation may control their activity (31). In addition, someproteins present in the nuclear pore may also be targets forA-kinase regulation (4). Therefore, different hypotheses can beproposed to account for the effect of A-kinase on nucleartransport. A-kinase-dependent phosphorylation could conceiv-ably modulate components of the nuclear pore matrix, NBPs,and/or some regulatory kinase or phosphatase. An importantquestion to be addressed is the determination of the potentialtarget(s) involved in the A kinase-dependent nuclear localiza-tion process of SRF and other actively transported nuclearfactors. The identification of SRF-NLS is an important step inthis direction, since the rapid nuclear translocation of thisprotein may be a useful model. Experiments to elucidate thenature of proteins potentially interacting with SRF-NLS andSV40-NLS are currently under way.

ACKNOWLEDGMENTS

We are grateful to J. Demaille and J. C. Cavadore for support of thiswork. We thank R. Treisman (Imperial Cancer Research Fund) forproviding us with SRF cDNA and J. Scott (Vollum Institute, Portland,Oreg.) for type II regulatory subunit. We also thank Jean Mery forpeptide synthesis and Jean-Paul Capony for mass spectroscopy analy-sis.This work was supported by the Association pour la Recherche

contre le Cancer and the Association Francaise contre les Myopathies.

REFERENCES1. Adam, S. A., T. J. Lobl, and L. Gerace. 1989. Identification of specific bindingproteins for a nuclear localization signal. Nature (London) 337:276–279.

2. Bornslaeger, E. A., P. Mattei, and R. M. Schultz. 1986. Involvement ofcAMP-dependent protein kinase and protein phosphorylation in regulationof mouse oocyte maturation. Dev. Biol. 114:453–462.

3. Chambers, A. E., S. Kotecha, N. Towers, and T. J. Mohun. 1992. Muscle-specific expression of SRF-related genes in the early embryo of Xenopuslaevis. EMBO J. 11:4981–4991.

4. Davis, L. I., and G. R. Fink. 1990. The NUP1 gene encodes an essentialcomponent of the yeast nuclear pore complex. Cell 61:965–978.

5. Dingwall, C., and R. A. Laskey. 1986. Protein import into the cell nucleus.Annu. Rev. Cell Biol. 2:367–390.

6. Dingwall, C., and R. A. Laskey. 1991. Nuclear targeting sequences—a con-sensus? Trends Biol. Sci. 16:478–481.

7. Dingwall, C., J. Robbins, S. M. Dilworth, B. Roberts, and W. B. Richardson.1988. The nucleoplasmin nuclear localisation sequence is larger and morecomplex than that of SV40 large T antigen. J. Cell Biol. 107:841–849.

8. Fernandez, A., J. Mery, M. Vandromme, M. Basset, J. C. Cavadore, and N.Lamb. 1991. Effective intracellular inhibition of the cAMP-dependent pro-tein kinase by microinjection of a modified form of the specific inhibitorpeptide PKI in living fibroblasts. Exp. Cell Res. 195:468–477.

9. Garcia-Bustos, J., J. Heitman, and M. N. Hall. 1991. Nuclear protein local-ization. Biochim. Biophys. Acta 1071:83–101.

10. Gashler, A. L., S. Swaminathan, and V. P. Sukhatme. 1993. A novel repres-sion module, an extensive activation domain, and a bipartite nuclear local-ization signal defined in the immediate-early transcription factor Erg-1. Mol.Cell. Biol. 13:4556–4571.

11. Gauthier-Rouviere, C., M. Basset, J. M. Blanchard, J. C. Cavadore, A.Fernandez, and N. J. C. Lamb. 1991. Casein kinase II induces c-fos expres-sion via the serum response element pathway and p67SRF phosphorylation inliving fibroblasts. EMBO J. 10:2921–2930.

12. Gauthier-Rouviere, C., Q. Q. Caı, N. Lautredou, A. Fernandez, J. M. Blan-chard, and N. J. C. Lamb. 1993. Expression and purification of the DNAbinding domain of SRF: SRF-DB, a part of a DNA binding protein whichcan act as a dominant negative mutant in vivo. Exp. Cell Res. 209:208–215.

13. Gauthier-Rouviere, C., J. C. Cavadore, J. M. Blanchard, N. J. C. Lamb, andA. Fernandez. 1991. p67SRF is a constitutive nuclear protein implicated in themodulation of genes required throughout the G1 period. Cell Regul. 2:575–

588.14. Gauthier-Rouviere, C., A. Fernandez, and N. J. C. Lamb. Unpublished data.15. Gilman, M. Z., R. N. Wilson, and R. A. Weinberg. 1986. Multiple protein

binding sites in the 59-flanking region regulate c-fos expression. Mol. Cell.Biol. 6:4305–4316.

