|Home | About | Journals | Submit | Contact Us | Français|
As obligatory parasites, viruses co-opt a variety of cellular functions for robust replication. The expression of the nonsegmented negative-strand RNA genome of respiratory syncytial virus (RSV), a significant pediatric pathogen, absolutely requires actin and is stimulated by the actin-regulatory protein profilin. As actin is a major contractile protein, it was important to determine whether the known functional domains of actin and profilin were important for their ability to activate RSV transcription. Analyses of recombinant mutants in a reconstituted RSV transcription system suggested that the divalent-cation-binding domain of actin is critically needed for binding to the RSV genome template and for the activation of viral RNA synthesis. In contrast, the nucleotide-binding domain and the N-terminal acidic domain were needed neither for template binding nor for transcription. Specific surface residues of actin, required for actin-actin contact during filamentation, were also nonessential for viral transcription. Unlike actin, profilin did not directly bind to the viral template but was recruited by actin. Mutation of the interactive residues of actin or profilin, resulting in the loss of actin-profilin binding, also abolished profilin's ability to stimulate viral transcription. Together, these results suggest that actin acts as a classical transcription factor for the virus by divalent-cation-dependent binding to the viral template and that profilin acts as a transcriptional cofactor, in part by associating with actin. This essential viral role of actin is independent of its contractile cellular role.
A number of viruses have been shown to use cytoskeletal proteins in various stages of their life cycle, but mainly for viral morphogenesis and budding (19, 32, 45). Interestingly, several nonsegmented negative-strand RNA (NNR) viruses require cytoskeletal proteins for viral RNA synthesis, i.e., transcription and replication of their genomes. The transcription complex of Newcastle disease virus irreversibly assembles and synthesizes viral mRNA on the cytoskeletal framework (21). Sendai (34), measles (35), and vesicular stomatitis (34) viruses require tubulin for transcription, whereas respiratory syncytial virus (RSV) (3, 7, 23) and human parainfluenza virus 3 (12, 13, 20), two clinically related but genetically diverse human pathogens, use actin.
As the predominant contractile protein of the living cell, actin is in dynamic equilibrium between its monomeric (globular [G-actin]) and polymeric (filamentous [F-actin]) forms. The polymerization (or filamentation) reaction is reversible, complex, and highly sensitive to myriad physiological regulators that include ADP/ATP, divalent cations, and regulatory proteins such as profilin (18, 43). Biochemical, biophysical, mutational, and crystallographic analyses of actin and its various complexes with the above-mentioned regulators have identified the amino acid residues critical for regulatory interactions. In stark contrast, no structure-function analysis of actin with respect to its viral transcriptional activation has been conducted. The negative-strand RNA viral genome is linear, single stranded, and intimately wrapped (encapsidated) within the viral nucleocapsid protein (N). The resultant N-RNA complex is the functional template for new RNA synthesis by the viral RNA-dependent RNA polymerase (RdRP). The RdRP holoenzyme in all NNR viruses minimally contains the viral large protein (L), which is the main polymerase subunit, and its transcription factor, the viral phosphoprotein (P) (4, 11, 15, 16). In RSV, an additional viral protein, M2-1, is essential for processive and robust transcription (11, 16). Here, we will refer to the transcriptionally active complex of NNR viruses consisting of the N-RNA template and the RdRP as the ribonucleoprotein (RNP) complex. To summarize, in this terminology, the RNP is the N-RNA template plus the RdRP, and the RdRP is proteins L, P, and M2-1. The viral NNR synthesis occurs in two modes (13, 40); whereas the transcription mode results in the synthesis of mRNAs from individual genes by a stop-and-start mechanism, the replication mode generates a full-length replica of the N-RNA template and requires the synthesis of new N protein (40). The fundamental RNA polymerization function of the RdRP is obviously essential for both replication and transcription; however, the biochemical distinction between the viral RNPs in these two modes is currently unclear, although cis-acting sites on the N-RNA template (46), specific domains of L (17), and the phosphorylation status of P (39) may all play a role. Previously, we used an in vitro-reconstituted viral transcription system to show that cellular actin is absolutely required for RSV transcription (7) and that the addition of profilin, the major physiological regulator of actin polymerization, further stimulates transcription by about threefold (8). Profilin alone, in the absence of actin, does not stimulate transcription, suggesting that it acts as an accessory to actin (8). We also showed that cytochalasin, an inhibitor of actin polymerization, inhibits RSV morphogenesis but not RSV transcription (7). Moreover, we found that small interfering RNA-induced knockdown of profilin has a greater effect on RSV morphogenesis, egress, and syncytium formation than on viral RNA synthesis (6). Knockdown of vasodilator-stimulated phosphoprotein, a regulator of profilin and actin, also inhibits RSV egress without inhibiting intracellular viral transcription (37). Similarly, actin and microtubules may work cooperatively to promote productive RSV growth (24). Together, these findings suggest a model in which both forms of actin can function as viral transcription factors and in which the filamentous form functions only as a scaffold and/or a vectorial force for virion assembly and budding.
