Ganciclovir and maribavir cross-resistance revisited: relative drug susceptibilities of canonical cytomegalovirus mutants

Abstract

Therapeutic use of maribavir for human cytomegalovirus infection has renewed attention to the extent of cross-resistance with ganciclovir as the existing standard therapy. Each drug selects in vivo for a characteristic set of resistance mutations in the viral UL97 kinase gene. To improve the calibration of relative susceptibilities to each drug, genetic variants at relevant UL97 codons were extensively phenotyped using the same baseline viral clone, cell culture conditions and growth readout. Ganciclovir-selected mutations at codons 460, 520, 592, 594, 595 and 603 conferred 2.8-fold (C603Y) to 12-fold (M460I) increases in ganciclovir 50% inhibitory concentrations (EC50) over wild type baseline, while conferring maribavir EC50 fold changes ranging from 0.21-fold (M460I) to 1.9-fold (A594V). Maribavir-selected mutations at codons 409, 411 and 480 conferred maribavir EC50 fold changes ranging from 17 (H411Y) to 210 (C480F), while conferring ganciclovir EC50 fold changes ranging from 0.7 (H411Y) to 2.3 (C480F). The P-loop substitution F342Y, selected by either drug, is confirmed to confer 4.7-fold and 6-fold increases in maribavir and ganciclovir EC50s respectively, and suggests this part of the ATP-binding domain of UL97 to be involved in moderate resistance to both drugs. The maribavir hypersensitivity of M460I and M460V may be advantageous.

1. Introduction

The nucleoside analog ganciclovir and its prodrug valganciclovir have long been used as first-line antiviral drugs for human cytomegalovirus (CMV) infection (Asberg et al., 2007; Crumpacker, 1996). The action of the viral UL97 kinase in an initial phosphorylation step contributes to the viral selectivity of ganciclovir triphosphate as a CMV DNA polymerase inhibitor. Myelotoxicity and drug resistance sometimes emerged to limit the therapeutic utility of ganciclovir, thus providing an impetus for alternative drugs that overcome these limitations. Maribavir, a benzimidazole L-riboside, is a potent CMV UL97 kinase inhibitor with a favorable adverse effect profile (Biron et al., 2002). Inhibition of UL97 kinase activity reduces viral fitness and ganciclovir phosphorylation, leading to an antagonistic interaction of the drugs (Chou et al., 2018). Because the mechanisms of action of both drugs involve UL97, the possibility of cross-resistance has been an open question, made more relevant by the FDA approval of maribavir as a CMV therapeutic in 2021 and its subsequent use as a salvage therapy in cases failing conventional treatment usually involving ganciclovir (Avery et al., 2022).

Seven specific UL97 amino acid substitutions (M460V/I, H520Q, C592G, A594V, L595S and C603W) are considered “canonical” ganciclovir resistance mutations because of their relative frequency in clinical diagnostic samples (Chou, 2020; Kleiboeker, 2023). A variety of other UL97 mutations have also been shown to confer ganciclovir resistance, concentrated in the codon range 590–607, including in-frame deletions of one or more codons. For maribavir, the corresponding canonical mutations were established more recently through large-scale clinical trials, and consist of UL97 substitutions T409M, H411Y and C480F, which are distinct from those for ganciclovir (Chou et al., 2023).

Existing data indicate a limited extent of ganciclovir-maribavir cross-resistance, although the information is incomplete and largely based on older literature and assay methods. Early studies reported no maribavir resistance of ganciclovir-resistance substitutions M460V, C592G, A594V and L595S (Drew et al., 2006), or ganciclovir resistance of maribavir-resistance substitutions T409M and H411Y (Chou et al., 2007; Chou and Marousek, 2008). However, recent clinical trials have identified instances of maribavir-ganciclovir cross resistance associated with UL97 substitutions F342Y and C480F (Chou et al., 2023).

With both ganciclovir and maribavir now in clinical use, it is timely to revisit the issue of cross-resistance by generating an updated and expanded list of comparative EC50 values for the mutations encountered after exposure to each drug.

The objective of this study was to determine EC50 values for canonical UL97 mutants and selected others at the same codon locations, in the context of a uniform viral genetic background and cell culture assay system, to include enough replicates for a confident differentiation from the baseline wild type control where applicable.

2. Materials and Methods

2.1. Antiviral compounds

Ganciclovir was sourced from Roche (Cytovene) and Sigma-Aldrich (PHR1593). Maribavir was sourced from Glaxo-Wellcome (1263W94) and MedChemExpress (HY16305).

2.2. Cell lines

Susceptibility assays were performed in ARPEp cells (Chou et al., 2017), which are ARPE-19 cells transduced to express the platelet-derived growth factor alpha receptor for viral entry. Cultures were grown in Dulbecco Minimal Essential Medium with 4.5g/L of glucose, glutamine and pyruvate, and fetal bovine serum or Hyclone Fetalclone III serum supplement (10% in the growth phase and 3% after confluency).