16. Gomez-Marquez, J., and F. Segade. 1988. Prothymyosin alpha is a nuclearprotein. FEBS Lett. 226:217–219.

17. Girard, F., U. Strausfeld, A. Fernandez, and N. J. C. Lamb. 1991. Cyclin Ais required for the onset of DNA replication in mammalian fibroblasts. Cell67:1169–1179.

18. Guihem, S., and N. J. C. Lamb. Unpublished data.19. House, C., and B. E. Kemp. 1987. Protein kinase C contains a pseudosub-

strate prototope in its regulatory domain. Science 238:1726–1728.20. Janknecht, R., R. A. Hipskind, T. Houthaeve, A. Nordheim, and H. G.

Stunnenberg. 1992. Identification of multiple SRF N-terminal phosphoryla-tion sites affecting DNA binding properties. EMBO J. 11:1045–1054.

21. Kalderon, D., W. D. Richardson, A. F. Markham, and A. E. Smith. 1984.Sequence requirements for nuclear location of simian virus 40 large T anti-gen. Nature (London) 311:33–38.

22. Klamut, H. J., S. B. Gangopadhyay, R. G. Worton, and P. N. Ray. 1990.Molecular and functional analysis of the muscle-specific promoter region ofthe Duchenne muscular dystrophy gene. Mol. Cell. Biol. 10:193–205.

23. Lamb, N., J. C. Cavadore, J. C. Labbe, R. A. Maurer, and A. Fernandez.1991. Inhibition of cAMP-dependent protein kinase plays a key role in theinduction of mitosis and nuclear envelope breakdown in mammalian cells.EMBO J. 10:1523–1533.

24. Lamb, N. J. C., A. Fernandez, A. M. Conti, R. Adelstein, D. B. Glass, W. J.Welch, and J. Feramisco. 1988. Regulation of actin microfilament integrity inliving non muscle cells by the cAMP-dependent protein kinase and themyosin light chain kinase. J. Cell Biol. 106:1955–1971.

25. Landford, R. E., P. Kanda, and R. C. Kennedy. 1986. Induction of nucleartransport with a synthetic peptide homologous to the SV40 large T antigentransport signal. Cell 46:575–582.

26. Li, R., Y. Shi, and J. O. Thomas. 1992. Intracellular distribution of a nuclearlocalization signal binding protein. Exp. Cell. Res. 202:355–365.

27. Maller, J., and E. G. Krebs. 1977. Progesterone stimulated cell division inXenopus oocytes: induction by regulatory subunit and inhibition by catalyticsubunit of adenosine 3959 monophosphate dependent protein kinase. J. Biol.Chem. 252:1712–1718.

28. McClure, D. B., M. J. Hightower, and W. C. Topp. 1982. Effect of SV40transformation on the growth-factor requirements of the rat embryo cell lineREF-52 in serum-free medium. Cold Spring Harbor Conf. Cell Proliferation9:345–364.

29. McVey, D., L. Brizuela, I. Mohr, D. R. Marshak, Y. Gluzman, and D. Beach.1989. Phosphorylation of large tumour antigen by cdc2 stimulates SV40DNA replication. Nature (London) 341:503–507.

30. Meier, U. T., and G. Blobel. 1990. A nuclear localization signal bindingprotein in the nucleolus. J. Cell Biol. 111:2235–2245.

31. Meier, U. T., and G. Blobel. 1992. Nopp140 shuttles on tracks betweennucleolus and cytoplasm. Cell 70:127–138.

32. Meijer, L., W. Dostmann, H. H. Genieser, E. Butt, and B. Jastorff. 1989.Starfish oocyte maturation: evidence for a cyclic AMP-dependent inhibitorypathway. Dev. Biol. 133:58–66.

33. Metz, R., and E. Ziff. 1991. cAMP stimulates the C/EBP-related transcrip-tion factor rNFIL-6 to translocate to the nucleus and induce c-fos transcrip-tion. Genes Dev. 5:175–176.

34. Minty, A., and L. Kedes. 1986. Upstream regions of the human cardiac actingene that modulate its transcription in muscle cells: presence of an evolu-tionarily conserved repeat motif. Mol. Cell. Biol. 6:2125–2136.

35. Mosialos, G., P. Hamer, A. J. Capobianco, R. A. Laursen, and T. D. Gilmore.1991. A protein kinase-A recognition sequence is structurally linked totransformation by p59v-rel and cytoplasmic retention of p68c-rel. Mol. Cell.Biol. 11:5867–5877.

36. Nakabeppu, Y., and D. Nathans. 1989. The basic region of fos mediatesspecific DNA binding. EMBO J. 8:3833–3841.

37. Nasmyth, K., G. Adolf, D. Lydall, and A. Seddon. 1990. The identification ofa second cell cycle control on the HO promoter in yeast: cell cycle regulationof SWI5 nuclear entry. Cell 62:631–647.

38. Norman, C., M. Runswick, R. Pollock, and R. Treisman. 1988. Isolation andproperties of cDNA clones encoding SRF, a transcription factor that bindsto the c-fos serum response element. Cell 55:989–1003.

