Sensitivity of the C-Terminal Nuclease Domain of Kaposi's Sarcoma-Associated Herpesvirus ORF29 to Two Classes of Active-Site Ligands

Kaposi's sarcoma-associated herpesvirus (KSHV), the etiological agent of Kaposi's sarcoma, belongs to the Herpesviridae family, whose members employ a multicomponent terminase to resolve nonparametric viral DNA into genome-length units prior to their packaging. Homology modeling of the ORF29 C-terminal nuclease domain (pORF29C) and bacteriophage Sf6 gp2 have suggested an active site clustered with four acidic residues, D⁴⁷⁶, E⁵⁵⁰, D⁶⁶¹, and D⁶⁶², that collectively sequester the catalytic divalent metal (Mn²⁺) and also provided important insight into a potential inhibitor binding mode.

ABSTRACT

Kaposi's sarcoma-associated herpesvirus (KSHV), the etiological agent of Kaposi's sarcoma, belongs to the Herpesviridae family, whose members employ a multicomponent terminase to resolve nonparametric viral DNA into genome-length units prior to their packaging. Homology modeling of the ORF29 C-terminal nuclease domain (pORF29C) and bacteriophage Sf6 gp2 have suggested an active site clustered with four acidic residues, D⁴⁷⁶, E⁵⁵⁰, D⁶⁶¹, and D⁶⁶², that collectively sequester the catalytic divalent metal (Mn²⁺) and also provided important insight into a potential inhibitor binding mode. Using this model, we have expressed, purified, and characterized the wild-type pORF29C and variants with substitutions at the proposed active-site residues. Differential scanning calorimetry demonstrated divalent metal-induced stabilization of wild-type (WT) and D⁶⁶¹A pORF29C, consistent with which these two enzymes exhibited Mn²⁺-dependent nuclease activity, although the latter mutant was significantly impaired. Thermal stability of WT and D⁶⁶¹A pORF29C was also enhanced by binding of an α-hydroxytropolone (α-HT) inhibitor shown to replace divalent metal at the active site. For the remaining mutants, thermal stability was unaffected by divalent metal or α-HT binding, supporting their role in catalysis. pORF29C nuclease activity was also inhibited by two classes of small molecules reported to inhibit HIV RNase H and integrase, both of which belong to the superfamily of nucleotidyltransferases. Finally, α-HT inhibition of KSHV replication suggests ORF29 nuclease function as an antiviral target that could be combined with latency-activating compounds as a shock-and-kill antiviral strategy.

INTRODUCTION

Kaposi's sarcoma-associated herpesvirus (KSHV) is the causative agent of Kaposi's sarcoma (KS), the most common AIDS-associated cancer, with significant morbidity and mortality in many subequatorial countries (1–4). KSHV belongs to the gammaherpesvirus subfamily, members of which are associated with tumorigenesis of lymphoma, epithelial cancers such as nasopharyngeal carcinoma, and vascular endotheliosarcomas such as KS (5). KSHV also causes primary effusion lymphomas and multicentric Castleman's disease. Effective virus-specific therapies are currently unavailable, leaving these frequently lethal diseases largely incurable. An antiviral strategy is also complicated by the fact that KSHV is primarily latent in the host in the form of viral episomes (6) which must be reactivated to enter the lytic cycle (7). KSHV latent transcripts contribute to oncogenesis by driving cell proliferation and preventing apoptosis, but lytic infection also plays an important role in KS pathogenesis (1). Thus, a strategy that combines KSHV activation with targeting important proteins/enzymes might be considered.

Herpesviruses have a linear double-stranded DNA (dsDNA) genome encased in an icosahedral capsid (8–12). Nascent viral DNA in host cells exists as multiple copies of the unit-length viral genome. Concatemers are resolved into genome-length units and inserted into a capsid by the virus-coded terminase (8, 10, 12) through a mechanism similar to that of tailed dsDNA bacteriophages. Bacteriophage terminase is typically comprised of a DNA recognition catalytic component, designated the small (TerS) and large (TerL) subunits (13–16). TerS proteins are thought to specifically bind concatameric viral DNA, triggering cleavage by nuclease activity in TerL, which generates a new DNA terminus. This terminus is threaded through the portal protein channel embedded in the procapsid, and DNA is pumped into the procapsid by TerL in an ATP-dependent manner. Once an appropriate genome-length unit is inserted, TerL cleaves the DNA again, dissociating from the procapsid to initiate the next packaging cycle. Despite clear similarities to tailed dsDNA bacteriophages in mechanisms of genome packaging, capsid assembly, and DNA injection into host cells (8, 11), herpesvirus DNA packaging exhibits extra complexity due to the tripartite composition of the packaging machineries (8, 11, 12). In the prototypic herpes simplex virus 1 (HSV-1), terminase comprises three subunits, namely, pUL15, pUL28, and pUL33 (12). Among them, pUL15 is a homolog of TerL (17) and pUL28 binds to the pac1 signal (18), and a role for pUL33 in assembly of the terminase holoenzyme has been proposed (19). pUL33, pUL28, and pUL15 interact in vitro and in infected cells (20, 21), and both pUL28 and pUL15 interact with the portal protein pUL6 (22, 23). pUL15, pUL28, and pUL33 form a complex in the host cytoplasm that is imported into the nucleus via the nuclear localization signal on pUL15 (19). Similar molecular interactions were observed for terminase subunits pORF25, pORF30, and pORF42 in varicella-zoster virus (24, 25) both among pUL51, pUL56, and pUL89 and between pUL56 and the pac motif in human cytomegalovirus (HCMV) (26–29).

