Two dimerization domains in the trans-activation response RNA-binding protein (TRBP) individually reverse the protein kinase R inhibition of HIV-1 long terminal repeat expression.

Trans-activation response (TAR) RNA-binding protein (TRBP) is a cellular protein that binds to the human immunodeficiency virus-1 (HIV-1) TAR element RNA. It has two double-stranded RNA binding domains (dsRBDs), but only one is functional for TAR binding. TRBP interacts with the interferon-induced protein kinase R (PKR) and inhibits its activity. We used the yeast two-hybrid assay to map the interaction sites between the two proteins. We show that TRBP and PKR-N (178 first amino acids of PKR) interact with PKR wild type and inhibit the PKR-induced yeast growth defect in this assay. We characterized two independent PKR-binding sites in TRBP. These sites are located in each dsRBD in TRBP, indicating that PKR-TRBP interaction does not require the RNA binding activity present only in dsRBD2. TRBP and its fragments that interact with PKR reverse the PKR-induced suppression of HIV-1 long terminal repeat expression. In addition, TRBP activates the HIV-1 long terminal repeat expression to a larger extent than the addition of each domain. These data suggest that TRBP activates gene expression in PKR-dependent and PKR-independent manners.


Trans-activation response (TAR) RNA-binding protein (TRBP) is a cellular protein that binds to the human immunodeficiency virus-1 (HIV-1) TAR element RNA. It has two double-stranded RNA binding domains (dsRBDs), but only one is functional for TAR binding. TRBP interacts with the interferon-induced protein kinase R (PKR) and inhibits its activity.
We used the yeast two-hybrid assay to map the interaction sites between the two proteins. We show that TRBP and PKR-N (178 first amino acids of PKR) interact with PKR wild type and inhibit the PKR-induced yeast growth defect in this assay. We characterized two independent PKR-binding sites in TRBP. These sites are located in each dsRBD in TRBP, indicating that PKR-TRBP interaction does not require the RNA binding activity present only in dsRBD2. TRBP and its fragments that interact with PKR reverse the PKR-induced suppression of HIV-1 long terminal repeat expression. In addition, TRBP activates the HIV-1 long terminal repeat expression to a larger extent than the addition of each domain. These data suggest that TRBP activates gene expression in PKR-dependent and PKR-independent manners.
HIV 1 replication employs the full variety of mechanisms available for the control of host cellular gene expression. The HIV-1 long terminal repeat (LTR) is regulated through a large number of cellular factors including nuclear factor B (NF-B) and SP1 (1). Because the basal transcriptional activity of HIV is very low, the viral Tat protein and host proteins are utilized to increase the transcription of the viral genome. In contrast to other known trans-activators, Tat acts through an RNA target called trans-activation response (TAR) located in the R region of the LTR. TAR RNA forms a stable stem-bulge-loop structure necessary for Tat trans-activation. For its trans-activating function, Tat is physically associated with cellular kinases that phosphorylate the C-terminal domain of RNA polymerase II. The resulting activity is an increase of transcriptional reinitiation and elongation through C-terminal domain phosphorylation by initiating the formation of an elongation-competent transcription complex (2)(3)(4)(5).
Similar to other viruses, HIV-1 relies on the cellular translational machinery for the synthesis of viral proteins. Initiation is the rate-limiting step of translation that determines the efficiency of protein synthesis from mRNA (6). Translation initiation begins with the binding of initiator methionyl-tRNA (Met-tRNA i ) to the 40S ribosomal subunit through the formation of an eIF2⅐GTP⅐Met-tRNA i ternary complex. Formation of this complex and the delivery of Met-tRNA i to the 40S ribosomal subunit constitute the rate-limiting step of translational initiation.
