Cidofovir Diphosphate Inhibits Adenovirus 5 DNA Polymerase via both Nonobligate Chain Termination and Direct Inhibition, and Polymerase Mutations Confer Cidofovir Resistance on Intact Virus

Human adenovirus (AdV) can cause fatal disease in immune-suppressed individuals, but treatment options are limited, in part because the antiviral cytidine analog cidofovir (CDV) is nephrotoxic. The investigational agent brincidofovir (BCV) is orally bioavailable, nonnephrotoxic, and generates the same active metabolite, cidofovir diphosphate (CDVpp).

ABSTRACT

Human adenovirus (AdV) can cause fatal disease in immune-suppressed individuals, but treatment options are limited, in part because the antiviral cytidine analog cidofovir (CDV) is nephrotoxic. The investigational agent brincidofovir (BCV) is orally bioavailable, nonnephrotoxic, and generates the same active metabolite, cidofovir diphosphate (CDVpp). However, its mechanism of action against AdV is poorly understood. Therefore, we have examined the effect of CDVpp on DNA synthesis by a purified adenovirus 5 (AdV5) DNA polymerase (Pol). CDVpp was incorporated into nascent DNA strands and promoted a nonobligate form of chain termination (i.e., AdV5 Pol can extend, albeit inefficiently, a DNA chain even after the incorporation of a first CDVpp molecule). Moreover, unlike a conventional mismatched base pair, misincorporated CDVpp was not readily excised by the AdV5 Pol. At elevated concentrations, CDVpp inhibited AdV5 Pol in a manner consistent with both chain termination and direct inhibition of Pol activity. Finally, a recombinant AdV5 was constructed, containing Pol mutations (V303I and T87I) that were selected following an extended passage of wild-type AdV5 in the presence of BCV. This virus had a 2.1-fold elevated 50% effective concentration (EC₅₀) for BCV and a 1.9-fold increased EC₅₀ for CDV; thus, these results confirmed that viral resistance to BCV and CDV can be attributed to mutations in the viral Pol. These findings show that the anti-AdV5 activity of CDV and BCV is mediated through the viral DNA Pol and that their antiviral activity may occur via both (nonobligate) chain termination and (at high concentration) direct inhibition of AdV5 Pol activity.

INTRODUCTION

Adenoviruses (AdVs) are nonenveloped viruses with linear double-stranded DNA (dsDNA) genomes ranging from 26 to 48 kbp in length. Human AdVs are classified into seven species (termed A-G) containing over 85 serotypes and cause gastrointestinal, respiratory, urinary, and conjunctival diseases of varying severity (1, 2). Adenoviral infections are more common in children and are most severe in immunocompromised individuals, including hematopoietic stem cell or solid organ transplant recipients, in whom AdV infections may lead to significant morbidity and mortality (1, 3–5).

Despite the recognized need for effective treatment, there is currently no FDA-approved antiviral for AdV infection. One drug that has shown activity in cultured cells against AdV is cidofovir (CDV), which has broad activity against dsDNA viruses in cell culture (6, 7). Unfortunately, the clinical use of CDV is limited by severe nephrotoxicity (8–10), as exemplified by a recent case report in which CDV was successfully used to control a disseminated adenovirus infection in an immune-suppressed patient but caused permanent renal disease requiring lifelong dialysis (11).

The need for a broad-spectrum antiviral has led to the development of brincidofovir (BCV, CMX001), a lipid-conjugate of CDV with greater oral bioavailability and significantly reduced nephrotoxicity (12). Clinical studies have demonstrated the potential utility of brincidofovir to treat severe AdV infection in immunocompromised transplant patients (13, 14).

Previous studies have shown that CDV and its active metabolite, cidofovir diphosphate (CDVpp), which is also the active metabolite of BCV, inhibit other dsDNA viruses, such as cytomegalovirus (CMV) and vaccinia virus (VV), by interfering with viral DNA replication (15, 16). However, despite the potent antiviral effect of BCV against AdV in the clinic, the mechanism by which BCV inhibits AdV replication remains poorly understood.

In the present work, we used biochemical and cell-based assays to elucidate the mechanism by which BCV inhibits the replication of a common human AdV, adenovirus 5 (AdV5). We found that CDVpp was incorporated into newly synthesized DNA by the purified AdV5 DNA polymerase (Pol), resulting in a large decrease in subsequent chain elongation. Consistent with this, misincorporated CDVpp was resistant to excision by the AdV5 Pol, unlike a conventional mismatched base pair. Interestingly, at elevated concentrations, CDVpp directly inhibited the enzymatic activity of the AdV5 Pol, suggesting two independent mechanisms of antiviral activity.