2.3. CMV clones and UL97 mutations

Methods used for construction of mutant viruses have been published (Chou et al., 2017; Chou et al., 2019). The CMV laboratory strain AD169 was cloned as bacterial artificial chromosome (BAC) BD1, modified to contain a CMV-IE promoter driven secreted alkaline phosphatase (SEAP) reporter gene at US3. Clone BD1 was further modified by replacement of the UL97 coding region with an ampicillin resistance cassette to prevent carryover of parental UL97 sequence into progeny clones, yielding clone BD13.

UL97 mutations were selected for phenotyping to represent amino acid substitutions M460I, M460V, H520Q, C592G, A594V, L595S and C603W (canonical ganciclovir resistance mutations), T409M, H411Y and C480F (canonical maribavir resistance mutations). Alternative amino acid substitutions that have been observed with lesser frequency in clinical isolates at codons 594 (A594P, A594T), 595 (L595F, L595W) and 603 (C603Y) were also represented. For comparison, the known cross-resistant mutant F342Y and the first-described maribavir-resistant mutant L397R (Biron et al., 2002) were included as well. Mutagenic PCR primers were used to introduce these amino acid substitutions into a plasmid transfer vector representing the strain AD169 genome from the NotI restriction site upstream of UL97 to the KpnI restriction site downstream of the gene (4.7 kb of CMV sequence spanning UL97). The vector was further modified upstream of UL97 with a bacterial kanamycin resistance selection cassette bounded by Frt motifs for subsequent removal. This mutant transfer vector, digested with NotI and KpnI, was used for conditional recombination into BD13 in a permissive SW105 E. coli strain (recombineering) (Warming et al., 2005). Qualifying recombinant clones were screened based on ampicillin susceptibility, an expected change in HindIII restriction pattern, and the kanamycin cassette was then removed by inducing the Flp recombinase present in the SW105 host. The resulting BAC clone was used to reconstitute live mutant CMV by transfection into ARPEp cells using Roche XtremeGene HP reagent. Cell-free stock was prepared for use in susceptibility testing. Mutant CMV clones were verified by Sanger DNA sequencing in both the transfer vector (mutagenized UL97 region) and final recombinant virus (entire UL97 gene), for presence of the desired mutation and absence of extraneous changes.

2.4. Reporter-based susceptibility assay

This assay has been extensively used for CMV drug resistance phenotyping (Chou and Bowlin, 2011; Chou et al., 2017; Chou et al., 2023). A calibrated viral inoculum is inoculated into each well of a row of 6 in a 24-well cluster plate containing 5- to 6-day old ARPEp culture monolayers, used 3 days after reaching confluence. After a 90-minute absorption period, the viral inoculum is removed and replaced with 1 mL of culture medium containing serial 2-fold drug dilutions or no drug in the control well. At 24 hours post-infection, a sample of culture supernatant is removed to check that the SEAP activity (assayed as relative light units emitted with a dioxetane substrate) is compatible with a low multiplicity of infection of 0.01–0.03. Drug susceptibility readouts are reported as the drug concentration that reduces supernatant SEAP activity by 50% (EC50) at 6 days (longer for growth-retarded strains), as measured by fitting an exponential curve to the SEAP activities measured at the various drug concentrations. Results are accepted for reporting if they meet quality control standards including the achieved reporter signal at time of readout, curve fit of SEAP activities to the range of drug concentrations used, and compatible EC50 results with control strains (a wild type strain and a resistant strain) of known susceptibility to the test drug. At least 7 valid replicates over at least 4 separate setup dates are used to allow for variation in cell culture conditions. Resistance is defined by an EC50 value ≥2x of matched wild type controls and hypersensitivity by an EC50 value ≤0.5x of controls.

3. Results

Either of the two sources (branded or generic) of ganciclovir and maribavir gave equivalent mean EC50 values for the wild type virus control under current cell culture conditions after a total of at least 14 replicates of assay with each drug and source (0.118 μM for maribavir and 1.32–1.35 μM for ganciclovir).

3.1. Phenotypes of ganciclovir-selected mutations

For ganciclovir, results listed in Table 1 confirm the published phenotypes of canonical resistance mutations (Chou and Bowlin, 2011; Chou, 2020; Komatsu et al., 2014; Kotton et al., 2018). UL97 substitutions M460I/V, H520Q, A594V, L595S and C603W all confer 5- to 12-fold increases in ganciclovir EC50 compared with the wild type baseline, and C592G confers a 3.2-fold increase. Alternative UL97 substitutions at codons 594 and 595, such as A594P and A594T, or L595F and L595W, also conferred ganciclovir EC50 increases in the 5- to 10-fold range. However, substitution C603Y, phenotyped here for the first time, conferred a lesser 2.8-fold increase in ganciclovir EC50.