Cells have evolved antiviral defenses that down-modulate protein translation in virus-infected cells. This response is mediated by the production of IFNs. IFN binding to specific receptors leads to the production of more than 250 proteins by the activation of signal transduction pathways (7). IFN␣ transcriptionally induces elevated cellular levels of PKR, a key cellular protein kinase (8,9). PKR is activated after binding to dsRNA through its two dsRNA binding domains (dsRBDs) and by dimerization. Both processes involve the ␣-helix structure of the dsRBDs through their charged residues (binding to dsRNA) and hydrophobic side (dimerization) (10). Another basic PKR domain located just upstream of the catalytic domain is also involved in PKR dimerization (11). Upon activation, PKR undergoes a conformational change that leads to its autophosphorylation. Once active, PKR phosphorylates its substrate, the translation initiation factor eIF2␣. When phosphorylated, eIF2␣ blocks the ability of eIF2B to renew the eIF2⅐GTP⅐Met-tRNA i ternary complex required for protein synthesis initiation (6). eIF2␣ phosphorylation dramatically alters the efficiency * This work was supported by grants from the Agence Nationale de Recherches sur le SIDA (ANRS), CNRS/ARC, and Ensemble contre le SIDA (Sidaction), Medical Research Council Grant 38112 (to A. G.) and ANRS Grant 61852 (to E. F. M.), and grants from the ANRS and Sidaction (to R. B.), a grant from ANRS (to C. V.), and National Health and Medical Research Council of Australia Grant 111700 (to D. F. P.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18  and rate of translational initiation and is a critical component of antiviral and cell growth pathways (12).
The importance of the PKR antiviral pathways is illustrated by the strategies that viruses have employed to avoid the activation of PKR. These include the expression of competitive inhibitory RNAs such as the VA RNA I of adenovirus, the EBER-1 and EBER-2 RNAs of the Epstein-Bar virus, and TAR RNA of HIV (13)(14)(15). Alternatively, viral proteins such as HIV Tat, vaccinia E3L and K3L, influenza NS1, reovirus 3, and hepatitis C virus NS5A can act as either direct inhibitors of PKR or competitive substrates for phosphorylation (16 -20). PKR activation may also be blocked by protein-protein interactions with particular cellular proteins such as the TAR RNAbinding protein (TRBP) (21)(22)(23), p58 IPK (11), and the 60S ribosomal subunit protein L18 (24). On the other hand, the cell has also a PKR activator, PACT, that has a high homology with TRBP (25).
TRBP, the first protein cloned as cDNA based on protein-TAR binding (26,27), is a cellular protein that binds HIV-1 TAR RNA. Two forms of the protein, TRBP1 (originally called TRBP) and TRBP2, coexist in the cell and are encoded from two alternatively initiated isoforms of mRNA that differ at their 5Ј ends. TRBP2 is 21 amino acids longer than TRBP1 (26,28,29). TRBP1 and TRBP2 belong to the family of double-stranded RNA-binding proteins with clearly defined dsRBDs (30 -33). TRBPs have two dsRBDs, but only one is functional for TAR binding because of the presence of a KR-helix motif. This 15amino acid peptide motif is the TRBP minimum TAR RNA binding motif, it destabilizes the TAR RNA structure, and binds to the upper stem/loop of TAR with high affinity (31,34,35). TRBPs are encoded by the tarbp2 gene, which has been mapped to human chromosome 12 and mouse chromosome 15 (36,37). Prbp, the murine homologue of TRBP, binds the 3Ј untranslated region of Prm1 protamine RNA, regulates its translation, and plays a physiological role in spermatogenesis (38,39). In the HIV context, TRBP acts in synergy with functional Tat to stimulate the expression of the HIV-1 LTR in human and murine cells (26,28).
In addition to binding TAR RNA in vivo (31) TRBP binds PKR (40,41), blocking the inhibitory effects of PKR on translation (23), HIV LTR expression (22), and HIV replication (21). All the available data suggest that TRBP facilitates viral replication by two mechanisms: direct activation of the LTR through TAR binding and inhibition of the host antiviral mechanisms leading to unhindered translation.
Although TRBP-PKR interaction has been documented between the full-length TRBP protein and the catalytically inactive PKR K296R mutant, no data are available about which regions in the proteins confer heterodimerization. In this paper, we show in a two-hybrid assay that TRBP also binds to wild-type PKR (PKRwt) and reverses its growth-inhibitory activity in yeast. This binding is mediated through each dsRBD in TRBP but is independent from the RNA binding motif of the protein. In an LTR-luciferase assay, we show that TRBP reverses the PKR-induced inhibition of expression of the HIV-1 LTR. Each dsRBD of TRBP mediates this function, but fulllength TRBP has a much stronger effect on gene expression than each individual domain.

EXPERIMENTAL PROCEDURES
Plasmids and Strains-The yeast indicator strains HF7c and SFY526 have been described previously (42,43). Yeast expression vectors pGBT9, carrying Gal4 binding domain (BD) and pGADGH carrying Gal4 activation domain (AD) have been described previously (44). PGBT11 is a modified version of pGBT9 that has a 4-base pair addition in the EcoRI site. TRBP2 cDNA was isolated as a BamHI fragment from the pBS-TRBP2 plasmid (29) and inserted into either pGBT11 or pGADGH to generate Gal4 BD-TRBP2 and Gal4 AD-TRBP2.
