Mutations Conferring Resistance to Viral DNA Polymerase Inhibitors in Camelpox Virus Give Different Drug-Susceptibility Profiles in Vaccinia Virus

Abstract

Cidofovir or (S)-HPMPC is one of the three antiviral drugs that might be used for the treatment of orthopoxvirus infections. (S)-HPMPC and its 2,6-diaminopurine counterpart, (S)-HPMPDAP, have been described to select, in vitro, for drug resistance mutations in the viral DNA polymerase (E9L) gene of vaccinia virus (VACV). Here, to extend our knowledge of drug resistance development among orthopoxviruses, we selected, in vitro, camelpox viruses (CMLV) resistant to (S)-HPMPDAP and identified a single amino acid change, T831I, and a double mutation, A314V+A684V, within E9L. The production of recombinant CMLV and VACV carrying these amino acid substitutions (T831I, A314V, or A314V+A684V) demonstrated clearly their involvement in conferring reduced sensitivity to viral DNA polymerase inhibitors, including (S)-HPMPDAP. Both CMLV and VACV harboring the A314V change showed comparable drug-susceptibility profiles to various antivirals and similar impairments in viral growth. In contrast, the single change T831I and the double change A314V+A684V in VACV were responsible for increased levels of drug resistance and for cross-resistance to viral DNA polymerase antivirals that were not observed with their CMLV counterparts. Each amino acid change accounted for an attenuated phenotype of VACV in vivo. Modeling of E9L suggested that the T→I change at position 831 might abolish hydrogen bonds between E9L and the DNA backbone and have a direct impact on the incorporation of the acyclic nucleoside phosphonates. Our findings demonstrate that drug-resistance development in two related orthopoxvirus species may impact drug-susceptibility profiles and viral fitness differently.

INTRODUCTION

Poxviruses are large double-stranded DNA viruses that replicate in the cytoplasm of the infected cell. The Orthopoxvirus (OPV) genus is composed of several virus species that can infect humans and animals (51). Variola virus (VARV), well known as the causative agent of smallpox (18), is now regarded as a potential bioweapon. Also, other OPV infections are emerging. Monkeypox virus is responsible for increasing outbreaks in Democratic Republic of the Congo (55), and increasing zoonotic infections of vaccinia virus (VACV), as well as of buffalopox, are being observed in Brazil and in India (1, 13, 34, 56). Similarly, emerging infections with cowpox virus (CPXV) are seen in humans and in domestic and zoo animals in Europe (9, 25, 35, 36, 49, 64). Moreover, three human cases of camelpox have been reported and laboratory confirmed, albeit camelpox virus (CMLV) has always been considered to be restricted to camels (11). From these observations and considering that routine smallpox vaccination was discontinued 30 years ago, it may be speculated that the world's population immunity against OPV infections became low or nonexistent (55).

In the last few years, various countermeasures have been developed for the prevention and the treatment of smallpox, and they could be envisaged for the management of other OPV-induced diseases. These countermeasures include, in particular, replication-deficient vaccines (8, 65) and antivirals (57). Today, three molecules are recognized as potent in vitro and in vivo inhibitors of OPVs, i.e., cidofovir or (S)-HPMPC [(S)-1-[3-hydroxy-2-(phosphonomethoxy)propyl]cytosine] (24), CMX001 (hexadecyloxypropyl-HPMPC) (42), and ST-246 [4-trifluoromethyl-N-(3,3a,4,4a,5,5a,6,6a-octahydro-1,3-dioxo-4,6-ethenocycloprop[f]isoindol-2(1H)-yl)-benzamide] (66). Each of these compounds has gained the investigational new drug status by the U.S. Food and Drug Administration (FDA) to permit their emergency use to treat smallpox vaccination-related adverse effects (6, 7, 41, 50).

Both (S)-HPMPC and CMX001 belong to the family of acyclic nucleoside phosphonates (ANPs). (S)-HPMPC is recognized as one of the therapeutics of choice for the treatment of poxvirus infections as demonstrated by its potent efficacy in several animal models of OPV infections (for a review, see references 20 and 57). This ANP has been approved by the FDA and by the European Medicines Agency for human use in the treatment of cytomegalovirus retinitis in AIDS individuals. (S)-HPMPC has to be administered by intravenous infusion. Pre- and posthydration, as well as probenecid, are required to manage its potential nephrotoxicity. (S)-HPMPC is also used off-label for the treatment of polyoma-, papilloma-, adeno-, herpes-, and poxvirus infections (20, 22, 23). The search for orally available ANPs led to the discovery of CMX001, a lipid conjugate of (S)-HPMPC. This molecule, compared to its parent counterpart (S)-HPMPC, exhibits a greater potency in vitro against various dsDNA viruses and no signs of nephrotoxicity, although gastrointestinal toxicity has been observed (45, 52). CMX001 is currently under development for the treatment of poxvirus infections following the “Animal Efficacy Rule” promulgated by the FDA (40).

Considerable numbers of ANPs have been and are still being synthesized in order to yield novel molecules with improved antiviral activities and pharmacokinetic properties (for a review, see references 19 and 21). Several of these derivatives have been evaluated in vitro and ex vivo against poxviruses including VACV, CPXV, CMLV, and orf virus replication (16, 17, 28, 42, 46, 61). As shown in Fig. 1, ANPs can be classified within three generations of molecules with the first generation, including (S)-HPMPC and its adenine counterpart, (S)-HPMPA, their cyclic forms and the 2,6-diaminopurine derivative [i.e., (S)-HPMPDAP]. The 6-[2-(phosphonomethoxy)alkoxy]-2,4-diaminopyrimidine derivatives are considered as the second generation of ANPs and include in particular (R)-HPMPO-DAPy, whose alkyl purine homolog is (S)-HPMPDAP, and PMEO-DAPy [2,4-diamino-6-[2-(phosphonomethoxy)ethoxy]pyrimidine], which belongs to the PME- (phosphonylmethoxyethyl-) subclass (39). These 2,4-diaminopyrimidine molecules, i.e., (R)-HPMPO-DAPy and PMEO-DAPy, appear to mimic an incomplete purine ring as the aliphatic phosphonate side chain is linked to the C-6 position instead of the N-1 position through an ether linkage and are seen as open-ring analogues (37). The molecule (R)-HPMPO-DAPy is of interest since it has been shown to be more effective than postexposure smallpox vaccination in a lethal model of MPXV infection in monkeys (63). The third generation of ANPs contains the triazine analogues derived from (S)-HPMPC, i.e., (S)-HPMP-5-azacytosine [(S)-HPMP-5azaC] and its cyclic form (44).

Fig 1.

Three generations of ANPs, with their abbreviations, chemical names, and structures.

The molecular mechanisms through which (S)-HPMPC inhibits VACV have been described (4, 10, 32, 43, 47, 48). The active form of (S)-HPMPC [i.e., (S)-HPMPC-pp] will act as an alternative substrate for the poxviral DNA polymerase (E9L protein of VACV) and, by this process, slows down and prevents viral DNA elongation and further virus replication. The study of the mode of action of ANPs requires also the characterization of drug-resistant viruses. The phenotyping of MPXV, VACV, CPXV, and CMLV resistant to (S)-HPMPC has been reported by Smee et al., and a 8- to 27-fold increase in 50% effective concentrations (EC₅₀) values toward (S)-HPMPC was observed (59). Other studies further described the genotyping of VACV resistant to (S)-HPMPC (HPMPC^R) (4, 10, 43) and to (S)-HPMPDAP (HPMPDAP^R) (32). Several amino acid changes, listed in Table 1, have been mapped to the E9L protein of drug-resistant VACV and MPXV, though the specific roles of some of these mutations have not been uniformly elucidated (4, 10, 30, 32, 43). Various degrees of resistance to (S)-HPMPC or (S)-HPMPDAP, ranging from 3- to 25-fold increase in EC₅₀s, may be achieved depending on the nature and the number of the mutated allele(s) (4, 10, 32, 43). Considering that CMX001 is a prodrug of (S)-HPMPC, it is expected that mutations linked to (S)-HPMPC resistance would also confer resistance to CMX001, as demonstrated by Kornbluth et al. (43). In vivo, VACV HPMPC^R and HPMPDAP^R viruses exhibited reduced virulence compared to wild-type (WT) viruses (4, 10, 32, 43).

