The Activation Domain of a Hormone Inducible HTLV-1 Rex Protein Determines Colocalization with the Nuclear Pore

EXPERIMENTAL CELL RESEARCH ARTICLE NO.

233, 363 –371 (1997)

EX973562

The Activation Domain of a Hormone Inducible HTLV-1 Rex Protein Determines Colocalization with the Nuclear Pore S. Rehberger,*,1 F. Gounari,*,1 M. DucDodon,† K. Chlichlia,* L. Gazzolo,† V. Schirrmacher,* and K. Khazaie*,‡,2 *Department of Tumour Immunology, German Cancer Research Centre, Im Neuenheimer Feld 280, Heidelberg 69120, Germany; ‡Section for Molecular Diagnosis and Therapy, Department of Surgery, German Cancer Research Centre, Im Neuenheimer Feld 116, Heidelberg 69120, Germany; and †Laboratory of Molecular and Cellular Immunovirology, UMR 5537-CNRS-Universite´ Lyon1, Faculte de Me´decine Lyon-RTH Lae¨nnec, Rue Guillaume Paradin, 69372 Lyon Cedex 08, France

Human T-cell leukemia/lymphoma virus type 1 (HTLV-1) Rex is an essential regulatory protein that acts at the posttranscriptional level to promote expression of unspliced and singly spliced genes of the virus. Rex functions have been attributed to at least three separate domains of the protein determining nuclear/ nucleolar accumulation and RNA binding (overlapping), multimerization, and nuclear export of Rex-responsive RNA. The steady-state intracellular localization of functional Rex molecules is mainly nucleolar. Fusions of wild-type Rex and the ligand binding domain of human estrogen receptor (ER) produced conditional molecules (ERRex and ERalaRex), which remained cytoplasmic in the absence of hormone and in response to hormone colocalized with the nuclear pore complex (NPC). These molecules induced in a hormone-dependent manner the expression of a Rex reporter plasmid and of the HTLV-1 Env protein and fusion of Env expressing cells. In contrast, activation domain mutants (ERRexD and ERRexGly) translocated from the cytoplasm and acquired a diffuse nuclear localization. These mutants did not associate with the NPC and failed to show any of the expected Rex functions. Rex functions were perturbed by inactivating the RNA binding domain (mutant ERM2) or the oligomerization domain (mutant ERM7). However, these two mutant fusion proteins exhibited a hormone-dependent NPC colocalization. These observations provide in vivo evidence that intranuclear translocation of intact Rex to the NPC is dependent exclusively on a functional activation domain and is not influenced by binding to the target RNA. q 1997 Academic Press

INTRODUCTION

A major function of the Rex protein of human T-cell leukemia virus (HTLV) is posttranscriptional regula1

Equally contributing authors. To whom correspondence and reprint requests should be addressed. 2

tion of splicing and export of unspliced or singly spliced pre-mRNAs encoding the viral structural genes. The mode of action, as well as the structural organization, of Rex has strong similarities to that of the Rev protein of HIV. Both proteins polymerize and contain an RNA binding domain and an interchangeable essential region known as the ‘‘activation domain,’’ which is speculated to interact with cellular factors. Direct binding to a Rex-responsive element within the 3* untranslated region of viral RNAs is necessary for the translocation of the corresponding viral mRNAs to the cytoplasm and their expression [1–3]. Two possible models have been put forward to explain the mode of action of these viral regulators. In one, Rex and/or Rev promote transport of unspliced viral RNAs through dissociation of splicesomes, before the splicing reaction is complete [4–6]. In the other, they establish a rapid, splicing-independent pathway for pre-mRNA transport [7 –11]. Other studies have shown a link between presence of splicing signals and the Rex or Rev dependence of viral transcripts, suggesting that export and splicing processes may be coupled [5, 12, 13]. The role of the activation domain in the nuclear export function was indicated recently by in vitro studies with limited sequences that define the activation domains of HIV Rev or HTLV-1 Rex. Thus, Rev activation domain conjugates with bovine serum albumin (BSAR) were actively exported from microinjected nuclei of cultured cells [14]. Biochemical approaches have identified cellular proteins interacting with the Rev and Rex activation domain sequences, which bear structural similarities to GLFG motif containing nucleoporins [15 –17]. However, targeting of the intact Rex (or Rev) protein in cells to the NPC and the extent to which RNA binding or oligomerization may influence this translocation have not been demonstrated. In the present investigation conditional versions of Rex were generated by fusion of the HTLV-1 rex cDNA, or mutants thereof, with the hormone binding domain of the human estrogen receptor cDNA (ERrex). The ER domain used here (HE14), although sufficient to confer

363

AID ECR 3562

/

6i20$$$401

05-28-97 12:02:40

0014-4827/97 $25.00 Copyright q 1997 by Academic Press All rights of reproduction in any form reserved.