TRBP-AC and -AB were generated by inserting mutations ⌬137-261 and ⌬270 -402 described before (31) into the full-length plasmids. PKR complete cDNA and PKR fragments were polymerase chain reaction-amplified from pcDNA1/Amp-PKR (45) with different sets of 5Ј primers and inserted into the XmaI-SalI site of either pGBT11 or pGADGH. TRBP fragments were generated by polymerase chain reaction from the pBS-TRBP2 plasmid and inserted into the XmaI-BamHI sites of pGBT9 and pGADGH. pLTR-Luc and pcDNA3-TRBP2 have been described previously (28). pcDNA3-PKR was obtained by inserting the HindIII-BamHI fragment bearing PKR cDNA from pcDNA1/Amp-PKR (45) into pcDNA3. All TRBP fragments were obtained by polymerase chain reaction and inserted into the BamHI-XbaI sites of pcDNA3. Each ATG was in the context of the Kozak sequence (CCACCATG) upstream of the initiation codon to allow for maximum translation efficiency (46). All inserts were verified by sequencing using the Bigdye terminator chemistry and automated sequencer ABI373A (Applied Biosystems). Control plasmids that express SNF1 linked to Gal4-BD and SNF4 fused to Gal4-AD have been described previously (47,48). The control plasmid that expresses uracil DNA glycosylase has been described previously (49).
Two-hybrid Assay-The SFY526 or HF7c yeast reporter strains containing the LacZ-Gal4-inducible gene were cotransformed as described and plated on selective medium lacking tryptophan and leucine (47). Double transformants were patched on selective medium, replicaplated on Whatman 40 filters, and tested for ␤-galactosidase activity (50).
Cell Transfections-HeLa cells were plated in 12-well plates at a density of 10 5 cells/well. Transfections were performed with the various plasmids using Fugene 6 reagent (Roche Molecular Biochemicals). 2 g of total DNA was mixed with serum-free medium to a total volume of 50 l. 6 l of Fugene 6 was added and incubated for 15 min at room temperature. The mixture was then applied to the cells in exponential phase. The resuspended cells were incubated in 2 ml of Dulbecco's modified Eagle's medium (Life Technologies, Inc.) supplemented with 10% fetal bovine serum (Hyclone). Transfected cells were lysed and assayed for luciferase activity 36 h post-transfection (luciferase assay system, Promega). Luminescence was measured on an EG & G Berthold luminometer. The protein concentration of the extracts was measured by Bradford assay (Bio-Rad), and all values were normalized to 1 g of total protein.

TRBP Forms Homo-and Heterodimers with PKR-N and
PKRwt-The PKR-inactive mutant K296R (51) and PKR p20 (first 184 amino acids) have been shown to form homodimers (10,41). TRBP1 forms homodimers and heterodimers with these two PKR forms (41). To verify these observations in the two-hybrid system with TRBP2 (28,29) and PKR wild type (8), we made fusions of the two proteins ( Fig. 1A) with GAL4 BD and GAL4 AD in plasmids pGBT9 and pGADGH, respectively. TRBP2 was first tested with itself and showed a similar bluecolor intensity as the positive control SNF1/SNF4 in two different yeast strains. No background was observed with a control uracil DNA glycosylase (49) or no protein. This result shows a strong homodimerization for the TRBP2 protein (Fig.  1B) as observed previously for TRBP1 (41). Because the strain SFY526 allowed the best coloration in the ␤-galactosidase assay, it was chosen in all subsequent two-hybrid assays.