Since viral gene expression is an absolute necessity in viral growth, it is important to understand the mechanism by which actin and profilin activate the viral RdRP, a unique enzyme that copies an RNA template to synthesize RNA. In this work, we continued our previous studies and created a series of mutant recombinant actins in which specific functional residues were altered. We also mutated the actin-binding surface residues of profilin to elucidate its relationship with actin in transcription. By studying the phenotypes of these mutants in a reconstituted RSV transcription system, N-RNA binding, and actin-profilin interaction, we provide a model in which both actin and profilin activate viral RdRP. In this model, actin is the primary, template-binding transcription factor and profilin acts as an accessory factor by being recruited via actin. A polymerization-deficient actin mutant was transcriptionally as active as the wild type (WT), ruling out a need for F-actin formation in transcription. Clearly, this novel role of actin as an RNA transcription factor is fundamentally different from its traditional contractile role in cellular processes.
The RSV RNP complex was purified from infected cells essentially as described previously (7, 8). As mentioned above, the RNP contains all the viral transcriptional components (the N-RNA template and the viral RdRP subunits, i.e., L, P, and M2-1) and requires only the host factors—actin and profilin—for optimal transcription (Table (Table11).
For actin- and profilin-binding experiments, the N-RNA template was prepared from the RNP by stripping off L, P, and M2-1 with a high-salt solution in a process similar to the procedures developed for the prototype NNR virus vesicular stomatitis virus (5). Briefly, 5 M NaCl was added to the RNP so that the final NaCl concentration was 0.5 M. The mixture was incubated on ice for 15 min and centrifuged at 120,000 × g through 15% sucrose as described previously (5) with a small cushion of 30% sucrose at the bottom. The N-RNA, landing on the bottom cushion, was collected by aspiration and dialyzed against buffer A (50 mM Tris-acetate [pH 7.5], 120 mM potassium acetate, 4 mM MgCl2, 0.2 mM CaCl2, 20% glycerol, 1 mM dithiothreitol).
Genes encoding chicken β-actin (7) containing an N-terminal His tag and human profilin I (8) containing an N-terminal FLAG tag (DYKDDDDK) were cloned into the pET-15b vector between NdeI and BamHI sites and NcoI and BamHI sites, respectively, with the aid of PCR. Subsequent mutations were created by using the QuikChange site-directed mutagenesis kit (Stratagene). To improve the solubility of the recombinant proteins, the plasmids were introduced into Escherichia coli BL21(DE3) containing the pET-compatible plasmid pG-KJE8 (which expresses the chaperones DnaK, DnaJ, GrpE, GroES, and GroEL) according to the instructions of the supplier (Takara) for antibiotic selection and induction. His-actin was purified by standard Ni2+ affinity chromatography using HisBind resin (Novagen). FLAG-profilin was purified using anti-FLAG M2 affinity gel (Sigma-Aldrich) and eluted in pH 2.5 glycine buffer, immediately neutralized by a premeasured volume of unbuffered 1 M Tris. For transcription reactions, the tags were removed by proteolytic digestion. The His tag was removed by digestion with thrombin-agarose (Sigma) that was subsequently separated away by low-speed centrifugation. The FLAG tag was removed by digestion with enterokinase (Novagen), and the latter was then removed by an enterokinase removal kit according to the protocols of the manufacturer (Sigma-Aldrich). The final profilin preparation was dialyzed in buffer A, and the actin was dialyzed in buffer A plus 0.2 mM Na2-ATP (43).
