|Home | About | Journals | Submit | Contact Us | Français|
The UL84 open reading frame of human cytomegalovirus encodes an essential multifunctional regulatory protein that is thought to act in the nucleus as an initiator of lytic viral replication. Nuclear trafficking of pUL84 is facilitated by a complex nonconventional nuclear localization signal (NLS) that mediates its interaction with the cellular importin-α/β pathway. Since binding of pUL84 to importin-α proteins mechanistically differs from that of cellular proteins containing a classical NLS, we assumed that specific interference with the nuclear import of pUL84 might be possible and that this could constitute a novel principle for antiviral therapy. In order to test this hypothesis, we employed peptide aptamer technology and isolated several peptide aptamers from a randomized peptide expression library that specifically bind with high affinity to the unconventional pUL84 NLS under intracellular conditions. Coimmunoprecipitation experiments confirmed these interactions in mammalian cells, and the antiviral potential of the identified peptide aptamers was determined using three independent experimental approaches. (i) Infection experiments with a recombinant human cytomegalovirus expressing green fluorescent protein demonstrated 50 to 60% decreased viral replication in primary human fibroblasts stably expressing pUL84-specific aptamers. (ii) A 50 to 70% reduction of viral plaque formation, as well as a 70 to 90% inhibition of virus release in the presence of pUL84-specific aptamers, was observed. (iii) Immunofluorescence analyses revealed a shift from an almost exclusively nuclear pUL84 staining pattern to a nucleocytoplasmic distribution upon coexpression of the identified molecules, indicating that interference with the nuclear import of pUL84 contributes to the observed antiviral activity of the identified pUL84-binding aptamer molecules.
Human cytomegalovirus (HCMV) is a widely distributed opportunistic betaherpesvirus with a 30 to 100% seroprevalence in the human population, depending on the socioeconomic status and geographic location of the country (5). Following primary infection, HCMV establishes lifelong latency and periodically reactivates, rarely causing symptoms in healthy individuals. In contrast, the virus still represents a major cause of morbidity and mortality in immunosuppressed patients receiving organ transplants or suffering from AIDS and tumors (5). Furthermore, HCMV is the leading viral pathogen of congenitally infected newborns (2). Although 90% of the congenitally infected infants are initially asymptomatic, a considerable proportion develop sequelae later in life, such as progressive sensorineural hearing loss. This is due to ongoing viral replication, indicating the urgent need for adequate antiviral treatment of these children (1). In addition, increasing evidence suggests that atherosclerotic vascular disease manifestations, such as coronary restenosis or transplant atherosclerosis, are linked to HCMV infection (5). Despite considerable diagnostic and therapeutic progress in recent years, the clinical application of all presently licensed anti-HCMV drugs is limited due to several drawbacks, including toxicity and the emergence of drug-resistant virus strains after prolonged therapy (27, 32). Consequently, new therapeutic strategies, as well as novel antiviral targets, are urgently required to improve the treatment options for life-threatening HCMV infections.
A new potential target candidate for antiviral therapy is the absolutely essential multifunctional regulatory protein encoded by the open reading frame UL84 of HCMV. pUL84 is a protein with nuclear localization that has been proposed to act during initiation of viral-DNA synthesis (25, 34, 45, 47, 48). Initially, pUL84 was identified as a direct binding partner of the regulatory protein IE2-p86, which is the major transcription-activating protein of HCMV (38). Studies concerning the functional consequences of the pUL84-IE2 interaction revealed on one hand that this interaction downregulates the transactivation of IE2 on some early promoters (17). On the other hand, it has been reported that this pUL84-IE2 complex is required for the activation of a bidirectional promoter located within the origin of lytic DNA replication (ori-Lyt) (46). Since pUL84 was the only noncore protein required for origin-dependent DNA replication in a transient-replication assay (35, 37) and it interacts directly with an RNA stem-loop sequence within the RNA/DNA hybrid region of ori-Lyt (11), pUL84 was proposed to act as an initiator protein for viral DNA synthesis of HCMV (46). Initiator proteins of some other herpesviruses were demonstrated to exert an inherent catalytic activity that may unwind a specific region of DNA within ori-Lyt, thus allowing the assembly of the DNA replication machinery. In line with this, pUL84 has been shown to display UTPase activity and to exhibit homology to the DExD/H box family of helicases (13). Furthermore, pUL84 shuttles between the nucleus and the cytoplasm, which is a characteristic that is shared by a subgroup of DExD/H-box-containing proteins (23).