ecal

364

REHBERGER ET AL.

dominant hormonal regulation on fusion proteins, lacks the DNA binding and nuclear localization signal (NLS) of the intact steroid receptor [18]. Although the mechanism underlying ‘‘protein inactivation’’ function of the steroid receptors is not completely understood, steroid receptor fusions have been extensively used to generate hormone inducible proteins. Inducibility of Rex function was analyzed in transient transfection assays, through apparent inhibition of splicing which led to the expression of an intron-encoded CAT cDNA, as well as through estrogen-dependent expression of HTLV-1 env and formation of syncytia. Hormone-induced activation of the fusion molecules was coincident with their change of intracellular distribution from the cytoplasm to the NPC. Colocalization with the NPC was dependent on an intact Rex activation domain, as disruption of this region led to diffusion

FIG. 1. Schematic representation of the rex-derived constructs and Western blot analysis of their synthesis in transiently transfected HeLa cells. (A) Schematic representation of the constructs used in this study. Black box represents the CMV promoter which drives the expression of wild-type rex and the seven ER fusions. The oval box represents the estrogen receptor moiety, while the empty box represents the rex sequences. Modifications of the rex sequence and the precise amino acid position of these modification in the protein are indicated over the relevant construct. The RNA binding domain mutant ERM2 contains a substitution of Asp and Leu (DL) for Lys and Arg, residues 14 and 15. The oligomerization domain mutant ERM7 contains a substitution of Asp and Leu (DL) for Tyr and Trp, residues 64 and 65. The activation domain mutants ERrexGly and ERrexD contain substitution of four Gly residues for residues 90– 93 or a deletion of residues 78–94, respectively. The SV40 polyadenylation signal is represented by an empty circle. The constructs were assembled in the expression vector pRK7 (Dr. David Goeddel, unpublished) (see Materials and Methods). (B) Western blot analysis of cells transiently transfected with the rex constructs. Lysates of 2 1 105 HeLa cells transfected with 10 mg DNA of the constructs described in (A) as indicated above the blot were probed with a rabbit polyclonal anti-Rex immune serum against a fused histidine–Rex protein (S. Hemaia and M. DucDodon, submitted for publication). A 27-kDa species corresponding to wild-type Rex was detected in the control cells, while cells transfected with the fusion molecules synthesized a 62-kDa species corresponding to the fusion of ER with Rex and mutants thereof. Molecular mass values of the markers are given in kDa.

AID ECR 3562

/

6i20$$$401

05-28-97 12:02:40

FIG. 2. Hormone inducible and constitutive Rex-dependent induction of CAT activity from Rex reporter plasmid. A schematic representation of the Rex reporter plasmid is shown in the upper part of the figure. SD, splice donor site; SA, splice acceptor site. CAT activity in extracts of cells transiently transfected with the indicated constructs was assayed. Absence (0) or presence (/) of estrogen in culture medium is indicated. The percentage of acetylation was quantitated with a linear TLC analyzer. Fold induction is the ratio of percentage of acetylation by extracts of cells cultured with compared to without hormone. Transient transfection of cells with Rex results in 93.3-fold stimulation of CAT activity (28% acetylation) with respect to that of cells transfected with the ER moiety (0.3% acetylation).

of the hybrid proteins into the nucleoplasm instead of the NPC. Translocation to the NPC was detected in the absence of RxRE encoding transcripts, suggesting that binding to target RNA was not required. These observations indicate that the fusion of ER and Rex dramatically changes the intracellular localization of the molecules in a Rex activation domain-dependent manner. The steady-state accumulation of the fusion proteins at the NPC reflects a change in the rate-limiting step of Rex intracellular transport, providing in vivo evidence that targeting of the HTLV-1 Rex to the NPC in intact human cells is determined by the activation domain alone. RESULTS

Expression of Rex and Fusions of Human Estrogen Receptor with Rex To generate inducible forms of HTLV-1 Rex, sequences encoding the ligand binding domain of the human estrogen receptor (ER) were introduced upstream and in frame with a series of wild-type and mutated forms of the rex cDNA (Fig. 1A). The previously described HE14 domain of ER (aa 282–595) was used in this study [18]. This domain, which includes the hormone binding and dimerization domains of the ER, does not contain a nuclear localization signal (NLS)