For the assessment of purity and accurate molecular weight determination, a volume of acetonitrile containing 2% (by volume) formic acid was added to an equal volume of each preparation and the mixtures were subjected to high-resolution electrospray ionization-time of flight mass spectrometry using a Mariner system (Perceptive Biosystems, Stafford, TX). In both cases, deconvolution of the m/z peaks provided mass values in agreement with the predicted amino acid composition within 0.01% accuracy. For actin, the only nonactin peptide corresponded to GroEL, which constituted less than 2% of the actin preparation. The GroEL most likely originated from the overproducer chaperone plasmid pG-KJE8, which likely helped actin to fold, but some of it remained associated with the client protein. The profilin preparation contained a trace amount of human keratin, obviously a contaminant. No other protein could be detected in either preparation.
Based on the consensus derived from previous publications (7, 8, 31), the complete, optimally active transcription reaction mixture (20 μl) contained the following: 1 μg of viral RNP; 200 ng of actin; 60 ng of profilin; 50 mg/ml bovine serum albumin (BSA); 2 μg/ml actinomycin D; 0.4 mM (each) ATP, GTP, and UTP; 50 μCi (50 μM) [α-32P]CTP; and 4% dimethyl sulfoxide in buffer A. The amounts of actin and profilin were saturating, as determined by prior optimization in which the amount of one was varied while the amount of the other was kept constant. Reaction mixtures were incubated at 30°C for 2 h, and the poly(A)+ viral mRNA was quantified as described by Mason et al. (31). First, the reactions were stopped with hybridization buffer (15 mM Tris-HCl [pH 7.5], 1.5 mM EDTA, 0.8 M LiCl, 0.8% sodium dodecyl sulfate [SDS], 120 mM NaCl, 4 mM KCl). A 20% slurry of magnetic oligo(dT) beads [Dynabeads Oligo(dT)25; Invitrogen Dynal AS] in a volume of 4 μl was then added to the reaction mixture, which was processed in a rotary shaker for 10 min at 4°C. Finally, the beads were magnetically separated, rinsed three times with the hybridization buffer, and mixed with scintillation fluid (ScintiSafe Plus 50%; Fisher), and the radioactivity was determined by liquid scintillation counting.
In testing the template-binding activities of the actin or profilin mutants, 1 μg of viral N-RNA was mixed with 200 ng of actin and/or 60 ng of profilin in a total volume of 20 μl containing buffer A plus 50 mg/ml BSA and 0.4 mM Na2-ATP. This composition simulated the transcription conditions described above without the occurrence of actual transcription due to three missing nucleotides. ATP was included because of its known role in regulating actin structure and function. The mixture was incubated at 30°C for 30 min, after which 2 μg of biotin-conjugated monoclonal RSV N antibody (MAB858-3B; Millipore) and 10 μl of a 20% slurry of streptavidin-coupled Dynabeads (Invitrogen) in buffer A were added. The mixture was incubated at room temperature in a rotary shaker for 30 min. The beads were separated magnetically as before, rinsed three times with a mixture of buffer A and 0.2 mM Na2-ATP, and boiled in standard SDS sample buffer. The material was spun at 10,000 × g for 5 min, and the supernatant was analyzed by SDS-polyacrylamide gel electrophoresis (PAGE) and Western blotting using appropriate primary antibody, horseradish peroxidase-conjugated secondary antibody, and detection with an enhanced chemiluminescence kit (Pierce).