Their essential role in virus replication illustrates herpesvirus terminase proteins as promising antiviral targets (12, 30–36). Structures of the TerL nuclease domains of HCMV (37) and HSV-1 (17) showed conserved folds resembling those of bacteriophage terminase large subunits (38) and RNase H-like nucleotidyltransferases (17). However, KSHV terminase proteins have not been experimentally characterized. In the present work, we purified and characterized the C-terminal domain of KSHV ORF29 (pORF29C) and mutants whose proposed active-site residues were substituted. Our combined biochemical analyses, including susceptibility to α-hydroxytropolone (α-HT) inhibition, confirm pORF29C as a member of the nucleotidyltransferase superfamily of nucleases.

Finally, since establishing KSHV persistent infection reflects a balance between latency, where most genes are silenced, and a lytic cycle, where nearly all are expressed, kick-and-kill strategies (39) to skew the latency-to-lytic cycle toward lytic activation are receiving increased attention as a means of activating host immune activation and viral clearance. As an example, KSHV reactivation following knockdown of Tousled-like kinases (40) supports development of small-molecule inhibitors targeting these kinases as lytic inducers. As a first step in this direction, we also demonstrate α-HTs as a novel class of KSHV replication inhibitors.

RESULTS

Modeling of the pORF29C active site.

An experimental three-dimensional (3D) structure of KSHV pORF29C is currently unavailable. However, since pORF29 shares 33.8% sequence identity with HSV-1 pUL15, we generated a homology model of pORF29C with I-TASSER (41) based on the HSV-1 pUL15C structure (17) as the template. The C-score reported by I-TASSER was 0.96, indicating good reliability of the model, showing an overall fold that is superimposable with its HSV-1 and HCMV homologs. The X-ray structure of bacteriophage Sf6 large terminase nuclease domain gp2C complexed with β-thujaplicinol (42) next was superimposed onto the KSHV pORF29C model (Fig. 1A). Overall folds of the two structures are superimposable, with a root mean square deviation of 1.05 Å for 45 Cα atoms. The pORF29C active site (Fig. 1B) comprises four acidic residues, D⁴⁷⁶, E⁵⁵⁰, D⁶⁶¹, and D⁶⁶², which are highly conserved in herpesvirus large terminases (17, 37) and fit well with D²⁴⁴, D²⁹⁶, and N⁴⁴¹ in gp2C (42). The pORF29C active site is located at the bottom of a crevice approximately 7 Å wide and 9 Å deep, surrounded by loop P⁴⁷⁷-G⁴⁸⁷, loop E⁵⁵⁰-S⁵⁵⁴, and loop Q⁶⁵⁰-S⁶⁶⁰ (Fig. 1B). The α-hydroxytropolone (α-HT) β-thujaplicinol occupies the crevice of the active site, making few, if any, direct contacts with adjacent protein atoms. This indicates that β-thujaplicinol binding to the pORF29C active site is mediated primarily by interactions with the two metal ions. Guided by this modeling exercise, pORF29C active-site residues D⁴⁷⁶, E⁵⁵⁰, D⁶⁶¹, and D⁶⁶² were selected for replacement with Ala followed by biochemical characterization.

FIG 1.

Modeling binding of small-molecule inhibitors at the predicted pORF29C active site. (A) Superimposition of the I-TASSER model of KSHV pORF29C (ribbon diagram in cyan) onto the Sf6 gp2C (ribbon diagram in gray) complexed with β-thujaplicinol (stick model in magenta). Active-site residues are shown in stick models in both structures. (B) Closeup of the active site. The view is 90° from that in panel A.

Expression and purification of the KSHV pORF29 C-terminal nuclease domain.