We next wanted to determine whether TRBP interacts with an intact PKR protein (PKRwt) and which domain of the protein is involved in the interaction. We constructed plasmids in which PKRwt, PKR expressing the first 178 amino acids (PKR-N), and PKR amino acids 179 -551 (PKR-C) are fused to the BD or the AD of Gal4 (shown in Fig. 1A). The results show that TRBP2 interacted with PKR-N and not with PKR-C (Fig. 1C), which extended previous results involving the first 184 amino acids (41). These two individual PKR fragments do not have the catalytic activity of PKR and do not exhibit toxicity when expressed in yeast. In contrast, PKRwt is known to inhibit yeast growth through the phosphorylation of the yeast homologue for eIF2␣ (52). For this reason, it has not been used previously by others in two-hybrid interactions. Accordingly, we too observed a strong growth inhibition when we used PKRwt on both pGBT11 and pGAD plasmids in the two-hybrid assay with, concomitantly, no detection of ␤-galactosidase activity. This led us to postulate that PKRwt could be useful in the two-hybrid assay, because only functionally relevant partners for PKR would be able to inhibit the kinase activity strongly and thus allow cell growth, interaction, and development of ␤-galactosidase expression. Indeed, transfection of TRBP2-Gal4 BD together with PKR-Gal4 AD in this assay reveals an interaction between the two proteins (Fig. 1C). Interaction in the opposite fusion combination could not be tested, because each transfection with PKR-Gal4 BD-expressing vector resulted in an inhibition of the yeast cell growth. Taken together, our results indicate that TRBP interacts with PKRwt through its N-terminal moiety (PKR-N, 178 residues). This binding inhibits PKR strongly and reverses its growth inhibitory activity in yeast.
All Deletion Mutants in TRBP Homodimerize and Interact with PKR-PKR-N contains the two dsRBDs of PKR. Therefore, its interaction with TRBP occurs through these domains. TRBP has two dsRBDs, and we wanted to determine whether similar regions were involved in its dimerization. We performed a series of deletions in the TRBP protein, which excluded dsRBD1 (fragment A), dsRBD2 (fragment B), or the C-terminal part (fragment C). Each protein was called by the fragments that it has conserved (see AB, AC, and BC in Fig.  1A). We coexpressed combinations of each mutant and the full-length protein in yeast and performed two-hybrid assays.
All TRBP-deleted proteins interacted with the full-length protein and each other, suggesting that two or more TRBP fragments are involved in the homodimerization (Table I).
These TRBP-deleted proteins were next assayed for their interaction with PKR-N and PKRwt for interaction and inhibition. Each deleted protein interacted with PKR-N in a similar profile as that with TRBP ( Fig. 2A). These results indicate that TRBP-PKR interactions are also mediated through at least two independent domains in TRBP. These TRBP proteins were also tested for their interaction with PKRwt (Fig. 2B). In this case, TRBP, TR-AB, TR-AC, and PKR-N could interact and inhibit PKR. The absence of interaction with TR-BC suggests an unfavorable folding that decreases the affinity of the protein with PKR. These results point out that PKR-N has a transdominant phenotype on PKR similar to the inhibitory effect of TRBP (Fig. 2B).
Each dsRBD in TRBP Mediates Homodimerization and TRBP-PKR Interaction Independently from the KR-helix RNA Binding Motif-To determine which fragments in TRBP are involved in the dimerization, we constructed plasmids that express each of fragments A, B, and C (Fig. 1A) fused to Gal4-BD, and we tested them for their interaction with TRBP2, PKR-N, and PKRwt. Fragments A and B, which encompass dsRBD1 and dsRBD2, respectively, interacted with TRBP2 and PKR-N. Fragment C did not interact with any protein. In addition, fragment B showed an interaction with PKRwt, which indicates that, in contrast to fragment A, it has a strong Numbering is given according to the TRBP2 protein (Gen-Bank TM accession number NP004169) and PKR (8). Gray boxes represent dsRBDs, and the dark boxes in TRBP2 represent basic regions within or outside the dsRBDs as indicated. The large dark box is the KR-helix motif that binds RNA (34,35). The hatched box in PKR represents the third basic region required for kinase activation (60, 61). B, homodimerization of TRBP2. Shown is the two-hybrid assay for TRBP2 in yeast strains SFY526 and Hf7c. The first protein cited is fused to Gal4 BD in pGBT9, and the second one is fused to Gal4 AD in pGADGH. SNF1/SNF4 is a positive control (47,48). SNF4 and uracil DNA glycosylase (UDG) (49) are negative controls for TRBP2. none represents the empty vector. C, TRBP2 interacts with PKR-N and PKRwt. Shown is the two-hybrid assay using TRBP2, PKR-N, PKR-C, and PKRwt in yeast strain SFY526. A representative experiment among three independent transfections and assays is shown here. inhibitory effect on PKR (Fig. 3). Fragments A, B, and C were also assayed when fused to the Gal4 AD. They showed similar results except that TR-B-Gal4 AD inhibited yeast growth, which prevented the testing of its inhibition of PKR (data not shown).