In testing the binding between actin and profilin, the tagged recombinants were used. FLAG-actin (100 ng) was mixed with His-profilin (30 ng) in a total volume of 10 μl containing buffer A plus 50 mg/ml BSA and 0.4 mM Na2-ATP. The mixture was incubated at 30°C for 30 min, and then standard immunoprecipitation (IP) was done with anti-FLAG M2 affinity gel (Sigma-Aldrich). The pellet was washed three times with 0.2 ml of buffer A and processed for Western blotting as described above to detect the His-profilin band by using anti-His primary antibody (GE Healthcare).
The ATP-binding activities of actin and its mutants were tested essentially as described originally for actin-related proteins (27). In brief, 0.8 μg of actin and 20 pmol of [α-32P]ATP (0.02 μCi) were mixed in 20 μl of buffer A minus the dithiothreitol, and the mixture was irradiated at 254 nm on ice from a distance of 2 cm for 40 min. Samples were subjected to SDS-PAGE, and the cross-linked ATP-actin band was visualized by autoradiography.
We hypothesized that examining the role of the well-characterized functional domains of actin may illuminate the mechanism by which it activates transcription. Significant crystal structure data for actin, including many structures of actin in complexes with various physiological regulators such as divalent cations (Mg2+ and Ca2+), ATP/ADP, and profilin, are now available. Our collective knowledge of the major amino acid residues of actin that interact or make contact with these entities is summarized in Fig. Fig.11 (2, 26, 38, 42, 43, 48). Regarding interaction with profilin, although 21 residues of actin contribute to this contact, Phe374 and His172 play particularly important roles (43). Phe374 makes the central contact such that the phenyl ring interacts with four profilin residues, namely, Ile73, His119, Gly121, and Asn124, three of which are labeled in Fig. Fig.1.1. In addition, the carboxyl group of the C-terminal Phe374 forms a salt bridge with Arg74 of profilin. We thus reasoned that mutating Phe374 alone may prevent actin-profilin interaction. This hypothesis is also supported by findings from biochemical studies of protonation (low pH) and proteolysis; for example, the removal of Phe374 by carboxypeptidase A digestion destabilizes the actin-profilin complex (30). The second major interaction occurs between His172 of actin and Phe59 of profilin (Fig. (Fig.1)1) in which the imidazole ring of the former stacks upon the phenyl ring of the latter. The divalent cation forms a complex with ATP/ADP, and both the cation and ATP/ADP also make direct contact with specific amino acid side chains of actin (26, 38, 42, 48). Ca2+ binds actin through water-mediated contacts with Asp10, Gln136, and Asp153, and ATP forms coordinate bonds through a series of domains that include residues Ser13, Gly14, Met15, Lys212, Glu213, Met304, Tyr305, and Lys335 (Fig. (Fig.11).
A biochemical approach to understanding the roles of these regulators in actin function would be to conduct reactions in their absence. Unfortunately, the biochemical requirements of actin and viral RdRP are very similar in that both need divalent cations and nucleotides (3, 23, 31). Although Mg2+ is used in RSV transcription and Ca2+ is the common divalent cation for actin, the tertiary structures of Mg2+-bound and Ca2+-bound actins did not reveal any obvious metal-specific differences (42). We therefore undertook a mutational approach in which we mutated the specific interacting residues mentioned above by means of recombination.
We first confirmed the yield of the recombinant proteins (Fig. (Fig.2A).2A). We also validated the activities of the WT recombinants as described before (8) by evaluating the polymerization competence of actin and the ability of profilin to bind to poly-l-proline. We then confirmed our previous observation (7, 8) that both actin and profilin are needed for maximal RSV transcription (set at 100% in Table Table1).1). Actin alone was responsible for about a quarter of maximal transcription (Table (Table1),1), and the addition of profilin further stimulated transcription about fourfold (to 100%). The authenticity of the transcription was also confirmed by using the inhibitory effect of foscarnet (phosphonoformic acid), a pyrophosphate mimic, which was shown to inhibit RSV transcription in vitro using the same poly(A)-based assay (31).
We then set out to determine the activities of the mutants in a reconstituted RSV transcription system as described below.