The wild-type (WT) KSHV pORF29 C-terminal domain (residues 442 to 687; pORF29C) and active-site mutants were expressed in Escherichia coli and purified by a combination of immobilized metal affinity and size exclusion chromatography. All proteins were estimated to be at least 90% homogeneous, migrating primarily as monomers when analyzed by size exclusion chromatography (SEC) (see Fig. S1A in the supplemental material). Although slightly higher levels of impurity were noted for mutant E⁵⁵⁰A (Fig. S1B), later data indicate that it is devoid of metal binding and nuclease activity.

Purified pORF29C exhibits sequence-independent nuclease activity.

Supercoiled plasmid DNA has been previously used to assay the nuclease activities of HCMV and HSV-1 large terminases (17, 37). However, the high sensitivity of an assay relying on cleavage of supercoiled DNA necessitates a complementary assay to eliminate the possibility of low-level contamination with bacterial nucleases. We therefore examined pORF29C nuclease activity on a short (27-nucleotide [nt]/28-nt) 5′, fluorescently end-labeled DNA duplex (Fig. 2A) in the absence and presence of an α-HT previously documented to inhibit activity of a related herpesvirus nucleotidyltransferase. In the absence of the αHT, pORF29C-derived cleavage products ranged in size from 20 to 29 nt, with 17-, 16-, 13-, and 9-nt products predominating (Fig. 2B). Based on the duplex DNA substrate, there appeared to be little sequence specificity. Under the same conditions, DNase I digestion yielded a significantly different hydrolysis pattern, with predominant products of 14, 7, and 5 nt. The pORF29C preparation was thus considered free of contaminating DNase. The absence of nuclease contamination was further strengthened when the sensitivity of pORF29C and DNase I to αHT treatment was examined. Under these conditions, nuclease activity of pORF29C was eliminated, while that of DNase I was unaffected.

FIG 2.

Sequence-independent cleavage of duplex DNA by WT pORF29C. (A) Sequence of the 27-nt/28-nt DNA oligonucleotide used as a linear pORF29C substrate. Duplex DNA was fluorescein labeled on the 5′ terminus of the upper strand, as indicated by the asterisk. (B) pORF29C cleavage profiles of the 27-nt/28-nt DNA substrate. Lanes C, control input DNA. Lanes O, incubation of duplex DNA with purified pORF29C. Lanes D, incubation of duplex DNA with DNase I. All incubations were for 15 min at 37°C. Panel notations −αHT and +αHT depict nuclease digestions performed in the absence or presence of compound 28a, an α-hydroxytropolone inhibitor of pORF29C, whose structure is indicated below the panel.

Binding of divalent metal to pORF29C variants.

Previously, we demonstrated differential scanning fluorimetry (DSF) as a facile strategy to evaluate both divalent metal and inhibitor binding to HIV RNase H (43) and HSV pUL15C (44) via stabilization against thermal denaturation. As a first step in characterizing pORF29C variants, we determined their thermal stability in the absence and presence of 5 mM MnCl₂ (Fig. 3A and B). For the wild-type enzyme, we observed a melting temperature (T_m) of 54.8°C in the absence of divalent metal and 59.5°C in its presence (ΔT_m of 4.7°C). Although thermal stability of mutant D⁶⁶¹A was slightly reduced in the absence of divalent metal (T_m of 50.1°C), inclusion of Mn²⁺ likewise afforded increased thermal stability (T_m of 56.9°C, ΔT_m of 6.8°C). T_m differences between WT and D⁶⁶¹A pORF29C suggest that while both support divalent metal binding, active-site architecture of the latter is compromised. A minor Mn²⁺-induced thermal shift was observed for mutant E⁵⁵⁰A (T_m of 52.9°C versus 53.9°C, ΔT_m of 1.0°C), while we observed a destabilizing effect for pORF29C mutants D⁴⁷⁶A (T_m of 59.7°C versus 59.1°C, ΔT_m of −0.6°C) and E⁶⁶²A (T_m of 57.4°C versus 56.2°C, ΔT_m of −1.2°C). Thus, to a first approximation, altering pORF29C active-site residues E⁵⁵⁰, D⁴⁷⁶, and E⁶⁶² leads to significant impairment of divalent metal binding. However, data in the next section illustrate that retention of divalent metal binding is not necessarily compatible with retention of catalytic activity.

FIG 3.

Active-site mutations affect enzyme stability and activity. (A) Representative thermal denaturation profiles for WT and mutant pORF29C in the absence (gray trace) and presence of 5 mM MnCl₂ (black trace). (B) Summary of thermal denaturation data.