39. Norris, J. L., and J. L. Manley. 1992. Selective nuclear transport of theDrosophila morphogen dorsal can be established by a signaling pathwayinvolving the transmembrane protein Toll and protein kinase A. Genes Dev.6:1654–1667.

40. Pollock, R., and R. Treisman. 1991. Human SRF-related proteins: DNA-binding properties and potential regulatory targets. Genes Dev. 5:2327–2341.

41. Prywes, R., and R. G. Roeder. 1987. Purification of the c-fos enhancerbinding protein. Mol. Cell. Biol. 7:3482–3489.

42. Rihs, H. P., D. A. Jans, H. Fan, and R. Peters. 1991. The rate of nuclearcytoplasmic transport is determined by the casein kinase II site flanking thenuclear localisation sequence of the SV40 T-antigen. EMBO J. 10:633–639.


43. Rivera, V. M., C. K. Miranti, R. P. Misra, D. D. Ginty, R. H. Chen, J. Blenis,and M. Greenberg. 1993. A growth factor induced kinase phosphorylates theserum response factor at a site that regulates its DNA-binding activity. Mol.Cell. Biol. 13:6260–6273.

44. Roberts, B. L., W. D. Richardson, and A. E. Smith. 1987. The effect ofprotein context on nuclear localization signal function. Cell 50:465–475.

45. Roux, P., J. M. Blanchard, A. Fernandez, N. Lamb, P. Jeanteur, and M.Piechaczyk. 1990. Nuclear localization of c-fos, but not v-fos proteins, isdependent upon extracellular stimulation. Cell 63:341–351.

46. Shi, Y., and J. O. Thomas. 1992. The transport of proteins into the nucleusrequires the 70-kilodalton heat shock protein or its cytosolic cognate. Mol.Cell. Biol. 12:2186–2192.

47. Silver, P., I. Salder, and M. A. Osborne. 1989. Yeast proteins that recognizenuclear localization sequence. J. Cell. Biol. 109:983–989.

48. Silver, P. A. 1991. How proteins enter the nucleus. Cell 64:489–497.49. Stojack, U., M. A. Bossie, K. Van Zee, A. M. Whalen, and P. A. Silver. 1993.

Analysis of conserved binding proteins for nuclear localization sequences. J.Cell Sci. 104:89–95.

50. Stojack, U., M. Osborne, T. Kurihara, and P. Silver. 1991. A yeast proteinthat binds nuclear localization signals: purification, localization and antibodyinhibition of binding activity. J. Cell Biol. 113:1243–1254.

51. Stojack, U., and P. A. Silver. 1992. A conserved phosphoprotein that spe-cifically binds nuclear localization sequences is involved in nuclear import. J.Cell Biol. 117:473–482.

52. Treisman, R. 1985. Transient accumulation of c-fos RNA following serumstimulation requires a conserved 59 element and c-fos 39 sequence. Cell

42:889–902.53. Treisman, R. 1987. Identification and purification of a polypeptide that binds

to the c-fos serum response element. EMBO J. 6:2711–2717.54. Treisman, R., and G. Ammerer. 1992. The SRF and MCM1 transcription

factors. Curr. Opin. Genet. Dev. 2:221–226.55. Vandromme, M., G. Carnac, C. Gauthier-Rouviere, D. Fesquet, N. Lamb,

and A. Fernandez. 1994. Nuclear import of the myogenic factor MyoDrequires cAMP-dependent protein kinase activity but not the direct phos-phorylation of MyoD. J. Cell Sci. 107:613–620.

56. Vandromme, M., C. Gauthier-Rouviere, G. Carnac, N. J. C. Lamb, and A.Fernandez. 1992. Serum response factor p67SRF is expressed and requiredduring myogenic differentiation of both mouse C2 and L6 muscle cell lines.J. Cell Biol. 6:1486–1500.

57. Vriz, S., J. M. Lemaitre, M. Leibovici, N. Thierry, and M. Mechali. 1992.Comparative analysis of the intracellular localization of c-Myc, c-Fos, andreplicative proteins during cell cycle progression. Mol. Cell. Biol. 12:3548–3555.

58. Wilson, L. A., H. L. Niman, R. A. Hougten, A. R. Cherenson, M. L. Connolly,and R. A. Lerner. 1984. The structure of an antigenic determinant in aprotein. Cell 37:767–778.

59. Yamasaki, L., P. Kanda, and R. E. Landford. 1989. Identification of fournuclear transport signal-binding proteins that interact with diverse transportsignals. Mol. Cell. Biol. 9:3028–3036.

60. Zeiling, C. E., and N. D. Goldberg. 1977. Cell-cycle related changes of39:59-cyclic GMP levels in Novikoff hepatoma cells. Proc. Natl. Acad. Sci.USA 74:1052–1056.


the serum response factor nuclear localization signal: general

Documents