ecal

FIG. 3. Rex-dependent expression of HTLV-1 Env protein and formation of multinucleated syncytia. (A) Schematic representation of the Env expressing plasmid, MT-env, is shown in the upper part of the figure. Lysates of HeLa cells transiently transfected with MT-env in combination with PRK7-rex (env / rex) or MT-env alone (env) were subjected to Western blot using an immune serum obtained from an HTLV-1 seropositive individual. The 67- and 21-kDa proteins detected in the (env / rex) lane correspond to the Env precleavage and the Env transmembrane cleavage products, respectively. Molecular mass values of the markers are given in kDa. (B) The number of syncytia in transiently transfected cells counted as described under Materials and Methods is shown in the y axis. In the x axis the constructs corresponding to each histogram as well as the presence (/ E) or the absence of estrogen are indicated. (env) denotes the MT-env construct, and controls are mock-transfected HeLa cells. Note that the number of syncytia obtained with the constitutively active Rex is similar to that obtained with the conditional ERalarex or ERrex in the presence of hormone. No syncytia were detected with ERrexGly and ERrexD. (Ca) Immunofluorescence staining of Env in HeLa cells 48 h after cotransfection of MT-env and ERalarex in the presence of estrogen. May– Grunwald staining of HeLa cells, (b) mock-transfected HeLa cells, (c) cells transfected with MT-env and ERrex in the absence of estrogen, and (d) HeLa cells transfected as in (c) but in the presence of 1 mM estrogen.

AID ECR 3562

/

6i20$$3562

05-28-97 12:02:40

ecal

366

REHBERGER ET AL.

and shows only weak preference for nuclear accumulation. Modified versions of the wild-type ERrex construct included, (i) the introduction of a double-stranded oligonucleotide encoding for 10 alanine residues at the fusion site separating the ER from the Rex moiety (ERalarex), (ii) the mutation of the RNA binding domain (ERM2) [19], (iii) the mutation of the oligomerization domain (ERM7) [19], (iv) the substitution of the critical LSLD (residues 90–93) motif of the activation domain by 4 glycine residues (ERrexGly) or, (v) the deletion of sequences encoding 17 amino acids (residues 78–94) spanning the entire Rex activation domain (ERrexD) (Fig. 1A, and Materials and Methods). To establish that the gene products of the chimeric cDNAs described above result in the synthesis of stable fusion proteins, the plasmids were transiently transfected into HeLa cells cultured in estrogen-free medium. After 2 days, extracts prepared from the transfected cells were subjected to SDS –PAGE and Western blot analysis. Rex (Ç27 kDa) and Rex fusion proteins (Ç62 kDa) were readily detected, using a rabbit polyclonal anti-Rex immune serum (Fig. 1B). As expected, the presence of estrogen did not influence synthesis of the transfected proteins (data not shown). These constructs were used to demonstrate the posttranslational mode of activation of the Rex fusion molecules. Conditional Expression of CAT Activity from a Rex Reporter Plasmid To quantitate Rex function, a biochemical assay was employed based on the ability of Rex to promote expression of otherwise spliced transcripts. For this assay a cytomegalovirus (CMV) promoter-driven reporter plasmid was used containing the CAT gene and the Rexresponsive element flanked by HIV splice donor and acceptor sites (Fig. 2). CAT activity from this plasmid is strictly Rex dependent. Cotransfection of HeLa cells with wild-type rex and the Rex reporter plasmid resulted in high levels of constitutive CAT activity. Similar cotransfection experiments using the ERrex or ERalarex plasmids resulted in estrogen-dependent CAT activity (Fig. 2). Induction of CAT activity upon hormone treatment was 60-fold over the background with ERalarex and 21-fold with ERrex. Fusion constructs with 10 leucine residues instead of the 10 alanine residues (the same oligo cloned in reverse orientation) produced inactive molecules (data not shown). As expected, the activation domain mutants ERrexGly and ERrexD, the RNA binding domain mutant ERM2, and the oligomerization domain mutant ERM7 did not promote any CAT activity. Similarly, the ER moiety alone did not promote any CAT activity (Fig 2). Comparable results were obtained by using the estrogen agonist 4-hydroxytamoxifen, as well as by transfecting Jurkat cells (data not shown). These results suggest that the fusion to ER brought Rex activity under hor-