To perform all these activities, pUL84 interacts with several proteins of cellular and viral origin (15). In addition to its capability to oligomerize (12) and to bind the viral transactivator IE2-p86 (38), pUL84 was demonstrated to interact with the viral polymerase accessory protein pUL44, which is part of the DNA replication machinery, and the tegument protein pp65 (15). Furthermore, pUL84 binds with high affinity to importin-α proteins, thus utilizing the cellular importin-α/β pathway for nuclear translocation (15, 25). Since herpesvirus DNA replication occurs within the nucleus, this interaction is a prerequisite for pUL84 to initiate viral-DNA replication. Interestingly, this interaction appears to be unique due to the fact that pUL84 uses a nonconventional nuclear localization signal (NLS) of 282 amino acids (aa) to bind to importin-α proteins via a domain that differs from the binding pocket for cellular proteins containing classical NLS motifs (25). Furthermore, pUL84 was shown to bind to the mitochondrial protein p32, as well as to casein kinase II (15, 16). Taking the data together, the ability of this viral factor to interact with proteins of viral and cellular origin appears to be essential for the performance of its functions. Thus, we hypothesized that blocking the capability to interact with viral and cellular binding partners should impede pUL84 functions and thus should result in inhibition of HCMV replication.
One approach to specifically interfere with protein-protein interactions is the recently developed peptide aptamer (PA) technology (28). PAs consist of short random peptide sequences that are displayed on the surface of an inert scaffold protein, resulting in a constrained conformation of the molecules (10, 21). Similarly to intracellular antibodies (8), PAs can bind with high affinity and specificity to target proteins under intracellular conditions (6, 7, 10) and represent a powerful method to inactivate protein functions in vitro and in vivo (20, 22).
In this study, we report on the selection of PAs directed against the NLS domain of pUL84 that efficiently inhibit both HCMV replication and virion production in primary human fibroblasts. Thus, these results contribute to the validation of the UL84 protein as a suitable target for antiviral therapy.
Primary human foreskin fibroblasts (HFFs) and HeLa and HEK 293T cells were cultured as described previously (18). HFFs stably expressing PAs were generated via retroviral transduction, using the Vira-Power lentiviral system according to the manufacturer's protocol (Invitrogen, Karlsruhe, Germany), and were maintained in Dulbecco's minimal essential medium supplemented with 10% fetal calf serum and 2 μg/ml blasticidin. Stocks of wild-type (wt) HCMV, strain AD169, and recombinant HCMV AD169-GFP (expressing enhanced green fluorescent protein [GFP]) were propagated using HFFs and titrated via IE1p72 fluorescence exactly as described previously (30, 31). HeLa and HEK 293T cells were transfected via the calcium phosphate coprecipitation procedure as published previously (18). MTS cell proliferation/cytotoxicity assays were performed as described by the manufacturer (Promega, Mannheim, Germany).
Oligonucleotides used for cloning and sequencing reactions were purchased from Biomers (Ulm, Germany), and the sequences are listed in Table Table1.1. Randomized 60-bp oligonucleotides for the generation of the PA library were kindly provided by F. Hoppe-Seyler (Heidelberg, Germany). To generate the bait plasmid pHM2319, the EcoRI/NheI-digested NLS sequence of UL84 was inserted into the EcoRI/SpeI-digested Saccharomyces cerevisiae vector pPC97 (43), thus destroying the SpeI site. Full-length UL84 was amplified by PCR with wt UL84 as a template using the oligonucleotides 5′-EcoRV-EcoRI-UL84 and 3′UL84XbaI. Subsequently, the EcoRI/XbaI fragment was introduced into the yeast bait vector pPC97 via EcoRI/SpeI, resulting in plasmid pHM2496. To construct suitable control plasmids, importin-α3 was cloned via BglII either into the yeast bait vector pPC97 or into the prey vector pPC86 (43), resulting in pHM2622 or pHM2335, respectively. Humanized GFP and UL43 were subcloned into pPC97 using SalI/NotI or BglII/NotI to obtain the plasmids pHM2652 and pHM2653. IE1, as well as an IE2 fragment comprising nucleotides 541 to 1876, was inserted into pPC97 via EcoRI or BglII restriction sites, resulting in plasmids pHM2655 and pHM2676, respectively. To generate pHM2678, the NLS domain of the simian virus 40 (SV40) large T antigen fused to β-galactosidase was introduced via EcoRI/BamHI into pPC97. The eukaryotic expression plasmids pCFN-beta-Gal, pcDNA-UL84, wtUL84 (pHM446), wtUL69 (pHM160), wtIE2 (pHM1766), and F-IE2 (pHM705) were described previously (17, 17, 24, 25, 44).
