|Home | About | Journals | Submit | Contact Us | Français|
Recombinant phenotyping of cytomegalovirus (CMV) pol region III mutations from clinical specimens showed that T813S and G841A each conferred foscarnet resistance and approximately threefold increased ganciclovir resistance; adding the UL97 mutation C592G increased ganciclovir resistance to approximately sixfold. Bacterial artificial chromosome CMV clones containing pol mutation L845P were nonviable unless repaired with the wild-type sequence.
Mutations in the cytomegalovirus (CMV) UL54 DNA polymerase (pol) gene may confer resistance to ganciclovir (GCV), foscarnet (FOS), or cidofovir (CDV) (9). Because GCV and FOS are currently the main alternative therapies, pol mutations that result in GCV-FOS cross-resistance are clinically important. In this study, we examined three previously uncharacterized pol region III (codons 805 to 845) mutations observed in clinical specimens. pol mutation G841A (in conjunction with UL97 mutation M460V  and pol mutation P522A) was reported in a CMV isolate from a patient with chronic leukemia (8). pol mutation T813S was found in conjunction with UL97 mutation C592G (4) in a heart transplant recipient. pol mutation L845P was found in conjunction with UL97 mutation M460V and pol mutation A809V (3) in a lung transplant recipient. Transplant recipients were CMV-seronegative recipients of a CMV-seropositive organ; both had received weeks to months of GCV and FOS, and viral mutations were detected by PCR amplification of CMV sequences directly from peripheral blood specimens without an accompanying viral culture.
Recombinant phenotyping (marker transfer) was performed to establish the phenotypes associated with the observed region III mutations. Strains and methods for performing these experiments have been previously described (4). The pol mutations were transferred individually or in combination with UL97 mutation C592G into reference strain T2211, which contains a secreted alkaline phosphatase (SEAP) reporter gene for viral quantitation by assay of supernatant SEAP activity with a chemiluminescent substrate (4). Recombinant viruses were checked for the presence of the desired mutation and plaque purified at least twice. Drug sensitivity was assayed as previously described (4), by determining the drug concentration required to reduce supernatant SEAP activity by 50% (EC50). Multiple assays (6 to 23 total replicates per strain) were performed to give mean EC50s and standard errors of each of the drugs GCV, FOS, and CDV.
Genotypes and phenotypes of the recombinant viruses are shown in Table Table1.1. The mutations A809V, T813S, and G841A all showed similar phenotypes of low-grade GCV resistance, three- to fivefold increased FOS resistance, and borderline or a slight decrease in CDV sensitivity. This phenotype is compatible with the original marker transfer studies on the A809V mutation (3), which were done with traditional plaque reduction assays and showed 6.3-fold, 2.6-fold, and 1.6-fold increases in the FOS, GCV, and CDV EC50s, respectively. When combined with the UL97 mutation C592G, which by itself confers low-grade GCV resistance (4), the pol mutations result in an overall increase in GCV resistance into the five-to sevenfold range, which approaches the level of GCV resistance conferred by the most common UL97 mutations M460V, A594V, and L595S (4).
When compared by using SEAP growth curves (Fig. (Fig.1),1), the pol mutants were attenuated in growth (A809V least, G841A most) in comparison with baseline strain T2233 but less so than UL97-defective strain T2266 (5). UL97 mutation C592G had little effect on growth.
In contrast to the other mutations studied, neither L845P nor the combination of A809V and L845P could be incorporated into live recombinant virus by the standard transfer methods repeatedly. Therefore, these mutations were transferred into the AD169-derived bacterial artificial chromosome (BAC) CMV clone pHB5 (1). Transfer was achieved by the galK selection and counterselection system with conditionally recombinogenic Escherichia coli strain SW102 and a protocol previously described (14). The UL54 pol region of pHB5 was replaced with a galK expression cassette amplified with galK plasmid PCR primers (14) fused to 48- to 50-bp CMV sequences from the beginning and end of pol. This was done by recombination in 42°C heat-induced SW102 containing pHB5, followed by selection on galactose medium. The desired pol mutations were then introduced by another round of recombination with a mutant transfer vector. Recombinant BACs were isolated by counterselection with 2-deoxygalactose (14) and verified as to their complete pol coding sequences and genomic restriction digest patterns. Live CMV is typically recovered by transfection of 2 μg of BAC DNA into subconfluent human foreskin fibroblast (HFF) monolayers with Fugene 6 reagent (Roche). The parental pHB5 BAC and other pHB5-derived BACs not part of this study routinely yielded infectious CMV (12 BACs on the first attempt and 1 on the second). However, neither of the BACs containing L845P yielded infectious virus after four or five attempts each.