AID ECR 3562

/

6i20$$$401

05-28-97 12:02:40

monal control and that the previously reported Rex inactivating mutations also inactivated the fusion molecules. Hormone Inducible Expression of HTLV-1 Env and Formation of Multinucleated Syncytia The biological role of Rex is to promote expression of the HTLV-1 singly spliced envelope and unspliced gag – pol transcripts. An assay to address this function was established, based on the property of HeLa cells expressing functional Env at their cell surface to fuse with neighboring cells and form multinucleated syncytia [20]. To this end, a Rex-dependent Env expressing plasmid (MT-env) was constructed based on pMTPX [21] (Fig. 3A, see Materials and Methods). The Rex responsiveness of this construct was controlled by transient transfection assays in HeLa cells. In these experiments the full-length 67-kDa Env and the 21-kDa transmembrane cleavage products were only detectable in the presence of Rex activity (Fig. 3A). It was further shown that cotransfection of HeLa cells with ERrex or ERalarex in combination with MT-env led to estrogen-dependent expression of envelope glycoproteins as detected by immunofluorescence staining of the cells (Fig. 3C, a). The MT-env vector was subsequently used in cotransfection experiments with the ERrex, ERalarex, ERrexD, and ERrexGly fusion constructs to observe and quantitate syncytia formation. The formation of syncytia was quantitated after staining with May– Gru¨nwald dye and counting cells with more than four nuclei. Control mock-transfected HeLa cells showed no syncytia formation (Figs. 3B and 3C, b). Cells transfected with the ERrex (Figs. 3B and 3C, c) and ERalarex (data not shown) in the absence of hormone exhibited a low level of background syncytia. Hormone stimulation of ERrex (Figs. 3B and 3C, d)- and ERalarex (data not shown)-transfected HeLa cells led to the formation of large multinucleated syncytia. Conditional formation of syncytia with ERrex (Fig. 3C) and ERalarex (data not shown) were at comparable numbers to those obtained with wild-type rex. The env encoding vector alone caused low to undetectable levels of fusion, compared to untransfected cells. Neither env expression nor syncytia formation were seen with cells expressing the activation domain mutants ERRexGly or ERRexD, confirming that these fusion proteins were, as expected, constitutively inactive (Fig. 3B). These observation provide independent evidence that the fusion to ER brought normal functions of HTLV-1 Rex under hormonal control. Hormone Activation Dramatically Alters the Subcellular Distribution of the ERRex Fusion Proteins Immunofluorescence staining of HeLa cells transiently transfected with ERrex chimeric constructs re-

ecal

NUCLEAR PORE COLOCALIZATION OF INDUCIBLE HTLV-1 Rex

367

vealed that hormonal induction of ERRex fusion proteins coincided with a dramatic shift in their subcellular distribution. In the absence of estrogen, the intact ERalaRex (Fig. 4c), ERRex (Fig. 4e), RNA binding domain mutant ERM2 (Fig. 4g), and oligomerization mutant ERM7 (Fig. 4i) were excluded from the nucleus, while the activation domain mutants ERRexD (Fig. 4k) and ERRexGly (Fig. 4m) exhibited both nuclear and cytoplasmic distributions. In the presence of hormone, the functional ERalaRex (Fig. 4d) and ERRex (Fig. 4f) proteins as well as the nonfunctional mutant ERM2 (Fig. 4h) became associated with the nuclear envelope, showing a prominent punctate perinuclear staining, and appeared to be excluded from the cytoplasm. Estrogen stimulation of the inactive mutant ERM7 induced a shift from completely cytoplasmic to punctate perinuclear as well as substantial cytoplasmic staining of the fusion protein (Fig. 4j). Conversely, the inactive ERRexD and ERRexGly mutant proteins, shifted from predominantly cytoplasmic in the absence of hormone (Figs. 4k and 4m, respectively) to exclusive diffuse nuclear localization, in the presence of estrogen (Figs. 4l and 4n, respectively). The subcellular distributions of the ERRex fusion proteins were visualized in the absence of RxRE encoding transcripts. No difference in localization was seen when the CAT reporter plasmids, which encoded RxRE, were included (data not shown). Constructs with ER attached carboxy terminal to Rex produced fusion proteins that localized constitutively to the nucleoplasm (data not shown). Control rex transfected cells showed mainly nucleolar localization of the Rex protein, although a weak perinuclear staining was also detectable in some cells (Figs. 4a and 4b). These results are consistent with the fusion molecules lacking an intact activation domain translocating to and accumulating in the nucleus. An intact activation domain was the only prerequisite for targeting to the nuclear periphery. This presumably followed translocation to the nucleus, as evidenced from the functionality of the intact Rex fusion molecules. An Intact Activation Domain Is Essential for Conditional Targeting of the Rex Fusion Molecules to the NPC Confocal microscopy was used to further characterize the unexpected redistribution of the hormone-induced ERRex fusion proteins. Thus, HeLa cells transfected with ERrex, ERalarex, ERM2, ERM7, ERrexGly, and ERrexD were induced with estrogen and 24 h later processed for immunofluorescence staining and confocal microscopy. The cells were doubly stained with antiRex antibodies (Figs. 5A –5E, panels a and panel d; green staining) and the 414 monoclonal antibody recognizing several nucleoporins of the NPC (Figs. 5A–5E, panels b and panel e; red staining). NPC staining at the focal plane of the nuclear surface is revealed as

AID ECR 3562

/

6i20$$$401

05-28-97 12:02:40

FIG. 4. Subcellular distribution of Rex and ER fusion proteins in the absence or the presence of estrogen. The distribution of wildtype Rex and the ER fusion molecules was examined in transiently transfected HeLa cells by fluorescence microscopy after staining with anti-Rex antibodies. Transfected cells were cultured either in the absence of hormone (a, c, e, g, i, k, m) or in the presence of 1 mM estrogen (b, d, f, h, j, l, n) for 24 h prior to staining. Cells were transfected with rex (a and b), ERalarex (c and d), ERrex (e and f), ERM2 (g and h), ERM7 (i and j), ERrexD (k and l), and ERrexGly (m and n) plasmids. Bar, 10 mm.

numerous little dots (Figs. 5A –5D, panels b), while in focal plains cutting through the nucleus it is revealed as a punctate, rim-like staining (Figs. 5A e and 5E b).

ecal

368

REHBERGER ET AL.