In order to generate eukaryotic expression plasmids of the isolated PAs, EcoRI/XbaI PCR fragments were produced with yeast plasmids containing thioredoxin, Ac, A27, A49, A56, A73, A110, A118, A4, or A8 as a template utilizing the oligonucleotides 5′Thioredoxin-SalI-EcoRI-2 and 3′TRX-rev-XbaI-2. Each fragment was introduced into a FLAG-containing derivative of the eukaryotic expression vector pcDNA3 (pHM971) (18), resulting in the following constructs: pHM2451 (A48), pHM2452 (A27), pHM2454 (A49), pHM2455 (A63), pHM2456 (A73), pHM2465 (A105), pHM2566 (A123), pHM2475 (trx), pHM2542 (Ac), pHM2735 (A56), pHM2738 (A110), pHM2739 (A118), pHM2760 (A4), and pHM2461 (A8). To generate plasmids for the expression of FLAG-tagged PAs in a lentiviral context, the oligonucleotides 5′-FLAG-TRX-SpeI and 3′-TRX-SacII were used for PCR amplification with the respective yeast plasmids as templates. Subsequently, the fragments were introduced into pHM2070, a derivative of the lentiviral vector pLenti6/V5-D-TOPO (Invitrogen, Karlsruhe, Germany), via SpeI/SacII, resulting in plasmids pHM2624 (A73), pHM2631 (Ac), pHM2689 (A49), pHM2715 (trx), pHM2719 (A27), and pHM2628 to -2631 (A56, A110, A118, A4, and A8).
PA screening was performed as described previously (6, 7). In brief, the yeast strain KF1 (MATa trp1-901 leu2-3,112 his3-200 gal4D gal80D LYS2::GAL1-HIS3 GAL2-ADE2 met2::GAL7-lacZ SPAL10-URA3) containing three selectable growth marker genes under the control of GAL4-binding sites was used (7). Wt UL84 or the NLS domain of UL84 (aa 226 to 508) was fused to the GAL4 DNA binding domain and served as bait. As prey, pADtrx was employed to generate a randomized 20-mer PA library fused to the GAL4 activation domain as described in a previous study (7). Briefly, randomized oligonucleotides of 60 bp were inserted into the sequence of Escherichia coli thioredoxin A, resulting in a combinatorial PA library of approximately 1 × 109 different clones. After transformation of KF1 with bait and prey plasmids, yeast clones were selected initially for growth in the absence of adenine. The interactions were verified by retransformation and reselection for growth in the absence of histidine and uracil. The binding specificities of the selected PAs were investigated via cotransformation with heterologous proteins into the yeast strain Y153 and subsequent yeast two-hybrid analyses as described previously (18).
For a GFP-based screening assay, 4 × 105 stably PA-expressing HFFs were seeded in triplicate into six-well plates and infected with HCMV AD169-GFP at a multiplicity of infection (MOI) of 0.025. Eight days postinfection (p.i.), the supernatants were harvested and stored at −80°C for further analyses. Subsequently, the infected cells were lysed and divided into two samples, which were subjected to a Victor 1420 Multilabel Counter (Perkin-Elmer Wallac GmbH, Freiburg, Germany) for automated GFP quantification as described previously (31). To investigate virus release from the stably PA-expressing HFFs, viral supernatants were titrated via IE1p72 fluorescence as described previously (30). For standard plaque assays, 3 × 105 stably PA-expressing HFFs were cultured in six-well plates and infected in triplicate with either 100 IE1-forming units (IEU) of AD169 or 150 IEU of AD169-GFP. After viral adsorption at 37°C for 90 min, virus supernatants were removed and 4 ml overlay medium containing 0.3% agarose was added to the wells. 14 days after infection, the viral plaques were stained with 1% crystal violet, and the number of plaques was determined with a microscope. Each assay was performed at least in triplicate, and standard deviations were calculated.
Coimmunoprecipitation was performed as described previously (26). Briefly, a 50% protein A-Sepharose suspension was incubated with the relevant antibody for 1 h at 4°C and washed three times with NP-40 lysis buffer (50 mM Tris-HCl, pH 8.0; 150 mM NaCl; 5 mM EDTA; 0.5% NP-40; 1 mM phenylmethylsulfonyl fluoride; 1 mg/ml each of aprotinin, leupeptin, and pepstatin) to remove nonbound antibody. Transfected HEK 293T cells were lysed in 800 μl of NP-40 lysis buffer and incubated for 20 min at 4°C, followed by centrifugation. Subsequently, the supernatant was added to the Sepharose beads coupled with antibody and incubated for 2 h at 4°C under rotation. The Sepharose-protein complexes were collected and washed six times in NP-40 lysis buffer. Finally, antigen-antibody complexes were recovered in sodium dodecyl sulfate sample buffer, boiled for 10 min at 95°C, and subjected to Western blot analysis.