Finally, we examined the functional significance of Pol mutations (V303I plus T87I) that arose following the prolonged passage of AdV5 in the presence of BCV. When these mutations were introduced into a recombinant AdV5 virus, they conferred 2.1-fold and 1.9-fold increased resistance to BCV and CDV, respectively. These findings are also consistent with a Pol-dependent mode-of-action for BCV’s antiviral activity.

RESULTS

Incorporation of CDVpp results in nonobligate chain termination.

We first set out to determine whether CDVpp is incorporated during DNA synthesis by the AdV5 DNA Pol and, if so, whether this results in chain termination. As seen in Fig. 1, the structure of the CDVpp terminated end varies from that of a natural nucleotide but still allows for the possibility of covalent attachment of an additional nucleotide. To determine whether attachment actually occurs, reactions were performed in biochemical assays on a synthetic primer-template in the presence of either dCTP or CDVpp and included the other three deoxynucleoside triphosphates (dNTPs) (Fig. 2). As seen in Fig. 3, in the absence of CDVpp, the DNA primer-template was quickly converted to a fully extended product. However, reaction mixtures containing CDVpp showed significant chain termination events at positions complementary to template guanosines. Nevertheless, some incorporation extended beyond the CDV insertion sites. The incorporation of a second CDVpp molecule within two nucleotides resulted in strong but still nonobligate chain termination, with a small amount of primer converted to a fully extended product. This effect was observed regardless of whether template guanosines were positioned adjacent to the primer terminus or further downstream in the template, allowing the growing primer to approach the CDV incorporation site with a “running start” (Fig. 3B).

FIG 1.

Proposed structure formed by CDV incorporation. The structure shown is a cytidine nucleoside with two molecules of CDV incorporated via phosphodiester bonds at the 3′ termini, emphasizing the possibility of continued elongation by additions of CDV molecules. The image was created using ChemDraw.

FIG 2.

Primer and templates. DNA sequences for primers and templates used in primer extension and exonuclease assays. “D” denotes incorporated cidofovir molecule.

FIG 3.

Cidofovir base pairs with guanosine and acts as a nonobligate chain terminator. Primer extension reactions were run for the indicated amount of time with wild-type AdV5 Pol (30 ng) in the presence of dNTPs (50 µM each) and in presence or absence of CDVpp (100 µM). An 18-bp primer was labeled on the 5′ end with ³²P and then annealed to one of four 30-bp templates. Primer (P1) and template pairs (T1, T2) were adopted from Xiong et al. (15) and shortened for purposes of consistency, with the first template guanosine at the n +1 position (A). Templates T3 and T4 have the first template guanosine at the n + 5 position, allowing the Pol to gain a “running start” (B). The primer/template sequences (Fig. 2) have been truncated to show only the non-base-paired segments that are extended in the assay. Template guanosines, where dCTP or CDVpp may be incorporated during elongation, are shown in bold. Similar results were obtained with 99% and 99.9% (UltraPure) pure dNTPs.

At higher concentrations, CDVpp directly inhibits AdV5 Pol.

We next asked whether CDVpp is capable of inhibiting the AdV5 Pol irrespective of CDV incorporation and chain termination. To do this, we designed templates in which the two unpaired guanosines were replaced by cytosines. As shown in Fig. 4A, with control templates having G residues, we again observed chain-termination events at positions corresponding to template guanosines. However, in the absence of template guanosines, we observed a dose-dependent inhibition of extension, with more of the primer remaining completely unextended as the CDVpp concentration was increased. Again, this effect was recapitulated with templates having different sequences (Fig. 4B).

FIG 4.

CDVpp inhibits AdV5 Pol directly at high concentrations in the absence of template guanosines. Primer extension reactions were run for 10 min with wild-type AdV5 Pol (30 ng) in the presence of dNTPs (5 µM each) and the indicated concentration of CDVpp. The 18-bp primer (P1) was labeled on the 5′ end with ³²P and then annealed to one of four 30-bp templates. The T1 template contains guanosines at the n + 1 and n + 3 positions, whereas these have been replaced by cytosines in T1C (A). Similarly, the T3 template contains guanosines at the n + 5 and n + 7 positions, which have been replaced by cytosines in T3C (B). (A and B, bottom) The primer/template sequences (Fig. 2) have been truncated to show only the non-base-paired segments that are extended in the assay.

Incorporated CDVpp is not easily removed by AdV5 Pol exonuclease.