As summarized above, residues of actin that form coordinate bonds with Ca2+ either directly or indirectly via ATP have been characterized. We targeted a number of them by site-directed mutagenesis and tested the mutants in RSV transcription assays. When actin residues important for divalent-cation binding were mutated even singly (in mutants 1 to 3), the transcriptional activity was reduced to near background levels. The activities with and without added profilin were equally low, consistent with the notion that actin is the primary transcription factor and that profilin is an accessory. Thus, when actin is nonfunctional, the reaction is equivalent to a profilin-only reaction, with little activity (8). The overall conclusion is that divalent-cation binding is important for actin's transcriptional activity.
A stretch of three Asp residues at the extreme N terminus of cytoplasmic actin has been noted, whereas striated-muscle actin contains four such aspartate residues. The role of acidic domains in activating DNA-templated transcription is well known (29). In vesicular stomatitis virus, a prototype NNR virus, the N-terminal acidic domain of the P protein was also shown to be important for viral transcription; interestingly, it can be replaced by β-tubulin that has a highly acidic C terminus (10). It was, therefore, reasonable to ask whether the N-terminal acidic domain of actin also contributes to RSV transcription. However, as shown in Table Table1,1, 1 deletion removing the whole cluster of three acidic residues (in mutant 4) had a negligible effect on transcriptional activity (yielding 20% ± 4% [mean ± standard deviation] of maximal transcription activity versus the WT level of 26% ± 2%). Its activity in the presence of profilin (88% ± 4% of WT activity) was also marginally affected. Thus, the short acidic domain at the N terminus contributes little to the transcriptional activity of actin.
A large number of actin residues participate to form coordinate bonds with the ribose, adenine, and phosphates of ATP. We mutated quite a few of these residues (in mutants 5 to 14) and confirmed the defects of a few representative mutants. As shown in Fig. Fig.2B,2B, mutants 6, 7, 8, and 11 were indeed highly defective in actin binding whereas control mutants 15 and 16 (respectively mutated in actin polymerization and profilin-binding residues) bound ATP like the WT.
Interestingly, when reconstituted in RSV transcription assays, all ATP-binding mutants were found to retain transcriptional activity. Only the mutant with a triple deletion (ΔS13, ΔG14, ΔM15; mutant 5) (Table (Table1)1) suffered considerable loss of activity, which we believe to be a result of gross structural alteration since the mutant not only had poor basal activity (14% ± 2% that of the WT) but also did not respond to profilin at all (exhibiting 12% ± 6% of WT activity in the presence of profilin).
Finally, as mentioned before, monomeric actin, or G-actin, polymerizes in a directional manner to form actin filaments, or F-actin, and it is primarily these filaments that participate in cellular properties and processes such as cell motility, transport, and cytokinesis. Much attention has been paid, therefore, to identifying the residues that make contact between two actin monomers in a polymer, and these residues include the two surface residues Ala204 and Pro243. Indeed, recombinant actin with the A240E-P243K double mutation is incapable of polymerization, and thus, its monomeric crystal structure has been studied extensively (42), confirming the role of the two residues. We tested the viral transcriptional activity of the double mutant (mutant 15) (Table (Table1),1), and found it to be essentially as active as the WT, providing direct evidence that the polymerization of actin is dispensable for its transcriptional function.
As profilin promotes optimal viral transcription in the presence of actin, there are two possible scenarios. In one, profilin directly binds to the viral N-RNA template and activates the RdRP; this scenario is analogous to a traditional DNA-dependent transcription factor's being recruited via its DNA-binding domain to its own enhancer sequence. In the alternative scenario, profilin is recruited by actin, much like the coactivator of a transcription factor. To distinguish between the two possibilities, we first created actin mutants with either of the two major profilin-interacting amino acids, His172 and Phe374, replaced. When tested in transcription, both actins (mutants 13 and 17) (Table (Table1)1) retained WT levels of transcriptional activity; however, unlike the activity of WT actin, the activities of the mutants could not be stimulated further by profilin. These results suggest that actin-profilin interaction is essential for profilin's ability to activate transcription.