Nuclease activity of pORF29C variants.

Although data shown in Fig. 2 indicate that small oligonucleotide duplexes can be used to interrogate pORF29C nuclease activity, degradation of supercoiled DNA has routinely been used by others as a measure of herpesvirus terminase nuclease activity (17, 37, 44). We therefore used this strategy to examine WT and mutant KSHV pORF29C variants, fractionating the products by agarose gel electrophoresis and visualizing by ethidium bromide staining (Fig. 4). The supercoiled plasmid substrate (ccc) contained only traces of open circular (oc) DNA, and short incubation with WT pORF29C gave rise to predominantly oc and linear (l) DNA. Following extended incubation, a smear of heterogeneously sized products was observed. Interestingly, although DSF predicted that mutant D⁶⁶¹A bound divalent metal at the active site, this mutant displayed significantly reduced nuclease activity, evident as an increase in the level of oc DNA and trace amounts of the linear form. Although this can only be answered by high-resolution structural analysis, the combined thermal stability and nuclease activity for D⁶⁶¹A pORF29C strongly suggests that while divalent metal occupies the active site, altered coordination geometry leads to severely reduced nuclease activity. Consistent with minor stabilization of mutant E⁵⁵⁰A by Mn²⁺, this also catalyzed a slight increase on the oc and linear plasmid forms, while the digestion profiles from mutants D⁴⁷⁶A and E⁶⁶² were suggestive of complete loss of nuclease function.

FIG 4.

Assessment of nuclease activity of pORF29C and active-site mutants by cleavage of supercoiled DNA (−) and resolution by agarose gel electrophoresis. Time points for each reaction are 20 and 35 min at 37°C (a and b, respectively). Migration positions of supercoiled (ccc), open circular (oc), and linear DNA (l) are indicated.

α-HTs inhibit pORF29C nuclease activity.

Extending our finding that αHTs were effective inhibitors of HSV pUL15C in vitro (44) and viral replication in vivo (45, 46), we examined inhibition of pORF29C by nine chemically synthesized β-thujaplicinol analogs (47–49) (Fig. 5). Visual inspection showed that compounds 4a, 4e, and 4f demonstrated dose-dependent inhibition of pORF29C nuclease activity. Compound 4a displayed modest inhibition at 0.8 μM (Fig. 5, 4a, lane 3), and at 4.0 μM, most of the DNA substrate was refractory to cleavage (Fig. 5, 4a, lane 2). Although data shown in Fig. 5 are qualitative, compound 4e displayed a similar inhibitory effect (Fig. 5, 4e, lanes 1 to 5). Based on recovery of covalently closed circular DNA at an inhibitor concentration of 4.0 μM, compounds 4f, 92a, and 92c were slightly less active, while minimal closed circular DNA accumulated with the remaining compounds. Although we have been unable to implement the quantitative dual-probe fluorescence assay we developed for HSV-1 pUL15C (44), data shown in Fig. 5 demonstrate that KSHV pORF29C nuclease activity is susceptible to αHT inhibition. However, future screening efforts would benefit from a more facile approach that can be conducted at higher throughput. DSF presents a viable option, and our recent work with HSV pUL15C demonstrated a very good relationship between ΔT_m and 50% inhibitory concentration (IC₅₀) (44). Inhibitors tested in Fig. 5 were therefore reexamined with respect to alterations they induce in thermal stability (Fig. 6A and B). A comparison of Fig. 5 and 6 indicates this relationship also holds true for KSHV pORF29C. Visual inspection of nuclease activity in Fig. 5 suggests α-HTs 4a and 4e most significantly affect nuclease activity, and Fig. 6B indicates these induce the most significant change in pORF29C thermal stability (ΔT_m of 7.98 ± 0.38°C and 6.57 ± 0.38°C, respectively). Conversely, agarose gel electrophoresis suggests compound 92d is the least effective α-HT, in keeping with only a modest alteration in thermal stability (ΔT_m of 0.82 ± 0.23°C). Thus, as we and others have proposed, DSF, which can be conducted on a 384-well basis, could represent a low-cost alternative to agarose gel electrophoresis for screening of herpesvirus nucleotidyltransferases.

FIG 5.

α-HTs inhibit DNase activity of WT pORF29C. Nuclease activity by select αHTs, analyzed by 1% agarose gel electrophoresis and SYBR gold staining, is presented. Reactions were terminated with EDTA after 35 min at 37°C. Lane a, plasmid DNA control. Lane b, no inhibitor. Each compound was tested at a final concentration of 20.0, 4.0, 0.80, 0.16, 0.032, and 0.006 μM (lanes 1 to 6, respectively). The migration positions of closed circular, open circular, and linear plasmid DNA are designated ccc, oc, and l, respectively. Structures of the compounds are detailed at the right of each gel panel.