As the AdV5 Pol contains a 3′ to 5′ nuclease function for exonucleolytic proofreading (17), we next asked whether the enzyme is able to remove CDVpp from the 3′ terminal ends. In order to characterize this activity, we designed both primer-terminal-matched (C-G pair) and mismatched (T-G pair) primer templates and generated a 3′-CDV-terminated primer by utilizing terminal transferase (TdT) and the CDVpp precursor. This primer was then annealed to a template for testing. As seen in Fig. 5, the mismatched 3′ base pair was quickly removed by the AdV5 Pol, whereas a matched base pair was more slowly removed. This was expected since a mismatched terminus is the preferred substrate for a proofreading nuclease. Notably, the rate of CDV removal by the 3′ to 5′ exonuclease activity of the AdV5 Pol was much slower than the 3′ nucleotide of a fully matched primer-template terminus, with most of the original primer still remaining after 2 h.

FIG 5.

Incorporated CDVpp is not easily removed by AdV5 Pol exonuclease activity. Exonuclease reactions were run with wild-type AdV5 Pol (50 ng) for the indicated amount of time in the absence of other dNTPs. The 18-bp P1 primer containing an additional cytosine (P2) or thymine (P3) was labeled on the 5′ end with P32 and then annealed to the 30-bp T1 template. P1 was also labeled on the 5′ end with P32 before the addition of a single CDV molecule to its 3′ terminus utilizing Terminal Transferase (TdT) forming P4, and subsequently annealing to the T1 template. The primer/template sequences (Fig. 2) have been truncated to show only the final residue of the base-paired region (matched, mismatched, or CDVpp) plus the non-base-paired segments.

Kinetic analysis of CDVpp inhibition.

In order to determine the kinetic properties of AdV5 polymerization and CDVpp inhibition, we next designed experiments such that initial rates of extension could be quantified. We first confirmed the validity of our kinetic measurements by running reactions in the absence of CDVpp, which yielded a linear Lineweaver-Burk plot indicating a K_m value with respect to dNTPs of 1.23 µM.

Next, we repeated the experiment in the presence of several concentrations of CDVpp. Using GraphPad Prism software, initial rates were fit with a competitive inhibition model (global r², 0.83), which yielded a K_i of 76.3 µM. Lineweaver-Burk plots were consistent with these measurements and support a competitive mode of inhibition (graphical data not shown).

Effect of brincidofovir (BCV) on replication of recombinant AdV5.

Previous work demonstrated that passaging a laboratory strain of AdV5 (AdVC5) in the presence of increasing concentrations of CDV leads to the development of mutations in the AdV5 Pol gene that render the virus more resistant to BCV inhibition (18). To determine the mechanism whereby these mutations confer resistance, we repeated this procedure using BCV as the selective agent. As seen in Fig. 6, following 15 passages in the presence of BCV, two mutations arose in the Pol gene (first V303I, followed by T87I). In order to confirm that these mutations alone render the virus more resistant to BCV and CDV, we generated a recombinant AdV5 virus containing the T87I and V303I mutations using the pAdEasy-1 cloning system. After purifying the recombinant virus, we tested its ability to replicate in the presence of BCV or CDV. We found that the presence of the T87I and V303I mutations resulted in a 2.1-fold increase in the EC₅₀ of BCV and 1.9-fold increase in the EC₅₀ of CDV, relative to wild-type recombinant AdV5 (Fig. 7), thereby confirming that Pol mutations can confer viral resistance to CDV and BCV.

FIG 6.

Changes in AdV5 DNA Pol sequence in BCV-passaged virus. A549 cells were infected with wild-type AdVC5 virus at an MOI of 0.001. The virus was passaged 15 times in the presence of increasing concentrations of BCV. Shown is an amino acid sequence alignment of the emergent mutations observed from passaged AdVC5 in the presence of BCV (passage 5 [p5], p10, p12, p14, and p15) along with the AdVC5 reference sequence. The X under amino acid 87 represents a mixture of threonine and isoleucine and under amino acid 303 represents a mixture of valine and isoleucine.

FIG 7.

T87I and V303I Pol mutations render AdV5 virus resistant to BCV and CDV. The 293A cells were infected with wild-type or T87I/V303I mutant virus at an MOI of 0.01 in the absence or presence of CDV (200, 100, 50, 25, 12.5, 6.25, 3.125, or 1.56 µM) (A) or BCV (200, 100, 50, 25, 12.5, 6.25, 3.125, or 1.56 nM) (B). Following a 3-day incubation, the medium was removed, cells were lysed, and the lysate was treated with proteinase K for 1 h at 65°C. Following heat inactivation, 2.5 µl of clarified lysate was used to perform qPCR with a standard curve ranging from 10³ to 10¹⁰ AdV5 genome copies/ml. Outliers were excluded as described (see Materials and Methods), and data were plotted in GraphPad Prism and fit with the log (inhibitor) versus response model (four parameters).