Because TRBP and PKR are both dsRNA-binding proteins, the interaction between the two proteins could be either direct or via an RNA substrate. We have shown previously that TRBP dsRBD2 harbors the main determinant for TAR RNA binding through a KR-helix motif, whereas dsRBD1 do not bind RNA significantly (34). The binding of both fragment A (including dsRBD1) and B (including dsRBD2) to TRBP and PKR shows that an RNA binding activity is not required for the interaction. To further define if the regions involved in the dimerization are located strictly within the dsRBDs, smaller domains were generated in fusion with the Gal4 BD. In a two-hybrid assay, fragments K and H representing strictly dsRBD1 and dsRBD2, respectively, interacted with TRBP and PKR-N but did not inhibit PKRwt. This suggests that the interaction of the smallest component of the dsRBD could bind PKR but in a manner that did not inhibit the activation of PKR. (Fig. 3). All together these data indicate that the TRBP has independent dimerization determinants located in each dsRBD. The interaction is independent from the KR-helix RNA binding motif, which is not present in TRBP dsRBD1. However, the inhibition of PKRwt by TRBP fragment B, but not H, shows that either a higher affinity interaction or a more intimate interaction is required to block the activation of PKR by TRBP (Fig. 3).

PKR Inhibits and TRBP Increases HIV-1 LTR Expression in the Absence and Presence of Tat-PKR influences gene expression by two main mechanisms that result in translational in-
hibition through the phosphorylation of eIF2␣ and transcriptional activation through the nuclear factor B pathway. The first mechanism depends on the kinase activity (53), whereas the second one occurs by a kinase-independent mechanism (54, 55). The effect of PKR on HIV-1 LTR activity measured in a reporter gene assay using an LTR-Luc plasmid is the result of both mechanisms, but the overall effect is an inhibition of expression (22,54). In contrast, TRBP1 and TRBP2 increase HIV-1 LTR expression in human and murine cells. This activity is caused by its binding to both TAR RNA and PKR (26,28). To quantify PKR and TRBP2 activity in our system, we measured independently their effect on the HIV LTR expression by a luciferase assay (Fig. 4). As expected, increasing amounts of PKR decreased LTR expression, whereas TRBP increased it. These effects were proportional to the amount of plasmid used to transfect the cells except that large amounts of TRBP2 inhibited LTR expression as observed previously (28). The Tat trans-activated level of expression from the HIV-1 LTR was inhibited to a similar proportion by PKR, whereas TRBP2 increased this expression (Fig. 4).

TRBP2 and its PKR-binding Fragments Reverse the Suppressed Expression of HIV-1 LTR by PKR-
The characterization of TRBP2-PKRwt interaction in the two-hybrid assay indicates that TRBP2 is a strong inhibitor of PKR function that reverses the growth inhibition phenotype of the kinase. We wanted to verify whether TRBP2 has a similar inhibitory effect on viral gene expression. 1 g of PKR plasmid was chosen for the reversion experiment because it inhibits LTR-luciferase expression 3-fold. We cotransfected HeLa cells with pcDNA3-PKR plasmid and increasing amounts of pcDNA3-TRBP2 and measured the luciferase activity expressed from the HIV-1 LTR. Reversion of the PKR inhibition was achieved with a TRBP2 plasmid amount 100 times lower than PKR (Fig. 5A,  lane 4). This activity was similar in the presence of Tat (Fig.  5B). The overall increase in LTR expression caused by TRBP2 was five times higher than the PKR-reduced expression at 0.1 and 0.25 g of TRBP-expressing plasmid. Higher amounts of TRBP2 inhibited LTR expression as shown previously in the absence of PKR (Fig. 4B).
The TRBP2 fragments previously tested in the two-hybrid assay were transferred on pcDNA3, and an ATG start codon was added when necessary. HeLa cells were transfected with these plasmids and TRBP2 in the presence of 1 g of pcDNA3-PKR. All the fragments that had a strong interaction with TRBP and PKR (A, B, H, and K) were able to reverse PKR inhibition in this assay at 0.25 g of plasmid in the absence and presence of Tat (Fig. 5,  A and B, lane 8). The maximum effect was reached at either 0.5 or 1 g of plasmid (Fig. 5, A and B, lanes 9 and 10), although they did not reach the strong activity of TRBP2. Fragment B was the most efficient in the absence of Tat and showed a decreased activity at high concentration similar to the whole protein. Fragment C did not have any effect on LTR expression in the absence or presence of Tat (Fig. 5). other Two-hybrid assays using TRBP, TR-AB, TR-AC, and TR-BC in pGBT9 (vertical lane) and in pGADGH (horizontal lane). ϩ indicates a positive ␤-galactosidase expression. SNF1 and SNF4 represent the positive (bottom right) and negative controls. Independent transfections and assays were performed at least three times.