Table 1.

Genotypes and phenotypes of CMV UL97 recombinant viruses

Straina	UL97 Genotypeb	Ganciclovir				Maribavir
		Mean EC50c	SDd	Ne	Ratiof	Mean EC50c	SDd	Ne	Ratiof
4200	Wild Type	1.33	0.21	103		0.118	0.028	94
4338	F342Y	7.91	1.74	32	6.0	0.577	0.18	61	4.7
4451	L397R	3.21	0.63	16	2.3	48.0	10.1	12	390
4352	T409M	1.64	0.21	11	1.3	9.90	2.33	36	80
4353	H411Y	0.96	0.16	12	0.71	1.8	0.53	22	17
4552	M460I	17.2	3.45	24	12	0.028	0.010	9	0.21
4455	M460V	13.4	2.44	16	9.1	0.038	0.008	11	0.30
4459	C480F	3.15	0.50	23	2.3	28.0	3.30	11	210
4452	H520Q	13.5	2.80	14	9.7	0.078	0.020	10	0.66
4207	C592G	4.30	0.92	79	3.2	0.091	0.016	13	0.78
4538	A594P	11.1	2.50	11	7.9	0.215	0.062	13	1.6
4548	A594T	7.23	1.95	19	5.0	0.098	0.015	11	0.73
4202	A594V	9.94	2.05	11	6.9	0.203	0.039	14	1.9
4545	L595F	7.73	2.17	17	6.2	0.179	0.043	12	1.3
4454	L595S	11.0	2.49	16	7.5	0.146	0.034	10	1.3
4547	L595W	13.1	2.02	15	9.0	0.162	0.024	11	1.2
4453	C603W	8.25	1.46	22	5.9	0.141	0.023	12	1.2
4546	C603Y	4.14	1.15	12	2.8	0.119	0.035	11	0.89

^aSerial number of mutant strain

^b.Amino acid substitution. WT = wild type (strain AD169)

^cDrug concentration in μM required to reduce viral SEAP growth by 50%

^dStandard deviation of EC50 values

^eNumber of replicates over at least 4 setup dates

^fEC50 of mutant virus / EC50 of matched wild type control strain

Values in bold denote EC50 ratio ≥2.0

Underlined values denote EC50 ratio <0.5

EC50 ratios <0.5 and >1.7 are significantly different from WT (p<1×10⁻⁵, Student t test, 2-tailed, unequal variances)

The maribavir susceptibilities (Table 1) of the same ganciclovir resistance mutations revealed hypersensitivity for UL97 M460I and M460V (EC50 at 0.2–0.3x of wild type), and a slight elevation of EC50 (1.9x of wild type) for A594V. Many of these maribavir-susceptible EC50 ratios are newly documented (for M460I, H520Q, A594P/T, L595F/W and C603W/Y). The more recently characterized P-loop substitution F342Y, inferred from clinical trial data to be selected after treatment with either valganciclovir or maribavir (Chou et al., 2023), shows dual resistance to maribavir and ganciclovir at 4.7- to 6-fold for each drug.

3.2. Phenotypes of maribavir-selected mutations

The most commonly encountered maribavir resistance substitutions T409M and H411Y conferred the expected 80- and 17-fold increases in maribavir EC50 over wild type (Chou et al., 2018) without ganciclovir cross-resistance (Table 1). The substitution C480F observed after maribavir treatment shows 210-fold increased maribavir EC50 and 2.3-fold increased ganciclovir EC50, as recently described (Chou et al., 2022). The in-vitro selected substitution L397R (Biron et al., 2002) likewise confers high-grade (390-fold) maribavir resistance and the same low-grade ganciclovir resistance, but has not yet been observed in clinical specimens.

Discussion

The introduction of maribavir as an alternative treatment for CMV infection has prompted an updated assessment of the degree of cross-resistance conferred by the UL97 mutations selected after exposure to either drug. The results of this study, generated using the same baseline clone, mutagenesis technique, cell culture assay and viral growth reporter system, indicate that the most common mutations selected by each drug do not confer cross-resistance to the other, but confirm important impacts on susceptibility for some specific mutations.

Among the UL97 drug resistance mutations that have been reported in clinical specimens, the P-loop substitution F342Y stands out for its 4.7- to 6-fold increases in maribavir and ganciclovir EC50s, preserved growth fitness, and cases of treatment failure of both drugs after emergence of this mutation (Chou et al., 2019; Chou et al., 2023). Thus, diagnostic genotyping needs to cover the UL97 P-loop codon range 337–345 and newly recognized emergent mutations in this codon range should be phenotyped.