The polyclonal antiserum directed against pUL84 of HCMV, as well as the monoclonal antibodies against pUL44, pUL86, and pUL99, have been described previously (17, 30). For detection of PAs, a rabbit polyclonal anti-thioredoxin A antibody from Sigma-Aldrich (Deisenhofen, Germany) was used. Monoclonal antibody p63-27 for detection of IE1, as well as monoclonal antibody 9E10 raised against the myc epitope, were described previously (3, 19). The anti-FLAG monoclonal antibody M2 was obtained from Integra Bioscience (Fernwald, Germany), whereas a monoclonal anti-β-galactosidase antibody was purchased from Roche (Penzberg, Germany). The monoclonal antibody AC-15, directed against β-actin, was obtained from Sigma-Aldrich (Deisenhofen, Germany). While anti-mouse, as well as anti-rabbit, horseradish peroxidase-conjugated secondary antibodies were purchased from Dianova (Hamburg, Germany), Alexa 488- and Alexa 555-conjugated secondary antibodies were ordered from Molecular Probes. For indirect immunofluorescence analyses, which were performed exactly as described previously (40), 3 × 105 transduced HFFs or 4 × 105 HeLa cells per six-well plate were grown on coverslips. Western blotting and chemiluminescence detection were also done as described previously (18). Briefly, extracts from transfected HEK 293T cells or stably PA-expressing HFFs were boiled for 10 min at 95°C in sodium dodecyl sulfate loading buffer, separated on sodium dodecyl sulfate-containing 12.5 to 17% polyacrylamide gels, and transferred to nitrocellulose membranes, followed by chemiluminescence detection.
Results were statistically analyzed using a program for calculation of Student's t test by applying independent samples (http://faculty.vassar.edu/lowry/VassarStats.html). Each assay was performed in triplicate, and experiments were repeated at least three times. A P value of <0.05 was considered to be statistically significant.
In order to identify PAs that specifically interact with the essential viral protein UL84 or its unique NLS domain in vivo, a PA-screening system was employed using the yeast strain KF1, which contains three selectable growth markers: GAL2-ADE2, GAL1-HIS3, and SPO13-URA3 (7). Either the complete 75-kDa UL84 protein (wt UL84) or the complex NLS domain of UL84 (aa 226 to 508; UL84-NLS) (Fig. (Fig.1A),1A), both fused to the GAL4 DNA binding domain, served as bait. A combinatorial randomized PA expression library with high complexity (1 × 109 different clones) fused to the GAL4 activation domain was used as prey. To obtain conformationally constrained PAs, the randomized 20-mers were inserted into the catalytically active loop of the scaffold protein thioredoxin A of E. coli. Two screening rounds using a total of approximately 1 × 108 yeast transformants led to the isolation of 12 different yeast clones exhibiting growth in the absence of adenine. Only two clones were identified using the complete UL84 protein as bait, whereas 10 PAs targeting the NLS domain of pUL84 were selected. The plasmids encoding these PAs were isolated and sequenced, and their corresponding amino acid sequences are listed in Fig. Fig.1B.1B. Retransformation of the identified plasmids into yeast cells, followed by replica platings, revealed the growth of all yeast clones in the absence of histidine (data not shown) and adenine (Fig. 1C and D, ADE2). Additionally, nine yeast transformants were able to grow in the absence of uracil (Fig. (Fig.1D,1D, URA3), which is the most stringent selection marker. Accordingly, one can assume that these molecules are strong interactors with their respective targets (7). The interaction of wt pUL84 or the NLS domain with importin-α3 served as a positive control (Fig. (Fig.1D),1D), whereas a PA randomly chosen from the library (Ac) or thioredoxin itself was employed as a negative control that did not induce growth under any of the selection conditions (Fig. (Fig.1D).1D). In order to test the specificities of the selected PAs for their targets, the isolated plasmids were retransformed into yeast cells in combination with either empty bait vector or heterologous proteins of viral and cellular origin. None of the identified molecules bound to the GAL4-binding domain alone or to any of the heterologous proteins, including importin-α, humanized GFP, SV40 NLS, IE1, or pUL43, in yeast-two-hybrid assays (data not shown). In order to elucidate whether the identified PAs were capable of binding to pUL84 in mammalian cells, we next performed coimmunoprecipitation analyses.