FIG. 2. Each TRBP-deleted mutant interacts with TRBP and PKR.
Two-hybrid assays using TRBP2 and deleted forms in homo-and heterodimerization with PKR-N (A) and PKRwt (B) in yeast strain SFY526. Independent transfections and assays were performed at least three times.
FIG. 3. Each dsRBD in TRBP2 binds TRBP and PKR independently. Two-hybrid assays using TRBP fragments A, B, C, H, and K in pGBT9 transfected with TRBP, PKR-N, or PKRwt in pGADGH are shown. SNF1 and SNF4 represent the positive (bottom right) and negative controls. Independent transfections and assays were performed at least three times.

DISCUSSION
PKR-PKR interactions have been described previously, are necessary for its kinase function, and have been shown to be either dependent on (41) or independent of (56 -58) the presence of dsRNA in vitro. TRBP has been shown to interact with itself, with a catalytically inactive PKR mutant (PKR K296R), and with PKRp20 (first 184 amino acids) in the two-hybrid system. This heterodimerization was found to be RNA-dependent in a far-Western assay (41). In contrast, coimmunoprecipitations of TRBP and PKR in the presence of RNase and in vitro interaction assays have shown that no RNA was required for the interaction between the two proteins. Furthermore, TRBP mutants defective for RNA binding interact with PKR and inhibit the kinase activity (21). These experiments led us to investigate which part of TRBP protein was involved in the homo-and heterodimerization and the role of these binding domains in PKR inhibition.
Homodimerization of TRBP2 was verified by a two-hybrid assay in two different yeast strains. The ␤-galactosidase activity had the same intensity in each TRBP-TRBP interaction as in the SNF1/SNF4 interaction control (Fig. 1B). The interaction with PKR-N (178 amino acids) was also observed clearly in both plasmid backgrounds (Fig. 1C). PKRwt (551 amino acids) has not been used before in the two-hybrid assay because of its toxicity in yeast cells. To verify whether TRBP was a strong inhibitor of PKR, we tested its effect on the wild-type version of the protein. Surprisingly, TRBP was able to interact with PKRwt, and this interaction allowed the reversion of the PKRmediated inhibition on yeast growth (Fig. 1C). This result is compatible with an inhibition of PKR autophosphorylation by TRBP observed previously (21). This is the first time that a two-hybrid assay has succeeded with PKRwt, and it shows that this method can be used to clone cDNAs coding for inhibitors of toxic proteins.
To investigate which site in TRBP2 interacts with itself and with PKR, we generated TRBP-deleted mutants encompassing two dsRBDs (TR-AB) or one of the dsRBD and the C-terminal part of the protein (TR-AC and TR-BC). All of them interacted with PKR-N, suggesting that at least two regions of TRBP were individually involved in the dimerization (Fig. 2 and Table I). Furthermore, AB and AC TRBP mutants also interacted with PKRwt, indicating a strong inhibitory potential of these variants. Individual fragments showed that A and B, which include dsRBD1 and dsRBD2, respectively, mediate the interaction and that B has the highest inhibitory effect on PKR. Inhibition of PKRwt by the AB and AC fragments suggested a predominant role for fragment A (Fig. 2), whereas the interaction data in Fig. 3 indicate an inhibitory potential for fragment B. This discrepancy might be caused by folding differences that favor the interaction or expose the inhibitory component in each case. We therefore concluded that both fragments A and B can interact with TRBP and PKR and can prevent PKR growthinhibitory effect in a specific context.
To further delineate the interaction site, the smaller fragments H and K that strictly have each dsRBD (Fig. 1A) were tested for their interactions with TRBP, PKR-N, and PKRwt. They both dimerized with TRBP and PKR-N but not with PKRwt. This result confirms that the dimerization site is within each dsRBD in TRBP. We have shown previously that only dsRBD2 (included in B and H) mediates RNA binding, whereas dsRBD1 (included in A and K) does not bind RNA (34). The present and previous experiments strongly suggest that TRBP and PKR do not require RNA to mediate their interaction in agreement with previous data (21). Similarly, PACT and its murine homologue, RAX, bind PKR in the absence of RNA (25,59). Therefore the requirement for dsRNA observed by others might be caused by the specific far-Western assay (41).