Biochemical analysis of the mechanism of resistance mutations in AdV5 Pol.

To better understand the mechanism by which the T87I and V303I mutations render the virus resistant to BCV, we tested several aspects of wild-type and mutant Pol activity in the presence of CDVpp, the active metabolite of BCV. In preliminary experiments, we noticed in reactions carried out with the mutant Pol a rapid accumulation of a labeled single-nucleotide product. We postulated that this product was the result of a contaminating 5′ nuclease, likely the well-characterized baculovirus alkaline nuclease (19). We found that the use of a structurally larger 5′ Cy5 fluorescent label in place of a ³²P label prevented this 5′-nuclease activity and allowed us to circumvent this issue (see Fig. S1 in the supplemental material).

Using the Cy5-labeled primer, we first investigated whether the T87I/V303I mutations allowed for the enhanced removal of incorporated CDV from the 3′ end, reactivating it for synthesis. In these and subsequent experiments, we compared equal amounts of the wild-type and mutant Pols by weight (unless otherwise stated). As shown in Fig. 8A, the rate of CDV removal by either wild-type or mutant AdV5 Pol was negligible. Next, we tested whether the mutations allowed the Pol to better extend a CDV-terminated substrate. We found that the rate of extension of the CDV-terminated primer was no different for the mutant Pol compared to the wild type (Fig. 8B). Another possibility was that the mutations change the kinetics of CDVpp incorporation. However, as shown in Fig. 9A, there was no significant difference in the rate of addition of CDVpp to a template in the absence of competing dNTPs. Lastly, we tested the hypothesis that the mutations allowed for the enhanced utilization of dCTP in the presence of competing CDVpp. Again, we did not find that the mutant Pol was able to better extend the primer in the presence of increasing concentrations of CDVpp (Fig. 9B).

FIG 8.

T87I/V303I mutations do not enhance excision or extension of a 3′-CDV-terminated primer. (A) Ability of the Pol to remove incorporated CDV from 3′ termini. Exonuclease reactions were run with 75 ng of Pol and 25-nM template for the indicated time in the absence of other dNTPs. A single CDV molecule was added onto the 3′ terminus of a 5′-Cy5-labeled P1 primer by TdT before the primer was annealed to the 30-bp T1 template. (B) Ability of Pol to extend a CDV-terminated primer. Extension reactions were run utilizing the 3′-CDV-terminated template generated in Fig. 6A for the indicated times, with 75 ng of Pol and 5 µM dNTPs. (B) The primer/template sequences have been truncated to show only the 5′ Cy5 label, the final residue of the base paired region (D = CDVpp), and the non-base-paired segment that is extended in the assay.

FIG 9.

Comparison of wild-type and T87I/V303I mutant Ad5 polymerase activity in the presence of CDVpp. (A) Kinetics of CDV incorporation. A single CDV molecule was added onto the 3′ terminus of a 5′-Cy5-labeled P1 primer by TdT before the primer was annealed to the 30-bp T1 template. Extension reactions were run for the indicated times utilizing a 5’Cy5-P1/T1 primer/template (25 nM), 30 ng of Pol, 10 µM CDVpp, and in the absence of all other dNTPs. (B) Inhibition of extension by CDVpp. Extension reactions were run for 5 min utilizing the 5’Cy5-P1/T1 primer/template (25 nM) with 30 ng of Pol, 5 µM dNTPs, and various concentrations of CDVpp. CDVpp concentrations were as follows: Lanes 1 and 7, 0 µM; lanes 2 and 8, 1.6 µM; lanes 3 and 9, 8 µM; lanes 4 and 10, 40 µM; lanes 5 and 11, 200 µM; and lanes 6 and 12, 1,000 µM. (B) The primer/template sequences have been truncated to show only the 5′ Cy5 label, and the non-base-paired segment that is extended in the assay.

Our next, and final, set of experiments used a longer template, with the rationale that a longer template might more faithfully recapitulate viral replication (since the AdV5 genome is approximately 36 kb). To this end, we performed extension assays utilizing a template having the sequence of the first 200 bp of the AdV5 genome. Again, we utilized a Cy5 5′ label in this experiment in order to prevent 5′ hydrolysis from obscuring our comparisons.