Table 1.

Summary of mutations associated with resistance to (S)-HPMPC and (S)-HPMPDAP^a

Mutation(s) identified in the viral DNA polymerase gene	Domain		Virus	Fold change to:		Hypersensitivity to PMEO-DAPy and PAA	Resistance confirmed by marker rescue	Growth rateb	Virulence in mice	Responding to ANP-based therapy	Source or reference
	Exonuclease	Polymerase		(S)-HPMPDAP	(S)-HPMPC
Single
ΔK174	x		VACV	–	3	–	Yes	↓	Attenuated	–	10
H296Y	x		VACV	–	–	–	No	–	–	–	43
A314T	x		VACV	10	8	Yes	Yes	Normal	Attenuated	–	4, 32
A314V	x		VACV	14.3	5.3 to 7	Yes	Yes	↓	Attenuated	–	10; This study
			CMLV	28.8	2.8				–	–	This study
H319N	x		VACV	–	–	–	No	–	–	–	43
R604S		x	VACV	–	–	–	No	–	–	–	43
M671I		x	VACV	–	5	–	Yes	↓	Attenuated	–	10
A684V		x	VACV	2	7	No	Yes	Normal	Attenuated	–	4
T831I		x	VACV	20.3	6.5	Yes	Yes	↓	Attenuated	–	This study
			CMLV	17	No			Normal	–	–
S851Y		x	VACV	10	No	Yes	Yes	↓	Attenuated	–	32
Combined
Y232H+A314T+A684V	x	x	VACV	–	25	–	No	Normal	–	–	4
H296Y+A314V	x		VACV	–	–	–	No	–	–	–	43
H296Y+S338F	x		VACV	–	2	–	Yes	–	–	–	43
H296Y+A314V+H319N+S338F+R604S	x	x	VACV	–	10 to 14	–	Yes	↓	Attenuated	Yes	43
A314T+A684V	x	x	VACV	20	15	No	Yes	Normal	Attenuated	Yes	4
A314V+A684V	x	x	VACV	38	13.4	Yes	Yes	Normal	Attenuated	–	This study
			CMLV	32	5.3			↓	–	–
A314V+A613T+A684T+T808 M	x	x	MPXV	–	19 to 27	–	No	↓	–	–	30, 59
A314T+T688A	x	x	VACV	20	12	Yes	Yes	↓	Attenuated	–	4
A684V+S851Y	x	x	VACV	25	10	No	Yes	Normal	Attenuated	Yes	32

^aThe mutations listed in column 1 numbered after VACV E9L.–, Not reported.

^bThe downward arrow (↓) means that the growth rate of mutant virus was decreased compared to the WT, and “Normal”means that the growth rates were equivalent between mutant and WT viruses.

Most of our knowledge of ANP resistance development in OPVs comes from studies performed with VACV and (S)-HPMPDAP or (S)-HPMPC. However, the recent identification of mutated residues in the DNA polymerase of MPXV HPMPC^R at amino acid positions that are similar to those seen in VACV HPMPC^R might predict for conserved genetic alterations within OPVs under (S)-HPMPC pressure (30). In order to extend our comprehension of how specific mutations in the DNA polymerase of different OPVs affect the susceptibility to ANPs, we studied the impact of homologous mutations within E9L of CMLV and VACV. After the selection of CMLV resistant to (S)-HPMPDAP, a potent anti-CMLV molecule in vitro and in vivo (26, 27), amino acid residues within the viral DNA polymerase of CMLV were identified as the cause of resistance to (S)-HPMPDAP. Recombinant CMLV and VACV were constructed by transfection of plasmids containing the desired mutation, and a detailed comparison of their drug-susceptibility profiles to viral DNA polymerase inhibitors was performed. The impact of the mutated residues on CMLV and VACV growth and on VACV pathogenicity was investigated. The effects of each of these amino acid changes on the structure of E9L were further examined.

MATERIALS AND METHODS

Cells and viruses.

The following virus strains were used: CMLV strain Iran and CMLV strain Dubai (CML1 and CML14, respectively, kindly provided by H. Meyer, Bundeswehr Institute of Microbiology [53, 54]); VACV strain Western Reserve (VACV-WR); and VACV-E9L-A684V, which is a recombinant VACV-WR harboring the A684V mutation within the E9L gene and known to give resistance to (S)-HPMPC and (S)-HPMPDAP (4, 32). All experiments with viruses were performed in biosafety level 2 with the approval of the Biosafety Committee of the KU Leuven (approval number AMV/22082011/SBB219.2011/0011). Human embryonic lung (HEL) fibroblast cells were grown in Earle's minimum essential medium (MEM Earle's; Invitrogen, Merelbeke, Belgium) containing 5% fetal calf serum (FCS), 0.2% serum substitute for animal cell culture (UltroserG; PALL Life Sciences, Cergy-Saint-Christophe, France), 1% HEPES, 1% nonessential amino acids, 1% sodium pyruvate, and 1% antibiotic/l-glutamine (Invitrogen) at 37°C under a 5% CO₂ atmosphere. Vero cells were grown in MEM Earle's containing 10% FCS and 1% HEPES, nonessential amino acid, sodium pyruvate, and antibiotic/l-glutamine at 37°C under a 5% CO₂ atmosphere. Viruses were grown on Vero cells for the selection of drug-resistant mutants. For all other experiments, HEL cells were used.

Compounds.

The sources of the compounds (see Fig. 1 for full names) were as follows. (S)-HPMPC and cyclic-(S)-HPMPC were obtained from Gilead Sciences (Foster City, CA). (S)-HPMPA, cyclic-(S)-HPMPA, 3-deaza-(S)-HPMPA [(S)-9-[3-hydroxy-2-(phosphonomethoxy)propyl]-3-deazaadenine], cyclic 3-deaza-(S)-HPMPA, (S)-HPMPDAP, (R)-HPMPO-DAPy, PMEO-DAPy [2,4-diamino-6-[2-(phosphonomethoxy) ethoxy]pyrimidine], (S)-HPMP-5-azaC, and cyclic-(S)-HPMP-5-azaC were synthesized at the Institute of Organic Chemistry and Biochemistry, Academy of Sciences of the Czech Republic, Prague, Czech Republic. Phosphonoacetic acid (PAA) was obtained from Sigma-Aldrich (Bornem, Belgium). ST-246 [4-trifluoromethyl-N-(3,3a,4,4a,5,5a,6,6a-octahydro-1,3-dioxo-4,6-ethenocycloprop[f]isoindol-2(1H)-yl)-benzamide] was kindly provided by D. E. Hruby from SIGA (Corvallis, OR).

Selection and purification of viruses resistant to (S)-HPMPDAP.

CML1 and CML14 strains were passaged repeatedly in Vero cells (∼30 rounds) in the presence of increasing amounts of (S)-HPMPDAP, starting at a concentration of 0.8 μM. The viruses that replicated in the presence of a final dose of 125 μM (S)-HPMPDAP were cultured twice more in drug-free medium. Plaque reduction assays were performed to evaluate the resistant phenotype of the viruses grown under selective pressure with (S)-HPMPDAP. The CML1- and CML14-(S)-HPMPDAP-resistant virus were named CML1 HPMPDAP^R and CML14 HPMPDAP^R, and from each of these mutant viruses, as well as from WT CML1 and CML14 strains, three to five clones were isolated by plaque purification and selected for further analysis.