To test the capacities of the selected PAs for interaction with full-length pUL84 in mammalian cells, the respective molecules, including the negative controls Ac and thioredoxin A, were (i) fused to a FLAG tag and (ii) coexpressed in HEK 293T cells, along with myc-tagged wt pUL84. The interaction of FLAG-tagged IE2 with wt pUL84-myc served as a positive control in these experiments (Fig. (Fig.2A,2A, lane 2). For coimmunoprecipitation analyses, either anti-FLAG (Fig. (Fig.2A)2A) or anti-myc (Fig. (Fig.2B)2B) antibodies were used, followed by Western blotting using anti-FLAG, anti-thioredoxin A, anti-pUL84, or anti-myc antibodies. Interestingly, all of the isolated PAs were able to coprecipitate wt pUL84 efficiently, regardless of whether they were selected against full-length pUL84 or targeted only the NLS domain (Fig. (Fig.2A2A and data not shown). Conversely, wt pUL84 was also able to coprecipitate the selected PAs (Fig. (Fig.2B).2B). Interestingly, these experiments demonstrated that PAs A4 and A8, which were initially selected in a yeast two-hybrid screen with full-length UL84, were coprecipitated after expression of the UL84 NLS domain (data not shown). This indicates that all of the isolated aptamers interact with the NLS domain of UL84. The observed interactions were specific, since neither the randomly chosen PA Ac nor thioredoxin A bound to wt pUL84 (Fig. 2A and B). However, since the expression levels of the individual PAs showed strong variations (Fig. 2A and B, bottom), the binding affinities of the selected molecules to their target pUL84 could not be compared by coimmunoprecipitation analyses.
Next, we addressed the question of whether the identified PAs exert inhibitory effects on HCMV replication. Therefore, the respective PA sequences were inserted into a lentiviral expression vector, and stably PA-expressing human fibroblasts were generated via lentiviral transduction. In addition, HFFs containing integrated sequences of the arbitrarily chosen PA Ac or of thioredoxin A without PA were produced as control cells. Selection of blasticidin-resistant cell populations was followed by verification of protein expression via indirect immunofluorescence analysis using an anti-thioredoxin A antibody. As depicted in Fig. Fig.3A,3A, expression of the PAs was detectable in the cytoplasm, as well as in the nuclei, of the transduced cells. We could not observe morphological alterations in PA-expressing HFFs, suggesting there was no major cytotoxicity due to the expression of the respective molecules. However, in order to investigate possible toxic effects of the expressed PAs in more detail, a standard MTS-based cell proliferation assay was conducted. As shown in Fig. Fig.3D,3D, none of the cell lines harboring pUL84-specific PAs showed a reduced rate of proliferation in comparison to control cells expressing thioredoxin A or Ac.
Although all cells exhibited resistance to blasticidin, the percentage of HFFs stably expressing PAs ranged from approximately 50% (Fig. (Fig.3A,3A, a1 to a3 and h1 to h3) to more than 90% (Fig. (Fig.3A,3A, j1 to j3) of the cell population, depending on the individual transduced peptide sequence. However, for some PAs (A48, A63, A105, A123, and A152), the generation of transduced HFFs was not successful, which was probably due to low-level expression of the respective PAs. This was deduced from the observation that PA expression was not detectable by immunofluorescence analyses, although these cells were resistant to blasticidin.
To further characterize the PA expression in the transduced fibroblasts, protein expression was tested by Western blotting, resulting in a notable variation of protein expression levels (Fig. 3B and C). Both findings, the variable number of cells expressing the respective constructs and the different levels of protein expression, were consistent with observations made in HeLa cells transiently transfected with the respective PAs, thus excluding the possibility that these variations were caused by differences in lentiviral transduction efficacy. For subsequent studies, only cell lines that expressed PAs in adequate amounts in more than 80% of the cell population were used. This was done in order to be able to compare the inhibitory effects of the selected PAs on HCMV replication.
As an initial screening assay for the antiviral activities of the PAs, we performed a GFP-based fluorimetric assay (31) that employed a GFP-expressing HCMV strain (AD169-GFP) for a fluorescence-based determination of viral replication in cell lysates. Eight days after infection of the transduced HFFs with AD169-GFP, herpesvirus replication was quantitated by the detection of GFP fluorescence. In addition, supernatants from the infected cells were collected. Remarkably, four PAs (A4, A8, A56, and A110) efficiently decreased HCMV replication up to 70% in comparison to control cells harboring thioredoxin A alone or the randomly chosen PA Ac (Fig. (Fig.4).4). While the antiviral activities of the aforementioned molecules were statistically highly significant, other PAs (A27, A49, and A118) did not influence viral replication in this assay. Since the levels of replication did not differ significantly in HFFs containing either thioredoxin A or Ac, we chose Ac-containing cells as the control cells in all subsequent studies.