If this conclusion is correct, then profilin mutants that are defective in actin binding should also fail to stimulate transcription. To test this possibility, we targeted three residues of profilin known to be involved in actin binding, and all three mutants (mutants 18 to 20) failed to stimulate transcription. Transcription was still actin dependent, as shown by the ability of actin to promote transcription at a low but normal level; thus, these reactions behaved essentially as if profilin was not added. A control profilin mutant (mutant 21) with an alteration of an irrelevant residue (Asn61 to Ala) (Fig. (Fig.1)1) stimulated actin-based transcription to the fullest extent.
To further dissect the roles of actin and profilin in viral transcription, we tested the abilities of the mutants to recognize and bind to the RdRP-free template. Essentially, N-RNA was immunoprecipitated by N antibody, and the bound actin and/or profilin was tested by immunoblotting (Western). Results (Fig. (Fig.3A)3A) show that WT actin and most actin mutants were capable of binding N-RNA with or without profilin, the only exception being those actin mutants that were mutated in the divalent-cation-binding domain (mutants 1 to 3) (Fig. (Fig.3A).3A). Notably, the addition of profilin to the reaction mixtures containing these actin mutants did not improve the dysfunctional transcription (Table (Table1).1). We conclude that profilin cannot assist RSV transcription if actin does not bind the template.
In testing for profilin binding (Fig. (Fig.3B),3B), we found that the profilin by itself did not associate with the N-RNA but did so if actin was present. Profilin mutants incapable of binding actin (mutants 18 to 20) did not exhibit this actin-dependent association with N-RNA. The control mutant (mutant 21), proficient in actin binding, did show actin-dependent N-RNA association, comparable to that of the WT. In a reciprocal experiment, actin mutants that were defective in N-RNA binding (mutants 1 to 3) (Fig. (Fig.3A)3A) also did not recruit profilin. Two other actin mutants, proficient in N-RNA binding (mutants 6 and 7) (Fig. (Fig.3A),3A), were able to recruit profilin (Fig. (Fig.3B3B).
Finally, to experimentally validate a direct interaction between actin and profilin mutants, we incubated purified FLAG-tagged actin with purified His-tagged profilin in vitro. We tested a mutant corresponding to each functional category of mutations listed in Table Table11 by co-IP and Western analysis. The results (Fig. (Fig.3C)3C) confirmed that the actin mutants lacking divalent-cation binding (mutant 1), an acidic terminus (mutant 4), nucleotide binding (mutants 6 and 8), and actin-actin interaction were all proficient in profilin binding. In contrast, mutant 16, mutated in a profilin-binding residue, failed to bind profilin. Reciprocally, a profilin variant with a mutation in an actin-binding residue (mutant 19) failed to bind WT actin. In short, mutations deduced based on crystal structure were experimentally confirmed, adding functionality to our transcriptional studies.
The main conclusions of this study are as follows: (i) the tripartite interaction among the viral N-RNA template and the two cytoskeletal proteins is linear, such that profilin binds actin and actin binds the template; (ii) mutational evidence confirmed our previous conclusion developed using cytochalasin that the transcriptional function of actin does not require its polymerization function; and (iii) of the four major functional domains of actin tested here, only the divalent-cation-binding domain is essential for RNP binding. This study represents the first large-scale mutational analysis of actin and the first attempt to map its transcriptionally important residues. In addition to these results, we previously showed that purified recombinant P and profilin bind to each other in vitro (6). Moreover, in A549 cells (i.e., in the presence of actin) that were either infected with RSV or transiently transfected with recombinant plasmids expressing P and N, both P and N could be coprecipitated with profilin. Finally, in earlier and recent studies, P protein was also shown to form complexes with itself as well as with N and N-RNA (9, 25, 28, 36, 44). We can now combine all these results together to obtain a working model of the RSV transcription complex (Fig. (Fig.4).4). This schematic model depicts the following interactions: P-P, P-N, actin-N-RNA, actin-profilin, and profilin-P.