The localization of the hormone-induced fusion proteins was compared to that of the NPC in the relevant focal planes. Hormone-induced ERRex (data not shown) and ERalaRex (Fig. 5A) showed extensive colocalization with the NPC at the focal plain of the nuclear surface (Figs. 5A and 5B, panel c; yellow staining). Adjacent (0.5 mm) but further intranuclear focal plains that showed a rim-like NPC staining (Fig. 5A, panel e; absence of red staining) still showed extensive ERalaRex staining in the form of numerous little dots (Fig. 5A, panels d and f; green staining), indicating that upon hormone induction ERalaRex shifts to an intranuclear NPC-associated localization. The RNA binding domain mutant ERM2 showed a strong NPC colocalization (Fig. 5B, panel c; yellow staining), while the oligomerization domain mutant ERM7 showed both colocalization with NPC and a localization in the cytoplasm (Fig. 5C, panel c; yellow and green stainings). These subcellular distributions indicate that binding of Rex to target RNA or Rex oligomerization did not influence targeting to the pore. Strikingly, upon hormone induction neither ERrexD nor ERrexGly localized at the NPC (Figs. 5D and 5E, panels c; lack of yellow staining), but rather unevenly dispersed into the nucleoplasm (Figs. 5D and 5E, panels a; green staining). The exclusion of these activation domain mutants from the NPC (Figs. 5D and 5E, panels c; absence of yellow staining) indicates that the activation domain was essential for targeting the protein to this compartment. DISCUSSION

In the present study fusion of Rex to the hormone binding domain of the human estrogen receptor was used to generate molecules that could be functionally modulated by estrogen. Two independent assays were used to show that fusion of ER to the amino terminal end of HTLV-1 Rex indeed brings Rex function under posttranslational hormonal control: a biochemical assay based on inhibition of splicing and expression of a CAT gene encoded within an intron, and a cellular assay based on Rex-dependent expression of HTLV-1 Env. In contrast, fusion of ER to the carboxy-terminal end of Rex failed to bring Rex activity under hormonal

control, while inclusion of 10 Leu residues between ER and Rex produced inactive molecules (data not shown). The RNA binding domain (ERM2), oligomerization domain (ERM7), and activation domain (ERrexD and ERrexGly) Rex mutants fused to the ER were, as expected, inactive in both the presence and the absence of hormone. All examined fusion molecules exhibited a hormonedependent translocation to the nucleus, but surprisingly none of them showed steady-state accumulation to the nucleolus. Both HTLV-1 Rex (Seiki et al., 1986) and HIV Rev have been mostly detected in the nucleolus and were shown to have affinity for nucleolar components [22, 23]. Although several other transient and second order preference subcellular sites of localization of Rev, including the nuclear periphery, have been reported [24], so far Rex function has always correlated with a steady-state accumulation in the nucleolus. Rev [25] is known to shuttle between nucleus and cytoplasm and to function even when not having a predominantly nucleolar localization [10, 26]. The steady-state NPC localization of the fusion Rex molecules is likely to be an ‘‘artifact’’ of the proximity of the ER to the Rex protein. Our observations indicate that the artifact arises because of a change in the ratelimiting step of Rex during translocation from nucleus to the cytoplasm. Indeed, fusion to the amino-terminal end of Rex was critical for producing this effect, as RexER did not show this localization. Furthermore, ERleuRex constructs with 10 leucine residues instead of the 10 alanine residues (the same oligo cloned in reverse orientation) produced inactive molecules (data not shown), indicating the critical role of the aminoterminal domain in changing the behavior of the Rex molecule. This change in the translocation rate-limiting step allowed visual analysis of targeting of Rex to the NPC, facilitating further studies on the structural requirements of Rex for nuclear export. The hormone-dependent activity of ERRex fusion molecules suggests that they are capable of entering the nucleus and mediating export of HTLV-1 LTRlinked RNAs. Indeed, the activation domain mutants ERRexD and ERRexGly responded to hormone by shifting from the cytoplasm to the nucleoplasm, indicating that import of Rex to the nucleus was not inhib-