Antiviral assays.

The antiviral activities of the compounds against in vitro selected drug-resistant viruses, WT viruses, and recombinant viruses were evaluated in HEL cells grown in 96-well microtiter plates. Confluent monolayers were infected at a multiplicity of infection (MOI) of 0.01 PFU/cell for 2 h at 37°C under a 5% CO₂ atmosphere. The medium consisted of MEM Earle's containing 2% FCS and supplemented with 1% HEPES, 1% nonessential amino acids, 1% sodium pyruvate, and 1% antibiotic/l-glutamine (Invitrogen). At 2 h postinfection, residual virus was removed and replaced with the medium containing serial dilutions of the test compounds (in duplicate). After 2 to 3 days for VACV and 4 to 5 days for CMLV, viral cytopathic effect (CPE) was recorded, and the EC₅₀ was defined as the concentration of a compound required to reduce the viral CPE by 50%. The EC₅₀s of the compounds for each viral strain were calculated as the means from at least three independent experiments.

Sequencing.

To facilitate the comprehension of the results, all genes cited in this article have been named after VACV strain Copenhagen nomenclature. Sequencing of E9L of WT, HPMPDAP^R clones, and recombinant viruses and of the two components of the processivity factor (A20R and D4R genes) of WT and HPMPDAP^R clones was performed as follows. DNA was extracted from virus-infected HEL cells using a QIAamp DNA Blood Mini apparatus (Qiagen Benelux B.V., Venlo, Netherlands) according to manufacturer's instructions. As shown in Table S1 in the supplemental material, the full-length E9L gene was amplified by PCR as five overlapping amplicons for CMLV, using the primer sets 1 to 5, or as four overlapping amplicons for VACV, using the primer sets A to D. The full-length A20R and D4R CMLV genes were amplified as two overlapping amplicons using the primer sets shown in Table S1 in the supplemental material. FastStart high-fidelity DNA polymerase (Roche Applied Science, Mannheim, Germany) was used to PCR amplify the E9L, A20R, and D4R genes. The PCR products were purified using QIAquick PCR purification kit (Qiagen) and sequenced with the use of the cycle sequencing kit BigDye Terminator kit version 3.1 (Applied Biosystems Europe BV, Halle, Belgium). Twenty, ten, and four virus-specific primers targeting both strands of the E9L, A20R, and D4R genes, respectively, were used for sequencing, and samples were further analyzed in an ABI 3730 sequencing system (Applied Biosystems Europe BV). The data were assembled and compared to the DNA sequences obtained for the WT CML1, CML14, and VACV clones using the software SeqScape version 2.7 (Applied Biosystems Europe BV) and Vector NTI Advance 11.0 (Invitrogen, Merelbeke, Belgium).

DNA cloning.

For the production of CMLV recombinant viruses, DNA was extracted from HEL cells infected with CML1/HPMPDAP^R/T831I and CML14/HPMPDAP^R/A314V+A684V using the QIAamp DNA Blood Mini (Qiagen Benelux B.V.). Portions of the E9L gene were further amplified with a FastStart high-fidelity DNA polymerase (Roche Applied Science), and cloned into a pCR2.1-TOPO plasmid (Invitrogen; for the primer set sequences, see Fig. 2 and Table S1 in the supplemental material). The primer set 5 was used to amplify the E9L portion from CML1/HPMPDAP^R/T831I, overlapping the nucleotide change (C→T) at position 2492, which was cloned to obtain TOPO-CML1-T831I. The primer set 2 was used to amplify the E9L portion from CML14/HPMPDAP^R/A314V+A684V, overlapping the nucleotide change (C→T) at position 941, which was cloned to get TOPO-CML14-A314V. The primer set 6 was used to amplify the E9L portion of CML14/HPMPDAP^R/A314V+A684V, overlapping the nucleotide changes (C→T) at positions 941 and 2051, which was cloned to get TOPO-CML14-A314V+A684V. In this last case, the PCR was performed using a KOD Hot Start DNA polymerase (EMD Chemicals, Inc., Gibbstown, NJ). The cloning reactions were done according to manufacturer's instructions, and chemically competent DH5α-T1^R were used. The production of recombinant VACV-WR bearing the single mutations A314V or T831I or the double mutation A314V+A684V required the cloning of portions of the WT E9L gene into a pCR2.1-TOPO plasmid. To this end, DNA was extracted from VACV-WR-infected HEL cells, and portions of the E9L gene were PCR amplified by using the primer sets given in Fig. 2; one amplicon covering the nucleotide position 941 was cloned into TOPO vector to obtain plasmid A and one amplicon covering the nucleotide position 2492 to obtain plasmid B. Each of these TOPO plasmids were used for site-directed mutagenesis.

Fig 2.

Production of recombinant viruses. (a) Overview of mutations reported in VACV and MPXV that confer resistance to (S)-HPMPC, (S)-HPMPDAP, and PAA (for references, see Table 1) and plotting of the mutations T831I and A314V+A684V identified in CML1 and CML14 HPMPDAP^R clones, respectively. (b) Schematic view of the CMLV PCR products used for the production of recombinant CML1 and CML14. The primer sequences used were as follows: forward (5′-ATG CAA TCG AAG AAG AAA TAA ACA ACG-3′) and reverse (5′-ATG GAT AAA CTG AAA CTA ACA AAG AGC-3′) primers for CML1-rcb-T831I, forward (5′-GAA ATA CAA GAA GCC GTC GAT AGA GG-3′) and reverse (5′-CCT TAT ACA TCT GTG CTA GAT CAA CG-3′) primers for CML14-rcb-A314V, and forward (5′-GAA ATA CAA GAA GCC GTC GAT AGA GG-3′) and reverse (5′-AGT CAG TAT CTC CGT ACA CGC TAC G-3′) for primers CML14-rcb-A314V+A684V. (c) Schematic view of VACV PCR products used for site-directed mutagenesis and further production of recombinant VACV. The primer sequences used were as follows: forward (5′-TAG AAT CGG TAC TAA ATG GAG CAG-3′) and reverse (5′-GAT AAA CTG AAT CTA ACA AAG AGC GAC G-3′) primers for VACV-WR-rcb-T831I and forward (5′-TGT TTA CGC TCA TTA ATG AAG AGA TG-3′) and reverse (5′-AGT CGG TAT CTC CAT ACA CGC TAC G-3′) primers for VACV-WR-rcb-A314V.

Site-directed mutagenesis.

The sequences of the primers used for VACV-WR site-directed mutagenesis were as follows: VACV-WR-C941T (5′-AGG GAG TAG GCG GCA TGG TCA ATA CTA CGT TTC ACG-3′) and VACV-WR-C2492T (5′-TAT AAA GGT ACT AGT GAA ATT AGA AGA GAT GTT TCC AAG T-3′), with the nucleotides indicated in boldface corresponding to the mutated allele. A QuikChange Lightning site-directed mutagenesis kit (Stratagene, Agilent Tech. La Jolla, CA) was utilized according to the manufacturer's recommendations. Plasmid A (100 ng) was used with the primer VACV-WR-C941T to produce TOPO-VACV-WR-C941T, and plasmid B was used with the primer VACV-WR-C2492T to produce TOPO-VACV-WR-C2492T. The sequence integrity after mutagenesis was confirmed by sequencing.

Recombinant constructs.