As shown in Fig. 10, even with the longer template, there were essentially no differences in the ability of the T87I/V303I mutant Pol to generate a full-length 200-bp product in the presence of CDVpp compared with the wild-type Pol. However, we did observe some slight differences in the banding pattern of intermediate extended products, with some bands more prominent with one protein and faded with the other. This result is representative of a sequence-specific effect on basic primer extension.

FIG 10.

CDVpp-mediated inhibition of primer extension over a 200-bp template is essentially the same for both wild-type and T87I/V303I mutant Pols. (A) Primer extensions. A 5′-Cy5-labeled primer (P5) was annealed to a 200-bp template (T5), having a sequence of the first 200-bp of the AdV5 genome. Concentrations of wild-type (50 ng) and T87I/V303I mutant (125 ng) Pols were adjusted in order to normalize the amount of template extension in the absence of CDVpp. Extension reactions were run for 2 h with 25 nM primer/template, 5 µM dNTPs, and in the absence or presence of 8 µM CDVpp. (B) Quantitation of fully extended 200-bp product. Densitometry was performed in Image Lab to measure the amount of fully extended 200-bp product in each reaction.

DISCUSSION

To our knowledge, this is the first characterization of the mechanism of inhibition of AdV5 replication by CDVpp, the active metabolite of BCV and CDV. We have demonstrated that the primary mechanism of inhibition involves the incorporation of CDVpp into the growing replicon and that the terminating adduct cannot be easily removed or extended. We also determined that at higher concentrations, CDVpp is capable of inhibiting the AdV5 Pol directly, irrespective of any incorporation; it is unknown to what extent this mechanism contributes to the inhibition of viral replication in cultured cells, as the concentration of CDVpp in cells may not reach the higher concentrations at which this phenomenon was observed in our biochemical assays.

Kinetic analyses indicate that there is competitive inhibition with respect to the dNTP substrates. We expected a competitive, rather than noncompetitive or mixed, mode because the primary mode of inhibition by CDVpp against other viruses is chain termination. CDVpp is an analog of the dCTP substrate, and its chain terminating mechanism necessitates use of the same active site as is used by the dNTP substrates, making it a competitor. Moreover, the secondary mode of inhibition is direct. This mode almost certainly involves binding of the CDVpp to the Pol. Although we do not know the exact binding site, since CDVpp is a dNTP analog, it most likely is binding the dNTP site, again as a competitor.

In agreement with previous studies of CDVpp inhibition of the CMV Pol, we found that the incorporation of a second CDV molecule greatly reduced the ability of the AdV5 Pol to further extend the primer. We also found that an incorporated CDVpp cannot be easily excised from the 3′ end by the 3′ to 5′ exonucleolytic proofreading activity of the viral Pol (15). McGee et al. reported a different result with the VV DNA polymerase, wherein a 3′-terminal CDV served as a good substrate for removal by the 3′ to 5′ exonuclease activity, but the extension of a CDV-terminated primer by one native nucleotide (n + 1) rendered it completely resistant to exonuclease activity (16). Interestingly, contrary to results with the CMV and VV DNA Pols, we did not observe a noticeable pause or chain termination event at the position n + 1 following CDVpp incorporation. Instead, we saw the greatest chain termination at the site of CDVpp incorporation, with more minor pauses at the n − 1 base immediately preceding a template guanosine. These differences likely represent variations in the three-dimensional structure of the Pols, but ultimately, we found chain termination to be the primary mechanism of CDVpp inhibition of AdV5.

While it is generally accepted that CDVpp is the active metabolite of both BCV and CDV, serial passaging of adenovirus in the presence of BCV yielded mutations that were distinct from those that arose after serial passaging of AdV5 with CDV, except for the shared V303I mutation (18). This result is not surprising, considering that serial passaging with CDV yielded 4 distinct variants, which shared at most a single one of up to three mutations dispersed throughout the Pol gene. Notably, neither the T87I or V303I mutation appears to fall within known domain features attributed to zinc fingers, nuclear localization, polymerization, or exonuclease activity (20).

In characterizing the biochemical properties of the T87I/V303I mutant Pol, we tested several hypotheses that might explain how these mutations render the virus resistant to BCV. The mutations did not allow the Pol to more easily remove incorporated CDV and reactivate terminated replicons (indeed, the rate at which a single CDV molecule was removed from a 3′ terminus was negligible for both wild-type and mutant Pols). Moreover, the rate at which a CDV-terminated primer was extended was nearly identical for the mutant and wild-type Pols.