FIG 6.

Thermal stabilization of WT ORF29C by active-site α-HT inhibitors. (A) Representative thermal stability profiles in the absence (C) and presence of αHTs 4a, 83a, and 92a (Fig. 5 shows the structures). (B) ΔT_m values for WT pORF29C in the presence of αHT inhibitors illustrated in Fig. 5. Results are averages from triplicate analyses.

Interaction of pORF29C variants with an α-HT inhibitor.

High-throughput screening identified the α-HTs β-thujaplicinol and manicol as active-site inhibitors of HIV-1 reverse transcriptase-associated RNase H activity (50), subsequent to which derivatives have been demonstrated to inhibit replication of herpes simplex virus 1 and 2 (45, 46) as well as hepatitis B virus (51). Mechanistically, and as proposed in Fig. 1, α-HTs sequester divalent metal at the active site. At the same time, α-HT binding to nucleotidyltransferases is in many cases accompanied by increased thermal stability, varying from 2.3 to 8.7°C (44). We therefore used DSF to interrogate binding of α-HT 4a (Fig. 5) to pORF29C mutants, the results of which are presented in Table S1. For WT pORF29C, inhibitor binding provided a ΔT_m of 7.8°C (59.7°C versus 67.5°C), while the ΔT_m for mutant D⁶⁶¹A (6.1°C) was of comparable magnitude (56.2°C versus 62.3°C). ΔT_m values for the remaining three mutants were only minimally affected in the presence of inhibitor, indicating that metal ion occupancy at the active site is a prerequisite for inhibitor binding. The combined divalent metal and inhibitor binding data for mutant D⁶⁶¹A indicate that active-site architecture needs only be slightly perturbed to influence catalysis.

pORF29C nuclease activity is also inhibited by HIV integrase inhibitors.

Raltegravir (RAL), elvitegravir (ELV), and dolutegravir (DOL) are first- and second-generation DNA strand-transfer inhibitors targeting HIV-1 integrase (IN) and are in clinical use (52, 53). Since HIV integrase supports a two-metal-ion catalytic mechanism analogous to HIV RNase H (54) and terminase nuclease domains from herpesviruses (17, 37) and dsDNA bacteriophages (42), we examined if KSHV pORF29C nuclease activity was also sensitive to these chemotypes. Under our experimental conditions, RAL exhibited dose-dependent inhibition that was clearly visible at a concentration of 0.16 μM (Fig. 7, lane 4). At 4 μM RAL, >50% of the supercoiled plasmid DNA remained uncut (Fig. 7, lane 6). DOL appeared to have a significantly stronger inhibitory effect (Fig. 7, lanes 8 to 12), displaying clear inhibition at a concentration as low as 0.032 μM (Fig. 7, lane 8), with most of the substrate DNA uncut at 4 μM (Fig. 7, lane 11). Curiously, although ELV conferred inhibition of pORF29 nuclease activity at a concentration as low as 0.032 μM (Fig. 7, lane 13), activity was minimally enhanced at elevated concentrations, suggesting a weak inhibitory effect. Collectively, our data indicate that of the currently available HIV-1 IN inhibitors, DOL has the strongest inhibitory effect on pORF29C nuclease activity. DSF analysis indicated that 20 μM DOL resulted in a ΔT_m of 2.5°C. The equivalent concentration of RAL afforded a smaller degree of stabilization (ΔT_m of 1.5°C). For DOL and RAL, we again observed a correlation between ΔT_m and inhibitory potency (Fig. 7). In keeping with its reduced efficacy in inhibiting pORF29C nuclease activity, ELV had no effect on enzyme thermal stability. Modeling data (Fig. S2) suggest that independent of the orientation in which ELV is docked into the pORF29C active site, it invokes a steric clash. Our observations are consistent with those of Nadal et al., who have investigated diketo acid-dependent inhibition of HCMV pUL89 and found ELV was inactive (37).

FIG 7.

Inhibition of KSHV ORF29C nuclease activity by HIV-1 IN antagonists. Raltegravir (RAL; lanes 3 to 7), dolutegravir (DOL; lanes 8 to 12), and elvitegravir (EVG; lanes 13 to 16). For each panel, IN antagonists were evaluated at final concentrations of 0.032, 0.16, 0.8, 4.0, and 20.0 μM. P, plasmid DNA; −I, pORF29C activity in the absence of IN inhibitor. Reactions were terminated with EDTA after 15 min at 37°C. ΔT_m values measured by DSF for pORF29C in the presence of IN inhibitor are provided with each structure.