FIG. 5. Confocal microscope images of doubly stained HeLa cell nuclei transiently transfected with the Rex fusion molecules. (A) ERalarex-transfected and estrogen-induced HeLa cell doubly stained with antibodies against Rex (a and d) and NPC (b and e); (c and f) The overlap of the two stainings showing colocalization; (a –c) Images of the focal plane near the nuclear surface showing the most prominent dot-like nucleoporin staining; (d, e, f) The images of a subsequent 0.5 mm farther inside the nucleus focal plain showing the punctate rimlike nucleoporin staining. (B) ERM2-transfected and estrogen-induced HeLa cell doubly stained with antibodies against Rex (a) and NPC (b); (c) The overlap of the two stainings, showing colocalization. (C) ERM7-transfected and estrogen-induced HeLa cell doubly stained with antibodies against Rex (a) and NPC (b); (c) The overlap of the two stainings, showing partial colocalization. (D) ERrexD-transfected and estrogen-induced HeLa cell doubly stained with antibodies against Rex (a) and NPC (b); (c) Overlap of the two stainings, showing no colocalization. (E) ERrexGly-transfected and estrogen-induced HeLa cell doubly stained for Rex (a) and NPC (b); (c) Overlap of the two stainings, showing no colocalization. The images in (E) represent an intranuclear focal plane where the Rex staining was more prominent. Bar, 10 mm.

AID ECR 3562

/

6i20$$$401

05-28-97 12:02:40

ecal

NUCLEAR PORE COLOCALIZATION OF INDUCIBLE HTLV-1 Rex

AID ECR 3562

/

6i20$$3562

05-28-97 12:02:40

ecal

369

370

REHBERGER ET AL.

ited by fusion to the ER moiety. The predominant NPC localization of fusion proteins with an intact Rex activation domain indicates that export through the pore may have become rate limiting. The exclusion of activation domain mutants from the nuclear periphery indicates that for export and therefore NPC localization an intact activation domain is needed. The Rex mutant M2 [19] was reported to distribute equally in the cytoplasm and the nucleus. NPC localization of ERM2 agrees with the export step becoming rate limiting. Since neither an intact RNA binding domain nor the presence of Rex Response Element-linked RNAs was necessary for NPC colocalization, one can conclude that binding to target RNA is an independent event from pore targeting and nuclear export. Similarly, NPC localization, and therefore presumably nuclear export, was not significantly perturbed by inactivating the Rex oligomerization domain (mutant ERM7). However, this mutant tended to stain also in the cytoplasm, suggesting that it may translocate more efficiently than its wild-type or RNA binding domain mutant counterparts. The observation invites speculation on the role of excessive oligomerization, due to the ER multimerization domain in hindering translocation out of the nucleus. Using a similar strategy, in a previous study the ligand binding domain of the glucocorticoid receptor was fused to HIV Rev to generate a glucocorticoid (GR)regulated Rev fusion protein [27]. In that case, the fusion did not interfere with the nucleolar localization of the modified Rev protein. In the work presented here, the ER moiety was attached immediately adjacent to the overlapping RNA binding/nucleolus localization domain of Rex. This may have upset the affinity of Rex for the nucleolus, helping to shift the steady-state localization of the fusion Rex proteins to the NPC. In contrast, in the case of Rev, the GR moiety was fused carboxy terminally and at least 60 amino acids apart from the Rev NLS. Indeed, fusion of ER carboxy terminal to Rex produced localization to the nuclear compartment and exclusion of the fusion molecules from the NPC (data not shown). The presented data indicate that the intranuclear targeting of HTLV-1 Rex to the NPC is exclusively dependent on a functional activation domain. Neither an intact RNA binding domain nor binding of Rex to its target RNA sequence appears to be required for translocation of Rex from the nucleoplasm to the NPC. At least in the presence of the ER multimerization domain, an intact Rex multimerization domain is also not required. Conditional targeting of HTLV-1 Rex to the NPC provides a new tool for investigating the intracellular transport of Rex and its targeting to the NPC. MATERIALS AND METHODS Plasmid constructions. The rex cDNA was cloned as a 600-bp HindIII/SmaI fragment into the pRK7 vector. For the ERrex plasmid