PKR has been shown to dimerize through the dsRBDs (41,57). Other results indicate that the first 280 amino acids do not mediate the interaction (56) and that the dimerization site occurs through amino acids 244 -296 excluding the dsRBDs (11). Our results indicate that the first 178 amino acids in PKR are fully competent to mediate homodimerization with itself and PKRwt. Interestingly, interaction with PKR-N also inhibited the growth defect of PKRwt, resulting in the detection of ␤-galactosidase expression. This expression indicates a transdominant phenotype for PKR-N over PKRwt (Fig. 2). We do not assess here the possibility of another homodimerization site in the catalytic domain (PKR-C) observed by others (11), but this site is not involved in the interaction with TRBP (Fig. 1C).
We have shown previously that the HIV-1 LTR is activated by TRBP in human and murine cells (26,28). PKR dual activities result in an inhibition of the LTR expression (22,45). These results were verified and quantified in our transfection assay in the absence and presence of Tat. The results are in agreement with previous data. TRBP increased LTR activity 10.4-and 5.2-fold in the absence and presence of Tat, respectively (Fig. 4B) with a maximum activity at 0.1 g of plasmid. PKR had an inhibitory effect that reaches 5.6-and 5.5-fold at 2.5 g of plasmid (Fig. 4A). When cells were cotransfected with PKR and TRBP2, only 0.01 and 0.025 g (in the absence or presence of Tat, respectively) of TRBP2 plasmid was necessary to counteract the effect of 1 g of PKR plasmid. Furthermore, the effect of TRBP was much greater than the simple inhibition of PKR, suggesting that TRBP also has a PKR-independent augmenting activity on LTR expression (Fig. 5, TRBP2). The TRBP-TAR binding properties may mediate this effect. Recent results indicate that TRBP prevents translational inhibitory activity of PKR in a cell-free translation assay and also suggest a PKR-independent effect of TRBP on translation. The activation of HIV-1 LTR expression by TRBP in PKR Ϫ/Ϫ murine cells supports this hypothesis. 2 High concentrations of TRBP2 plasmid (Ͼ0.1 g) reproducibly inhibited HIV-1 LTR expression (28). This inhibition was also observed at the highest concentrations of some TRBP fragments (Figs. 4 and 5). Although no definitive explanation can be reached, it is likely that only low concentrations of TRBP added to the cells reflect physiological conditions. Therefore, this inhibition at high amounts may be caused by (i) TRBP cellular toxicity through its oncogenic potential observed previously (21), (ii) a high concentration of TRBP binding to TAR that will relocate the TAR-Luc RNA to a cellular compartment not suitable for translation, or (iii) a combination of both mechanisms.
Domains of TRBP interact with PKR dsRBDs in a two-hybrid assay (Fig. 3). Among them, only B and H have the KR-helix motif, which mediates RNA binding in dsRBD2. Therefore, these fragments provided a way to distinguish between the effects of PKR inhibition with and without TAR binding activity. All the fragments that showed a clear interaction with PKR (A, B, H, and K) were able to reverse PKR-inhibitory effects on HIV-1 LTR expression in the absence and presence of Tat. Among them, fragment B has a stronger activity on the basal 2 D. Dorin, M. Bonnet, S. Bannwarth, A. Gatignol, E. Meurs, and C. Vaquero, manuscript in preparation. LTR expression (Fig. 5). None of the fragments or the sum of two individual fragments reached the same activity as TRBP2 at low concentration, which supports the idea that the entire protein has an additional activity not found in its individual components. Fragment B has the strongest effect on the reversion of PKR inhibition compared with the other fragments, which is in agreement with its interaction with PKRwt observed in Fig. 3. This activity may be caused by its RNA binding properties or to amino acids 98 -152 located between each dsRBD. The absence of a similar behavior with the H fragment favors the second hypothesis but remains to be confirmed. Overall, the experiments indicate that each dsRBD in TRBP reverses the PKR-mediated LTR inhibition of expression through direct interaction with PKR dsRBDs. This mechanism is one part of the entire effect of TRBP on gene expression.