We also ruled out the possibility that these mutations altered the enzyme in a way that slows down the utilization and incorporation of CDVpp. Reactions with wild-type versus mutant Pol performed with a 30-bp template showed no difference in extension kinetics in the presence of competing dCTP and CDVpp. We also conducted reactions with a longer, more physiologically relevant 200-bp template, with the anticipation that they might reveal the basis of resistance. However, few differences were evident. The longer template did bring to light some sequence-specific differences in pause site intensities that were not evident with the shorter template. However, upon repeating these reactions several times, these intermediate products appeared inconsistently and often at sites that did not represent a competition of dCTP and CDVpp for incorporation.

After considering many possible mechanisms, we conclude that any differences in biochemical activity due to the T87I/V303I mutations are very slight and not reliably detectable in the biochemical assays we employed here. The cellular environment, and natural setting for AdV5 replication, is impossible to emulate exactly in these biochemical assays. Indeed, all reconstitutions of efficient DNA replication in biochemical assays require much lower ionic strength and protein concentrations and different ions than those employed by viruses and cells (21). Furthermore, it is plausible that the T87I/V303I mutations alter the interaction of the Pol with other replication machinery, such as the viral-DNA-binding protein or initiation factors, which are not included in these reactions. A notable feature of adenovirus DNA replication is that it is initiated from a terminal protein, which is attached to a dCMP residue that serves as the initiator primer (22). The presence of cidofovir, as a dCMP-specific competitor, may either inhibit attachment of the C residue or substitute for it. This idea raises the possibility that a component of the inhibitory effects of cidofovir occurs during priming and that our resistance mutations partially circumvent this inhibition.

It is also possible that small differences in enzymatic activity, which may be unobserved in these assays, would interact and rise exponentially in aggregate as replication proceeds through the 36-kb genome in the presence of CDVpp. This possibility explains why even our 200-bp template requires micromolar amounts of CDVpp to see inhibition, whereas only nanomolar amounts of BCV are needed in cultured cells for EC₅₀ measurement.

In conclusion, we found that CDV inhibits AdV5 DNA replication largely by a nonobligate chain termination mechanism but also by direct inhibition of the AdV5 DNA Pol. Moreover, the repeated passage of the virus in BCV, the metabolic precursor of CDV, produced a resistant mutant Pol. However, the mutations did not alter a comprehensive panel of biochemical primer elongation and exonuclease assays for CDV-based inhibition. This finding indicates that the mechanism of resistance involves molecular interactions occurring in the cell that are not emulated in these biochemical assays.

MATERIALS AND METHODS

Materials and reagents.

Restriction endonucleases, T4 polynucleotide kinase, and TdT were from New England Biolabs (NEB). Gene fragments containing mutations of interest were synthesized by GeneArt (Thermo Fisher Scientific) or Integrated DNA Technologies (IDT; gBlocks). BacPAK6 baculovirus DNA, BacPak baculovirus rapid titer kit, Adeno-X maxi purification kit, Talon metal affinity resin, and all nucleic acid isolation kits (Nucleotrap, Nucleospin, and Nucleobond Xtra) were from Clontech. UltraPure dNTPs were from Bioline. All primers and templates for biochemical assays were synthesized and HPLC purified by IDT. Cellfectin and SimplyBlue SafeStain were from Invitrogen.

Production and purification of AdV5 Pol from baculovirus.

The Baculovirus transfer vector pAcGP67A-AdPol6xHis containing the AdV5 viral Pol region was generated previously (23). Gene segments containing the T87I (GeneArt) and V303I mutations (IDT) were digested with BamHI-SbfI and SbfI-BlpI endonucleases, respectively, and subsequently cloned into the pAcGP67A-AdPol6xHis vector. Recombinant constructs were sequence verified and cotransfected with baculovirus DNA (Clontech) into Spodoptera frugiperda (SF9) insect cells, as described (23). Baculovirus was passaged 4 to 6 times, the titer was determined with the BacPak rapid titer kit (Clontech), and the Pol was purified essentially as described by Capella et al. (23), with the following modifications. Nonspecifically bound protein was removed with successive washes with 5 mM and 40 mM imidazole. Following dialysis, purified Pol was concentrated using an Amicon Ultra-15 centrifuge filter with a 10-kDa cutoff (Millipore) and stored at −80°C. The purity and concentration of Pol were determined by running SDS-PAGE gels alongside bovine serum albumin (BSA) standards (Bio-Rad) and staining with Simply Blue SafeStain (Invitrogen). Densitometry was performed in ImageLab.