AID ECR 3562

/

6i20$$$401

05-28-97 12:02:40

rex cDNA [19] was PCR amplified and inserted as a 600-bp XhoI/ EcoRI fragment downstream from the HE14 [28] fragment encoding amino acids 282– 595 of the human estrogen receptor in pRK7. To generate the ERM2 and ERM7 constructs the M2 and M7 mutants [19] were PCR amplified and inserted as a 600-bp XhoI/XhoI fragment downstream from the HE14 in PRK7. To separate the two functional moieties the 5*-CG(GCA)10-3* double-stranded oligonucleotide coding for a 10-residue poly alanine linker was inserted between the ER and rex coding regions, resulting in ERalarex. The two activation domain mutants of Rex were generated by a threestep PCR mutagenesis on ERrex. In ERrexGly sequences encoding amino acid residues the 90 to 93 LSLD were changed, using the 5*CAG CTC TAC AGT TCA GGT GGC GGC GGC TCC CCT CCT TCC CC-3* and 5*-CCG CCG CCA CCT GAA CTG TAG AGC TGA GC-3 * mutagenic oligos. To generate ERrexD sequences encoding amino acid residues 78 to 94 were deleted using the 5*-GAT CAC CTG GTA CCC CAC CTC CTT CCC CAC CC-3 * and 5*-GGA GGT GGG GTA CCA GGT GAT C-3* mutagenic oligos. After construction all the expression plasmids were sequenced using rex- and ER-specific primers. The Rex reporter plasmid was derived from the pDM138 [29] (kind gift of T. Parslow) which contains the CAT cDNAs and Rexresponse sequences in between two splice sites, except that the promoter was changed from SV40 to CMV. To generate MT-env, the internal XhoI fragment deleted from the env reading frame in PMTPX [21] was repaired. Expression of tax gene was abolished by a 100-bp StuI deletion in the tax coding sequence. Expression of rex gene was inhibited by removal of four nucleotides through exonuclease digestion and blunting of the SphI site overlapping the rex AUG start codon. Transient transfections and CAT assays. HeLa cells (1 1 104 cells/ 35-mm-diameter petri dish) were cotransfected with 1 mg of the Rex reporter plasmid and 1 mg of the indicated Rex-expressing plasmid using the calcium phosphate precipitation method [30]. After 24 h, 1 mM estrogen (17b-estradiol) was added, and cells were incubated for a further 24 h at 377C. Assays for CAT activity were performed as described [31]. Acetylation of input [14C]chloramphenicol was quantitated with an automatic TLC linear analyzer (Berthold). HTLV-1 env-mediated cell fusion assay. HeLa cells were transfected with a total of 10 mg DNA using the calcium phosphate precipitation method. Transfected cells were incubated with (/ E) or without 1 mM estrogen for 48 h before staining with May– Gru¨nwald/ Giemsa. Briefly, the cells were incubated with 1 ml May– Gru¨nwald stain (0.25% methanol) for 10 min. An equal volume of distilled water was added for an additional 10 min. The stain was replaced by Giemsa dye (1/25 dilution of a 0.4% stock) for another 30 min. Afterward the plates were washed with distilled water and dried. Multinucleated cells containing more than four nuclei were scored and averaged from three fields defined by a 2-mm 2 grid at 1100 magnification. Western blot analysis. Transiently transfected HeLa cells were lysed in Laemmli buffer. After heating to 957C lysates were subjected to SDS–PAGE in 12% gels and blotted onto nitrocellulose membranes. After probing with a rabbit polyclonal anti-Rex immune serum against a fused histidine – Rex protein (S. Hemaia and M. DucDodon, submitted for publication) and the blots were developed using chemiluminescence and according to the ECL protocols of Amersham. Immunofluorescence and confocal microscopy. HeLa cells were plated in 35-mm plates onto glass coverslips and transfected as described before. At 48 h after transfection the cells were washed with PBS and fixed for 15 min in 3.7% formaldehyde in PBS and fixing was quenched with 10 mM glycine for 5 min. The specimens were blocked in 250 mM KCl, 20 mM Hepes, pH 7.9, 1% BSA, 0.1% gelatin, and 0.5% Triton in PBS (blocking buffer) for 5 min and incubated with appropriate dilutions of the primary antibody in blocking buffer for 1 h at room temperature. Excess antibody was washed three times with blocking buffer and the specimens were incubated with a 1/50 dilution of fluorescent-labeled secondary antibodies for 45 min. After 31 washing with blocking buffer the coverslips were dried and

ecal

NUCLEAR PORE COLOCALIZATION OF INDUCIBLE HTLV-1 Rex mounted on slides with a mounting solution containing 90% glycerin in PBS and 2 mg/ml DAPCO. For double staining the two primary antibodies were incubated independently, while the two secondary antibodies were used simultaneously. To maintain the thickness of the specimens for confocal microscopy Ç15-mm-thick drops of nail polish were applied on the slide, forming ‘‘feet’’ which supported the specimen and mounting medium. The antibodies used in this study included a rabbit polyclonal anti-Rex immune serum against a fused histidine– Rex protein (S. Hemaia and M. DucDodon, submitted for publication), the 414 anti-nucleoporin monoclonal antibody [32], an anti-fibrilarin monoclonal antibody, and an anti-coilin monoclonal antibody (kind gifts of Dr. Santama). We are in debt to Dr. M. Busslinger for his long-standing advice and support. Dr. I. Mattaj is particularly thanked for his interest and invaluable help. Dr. G. Simos, Dr. E. Hurt, Dr. P. Grandi, and Dr T. Parslow have kindly provided us with their critical comments and material. Dr. J. King and Dr. Marei Sammar are acknowledged for advice and encouragement. Hydroxytamoxifen (Zeneca) was kindly provided by Dr. B. Furr. This work was supported by the Dr. Mildred Scheel Stiftung and the Deutsche Forschungsgemeinschaft (to K.K.) and by the Agence Nationale de Recherches sur le SIDA, France (to L.G.).