A central feature of the model (Fig. (Fig.4)4) derived from the present study is that monomeric actin is a bona fide N-RNA-binding, polymerase-activating transcription factor for the RNA virus. The divalent cation may either impart the proper conformation to the template-binding domain or directly coordinate between actin and the template nucleotides and/or phosphates. At this time, we do not know if actin interacts with N or RNA or both. However, there is currently no report that actin binds RNA. F-actin binds the N protein of influenza virus (a segmented negative-strand RNA virus), and this binding is essential for the late-stage cytoplasmic retention and assembly of the viral RNP in virion particles (14). As mentioned previously, direct interaction studies with purified NNR virus N have not been possible due to the technical difficulty of obtaining pure N protein in soluble form (1).
Actin binding to the N-RNA template does not seem to require other host proteins, including profilin, as we used bacterially produced recombinant actin free of mammalian proteins. We have also shown that profilin by itself has no affinity for N-RNA but uses actin as the docking protein. We conclude that actin serves in two roles: it binds and activates viral RdRP directly, and it recruits profilin to serve as an accessory transcription factor, which then also activates RdRP (Fig. (Fig.4).4). Although profilin and P directly bind in a purified system, as mentioned above (6), we do not know whether the N-RNA-bound P actually helps in recruiting profilin when actin is also present.
Unlike the in vitro transcription system, no in vitro replication system, reconstituted with purified proteins, currently exists for NNR viruses. Nevertheless, as actin and profilin were essential for the basic RNA synthetic process, we presume that they will be important for the viral replication mode as well.
The highly regulated actin dynamics is the driving force behind various cellular properties and processes, including motility, membrane dynamics, and cell-pathogen interactions. These functions require reversible polymerization and depolymerization of cytoplasm actin. The role of actin in gene expression is relatively uncharted territory. Nonetheless, a large body of literature has accumulated over the years, establishing roles for monomeric and polymeric actin as well as a number of actin-regulatory proteins, including profilin, cofilin, gelsolin, supervillin, and thymosin-β4, in nuclear transcription (22, 33, 41). Together, these proteins are believed to regulate cellular chromatin structure and function, perhaps also aiding the motor (translocation) activity of RNA polymerase on the DNA template. It is tempting to assume that they may also stimulate the transcription of large viral DNA genomes such as those of herpes and pox viruses, adenoviruses, and simian virus 40. Overall, the exact mechanism by which actin activates DNA-dependent RNA polymerase is currently unknown. Thus, the domains of actin and profilin that we found to be relevant for the transcription of a cytoplasmic viral RNA genome by the RdRP are the first domains in actin and profilin identified to be involved in transcription.
Profilin has interesting dualism in cytoplasmic actin polymerization: it inhibits the spontaneous nucleation of actin filaments, but in the presence of free barbed ends, it also promotes actin filament assembly (47). Since actin functions as a monomer in viral transcription, the role of profilin in our model (Fig. (Fig.4)4) is independent of its role in actin filamentation, although the same interactive amino acids used in actin filamentation are used to recruit profilin to actin. Our model makes the prediction that profilin, as well as actin, contains a specific polymerase activation domain(s) that makes direct contacts with the viral RdRP (Fig. (Fig.4).4). In the alternative mechanism, profilin simply changes the conformation of actin but does not directly interact with the RdRP. We consider this mechanism unlikely, since current evidence suggests that profilin does not change the structure of monomeric actin (26). However, we do not rule out the possibility that a composite surface, consisting of both actin and profilin residues, makes contact with and activates the RdRP. The absence of a role of the N-terminal acidic domain of actin in activation indeed hints at a potentially novel activation domain(s) that is awaiting discovery. Further mutational analyses to identify such a domain(s) and the exact binding partners are in progress.
This research was supported in part by an NIH grant to S.B. (AI059267), who is also a member of Lions-USA Eye Research Group. M.H. and T.B. were supported in part by a National Science Foundation-Research Experiences for Undergraduates site project on Structure and Function of Proteins (NSF-REU no. 0353562) at the University of South Alabama.
Published ahead of print on 26 August 2009.