A second category of ganciclovir-maribavir cross-resistance includes other published mutants that are significantly impaired in growth and UL97 kinase activity (Table 2). For example, mutations at the critical K355 codon that completely knock out UL97 kinase activity simulate the mechanism of action of maribavir and are also ganciclovir-resistant as inferred from the loss of ganciclovir phosphorylation. All of the highly maribavir-resistant mutants in Table 2 were reported to have severely reduced growth fitness (Chou et al., 2013; Komazin-Meredith et al., 2014). The first maribavir-resistant mutant observed in vitro was L397R (Biron et al., 2002), which confers high-grade maribavir resistance, moderate growth impairment (Chou et al., 2007) and considerably decreased UL97 kinase activity (Shannon-Lowe and Emery, 2010) explaining the low-grade ganciclovir cross-resistance (Table 1). The only clinically authenticated mutation in this category (reproducibly detected in treated patients) is the substitution C480F, which has a similar degree of ganciclovir cross-resistance as L397R and moderate growth impairment (Chou et al., 2022). In the phase 3 trial of maribavir in those who failed conventional therapy for CMV, C480F was often detected in those who did not respond to maribavir (Chou et al., 2023). However, 8 cases of successful valganciclovir or ganciclovir re-treatment in that trial suggested that the combination of growth impairment and low-grade resistance of C480F does not rule out a response to ganciclovir-based treatments, although close virologic monitoring is advisable.

Table 2.

Other published ganciclovir-maribavir cross-resistant mutants

UL97 Genotypea	Published EC50 ratiob		Reference
	Ganciclovir	Maribavir
F342S	7.8	18	Chou S, et al., 2013
K355del	16	304	Chou S, et al., 2013
V356G	5.5	108	Chou S, et al., 2013
D456N	12	278	Komazin-Meredith G, et al., 2014
V466G	11	321	Chou S, et al., 2013
C480R	9	243	Komazin-Meredith G, et al., 2014
P521L	17	428	Chou S, et al., 2013
Y617del	10	372	Komazin-Meredith G, et al., 2014

^aAmino acid substitution (del = in-frame codon deletion)

^b.EC50 of mutant virus / EC50 of wild type control virus

Among the well-known UL97 resistance mutations selected after exposure to ganciclovir, absence of cross-resistance to maribavir is the rule (Table 1). Extensive replicates of testing show some statistically significant differences in maribavir susceptibility that await clinical correlation. The clearest examples are M460I and M460V, which are hypersensitive to maribavir (Table 1) and may favor its use when detected after ganciclovir treatment. This phenotype is consistent with the maribavir hypersensitivity of biochemically expressed M460I mutant kinase (Shannon-Lowe and Emery, 2010). The uncommon UL97 substitutions V345I and E362D have also been reported to confer maribavir hypersensitivity (Chou et al., 2021). One of the most common ganciclovir-selected mutations is UL97 A594V, which has a slightly elevated maribavir EC50 ratio of 1.9, previously estimated at 2.1 (Drew et al., 2006), now shown to be statistically significant but not classified as resistant or associated with maribavir treatment failure in clinical trials. Other variants at codon 594 (Table 1) show EC50s closer to wild type.

The extensive retesting of known ganciclovir-resistant UL97 mutants offers a fresh benchmarking of the ganciclovir EC50 values of a wide range of historically reported mutants, some of which were phenotyped using insufficient replicates of older assays. Yet in most cases, the ratios in Table 1 are consistent with those used in current clinical practice (Kotton et al., 2018). Ganciclovir-selected UL97 mutants typically confer a 5- to 15-fold increased ganciclovir EC50 value, except C592G at 3.2-fold. The substitution A594T, earlier reported by plaque reduction assays as conferring 2.7- to 3.8-fold increased ganciclovir EC50 (Chou et al., 2002; Fischer et al., 2016), now has an updated ratio of 5. The newly phenotyped substitution C603Y has a ganciclovir EC50 ratio of less than 5, similar to C592G and A591V (Chou et al., 2017).

In conclusion, results of the current study support the lack of ganciclovir-maribavir cross-resistance of most mutations preferentially selected after treatment with either drug, but do highlight the potential for clinically significant cross-resistance especially of the P-loop mutation F342Y, and the possibly advantageous maribavir hypersensitivity of M460I and M460V.

Highlights.

• Cytomegalovirus UL97 mutations selected after ganciclovir or maribavir exposure were phenotyped to assess cross-resistance

• Mutations preferentially selected by either drug seldom confer resistance to the other

• The P-loop substitution F342Y confers clinically significant ganciclovir and maribavir cross resistance

• Ganciclovir-selected substitutions M460I and M460V confer maribavir hypersensitivity

• The low-grade ganciclovir cross resistance of C480F may not prevent its use as alternative therapy