In order to confirm the results of the GFP-based assay in alternative test systems, standard plaque assays were conducted using PA-expressing cells infected with either AD169-GFP (150 IEU/well) (Fig. (Fig.5A)5A) or wt AD169 (100 IEU/well) (Fig. (Fig.5B).5B). Subsequently, the number of plaques that developed after infection of HFFs expressing the respective PAs was determined 14 days p.i. In line with the results obtained in the GFP assay, 50 to 75% fewer viral plaques were detected in cells containing the PA A4, A8, A56, or A110 than in control cells. Again, no inhibitory effect on HCMV replication was observed for PA A27. Furthermore, the amount of virus released from infected PA-expressing cells was analyzed. After incubation of native HFFs with supernatants obtained in the GFP assays, the viral titers of the inocula were determined via quantification of IE1-positive cells. Consistent with the findings of GFP and plaque assays, infection of HFFs transduced with PAs A4, A8, A56, and A110 led to a statistically highly significant reduction of virus release, whereas no significant difference was observed for PA A27 in comparison to control cells.
Finally, we wanted to investigate whether the accumulation of viral proteins was affected in the presence of pUL84-binding PAs. For this, PA-expressing or control cells were infected with HCMV AD169 (MOI, 0.5). Cell lysates were prepared at 12, 24, 48, and 72 h p.i. and analyzed by Western blotting using various antibodies against viral proteins. As shown in Fig. Fig.6,6, no major difference in the accumulation of the immediate-early (IE) protein IE1p72 could be observed after infection of PA-expressing and control cells. This excludes the possibility of a difference in the infectibilities of the various cell lines, as well as an effect of PAs on IE gene expression. Interestingly, however, we detected a significantly reduced amount of pUL84 protein in cells expressing the PAs with antiviral activity (A4, A8, A56, and A110) that had already started at 12 h p.i. but could be detected throughout the HCMV replicative cycle (Fig. 6A to D, lanes 4, 5, 7, and 8). In addition, the expression of the viral polymerase accessory protein pUL44, as well as the true late proteins pUL86 and pUL99, was also significantly decreased in A4-, A8-, A56-, and A110-expressing cells compared to either the control cells or A27-expressing cells (Fig. 6B to D, compare lanes 4, 5, 7, and 8 with 2, 3, and 6). Thus, the pUL84-binding aptamers with antiviral activity apparently affect pUL84 protein accumulation, as well as the accumulation of early-late and late proteins, in HCMV-infected cells.
Since all four PAs exhibiting significant antiviral activity were directed against the NLS domain of pUL84, we assumed that these PAs might interfere with the nuclear translocation capacity of pUL84. Thus, in order to elucidate the effects of these PAs on the nuclear import of pUL84, the subcellular distribution of this essential viral protein was visualized in PA-expressing cells by using indirect immunofluorescence. pUL84 appears to be localized in the nuclei of host cells, whereas all PAs were present both in the nucleus and in the cytoplasm. However, upon coexpression in HeLa cells, we observed that all four inhibitory PAs, A4, A8, A56, and A110, were able to substantially alter the intracellular localization of pUL84 from an almost exclusively nuclear to a nucleocytoplasmic staining pattern (Fig. (Fig.7A).7A). This mislocalization phenotype of pUL84 was observed in up to 85% of transfected cells in the presence of pUL84-specific PAs (Fig. (Fig.7B).7B). In contrast, neither thioredoxin A nor Ac mediated changes in the subcellular distribution pattern of pUL84 (Fig. (Fig.7B).7B). To further validate the specificities of the selected PAs, the respective constructs were coexpressed with different nuclear proteins comprising classical mono- or bipartite NLS motifs. Neither the nuclear localization of the viral proteins IE2 and UL69 nor that of a fusion protein consisting of SV40 T antigen-NLS fused to beta-galactosidase was affected by the UL84-specific PAs (Fig. (Fig.7C).7C). Taken together, these findings indicate that four of the isolated PAs (A4, A8, A56, and A110) are indeed able to affect the nuclear import of pUL84 in a highly specific manner.