For all transfection experiments, cells were grown or infected in medium without antibiotics. HEL cells were grown to confluence in six-well plates and then infected for 1 h with CML1, CML14, or VACV-WR at MOIs of 0.5 and 1 in 1 ml of MEM–2% FCS. Cells were replaced in 2 ml of warm medium and returned to the incubator for 2 h. For each construct, i.e., TOPO-CML1-T831I with CML1, TOPO-CML14-A314V with CML14, TOPO-CML14-A314V+A684V with CML14, TOPO-VACV-WR-C941T with VACV-WR, or VACV-E9L-A684V and TOPO-VACV-WR-C2492T with VACV-WR, three transfection conditions were used as follows: 2 μg of plasmid DNA for the MOIs of 0.5 and 1 and 4 μg of plasmid DNA for the MOI of 1 using 10 μl of Lipofectamine 2000 (Invitrogen) per well. The plates were incubated for 5 h, and the transfection solution was replaced with 2.5 ml of warm medium. Cells were further cultured for 24 h at 37°C and 5% CO₂. Each of the HPMPDAP^R recombinant viruses were released by freeze-thawing and further grown, at a low MOI, for two passages in HEL cells in the presence of at least 3 μM HPMPDAP in order to eliminate the majority of background WT virus since the chosen concentration was known to completely inhibit WT virus growth. After a last passage in HEL cells in the absence of the drug, each viral construct was titrated and characterized phenotypically. The recombinant constructs exhibiting a resistant phenotype were subjected to two rounds of plaque purification. Plaque-purified recombinant viruses were further grown on HEL cells and titrated, and their entire E9L genes were sequenced to confirm the presence of the required mutated allele. The double mutant VACV-WR-rcb-A314V+A684V should have been obtained by transfecting TOPO-VACV-WR-C941T into cells infected with a VACV-E9L-A684V. However, in the process of recovering VACV-WR-rcb-A314V, we also obtained a population of double-mutant A314V+A684V VACV clones that were finally used for these studies. We speculated that the two passages performed in 3 μM (S)-HPMPDAP in order to facilitate the recovery of resistant viruses may have favored to appearance of the secondary A684V mutation. One isolate of each recombinant construct, namely, CML1-rcb-T831I, CML14-rcb-A314V, CML14-rcb-A314V+A684V, VACV-WR-rcb-T831I, VACV-WR-rcb-A314V, and VACV-WR-rcb-A314V+A684V, was retained for further analysis.

Growth curves of recombinant viruses.

To measure the growth rates of recombinant viruses, HEL cells were allowed to grow until confluence in 24-well plates and infected with the strain of interest at an MOI of 0.01. At different time points, i.e., 3, 6, 9, 24, 30, 48, and 72 h postinfection, the cells were harvested (in triplicates), and the viruses were released by freeze-thawing for further titrations by plaque assays in HEL cells. Virus titers, expressed in PFU/ml, are presented as the means from two independent experiments.

In vivo experiments.

All animal work was approved by the Katholieke Universiteit Leuven Ethics Committee for Animal Care and Use (permit P044-2010). All animal guidelines and policies were in accordance with the Belgian Royal Decree of 14 November 1993 concerning the protection of laboratory animals and European Directive 86-609-EEC for the protection of vertebrate animals used for experimental and other scientific purposes. Infections were performed under anesthesia using ketamine-xylazine in saline and, when required, euthanasia was done by administration of pentobarbital sodium. Female NMRI mice [Rj:NMRI(Han)], 5 weeks old, were purchased from Elevage-Janvier (Le-Genest-St-Isle, France). All experimental and animal experimentations were completed at biosafety level 2. Groups were defined as uninfected (mock infected) or as virus infected, i.e., VACV-WR WT, VACV-WR-rcb-A314V, VACV-WR-rcb-A314V+A684V, and VACV-WR-rcb-T831I, and five animals per group were used. Mice were anesthetized using a cocktail of ketamine-xylazine in saline and inoculated intranasally with 10 μl of phosphate-buffered saline (PBS) (uninfected) or with 10 μl of PBS containing 4,000, 400, or 40 PFU of the virus of interest (5 μl per nostril). Cohorts were monitored for body weight, morbidity, and mortality for 30 days.

Statistical analyses.

All statistical analyses were done with GraphPad Prism version 5 Software (GraphPad Software, Inc., La Jolla, CA). Unpaired t tests were used to compare mean EC₅₀s, obtained from at least three independent experiments, between WT and drug-resistant viruses or between WT and the corresponding recombinant virus. A one-way analysis of variance test associated with a Dunnett's multiple-comparison test was used to compare mouse weight loss. Statistical significance was determined for these tests as follows: ***, P < 0.001 (extremely significant); **, P < 0.01 (very significant); *, P < 0.05 (significant); or “ns,” P > 0.05 (not significant).

The growth rates between WT viruses and recombinant constructs were subjected to statistical analysis as previously described elsewhere (4, 32). Briefly, best-fit curves were obtained by transforming the virus titers to their natural logarithm (ln[PFU/ml]). Fit lines derived from growth curves were then analyzed by linear regression and P values for slopes, and intercepts were calculated. P values for intercepts could only be calculated when the P values for slopes were not significant, according to the software instructions, and statistical significance was only considered for P ≤ 0.01.

RESULTS

Selection and characterization of HPMPDAP^R CMLV isolates.

Two different strains of CMLV, namely, CML1 and CML14, were repeatedly passaged in increasing concentrations of (S)-HPMPDAP for approximately 30 rounds. The susceptibility phenotype of each viral stock was then assessed by a CPE reduction assay, and a 28- to 46-fold increase in (S)-HPMPDAP EC₅₀s was observed, these concentrations being 25 μM for CML1 HPMPDAP^R versus 0.9 μM for WT and 28 μM for CML14 HPMPDAP^R versus 0.6 μM for the WT. Clones of each drug-resistant and WT parent viruses were further isolated, and their susceptibility profiles to ANP molecules and to unrelated reference drugs (PAA and ST-246) were investigated. Two CML1 HPMPDAP^R and two CML14 HPMPDAP^R clones were enrolled in these antiviral assays. As depicted in Fig. 3, both CML1 and CML14 HPMPDAP^R clones appeared to be resistant to (S)-HPMPDAP (EC₅₀s of, respectively, 5.4 ± 1.9 μM versus 0.28 ± 0.09 μM for CML1 WT clones [a 19-fold change] and of 17.3 ± 9.7 μM versus 0.45 ± 0.17 μM for CML14 WT clones [a 38-fold change]), as well as to its parent molecule (S)-HPMPA and to cyclic-(S)-HPMPA, albeit the CML14 resistant clones exhibited higher levels of resistance. Cross-resistance toward (R)-HPMPO-DAPy was also seen, although the level of resistance observed with CML14 HPMPDAP^R clones was higher than those of CML1 (12.5-fold change versus 4.4-fold). Interestingly, four ANP molecules, i.e., (S)-HPMP-5-azaC, 3-deaza-(S)-HPMPA and their cyclic forms, had antiviral activities against CML1 and CML14 drug-resistant viruses similar to the WT viruses (Fig. 3). In contrast, only the CML14 HPMPDAP^R clones showed increased EC₅₀s toward (S)-HPMPC (32.4 ± 9.5 μM versus 6.0 ± 3.2 μM for WT, a 5.4-fold change) and cyclic-(S)-HPMPC (28.2 ± 9.1 μM versus 5.6 ± 2.0 μM for WT, a 5.0-fold change) (Fig. 3). Both HPMPDAP^R strains displayed hypersensitivity to PMEO-DAPy and PAA, with CML1 HPMPDAP^R emerging as the most sensitive virus to both compounds. ST-246, which acts at the level of virus egress, conserved its antiviral activity against HPMPDAP^R viruses. These results demonstrated that, following a long process, resistance to (S)-HPMPDAP could be achieved in CMLV and that the phenotypic differences observed between the two drug-resistant viruses might reflect a different genetic basis for resistance.