α-HT inhibition of KSHV replication.

In previous studies, we and others demonstrated α-HT inhibition of both human (45) and animal herpes simplex virus replication (55). To date, however, their activity against gammaherpesviruses, and in particular KSHV, had not been examined. Based on our HSV studies, sensitivity of KSHV to inhibition by α-HT 28a (structure indicated in Fig. 2) was examined in the inducible TREx BCBL-1-Rta cell line, the results of which are provided in Fig. 8. TREx BCBL-1-Rta cells contain a doxycycline (DOX)-inducible replication and transcription activator (Rta) that activates viral lytic replication (56). As predicted, treatment of TREx BCBL-1-Rta cells with DOX, either alone or in combination with the histone deacetylase inhibitor sodium butyrate (NaB), leads to reactivation, which is significantly reduced in the presence of the DNA polymerase inhibitor phosphonoacetic acid (PAA) at a final concentration of 500 μM after 24 and 72 h. For both DOX and the DOX-NaB combination, α-HT 28a at a final concentration of 5 (Fig. 8) or 20 μM (data not shown) provided equally significant suppression of virus replication in both cell pellets and DNase I-treated supernatants. The effect of α-HT 28a on viral replication was not due to cytotoxicity, as this compound induced no cell death at 5 μM and only moderate effects at 20 μM (Fig. S3).

FIG 8.

Effects of α-HT 28a on KSHV lytic replication. Latently infected TREx BCBL-1-Rta cells were reactivated with 1 μg/ml DOX or a combination of DOX and 1 mM NaB and treated with the known DNA polymerase inhibitor PAA (500 μM) or α-HT 28a (5 μM) for 24 h (black bars) or 72 h (gray bars). KSHV viral copies were determined by a quantitative real-time assay in cells (A) and DNase I-treated supernatants (B). Statistical significance was determined by t test: *, P < 0.05; **, P < 0.01, ***, P < 0.001; ns, not significant.

DISCUSSION

Homology between proteins involved in capsid assembly of alphaherpesviruses (HSV-1 and HSV-2), betaherpesviruses (HCMV), and gammaherpesviruses (KSHV) has spawned the notion that broadly acting antiviral agents could be developed to treat their associated diseases, which range in severity from mild rashes to nasopharyngeal cancer (12). Of particular interest is the viral terminase, a tripartite “molecular motor,” which cleaves concatemeric DNA into unit genome lengths as a prerequisite to their insertion into the capsid. Structural studies of the C-terminal terminase nuclease domains of HSV-1 (pUL15C) (17) and HCMV (pUL89C) (37) have identified an RNase H-like fold, indicating they are derived from the nucleotidyltransferase (NTase) superfamily of nucleases that use a 2-metal hydrolytic mechanism (57). Based on the efforts invested in developing potent and selective antiviral agents against HIV integrase and reverse transcriptase-associated RNase H, chemotypes inhibiting these NTases would seem a reasonable starting point for developing antiherpesvirus agents. Supporting this notion, Ireland et al. have demonstrated that synthetic α-HTs potently inhibit replication of wild-type and acyclovir-resistant HSV-1 and HSV-2, with an 50% effective concentration (EC₅₀) in the submicromolar range and therapeutic indices ranging from 170 to >1,200 (45). This study is particularly significant because it opens the possibility of implementing a combination therapy in conjunction with current anti-HSV agents to prevent viral shedding and severely curtail the development of drug-resistant mutants. The present study extends these findings, combining α-HT inhibition of KSHV pORF29C nuclease activity in vitro with inhibition of virus replication in cell culture.

A significant challenge when purifying recombinant herpesvirus nucleases is verifying that they are devoid of contaminating bacterial nucleases, since even in trace amounts these would elicit cleavage of supercoiled DNA. In this study, we coupled altered digestion patterns of a model DNA duplex with sensitivity to α-HT inhibition to both confirm the purity and demonstrate the activity of WT pORF29C. Using modeling studies, we have also constructed, purified, and characterized pORF29C variants containing Ala substitutions at proposed active-site residues. As confirmation of their catalytic role, neither divalent metal (Mn²⁺) nor an active-site α-HT inhibitor stabilized mutants D⁴⁷⁶A, E⁵⁵⁰A, and D⁶⁶²A against thermal denaturation. Concomitantly, these mutants displayed minimal nuclease activity on a supercoiled DNA substrate. However, a surprising observation was that although nuclease activity of mutant D⁶⁶¹A was also severely impaired, thermal denaturation studies indicated it bound divalent metal and an active site inhibitor as efficiently as the wild-type enzyme. This could reflect one of two possibilities, namely, (i) a subtle change in metal ion coordination geometry induced by Ala substitution that permits ligand binding but is incompatible with catalysis or (ii) D⁶⁶¹ is not itself involved in catalysis, but its replacement with Ala has a direct impact on the coordination geometry of the neighboring residue, D⁶⁶². Crystallization studies with WT and D⁶⁶¹A pORF29C are under way to resolve these questions.