In order to investigate whether this could also be observed during HCMV infection, fibroblasts expressing either PA A56, which exhibits antiviral activity, or the control aptamer Ac were infected with HCMV AD169 (MOI, 0.1), followed by the detection of pUL84 and pUL44 by indirect immunofluorescence analyses at 24 h p.i. As shown in Fig. Fig.7D,7D, pUL84 was retained in the cytoplasm of A56-expressing cells, while this was not detected in Ac-expressing cells (Fig. (Fig.7D,7D, compare a-2 and b-2). This mislocalization of pUL84 was present in more than 80% of A56-expressing fibroblasts (Fig. (Fig.7E).7E). Thus, we assume that the capacity of the isolated PAs to retain viral pUL84 in the cytoplasm of host cells contributes to their inhibitory impact on HCMV replication.
It is generally accepted that the still unresolved clinical problems with HCMV infections in transplant patients, necessitating repeated and prolonged treatment courses, as well as the requirement for therapy of congenital infections, creates an increasing need for new antiviral drugs. HCMV is the most complex human-pathogenic herpesvirus and encodes approximately 180 proteins, of which 45 were categorized as absolutely essential and 35 were assumed to be critical for efficient viral replication in cell culture (14, 33, 48). Thus, in addition to the well-established antiviral target molecules comprising the viral protein kinase UL97 and the DNA polymerase UL54, HCMV encodes several more target proteins that deserve further evaluation and validation (32). In this report, we focused on the gene product encoded by the open reading frame UL84, which has previously been shown to be essential for the initiation of viral lytic DNA replication (34, 35, 37). Although the exact mechanism of this viral replication factor has yet to be determined, several recent reports have emphasized the importance of protein interactions for functional activity (12, 15, 16, 25). Thus, we hypothesized that interference of PAs with UL84 protein interactions could be utilized to inhibit HCMV replication.
The capability of PAs to interfere with viral replication has previously been investigated for several other viral systems (6, 20, 29, 36, 41, 42). For instance, Butz and colleagues demonstrated the PA-mediated inhibition of hepatitis B virus (HBV) replication by targeting the HBV core protein, which resulted in an abrogation of capsid formation (6). In this study, in vivo interference with protein interactions was assumed to be the underlying mechanism of inhibition. Interestingly, a subsequent publication suggested that PAs may also act via the sequestration of bound proteins into aggresomes, which correlated with a dramatic intracellular redistribution of the respective target protein into perinuclear inclusion bodies (41). However, not only structural proteins, but also regulatory factors are amenable to PA-mediated inhibitory approaches, as was recently shown for the replication proteins AL1 of the geminivirus tomato golden mosaic virus and the A20 protein of the vaccinia virus replication complex (29, 36).
In this study, we reported the isolation of 12 PAs that specifically bind to the replication factor pUL84 of HCMV, as demonstrated both by yeast interaction studies and by coimmunoprecipitation analyses performed with proteins expressed in mammalian cells. All of the isolated PAs interacted with a domain of pUL84 that has previously been defined as the minimal region of the protein required both for interaction with importin-α proteins and for nuclear localization, thus serving as a complex, nonconventional nuclear localization domain (25). Sequence comparison of the isolated PAs revealed neither significant homology to known viral or cellular proteins nor detectable homology to each other. The latter finding may indicate that the individual aptamers bind to different epitopes within the large pUL84 NLS domain, which comprises aa 226 to 508. Unfortunately, fine mapping of aptamer binding sites within pUL84 has not been possible so far, since further N- or C-terminal deletions of the NLS domain completely abrogated the interaction with all isolated PAs, as well as binding to importin-α (reference 25 and data not shown). This strongly suggests the existence of a folded protein structure that is destroyed by N- or C-terminal deletion mutagenesis. Thus, additional structural information on the UL84 NLS domain will be required to further localize individual aptamer binding sites.
In order to investigate whether the isolated PAs were able to inhibit HCMV replication, stably PA-expressing primary human fibroblasts were generated via retroviral transduction. This revealed considerable variability in the protein expression levels of different PAs, which was also observed in transient-transfection experiments. The reason for this is not entirely clear but may be related to the fact that PA-encoding sequences were selected from random nucleotide sequences, and thus, the codon usage of some aptamers may not be adapted to eukaryotic cells. Alternatively, since we noted that the prokaryotic expression levels of these proteins also varied (data not shown), one may speculate that the insertion of specific sequences into the catalytic domain of the scaffolding protein thioredoxin A is associated with a destabilization of the respective fusion protein. Variations of cellular expression levels of thioredoxin A scaffolds presenting different peptide moieties were also noted in a recent study investigating aptamers binding to the duck HBV core protein and are thus not confined to pUL84-binding PAs (41).