Fig 3.

Drug-resistance properties of plaque-purified CML1 (A) and CML14 (B) HPMPDAP^R clones. Two CML1 and CML14 HPMPDAP^R clones were used for plaque reduction assays, and at least three to four independent experiments were performed for each test compound. The data are presented as a dot plot of the EC₅₀s of the HPMPDAP^R clones (empty symbols) versus the EC₅₀s of the WT parent clone (filled symbols). At the top of each graph are shown the fold changes in EC₅₀ concentrations that have been calculated as the ratio of the mean EC₅₀s of the HPMPDAP^R clones divided by the mean EC₅₀s of the WT clones. The results are presented as means ± the standard deviations (SD). The statistical significance is indicated. “∼WT” indicates that the drug sensitivity was not significantly different from that of the WT (P > 0.05).

Mapping of mutations involved in drug resistance.

We further sequenced the viral DNA polymerase gene (E9L, named after VACV nomenclature), as well as the two genes A20R and D4R coding for the heterodimeric processivity factor of the E9L protein, to investigate whether (S)-HPMPDAP resistance was linked with allele changes in these genes. Four clones of each HPMPDAP^R strain, as well as WT CML1 and CML14, were sequenced. Each of the four CML1 HPMPDAP^R clones exhibited a single amino acid change in the polymerase domain of the E9L DNA polymerase at position 831, replacing a threonine by an isoleucine (T831I). This amino acid change was confirmed as a mutation point (i.e., not related to polymorphism) following protein blast analysis (www.ncbi.nlm.nih.gov/guide/). Each of the four CML14 HPMPDAP^R clones showed two mutations in the viral DNA polymerase at positions 314 (exonuclease domain) and 684 (polymerase domain), both resulting in a “an alanine-to-valine change” (A314V+A684V). These two mutation points had already been reported individually or accompanied by other substitutions in VACV and MPXV (Table 1) and were therefore not related to polymorphisms. Examination of the A20R and D4R genes of all CML1 and CML14 HPMPDAP^R clones did not reveal any nucleotide change compared to the respective WT clones. In order to demonstrate that the mutated alleles identified in the E9L gene were responsible for imparting resistance to (S)-HPMPDAP, we generated recombinant viruses by transferring the identified mutation into WT CML1 and CML14. As shown in Fig. 2, PCR amplicons overlapping the mutations were produced and used to create the recombinant viruses CML1-rcb-T831I, CML14-rcb-A314V, and CML14-rcb-A314V+A684V. The CML14 A684V recombinant virus could not be isolated despite several attempts.

As depicted in Fig. 4A, CML1-rcb-T831I was resistant to (S)-HPMPDAP, (S)-HPMPA, cyclic-(S)-HPMPA, and (R)-HPMPO-DAPy (with 17-, 6.3-, 5.7-, and 4.1-fold changes in the EC₅₀s, respectively). This T831I recombinant clone, similar to the HPMPDAP^R virus, did not show cross-resistance to the cytosine derivatives, to 3-deaza-(S)-HPMPA, to cyclic-3-deaza-(S)-HPMPA, and to the unrelated molecule, ST-246 but demonstrated hypersensitivity to PMEO-DAPy and PAA.

Fig 4.

Drug-susceptibility profiles of CML1-rcb-T831I (A), CML14-rcb-A314V (B), and CML14-rcb-A314V+A684V (C). Recombinant CMLV were obtained as detailed in Materials and Methods. At least three independent antiviral assays were performed for each test compound. The data are presented on a logarithmic scale as a dot plot of the EC₅₀s of the recombinant clones versus the EC₅₀s of WT CML1 and CML14 clones (filled symbols). At the top of each graph are shown the fold changes in the EC₅₀s that have been calculated as the ratio of the EC₅₀ of a recombinant clone divided with the EC₅₀ of the corresponding WT clone. The results are presented as means ± the standard errors of the mean (SEM). The statistical significance is indicated. “∼WT” indicates that the drug sensitivity was not significantly different from that of the WT (P > 0.05).

The single mutation A314V caused resistance to (S)-HPMPDAP (28.8-fold increase in EC₅₀s compared to WT) and to purine-related molecules, including (S)-HPMPA, cyclic-(S)-HPMPA, and (R)-HPMPO-DAPy (Fig. 4B). On the contrary, both 3-deaza analogues conserved their antiviral potency against this mutant. The change A314V also caused a 2.8- and 3.4-fold increases in the EC₅₀s to (S)-HPMPC and cyclic-(S)-HPMPC, while both triazine analogues (i.e., 5-azaC derivatives) kept their antiviral activities. CML14-rcb-A314V resulted in a pronounced sensitivity to PMEO-DAPy and PAA inhibition (P < 0.001). Adding the A684V to the A314V change did not drastically modify the level of resistance toward either (S)-HPMPDAP (a 32-fold increase in EC₅₀s) or the cross-resistance degrees to the cytosine derivatives (S)-HPMPC (a 5.3-fold increase) and cyclic-(S)-HPMPC (a 3.7-fold increase) compared to the single A314V modification (Fig. 4C). Hypersensitivity to PMEO-DAPy and PAA was seen with CML14-rcb-A314V+A684V but at levels lower than those seen with A314V. These results demonstrated that the single mutation T831I and the double mutation A314V+A684V were sufficient to induce resistance or hypersensitivity to drugs in CMLV, at levels similar to those seen with the parent HPMPDAP^R viruses depicted in Fig. 3. Furthermore, we showed that the single mutation A314V could also select for drug resistance or hypersensitivity in CMLV.

Growth properties of recombinant HPMPDAP^R CMLV clones.

We then explored the growth rates of these marker rescued viruses compared to WT CML1 and CML14 (Fig. 5). CML1-rcb-T831I grew as efficiently as its WT counterpart over the 72 h assessed. In contrast, the single change A314V significantly hampered CML14 growth compared to the WT (P < 0.0001). Adding the A684V mutation, as in CML14-rcb-A314V+A684V, resulted only in a slight reduction in virus yield that was not appreciably different than the WT virus (P value of intercepts of 0.032). It may be that the mutated allele A684V could compensate the reduced fitness engendered by the A314V mutation. Of note, the morphology of the CPE induced by the three recombinant viruses was different from that of the WT viruses: all HPMPDAP^R recombinant viruses produced cell rounding, as already observed by Becker et al. with VACV HPMPC^R viruses (10), with few syncytia in contrast to WT CML1 and CML14 that yield a typical CPE (i.e., syncytia) in HEL cells.

Fig 5.

Growth features of recombinant HPMPDAP^R CML1 (AB) and CML14 (CF). HEL cells were infected with either WT viruses, CML1-rcb-T831I (A), CML14-rcb-A314V (C), or CML14-rcb-A314V+A684V (E) at an MOI of 0.01. (A, C, and E) Samples were harvested at the indicated number of hours postinfection (hpi). Virus titrations were done on HEL cells and expressed in PFU/ml. The results originated from triplicates of two independent experiments and are presented as means ± the SD. (B, D, and F) To compare the growth rates between WT and recombinant constructs, “best-fit” curves were fit from h 9 to 72, and slopes and intercepts were analyzed by linear regression (see Materials and Methods). P values for slopes and intercepts are indicated on the graph, and statistical significance demonstrates a slow-growing phenotype of the recombinant virus compared to WT. Adding the A684V mutation appeared to attenuate the slow-growing phenotype of the A314V single mutant.

Introduction of T831I, A314V, and A314V+A684V in VACV confers resistance not only to (S)-HPMPDAP and but also to (S)-HPMPC.