Although we demonstrate α-HT inhibition of KSHV replication, it is important to recognize that KSHV is primarily latent in the infected host in the form of viral episomes (6) and must be reactivated to enter the lytic cycle (7). Stated differently, an antiviral agent targeting an essential KSHV enzyme would only be effective when used in conjunction with a lytic activator, recapitulating the kick-and-kill strategy that has been proposed to clear the latent HIV reservoir (39). The feasibility of this approach has been demonstrated by the ability of short interfering RNAs directed against Tousled-like kinase to induce robust KSHV reactivation (40). Our own studies have indicated that small molecules directed against a triple helix encoded by KSHV lncRNA PAN likewise lead to virus reactivation (J. Sztuba-Solinska, F. A. Abulwerdi, T. Kenderdine, E. M. Cornejo Castro, V. A. Marshall, N. Labo, J. S. Schneekloth, G. T. Pauly, D. M. Sigano, R. L. LeBlanc, O. Becette, D. Fabris, D. Whitby, and S. F. J. Le Grice, unpublished data). The availability of a KSHV reactivation strategy that targets a viral component rather than a cellular function suggests that development of small molecules designed to target essential viral enzymes should be continued. Based on their promising antiviral activity against wild-type and mutant forms of HSV-1 and HSV-2, the α-HTs reported here represent important lead compounds for such studies.

Finally, mapping of RAL resistance to the HSV DNA polymerase processivity factor pUL42 (58), in combination with a report from Yan et al. that the HIV IN antagonist XZ45 inhibited DNA binding activity of HSV-1-infected cell protein 8 (ICP8) in vitro and inhibited alpha-, beta-, and gammaherpesviruses in vivo (59), indicates that herpesviruses encode several members of the NTase superfamily. This raises the possibility that a single NTase antagonist, such as the α-HTs described here, have more than one viral target, providing a greater barrier to the emergence of drug resistance. While speculative, supportive evidence comes from observations of Tavis et al. that the α-HTs may block events at an early, postentry stage as well as a later phase of HSV replication (46).

MATERIALS AND METHODS

Protein expression and purification.

Plasmids encoding WT and mutant variants of ORF29C in pET15b, appended with an N-terminal (His)₆ tag and thrombin cleavage site, were transformed in E. coli strain BL21(DE3) (Novagen). Bacteria were grown at 37°C with agitation (225 rpm) for ∼3 h until an optical density at 600 nm (OD₆₀₀) of 0.7 to 1.0 and induced by adding 0.2 mg/ml isopropyl β-d-1-thiogalactopyranoside (IPTG) (Fisher Scientific). Cultures were switched to 16°C, incubated overnight with agitation, and then centrifuged at 4,500 rpm for 15 min at 4°C. Pellets were frozen at −20°C until needed. For purification, pellets were resuspended in 50 mM NaH₂PO₄–Na₂HPO₄ buffer, pH 7.8, containing 1 mM phenylmethylsulfonyl chloride (PMSF). Lysozyme was added to a final concentration of 0.5 mg/ml to the cell suspension and incubated at 4°C with gentle rocking for 15 min. After adding NaCl to a final concentration of 0.3 M, cells were further lysed by sonic disruption and the lysate clarified by high-speed centrifugation. Supernatant was loaded on a nickel-nitrilotriacetic acid (Ni-NTA)-Sepharose column equilibrated in 50 mM NaH₂PO₄–Na₂HPO₄, pH 7.8, 0.3 M NaCl, 10% glycerol, 10 mM imidazole (buffer A), washed with buffer A, and eluted with buffer B (buffer A plus 500 mM imidazole). Fractions containing pORF29C were pooled and concentrated using an Amicon ultra-15 centrifugal concentrator (3-kDa molecular weight cutoff [MWCO]). Concentrated sample was loaded onto a HiLoad 16/60 Superdex 75 prep-grade gel filtration column equilibrated in 20 mM Tris, pH 8.5, 100 mM NaCl, 2 mM MgCl₂, and 1 mM dithiothreitol (DTT). Peak fractions were pooled and concentrated using an Amicon ultra-15 centrifugal concentrator (3-kDa MWCO and frozen in single-use aliquots at −80°C.