Enzymatic assays with AdV5 Pol.

All extension and exonuclease assays were run at 37°C in standard 20-µl reactions unless otherwise stated in figure legends. The addition of Pol into each reaction varied from 10 to 100 ng, depending on the goal of the experiment. Primers labeled at the 5′ end with ³²P or Cy5 were used at final concentrations of 5 nM or 25 nM, respectively. The 5′ ³²P label was generated with a T4 polynucleotide kinase according to the manufacturer’s instructions (NEB). The 3′-CDV-terminated primer was generated by reaction with TdT (NEB), utilizing a molar ratio of CDVpp:oligomer of 1,000:1 and a reaction time of 2 h. Following the addition of ³²P or CDVpp, primers were purified through illustra MicroSpin G-25 columns (GE Healthcare) to remove excess reagents.

Initial extension reactions with wild-type Pol were run in buffer containing 50 mM Tris-HCl (pH 7.5), 2 mM dithiothreitol (DTT), 25 µg/ml BSA, 1 mM ATP, 4 mM MgCl₂, and 75 mM NaCl. Exonuclease assays were performed in a slightly different buffer adopted from King et al. (17), containing 50 mM Tris-HCl (pH 7.5), 4% glycerol, 1 mM DTT, 100 µg/ml BSA, 1 mM MgCl₂, and 50 mM NaCl. All assays performed to compare wild-type and mutant Pol activity were performed in the latter buffer. Reactions were stopped by the addition of one-half volume of stop buffer (40 mM EDTA, 99% formamide). Reactions with 20 to 30-bp templates were run on 20% polyacrylamide gels containing urea (Sequagel, National Diagnostics), while reactions run with 200-bp templates were loaded onto 12% polyacrylamide urea gels.

For kinetic experiments, a reaction duration of 2 mins was chosen to represent the initial reaction rates, as this yielded a more linear Lineweaver-Burk plot than other sampling times tested. Volumes of kinetic reactions were doubled to 40-µl reactions (24-ng Pol/reaction) in order to reduce variability inherent to shorter reaction times. To measure the kinetics of inhibition, dNTPs were used at 0.5, 1, 2, 4, and 8 µM, while CDVpp was varied at 0, 50, 100, 200, and 400 µM.

Resistance passaging of AdV5 virus.

Tissue culture flasks (75 cm²) were seeded with 2 × 10⁶ A549 cells, and cells were allowed to adhere overnight. Medium was aspirated, and flasks were infected with AdV strain AdVC5 (ATCC catalog no. VR-5) at an MOI of 0.001 for 2 h in 4 ml medium containing 2% fetal bovine serum (FBS). After 2 h, the infection mixture was aspirated, cells were rinsed with medium, 20 ml of fresh medium containing 2% FBS along with the appropriate concentration of BCV was applied to the flasks, and the flasks were incubated at 37°C. The medium was replaced every 4 days until cytopathic effects (CPEs) were visible. Incubation was continued until distinct CPEs were seen over the entire flask, generally between days 6 to 13 postinfection. The concentration of BCV used over the 15 passages was increased stepwise from 1× to 21× EC₅₀. Flasks displaying 50% to 100% CPE were frozen at −80°C and subjected to 3 freeze-thaw cycles to disrupt cells and release intracellular virus. After the third thaw, the cell debris was removed by centrifugation at 250 × g for 10 min. The clarified cell culture fluid was distributed into 2-ml tubes and stored frozen at −80°C. The DNA Pol region was sequenced across several passages to identify any changes at the nucleotide level and to determine the evolution of any such changes.

Production and purification of recombinant AdV5 virus.