We further produced recombinant VACV bearing T831I, A314V, or A314V+A684V in order to study the impact of each mutated allele in a VACV backbone. Our group previously reported the characterization of VACV bearing the A684V mutation (4), and two studies identified controversial effects of the A314V allele in (S)-HPMPC resistance development (10, 43). It was therefore of interest to construct a recombinant VACV bearing the A314V mutation and investigate its drug sensitivity profile. The phenotyping of the different VACV recombinant clones was performed (Fig. 6). All three recombinant constructs demonstrated high levels of resistance toward (S)-HPMPDAP with EC₅₀ increased by 14.3- to 38-fold compared to WT.

Fig 6.

Resistance profiles of recombinant VACV T831I (A), A314V (B), and A314V+A684V (C). Recombinant VACVs were obtained by site-directed mutagenesis as explained in Materials and Methods. At least three independent antiviral assays were performed for each test compound. The data are presented on a logarithmic scale as a dot plot of the EC₅₀s of the recombinant VACV clones versus the EC₅₀s of WT VACV (filled symbols). At the top of each graph are shown the fold changes in EC₅₀ concentrations that have been calculated as the ratio of EC₅₀ of a recombinant clone divided with the EC₅₀ of WT VACV. The results are presented as means of at least three independent experiments ± the SEM. The statistical significance is indicated. “∼WT” indicates that the drug sensitivity was not significantly different from that of the WT (P > 0.05).

The mutated allele T831I conferred resistance to (S)-HPMPA, to cyclic-(S)-HPMPA, and to the pyrimidine counterparts, as well as to (R)-HPMPO-DAPy, whereas no resistance was observed with the 5-azaC derivatives, the two 3-deaza analogues, and the unrelated compound ST-246. Hypersensitivity to PMEO-DAPy and PAA was also seen (Fig. 6A). The single mutation A314V caused clear resistance to (S)-HPMPDAP, (S)-HPMPA, cyclic-(S)-HPMPA, (S)-HPMPC, cyclic-(S)-HPMPC, and (R)-HPMPO-DAPy (Fig. 6B). No cross-resistance to triazine analogues (i.e., 5-azaC molecules) was seen, and a pronounced hypersensitivity to PMEO-DAPy and PAA was observed with EC₅₀s for A314V of 8.2 ± 2.3 μM and 107.9 ± 33.5 μM, respectively, compared to 98.7 and 658.1 μM for the WT (0.1- and 0.2-fold changes). A trend of hypersensitivity to the two 3-deaza derivatives was seen with this mutant, but this was not significant. The double mutation A314V+A684V resulted in high levels of resistance to both purine and pyrimidine molecules (8- to 56.4-fold increases in EC₅₀s), although the 3-deaza and cyclic-3-deaza-(S)-HPMPA retained marked antiviral activities (Fig. 6C). Adding the A684V mutation to A314V reduced the levels of hypersensitivity to PMEO-DAPy and PAA (decreases from 0.1- to 0.4-fold in the EC₅₀s for PMEO-DAPy and from 0.2- to 0.3-fold for PAA). ST-246 conserved its antiviral potency against the three recombinant viruses. These findings demonstrate that the double mutation A314V+A684V as well as the single mutations T831I and A314V could select for (S)-HPMPDAP and (S)-HPMPC resistance in VACV.

Comparison of the drug-resistance levels observed between CMLV and VACV recombinant viruses.

In terms of similarities, we observed that each mutation (A314V, A314V+A684V, and T831I) in its respective virus backbone (CML1 or CML14 and VACV) induced resistance to (S)-HPMPDAP, (S)-HPMPA, cyclic-(S)-HPMPA, and (R)-HPMPO-DAPy (Fig. 7). Furthermore, all recombinant constructs showed hypersensitivity to both PMEO-DAPy and PAA. Importantly, both 3-deaza derivatives showed antiviral activities against all recombinant viruses in the range of the WT viruses. The unrelated molecule ST-246 retained antiviral efficacy against all mutant viruses.

Fig 7.

A314V (A), A314V+A684V (B) and T831I (C) mutations select for different drug-resistance levels in CMLV and in VACV. The figures compare the activities of various ANP molecules and PAA against the recombinant HPMPDAP^R CMLV and VACV encoding the indicated mutations. The data are presented as the ratio of the EC₅₀ of the recombinant clone versus the EC₅₀ of the parent WT. Fold changes are plotted on a logarithmic scale to facilitate the comparison between drug-resistance levels with ratios of >2 corresponding to a resistant profile and ratios <1 corresponding to hypersensitivity. The raw data presented in Fig. 4 and 6 were used to construct this figure, and means ± the SEM of at least three independent experiments are shown.

Besides these similarities, it is interesting that discrepancies were seen between VACV and CMLV in the pattern of sensitivity to the pyrimidine analogues. Although the mutation A314V in CMLV and VACV induced resistance to (S)-HPMPC and cyclic-(S)-HPMPC and not to the triazine molecules [(S)-HPMP-5-azaC and cyclic-(S)-HPMP-5azaC], adding A684V to A314V rendered VACV but not CML14 resistant to all pyrimidine analogues (Fig. 7). Furthermore, while the T831I change did not confer resistance to (S)-HPMPC and cyclic-(S)-HPMPC in CML1, it did do so in VACV.

In terms of levels of resistance or hypersensitivity, differences were also noted between recombinant CMLV and VACV. Thus, CML1-rcb-A314V was 1.9- and 3.1- fold less resistant to (S)-HPMPC and cyclic-(S)-HPMPC and 4-fold more sensitive to PMEO-DAPy and PAA than its VACV counterpart (Fig. 7A). In addition, adding A684V to A314V resulted in a 2.5- to 8.3-fold increased level of resistance in VACV compared to CML14, whereas both viruses had similar degrees of resistance to (S)-HPMPDAP or hypersensitivity to PAA (Fig. 7B). When analyzing the effect of the T831I change, the drug-resistance levels induced by VACV with the HPMP molecules were 2.4- to 4.5-fold higher than those seen with CML1 [with the exception of (S)-HPMPDAP] (Fig. 7C).

Overall, the degrees of drug resistance were more elevated in VACV than in CMLV. The double mutation A314V+A684V was responsible for the highest levels of drug-resistance in VACV, while in CMLV the mutation A314V alone or with A684V gave comparable degrees of resistance to HPMP drugs.

Growth rates of VACV recombinant viruses.

We further studied the impact of the mutations on VACV growth. The substitution T831I hampered VACV replication in cell culture, which was not seen with CML1-rcb-T831I (Fig. 8A and B). Similarly to what was observed with CMLV, the single mutation A314V drastically hindered the propagation of VACV since the amount of infectious viruses produced at 72 hpi by VACV-WR-rcb-A314V was reduced by >10-fold (Fig. 8C and D). Adding the A684V to A314V appeared to compensate for the slow-growing phenotype of the single mutant and resulted in a virus yield comparable to the WT (Fig. 8E and F), which was similar to what was seen with CML14-rcb-A314V+A684V (Fig. 5).

Fig 8.

Growth curves of recombinant VACV bearing T831I (A and B), A314V (C and D), or A314V+A684V (E and F). HEL cells were infected with either VACV-WR WT virus, VACV-WR-rcb-T831I, VACV-WR-rcb-A314V, or VACV-WR-rcb-A314V+A684V at an MOI of 0.01. (A, C, and E) Samples were harvested at the indicated hpi; virus titrations were done on HEL cells and expressed in PFU/ml. The results originated from triplicates of two independent experiments and are presented as means ± the SD. (B, D, and F) To compare the growth rates between WT and recombinant constructs, “best-fit” curves were fit from h 9 to 72, and slopes and intercepts were analyzed by linear regression (see Materials and Methods). P values for slopes and intercepts are indicated on the graph, and the statistical significance demonstrates a slow-growing phenotype of the recombinant virus compared to the WT.