In vitro nuclease activity assay.

Supercoiled plasmid DNA was incubated with pORF29C in a buffer of 20 mM HEPES, pH 7.5, 10 mM NaCl, 5 mM MnCl₂, and 1 mM DTT. Incubation times were 20 and 35 min, after which the reaction was quenched with EDTA at a final concentration of 46 mM. Inhibition of pORF29C nuclease activity by α-HT and HIV IN strand transfer inhibitors was performed in the above-described buffer with 10 nM plasmid DNA, 1 μM pORF29C, and increasing concentrations of inhibitor (dissolved in 100% dimethyl sulfoxide) at 37°C and terminated with EDTA. Reaction products were separated by 0.8% agarose gel electrophoresis and visualized by ethidium bromide staining.

Fluorescence-based thermal shift assay.

Thermal shift assays were performed with purified pORF29C variants at a concentration of 1 μM in a buffer containing 20 mM HEPES, pH 7.5, 100 mM NaCl, 10 mM MnCl₂, 1 mM DTT, and SYPRO Orange (diluted 2,000-fold from the 5,000× stock solution; Invitrogen) using procedures essentially as described previously (44). Reaction mixtures were prepared in duplicate in a 96-well fast PCR plate at a final volume of 25 μl. Melting curves were measured in the temperature range of 20°C to 80°C at 1°C/min on an LC480 real-time PCR machine (Roche). Melting temperatures were obtained at the minima of the derivatives of SYPRO Orange fluorescence curves with the LC480 software, v1.5.1.62. Each measurement was performed in duplicate.

KSHV lytic induction.

TREx BCBL-1-Rta cells (kindly provided by Joseph Ziegelbauer, NCI) were grown and maintained in RPMI 1640 medium supplemented with 10% heat-inactivated fetal bovine serum, penicillin (100 U/ml), and streptomycin (100 μg/ml; all from Gibco) at 37°C under 5% CO₂. Selection for the ORF50 gene (Rta) was maintained by addition of hygromycin B (100 μg/ml; Thermo Fisher Scientific). To induce KSHV lytic replication, cells were seeded at 10⁶ cells/ml and treated with doxycycline (1 μg/ml; DOX; Sigma-Aldrich) or both DOX and sodium butyrate (1 mM; NaB; Sigma-Aldrich). Where indicated, cells were also treated with phosphonoacetic acid (500 μM; PAA; Sigma-Aldrich), a known inhibitor of herpesvirus DNA polymerase (60), or α-HT 28a (5 or 20 μM). At 24 or 72 h, cells and supernatants were collected for the detection of viral genome copies. All experiments were performed as biological triplicates. Differences were assessed using the t test.

DNA extraction and KSHV viral load detection.

Cell-associated DNA was extracted using a Qiagen body fluids minikit by following the manufacturers' instructions. Supernatants were treated with DNase I (Zymo) to remove any unencapsidated viral DNA. Extraction was performed using the QIAamp MinElute virus spin kit (Qiagen) according to the manufacturer's protocol. KSHV viral load (VL) was assessed by quantitative real-time PCR as described before (61, 62). Briefly, supernatant DNA was tested using 10 μl of the elution in triplicate reaction mixtures. The average KSHV copy was adjusted to copies per milliliter based upon fraction of material tested. The cellular DNA was tested in triplicate reactions by TaqMan quantitative PCR to both KSHV, specifically targeting the K6 gene region, and also the human endogenous retrovirus 3 gene (ERV-3), which is used as a cell quantitation assay. KSHV copies were normalized to one million cellular estimates, as previously described.

Cytotoxicity determination.

Cytotoxicity was determined with the CytoTox-Glo cytotoxicity assay kit (Promega). Quadruplicate samples of TREx BCBL-1-Rta cells were plated in clear-bottomed 96-well plates with white walls at 5,000 cells/well with or without α-HT 28a at 1, 2, 5, and 20 μM. Plates were processed at 24 h according to manufacturer's directions and read in a Tecan Infinite M1000Pro plate reader.

Homology modeling.

The 3D model of pORF29C was generated with I-TASSER (41), which used the HSV-1 pUL15C structure (17) as the template. To model the compound and two metal ions bound to the active site, the bacteriophage Sf6 large terminase gp2 nuclease domain structure in complex with β-thujaplicinol was superimposed onto the pORF29C model with the MatchMaker function in the program UCSF Chimera (63). The resulting coordinates of β-thujaplicinol and the two metal ions were taken as being in the bound state. The RAL structure was superimposed onto the β-thujaplicinol structure so that the hydroxyl groups and carbonyl group that coordinate with the two metal ions were maintained.