The adenoviral vectors were generated using the AdEasy system (Stratagene), utilizing the pLitmus V.1 and pAdTrack-E1SV40/eGFP vectors that were generated previously (23). The pLitmus V.1 cloning vector was used to introduce mutations into the AdV5 Pol region. Gene segments containing the T87I (GeneArt) and V303I (IDT) mutations were cloned into the pLitmus V.1 using AleI-SbFI and SbfI-SapI restriction endonucleases, respectively. The modified pLitmus V.1 cloning vector was then digested with ClaI-PmeI, and this fragment was ligated into the AdEasy-1 ΔE1/E3 shuttle vector and subsequently cloned into pAdTrack-E1SV40/eGFP via homologous recombination in BJ5183 cells as described (23). This produced a full-length adenovirus genome, with the AdV5 Pol region containing the T871/V303I mutation. Final constructs were linearized and transfected into 293 A cells (Stratagene) in T75 flasks. Media was replaced every 4 to 5 days until cytopathic effects (CPEs) had reached 75% of the monolayers, at which time the supernatant fluid was removed and cell debris was removed at 2000 × g for 10 min. For propagation, virus was passaged 1 to 5 times onto fresh monolayers of 293 A cells. Virus was purified from clarified supernatant fluid via CsCl₂ gradient or utilizing an Adeno-X maxi purification kit (Clontech, catalog no. 631532). Purified virus was dialyzed in buffer containing 10 mM Tris-HCl (pH 7.5), 1 mM MgCl₂, and 10% glycerol at 4°C in a 10,000-molecular weight (MW)-cutoff dialysis cassette (Thermo Scientific, catalog no. 55380) and stored at −80°C. Viral DNA was purified and the presence of the desired mutations was confirmed by Sanger sequencing. The virus titer was determined by 50% tissue culture infective dose (TCID₅₀).

Viral assays.

The AdV phenotyping assay is a cell-based virus replication assay that utilizes a quantitative real-time PCR (qPCR) readout to quantify the amount of viral DNA produced in infected cells. The assay measures the compound concentration that inhibits virus replication by 50% and reports this EC₅₀, as well as the fold change relative to a laboratory strain reference virus (wild-type AdV).

The assay was run in a 96-well format using 293 A cells. Cells were seeded at 2.0 × 10⁴ cells/well on poly-d-lysine-coated plates and were allowed to adhere for 16 h. The medium was removed, and the cells were infected at an MOI of 0.01 in Dulbecco’s modified Eagle medium (DMEM) containing 2% FBS. Compound (BCV or CDV) dilutions were prepared at 2× of the desired final concentration in 2% FBS/DMEM, and immediately added to wells. Plates were incubated at 37°C in a CO₂ incubator for 3 days and then harvested as described (24). Briefly, the medium was aspirated and the cells were rinsed carefully twice with 1× PBS and then were lysed with 175 µl of lysis buffer containing 10 mM Tris–HCl (pH 8.3), 50 mM KCl, 2.5 mM MgCl₂, 0.45% (wt/vol) NP-40, 0.45% (wt/vol) Tween 20, and proteinase K freshly added to final concentration of 0.5 mg/ml. Plates were incubated at 37°C for 30 min, the lysis buffer in each well was thoroughly mixed, and 100 µl of lysate was transferred to 96-well PCR plates. Incubation was continued at 65°C for 1 h, followed by a 20-min incubation at 98°C to heat inactivate the proteinase K.

AdV DNA was quantified from the processed sample using the Bio-Rad CFX Connect real-time system. The TaqMan gene expression master mix (Life Technologies, catalog no. 4369016) was used with a pair of PCR primers and a TaqMan probe. The probe had a 6-carboxyfluorescein (FAM) dye label at the 5′ end and a minor groove binder (MGB) and nonfluorescent quencher (NFQ) at the 3′ end. The primers and probe sequences along with the final concentrations used in the reaction were as follows: forward primer, 5′-ACC TGG GCC AAA ACC TTC TC-3′ (0.3 µM); reverse primer, 5′-CGT CCA TGG GAT CCA CCT C-3′ (0.3 µM); and probe, 5′-AAC TCC GCC CAC GCG CTA GA-3′ (0.2 µM). A reference primer/probe set measuring RNase P copies (Life Technologies, catalog no. 4403328) was used to quality control each plate.

The processed sample (2.5 µl) was added to the qPCR reaction mixture (TaqMan gene expression master mix) in a total volume of 25 µl/reaction in a 96-well plate. The qPCR cycling parameters included an initial denaturation cycle at 95°C for 10 min, following by 40 cycles at 95°C for 15 s and 60°C for 1 min. The AdV DNA copy number was quantified by utilizing a DNA standard prepared from plasmid DNA with a dynamic range of 10¹⁰ to 10³ DNA copies/ml and generation of a standard curve with quantification cycle (C_q) values in GraphPad Prism 7.

For quality control of each plate, we calculated the average and standard deviation of all C_q values for the RNase P assay. We then defined outliers as well with C_q values that were more than 1 standard deviation away from the mean. If more than half of the uninfected wells were outside one standard deviation, the plate was not scored and the assay was repeated. Remaining data were plotted with GraphPad Prism 7, using a 4-parameter curve fitting function, log(inhibitor) versus response – variable slope, to determine EC₅₀ values. Data are expressed as fold increase in EC₅₀ over that of the wild-type AdV5.