REFERENCES 1.

Bogerd, H. P., Tiley, L. S., and Cullen, B. R. (1992) J. Virol. 66, 7572 –7575.

2.

Grone, M., Hoffmann, E., Berchtold, S., Cullen, B. R., and Grassmann, R. (1994) Virology 204, 144– 152.

3.

Seiki, M., Inoue, J., Hidaka, M., and Yoshida, M. (1988) Proc. Natl. Acad. Sci. USA 85, 7124-7128.

4.

Black, A. C., et al. (1991) J. Virol. 65, 6645 –6653.

5.

Chang, D. D., and Sharp, P. A. (1989) Cell 59, 789– 795.

6.

Stutz, F., and Rosbach, M. (1994) EMBO J. 13, 4096 –4104.

7.

Black, A. C., et al. (1991) Virology 181, 433– 444.

8.

Felber, B. K., Hadzopoulou-Cladaras, M., Cladaras, C., Copeland, T., and Pavlakis, G. N. (1989) Proc. Natl. Acad. Sci. USA 86, 1495 –1499.

9.

Fischer, U., et al. (1994) EMBO J. 13, 4105 – 4112.

10. Malim, M. H., and Cullen, B. R. (1993) Mol. Cell. Biol. 13, 6180 – 6189.

11. Seiki, M., Hikikoshi, A., and Yoshida, M. (1990) Virology 176, 81–86. 12. Barksdale, S. K., and Baker, C. C. (1995) Mol. Cell. Biol. 15, 2962 –2971. 13. Hammarskjo¨ld, M. L., Li, H., Lu, X., and Prasad, S. (1991) in Genetic Structure and Regulation of HIV (Haseltine, W. A., and Wong-Staal, F., Eds.), pp. 345– 353, Raven Press, New York. 14. Fischer, U., Huber, J., Boelens, W. C., Mattaj, I. W., and Lu¨hrmann, R. (1995) Cell 82, 475– 483. 15. Fritz, C. C., Zapp, M. L., and Green, M. R. (1995) Nature 376, 530–533. 16. Bogerd, H. P., Fridell, R. A., Madore, S., and Cullen, B. R. (1995) Cell 82, 485 –494. 17. Stutz, F., Neville, M., and Rosbash, M. (1995) Cell 82, 495– 506. 18. Picard, D., Kumar, V., Chambon, P., and Yamamoto, K. R. (1990) Cell Regul. 1, 291–299. 19. Rimsky, L., Dodon, M. D., Dixon, E. P., and Greene, W. C. (1989) Nature 341, 453–456. 20. Girerd, Y., Casse, H., Duc Dodon, M., and Gazzolo, L. J. (1995) Gen. Virol. 76, 1021 – 1024. 21. Seiki, M., Inoue, J., Takeda, T., and Yoshida, M. (1986) EMBO J. 5, 561– 565. 22. Cochrane, A. W., Perkins, A., and Rosen, C. A. (1990) J. Virol. 64, 881–885. 23. Cullen, B. R., et al. (1988) J. Virol. 62, 2498 –2501. 24. Kalland, K. H., Szilvay, A. M., Langhoff, E., and Haukenes, G. (1993) J. Virol. 68, 1475 – 1484. 25. Kalland, K. H., Szilvay, A. M., Brokstad, K. A., Saetrevik, W., and Haukenes, G. (1994) Mol. Cell. Biol. 14, 7436 – 7444. 26. McDonald, D., Hope, T., and Parslow, T. G. (1992) J. Virol. 66, 7232 –7238. 27. Hope, T. J., Huang, X. J., McDonald, D., and Parslow, T. G. (1990) Proc. Natl. Acad. Sci. USA 87, 7787 – 7791. 28. Kumar, V., Green, S., Staub, A., and Chambon, P. (1986) EMBO J. 5, 2231 – 2236. 29. Hope, T. J., Bond, B. L., McDonald, D., Klein, N. P., and Parslow, T. G. (1991) J. Virol. 65, 6001 – 6007. 30. Chen, H., Boyle, T. J., Malim, M. H., Cullen, B. R., and Lyerly, H. K. (1992) Proc. Natl. Acad. Sci. USA 89, 7678 – 7682. 31. Chlichlia, K., et al. (1994) Oncogene 10, 269–277. 32. Davis, L. I., and Blobel, G. (1987) Proc. Natl. Acad. Sci. USA 84, 7552 –7556.

Received February 1, 1997 Revised version received February 17, 1997

AID ECR 3562

/

6i20$$$401

05-28-97 12:02:40

371

ecal