T831I, A314V, and A314V+A684V mutations render VACV less pathogenic.

HPMPDAP^R and HPMPC^R VACV have been shown to display an attenuated pathogenicity in mice (Table 1) (4, 10, 32, 43), but no data concerning the virulence linked to the mutations T831I and A314V+A684V have been published. We therefore evaluated the lethality induced by each of the mutant VACV viruses in a mouse model of intranasal infection. Three virus doses were assessed: 4,000, 400, and 40 PFU per animal.

VACV-WR WT led to 100 and 60% mortality when administered at doses of 4,000 and 400 PFU, respectively, whereas no lethality was reported in the group infected with 40 PFU (Fig. 9A and B). None of the VACV-WR recombinant viruses was lethal to mice at the highest dose tested (Fig. 9C and F). In terms of body weight loss, mice exposed to the WT virus exhibited strong body weight loss at virus doses of 4,000, 400, and 40 PFU (P < 0.05). In contrast, animals exposed to each of the recombinant viruses bearing 314V, A314V+A684V, or T831I exhibited a very reduced or no body weight loss at all of the virus doses assessed (P > 0.05) (Fig. 9C and F). These data suggest that the mutations T831I, A314V, and A314V+A684V resulted in attenuation of VACV virulence in mice. Also, whereas the A314V hardly affected the mice, compared to PBS-treated controls, adding the A684V mutation seemed to slightly enhance the pathogenicity of the single mutated A314V virus, albeit no statistical significance was observed (P > 0.05) (Fig. 9C and D).

Fig 9.

HPMPDAP^R VACV are less pathogenic than the WT virus. NMRI mice were inoculated intranasally either with VACV-WR (A) or one of the recombinant VACVs (A314V [C], A314V+A684V [D], or T831I [E]). Uninfected groups were inoculated with PBS (●), and virus was administered at a doses of 4,000 PFU (○), 400 PFU (□),] or 40 PFU (♢). Animals were monitored for 25 days for body weight (A, C to E) and survival (B and F). The body weight evolution is shown as the percentage of the change in average weight for each group of mice.

DISCUSSION

In the few last years, our group has been involved in the establishment of the camelpox model (26–28) as a surrogate model of VARV due to their close phylogenic link (2, 33). We have observed that the molecule (S)-HPMPDAP shows a potent antiviral activity against CMLV in vitro and in vivo, and we therefore decided to select for CMLV resistant to this compound. In the present study, we describe the selection and characterization of (S)-HPMPDAP mutants of CMLV. Sequencing and marker rescue experiments revealed that three amino acid changes located in (i) the 3′-5′ exonuclease domain (A314V), (ii) the polymerase domain (T831I), and (iii) both 3′-5′ exonuclease and polymerase domains (A314V+A684V) of the viral DNA polymerase were responsible for the development of CMLV resistance toward (S)-HPMPDAP. At the gene level, each of these changes was associated with a single-nucleotide change from C to T. We further confirmed that each of the mutated alleles conferred (S)-HPMPDAP resistance in VACV.

The A314V mutation was obtained together with A684V, and this association may not be surprising since mutated amino acids at identical positions have been linked to HPMPC^R and HPMPDAP^R. In previous studies, the A314V change (Table 1) was found in VACV HPMPC^R (i) as a single point mutation (10) or (ii) associated with other amino acid modifications, including H296Y, H319N, S338F, and R604S (43). In HPMPC^R MPXV, the mutated residue A314V has been reported with another amino acid change at position 684 (A684T), and they were concomitant to A613T and T808M (30). The A684V change has already been selected with another amino acid change at position 314 (A314T), and these mutations were involved in (S)-HPMPC and (S)-HPMPDAP resistance development in VACV (Table 1) (4, 32). Also, the allele A684V has been shown to induce (S)-HPMPDAP resistance in VACV when appearing with the S851Y change (32). By producing recombinant viruses, we demonstrated that the single change A314V and the double mutation A314V+A684V were responsible for HPMPDAP^R in CMLV and VACV. In this report, we also identified a novel allele localized in the polymerase domain, T831I, responsible for HPMPDAP^R development in CMLV and VACV. It is interesting to observe that this amino acid change is located in the same region as the mutation S851Y identified in HPMPDAP^R VACV (32).

In addition to E9L sequencing, it was of interest to sequence the A20R and D4R genes since they encode the heterodimeric processivity factor of E9L (62). It has been observed that the A20R protein interacts with E9L, though the domains involved remain unknown, and it is believed that the A20R protein may serve as a bridge, within the DNA polymerase complex, binding to both E9L and D4R (15). Further, the sequencing of HPMPC^R MPXV revealed an amino acid change in A20R (S216L), although its involvement in drug resistance was not studied (30). Based on these data, we investigated whether (S)-HPMDAP could select for mutations in one of these proteins. However, no sequence alterations were observed in A20R and D4R of various HPMPDAP^R CMLV clones.

CMLV bearing T831I, A314V, or A314V+A684V were cross-resistant to the purine analogues (S)-HPMPA and cyclic-(S)-HPMPA. Also, the 5-azaC derivatives and the two 3-deaza analogues retained efficacy against each HPMPDAP^R CMLV mutant. The A314V mutation alone or in combination with the A684V mutation, induced resistance to (S)-HPMPC and cyclic-(S)-HPMPC, but this was not seen with the T831I mutation. (R)-HPMPO-DAPy, whose base structure is a pyrimidine, did not result in drug profiles similar to those of pyrimidine derivatives such as (S)-HPMPC. Our results suggest that this nucleobase displays a behavior closer to that of purine-like compounds, e.g., (S)-HPMPDAP (Fig. 7), which is in line with previous observations made with HPMPDAP^R VACV (32). Similarly, HIV-1 reverse transcriptase has been found to recognize another second-generation ANP, the 2,4-diaminopyrimidine diphosphate derivative or PMEO-DAPy-pp, as a purine base instead of a pyrimidine base (37). Each of the three CMLV mutants displayed a resistance profile to (R)-HPMPO-DAPy similar to that of the purine derivatives (Fig. 7).

Hypersensitivities to PMEO-DAPy and PAA have been seen with the three recombinants CMLV. Although both molecules had weak antiviral activity against the WT viruses, their efficacies increased against HPMPDAP^R CMLV. The two molecules have different modes of action: PMEO-DAPy, once phosphorylated, acts as a purine analogue and compete with deoxynucleoside triphosphates (dNTPs) for the binding on the viral DNA polymerase (37) and PAA is a structural mimic of inorganic pyrophosphate (PP_i) that selectively binds to the pyrophosphate binding site on various viral DNA polymerases, thus preventing PP_i cleavage from incoming dNTPs (12, 14). The mechanisms involved in this hypersensitivity development remain unclear, but comparable observations have been reported with HPMPC^R and HPMPDAP^R VACV bearing the single mutation A314T or S851Y, as well as the double mutation A314T+T688A (Table 1) (4, 32), and with herpes simplex viruses resistant to (S)-HPMPC (3, 5). Also, cross-resistance or cross-hypersensitivity between PME derivatives and pyrophosphate analogues (PAA and foscarnet) has been described for herpes simplex virus 1 (3, 5). These findings suggest that although PMEO-DAPy and PAA are not structurally related, they may have a common site of interaction with the viral enzyme. Our group previously reported the effect of the double mutation A314T+A684V in VACV, and it is interesting that both PMEO-DAPy and PAA exhibited EC₅₀s in the range of the WT virus: no hypersensitivity was seen with this mutant (4). We may therefore assume that, in combination with A684V, the change A→V, as in A314V, allowed E9L for accepting PMEO-DAPy and PAA, while the change A→T, as in A314T, did not. ST-246 was also included in our studies and demonstrated a conserved antiviral activity toward WT and drug-resistant viruses, which may not be surprising since it targets another step of orthopoxvirus propagation, the egress of the virus (29).