In infection experiments with stably PA-expressing fibroblasts, we detected statistically significant inhibition of viral replication by PAs A4, A8, A56, and A110 in three independent assay systems. Inhibition of viral replication was more pronounced in plaque reduction and virus release assays than in the GFP-based antiviral test system that was used for the initial screening of PA-mediated antiviral effects. This may be due to the fact that the GFP-based antiviral assay makes use of a recombinant CMV that expresses the gene for enhanced GFP under the control of the major IE promoter of HCMV (31). Since pUL84 is assumed to be important mainly during the late phase of the viral replication cycle (34), PA-mediated UL84 inhibition may be less distinct in a test system that is based on the quantification of IE gene activation. However, compared to the antiviral effect exerted by PAs in other viral systems, such as the recently described PA-mediated inhibition of vaccinia virus replication, which resulted in a maximum virus yield reduction of approximately 60% (36), the inhibitory effects of pUL84-binding PAs were even stronger. Since cell proliferation assays excluded the possibility that the intracellular expression of PAs affected the viability of cells, we conclude that the pUL84-binding PAs A4, A8, A56, and A110 exhibit strong anti-HCMV activity.
Our tests revealed that only a subset of pUL84-binding PAs significantly impeded HCMV replication. This was not a surprising finding, since PAs have the benefit of recognizing individual domains of a pleiotropic protein; however, only a subset of binding aptamers may be able to interfere with specific activities of a protein. For instance, Butz et al. described the selection of eight PAs that interact with the HBV core protein, but only one of them efficiently inhibited viral capsid formation and, consequently, HBV replication (6). Thus, we assume that the antiviral activities of PAs A4, A8, A56, and A110 indicate interference with functions or interactions of pUL84 that are critical for efficient HCMV replication.
Consistent with the assumption that PA-mediated interference with important pUL84 functions should inhibit late viral-gene expression, we observed severely decreased expression of the true late proteins UL86 and UL99 after infection of A4-, A8-, A56-, or A110-expressing cells in comparison to control cells or cells expressing aptamers without antiviral effects (Fig. (Fig.6).6). The fact that accumulation of the IE protein IE1p72 was not diminished indicates that there was no general decrease in the infectibility of the generated primary fibroblast lines. Interestingly, we observed significantly decreased protein levels of pUL84 in the presence of UL84-binding PAs with antiviral activity. This was already detected at early times of the replicative cycle, suggesting that UL84-binding PAs may be able to modulate either the expression or the stability of pUL84.
Since all aptamers were directed against the nuclear localization domain of pUL84, we tested whether the nuclear import of pUL84 was impeded. Indeed, cytoplasmic retention of pUL84 was detected after coexpression of pUL84-binding aptamers and after infection of PA-expressing fibroblasts, which was not observed for several other nuclear proteins, indicating specific mislocalization of pUL84 by interacting PAs. Of note, the cytoplasmic pUL84 accumulations were diffuse and morphologically distinct from aggresomes, thus excluding the recently described PA-mediated aggresome formation as the inhibitory mechanism (41). Rather, we propose that PAs binding with high affinity to the UL84 NLS domain may abrogate the interaction with importin-α proteins and thus block the nuclear translocation of this viral replication factor. However, the possibility that other protein interactions of pUL84 are also affected by pUL84-binding PAs cannot be excluded, and this may well contribute to the observed antiviral activity. For instance, recent publications have described the interaction of pUL84 with the DNA polymerase processivity factor UL44; however, the UL84-binding domain that mediates this interaction has not been determined (15, 39). Thus, additional experiments will be necessary to determine to what extent specific protein interactions of pUL84 are abrogated by antiviral PAs.
Although the direct application of the identified PAs as biotherapeutic molecules, which would require efficient intracellular delivery to HCMV-infected cells, appears unlikely, UL84-binding PAs could be extremely useful to guide the discovery of small-molecule drugs. Since aptamers with antiviral activity define a specific molecular surface with critical function on a target protein, a high-throughput screening assay to identify small molecules that disrupt the target-aptamer interaction could be performed (4, 9). In this way, small molecules that trigger the same biological effect as the aptamer might be selected (9). In this respect, a further characterization of the binding pockets of pUL84-binding aptamers appears to be highly desirable. In summary, this study describes the identification of pUL84-binding PAs that show a clear antiviral effect. This further contributes to the validation of pUL84 as a valuable target protein for antiviral approaches and opens new perspectives for the development of novel screening strategies to identify small-molecule inhibitors of HCMV replication.
We thank K. Butz (Heidelberg, Germany) and F. Hoppe-Seyler (Heidelberg, Germany) for providing reagents and for valuable advice concerning the establishment of the aptamer-screening system.
This work was supported by the Wilhelm Sander Stiftung, the IZKF Erlangen, SFBs 473 and 796, and GRK1071.
Published ahead of print on 9 September 2009.