To prove that the L845P mutant BAC was not defective elsewhere in the CMV genome, its DNA was cotransfected with 6.6 kb of plasmid-derived DNA representing the pol region wild-type sequence (nucleotides 74828 to 81436 of the AD169 sequence with GenBank accession no. X17403). This cotransfection resulted in the recovery of live virus in HFF culture. When this virus was plaque purified and evaluated, each of eight plaques showed a wild-type configuration at codon 845 (L845); no plaques containing L845P were isolated. Similarly, when the BAC containing both L845P and A809V was cotransfected with a shorter plasmid DNA segment (representing wild-type pol codons 239 to 1107), live virus was recovered, and sequencing of each of six isolated plaques showed the wild-type pol configuration and loss of both the L845P and A809V mutations. Rescue of live virus after reversion of L845P to the wild type in the CMV BAC clones is interpreted as strong evidence that the L845P mutation causes loss of viral viability, since the required recombination repair step is a low-frequency event.
Because of an increasing number of recognized mutations, pol region III has become more important in the genotypic diagnosis of CMV drug resistance. The usual phenotype of region III mutations is severalfold increased FOS resistance, as shown here by the mutations A809V, T813S, and G841A. Historically, L802M (2, 6) and A809V (3) have been the most common resistance mutations in this region; other validated ones include V812L (7), T821I (6), A834P (11), and T838A (12). Exceptionally, mutation K805Q was reported to confer FOS hypersensitivity (6). Although some region III mutations by themselves (e.g., T821I and A834P) are also reported to confer approximately fivefold increased GCV resistance, the more commonly associated two- to threefold increases in GCV resistance are mainly of concern when combined with UL97 mutations. This leads to a viral phenotype of clinically significant resistance to both of the drugs (GCV and FOS) commonly used to treat CMV disease. Most patients developing region III mutations have received extensive GCV therapy in addition to FOS and harbor UL97 mutations, as illustrated by all of the cases studied here and in previously published case reports (3, 12).
As discussed in recent publications (10, 11), modeling analyses based upon available homologous polymerase crystal structures strongly suggest that region III mutations potentially affect FOS, GCV, and CDV sensitivity because the functions of this region include pyrophosphate binding and recognition of the incoming base. Despite these critical functions and many conserved residues, the virus is clearly tolerant of a variety of mutations with an acceptable compromise in viral fitness (Fig. (Fig.1).1). The increased GCV resistance of UL97-pol double mutants is attributed to the added effect of decreased production of GCV triphosphate due to UL97 mutation (13).
Codon 845 is highly conserved among homologous DNA polymerases, but this is true of many other pol residues mutated in drug-resistant CMV strains (9), so there was no expectation that mutation L845P would cause loss of viral viability. Indeed, this is the first such mutation found in the CMV DNA polymerase. Why it should appear in a clinical specimen is unclear. Interpretation is confounded by recent diagnostic practices that commonly yield no live CMV isolates, no serial genotypic analyses, and no residual specimen for reanalysis after an unusual circumstance is discovered. A distinct possibility is that the L845P diagnosis was a PCR or sequencing artifact. L845P was detected in 60 to 75% of the sequence population when the PCR product was sequenced in both directions. Artifacts may be more likely if few intact viral copies exist in the clinical specimen, and extensive PCR amplification is required to produce a template for sequencing. Another possibility for which no evidence exists is that an L845P mutant may be viable in vivo when complemented by genetic change elsewhere in the clinical CMV isolate. The current practice of diagnosing CMV resistance by sequencing without reference to live clinical isolates introduces new technical factors that must be taken into account in interpreting viral genotypes and reinforces the need for phenotypic validation of newly recognized mutations.
We thank Martin Messerle, Gabriele Hahn, and Ulrich Koszinowski for providing the pHB5 BAC, Donald Court and Neal Copeland for providing E. coli strain SW102 and the pGalK plasmid, and Heather Lichy for technical assistance.
This work was supported by NIH grant AI39938 and Department of Veterans Affairs research funds.
Published ahead of print on 20 August 2007.