The study of the impact of T831I, A314V, and A314V+A684V mutations in a VACV-WR backbone was of interest to understand the mechanisms of drug-resistance development in two related species. Comparison of the drug-resistance profiles of CMLV and VACV bearing homologous mutations revealed similarities and discrepancies in terms of (i) levels of resistance to drugs and (ii) cross-resistance to (S)-HPMPC, cyclic-(S)-HPMPC, (S)-HPMP-5-azaC, and cyclic-(S)-HPMP-5-azaC. The phenotype (i.e., as determined by antiviral assays) and the viral kinetics of the A314V recombinant VACV were similar to those of CMLV-rcb-A314V, although a variability in the levels of drug resistance was seen. However, it was surprising to observe differences in terms of phenotype between VACV and CMLV bearing the double mutation A314V+A684V, while their viral kinetics were comparable to those of the WT parents. In addition, discrepancies were seen between both viruses containing the T831I change in terms of drug sensitivity and growth. From our data, we showed that the mutation A314V+A684V induced cross-resistance to the 5-azaC analogues in VACV-WR but not in CMLV. Also, in contrast to CMLV-rcb-T831I, the residue T831I in VACV led to resistance to (S)-HPMPC and cyclic-(S)-HPMPC. We hypothesized that CMLV and VACV could have species-specific fitness that may account for these discrepancies. Growth curve comparisons of CML1 versus VACV and of CML14 versus VACV only revealed significant differences between CML14 and VACV virus yields (data not shown). Therefore, if this species-specific fitness hypothesis may be a possible explanation to understand the differences seen between the two double mutants, it is not the case for CML1- and VACV-WR-rcb-T831I. We speculate that (i) the minor differences seen in the percent identities between CMLV and VACV-WR viral DNA polymerases (Fig. 10A) and (ii) the genomic diversity of the two viruses may account for the discrepancies observed in the drug profiles. To date, in the absence of a tridimensional structure of VACV E9L, it remains problematic to further address these dissimilarities, but putative models of VACV E9L can be used.

Fig 10.

Models of WT and T831I VACV E9L complexed to DNA. (A) Amino acid alignments of a portion of the E9L sequence of VACV-WR (YP_232947) and CML1 or CML14. (B-E) The tridimensional structure of the bacteriophage RB69 DNA polymerase (PDB code: 3RMD) was used to build the models of WT and T831I VACV-WR E9L. (B) Overview of VACV-WR E9L complexed to DNA (orange). (C) Superimposition of WT and T831I VACV-WR E9L. At position 831, the threonine residue is represented as blue sticks and the isoleucine as red sticks. The hydrogen bonds between T831 and the phosphate moiety of the DNA backbone are shown as dashed black lines. (D) Closer view of the position 831 showing the interaction between the two hydroxyl groups (dashed black lines) of the threonine (red ball) and of the side chain of the DNA backbone. (E) Visualization of the positions 535, 807 and 845 in the thumb domain of VACV-WR E9L. The polymorphisms observed at positions 535, 807, and 845 between VACV-WR and CML1 or CML14 E9L might have a direct effect on the interaction of the residue 831 with the DNA molecule. The figures were generated using PyMol software.

The potential roles of mutated residues at positions 314 and 684 have previously been discussed in VACV (4, 31, 32). Mutations localized in the exonuclease domain at position 314 may confer on VACV E9L an enhanced ability to excise a drug residue incorporated in the DNA strand, which was the case of the A314T VACV mutation, since this amino acid is localized on a site interacting with the DNA molecule (31). This may apply for the A314V mutation since both A314V and A314T changes induced similar drug-resistance profiles in VACV (32). The A684V mutation, localized in the polymerase domain, has been suggested to (i) reduce the enzyme capability to use diphosphorylated ANPs as a substrate or (ii) affect its capacity to accommodate templates bearing (S)-HPMPC (4, 31). This mutation is indeed localized in an α-helix of E9L (from amino acids 647 to 672) close to the DNA molecule. Since both A314V and A684V are in close contact with the DNA molecule, they may, when combined, exacerbate drug-resistance levels. In order to determine the influence of the T831I mutation, we used a predictive tridimensional model of VACV E9L bound to a DNA molecule (38). As shown in Fig. 10B, the residue 831 is located in the E9L thumb domain of the polymerase region. The T→I amino acid change results in the replacement of a polar residue toward a highly hydrophobic amino acid. This may generate a local disturbance in the E9L portion that binds to the DNA by abolishing hydrogen bonds with the phosphate moiety of the DNA backbone (Fig. 10CD). Consequently, (i) the affinity of E9L for phosphonate molecules or (ii) the exonuclease activity of E9L may be affected. The homologous amino acid of the T831 residue in bacteriophage RB69 DNA polymerase is S735, which is known to play a role not only in polymerase activity but also in exonuclease activity (38).

As mentioned earlier, variations in the amino acid sequences of VACV and CMLV E9L may account for the discrepancies seen between the resistance profiles of drug-resistant viruses. Comparison of VACV and CMLV E9L amino acid sequences revealed 15 polymorphisms from which 3, at positions 535, 807, and 845 (Fig. 10A), could potentially alter the structure of the E9L loop that bears the residue 831 (Fig. 10E). Additional biochemical studies may help us to understand these differences.

In vivo experiments were performed to evaluate the pathogenicity of these drug-resistant viruses. We observed that none of the recombinant VACV bearing A314V, A314V+A684V, or T831I induced mortality at the highest dose evaluated (4,000 PFU). Also, no significant body weight losses were recorded in each cohort infected with the recombinant viruses as observed at day 25 postinfection. These results are in line with previously reported observations: mutations associated with HPMPC^R and HPMPDAP^R in VACV E9L are linked with an attenuated phenotype in vivo (Table 1) (4, 10, 32, 43, 60). We also looked at a potential link between the localizations of the mutations in the E9L, i.e., within the exonuclease and/or polymerase domains, and this attenuated pathogenicity, but no conclusions could be drawn since all of the recombinant viruses showed comparable degrees of attenuation. Tissue virus titers were not examined, but from our experience we observed that there was no clear correlation between in vitro growth, in vivo attenuation, and the lung virus titers of mutant VACV compared to WT VACV (4, 32). Whether or not the VACV mutants described here could have a restricted growth in mouse tissues will be studied in further experiments.

The potential treatments that may be envisaged in the case of drug-resistance emergence were not investigated here, but others have (Table 1) (4, 32, 43, 59). Immunocompetent mice challenged with 5 × 10⁶ PFU of HPMPC^R VACV-WR and treated at days 1 and 3 postinfection with 100 or 50 mg of (S)-HPMPC or CMX001/kg, respectively, did show significant reductions of virus replication in the lungs and snout (59). Other studies presented similar observations with genotypically characterized HPMPC^R and HPMPDAP^R VACV and demonstrated the antiviral efficacies of (S)-HPMPC, (S)-HPMP-5-azaC and CMX001 (4, 32, 43). Recently, a PAA-based cream therapy reduced primary lesion area and delayed the mortality of immunosuppressed mice infected cutaneously with VACV-WR (58). Therefore, it might be valuable to investigate the inhibitory effect of PAA against HPMPDAP^R and HPMPC^R VACV or CMLV since such viruses exhibited hypersensitivity to this molecule.

In summary, this is the first genotypic and phenotypic characterization of two related orthopoxviruses bearing homologous mutated alleles in E9L. Our findings demonstrated that the levels of drug resistance may be exacerbated in a VACV-WR backbone in comparison to CMLV. We provide further evidences that drug-related mutations occurring in E9L contribute to an attenuated phenotype in vivo.