Ganciclovir and maribavir susceptibility phenotypes of cytomegalovirus UL97 ATP binding region mutations detected by expanded genotypic testing

Abstract

Because ganciclovir resistance mutations in the cytomegalovirus UL97 gene most commonly occur at codons 460, 520 and 590-607, diagnostic genotyping for drug resistance has often omitted the analysis of codons below 440. However, the UL97 kinase inhibitor maribavir selects for distinctive resistance mutations at codons 409 and 411, and ganciclovir/maribavir resistance mutations have also been described in the ATP binding region starting at codon 335. Expanded genotypic testing of UL97 codons 335-440 in 1535 clinical specimens disclosed 10 uncharacterized sequence variants that were phenotyped for ganciclovir and maribavir susceptibility. Notable findings included low-grade ganciclovir resistance conferred by amino acid substitutions K359N and E362D, decreased maribavir susceptibility of L348V, and maribavir hypersensitivity of V345I and E362D. Recently published substitutions F342Y and K359E/Q were also confirmed. The data indicate that mutations in the UL97 ATP binding region may arise in clinical specimens to affect the interpretation of ganciclovir and maribavir resistance. This region should now be included in the standard diagnostic genotyping of UL97, especially with the introduction of maribavir into therapeutic use.

1. Introduction

The prevention and treatment of cytomegalovirus (CMV) infection is a high priority in immunocompromised individuals and has relied on ganciclovir and its prodrug valganciclovir as treatments of choice (Kotton et al., 2018). Adverse hematologic effects and occurrences of drug resistance led to foscarnet and cidofovir as alternative therapy targeting the viral DNA polymerase, and to the development of letermovir and maribavir as new antivirals with different drug targets (Limaye et al., 2020). Letermovir targets the viral terminase complex and has been approved for prophylaxis of CMV infection in stem cell transplant recipients, while maribavir targets the same viral UL97 kinase that phosphorylates ganciclovir and has successfully completed Phase 3 clinical trials for the treatment of CMV infection refractory to standard therapy in transplant recipients (ClinicalTrials.gov identifier NCT02931539).

Ganciclovir resistance results commonly from mutations in the viral UL97 kinase that phosphorylates ganciclovir, and sometimes from mutations in the UL54 viral DNA polymerase target (Chou, 2020). Since the UL97 ganciclovir resistance mutations cluster strongly at codons 460, 520 and 590-607, diagnostic genotyping has often been limited to codon ranges such as 440-640 (Kleiboeker et al., 2014). However, new developments suggest the need for expanded testing. First was the introduction of the UL97 kinase inhibitor maribavir into clinical trials as CMV therapy. In vitro and in vivo data have mapped maribavir resistance to the UL97 ATP binding domain beginning at codon 337, and extending into the catalytic domain at codon 480 (Chou, 2020). Prominently involved amino acid substitutions include T409M, H411Y and C480F, with C480F also conferring low-grade ganciclovir cross-resistance (Chou et al., 2020). Secondly, UL97 P-loop mutations at codon 342 were detected both in vitro and in vivo. F342S was selected in vitro under the UL97 kinase-activated guanosine analog filociclovir (Hussein et al., 2020) to confer cross-resistance to ganciclovir and maribavir (Chou et al., 2013). F342Y was observed after ganciclovir therapy to confer dual ganciclovir and maribavir resistance (Chou et al., 2019). L337M was selected in vitro under maribavir to confer low-grade resistance (Chou et al., 2012). Finally, sequencing of the UL97 ATP-binding domain in patients failing ganciclovir therapy and before maribavir exposure revealed newly characterized K359E and K359Q mutations that confer ganciclovir resistance but are of unknown frequency in clinical practice (Chou et al., 2019).

The purpose of this study was to determine the relative frequency and phenotypes of UL97 sequence variants in the codon range 335-440 that includes the ATP-binding region covered by expanded testing in use at a national reference laboratory over the past few years. Frequency of detection was compared with that of well-established (canonical) ganciclovir resistance mutations. Uncharacterized mutations were phenotyped for ganciclovir and maribavir susceptibility.

2. Materials and Methods

2.1. Diagnostic samples

Plasma samples (n = 1535) submitted to the reference laboratory (Eurofins-Viracor) for routine diagnostic CMV resistance genotyping in 2019 and 2020 were studied. Testing is usually ordered on the basis of suspected poor response to CMV therapy, such as in transplant recipients. Samples were de-identified prior to the current analysis, and there was no interaction with human subjects or their clinical data.

2.2. Genotyping protocol

The protocol for sample processing and Sanger sequencing was updated from a prior publication (Kleiboeker et al., 2014) by upstream extension of PCR and sequencing primers to provide bidirectional coverage of UL97 codons 248 to 640 using two forward and two reverse sequencing primers. Routine diagnostic coverage of gene UL54 was unchanged from the prior publication (Kleiboeker et al., 2014).

Total nucleic acid extraction was performed from plasma using Qiagen DNA Blood Mini kits (Qiagen, cat. no. 51104), followed by PCR amplification with a single forward and single reverse primer (Table S1) using proof-reading FailSafe Taq polymerase with pre-mix D (Epicentre, cat. no. FSP995D and FSE5101K). A single UL97 gene fragment was amplified. Agarose gel electrophoresis of the PCR products demonstrated that amplicons of the correct size and acceptable purity were generated. PCR amplicon purification was performed with Qiagen QIAquick PCR Purification Kit (Qiagen, cat. no. 28183). Sanger sequencing was performed using the BigDye® Terminator v3.1 Cycle Sequencing Kit (Life Technologies, cat. no. 4337456) with reactions for two forward and two reverse UL97 sequencing primers (see Table S1) run on an ABI 3730xl capillary sequencer with ABI POP-7 Polymer (Life Technologies; cat. no. 4363929). Acceptance criteria were trace scores ≥30; Quality values ≥20, 95% bi-directional coverage, and control sequence matching 100% with the reference sequence in the reportable range. Sequences were analyzed with ABI’s SeqScape 3 software v3.0.

2.3. Variant analysis

For this study, analysis was focused on the codon range 335-440 of the UL97 kinase ATP-binding region, starting with the conserved kinase P-loop (codons 335-346)(Chou et al., 2013), extending past the invariant lysine at codon 355 and ending at the codon 440 boundary where previous diagnostic genotyping coverage started (Kleiboeker et al., 2014). Translated nucleotide sequence data were compared with the reference CMV strain AD169 UL97 amino acid sequence using the BlastX program. All amino acid substitutions called out in the genotyping protocol were reviewed and previously unknown variants were further authenticated to exclude false-positive mutation readouts as encountered in prior studies (Chou, 2020; Chou et al., 2020). This included re-inspection of bidirectional sequencing chromatograms and resequencing of samples representing isolated instances of novel variants where residual sample was available. Remaining variants were then classified as known polymorphisms (documented by Genbank data or published literature), known resistance mutations, or uncharacterized variants for consideration of phenotyping. For comparison, the frequency of various mutations detected at codons 460, 520, 592, 594, 595 and 603 canonically associated with ganciclovir resistance (Chou, 2020) was tabulated for the 1535 diagnostic samples.

2.3. Recombinant phenotyping

Novel amino acid substitutions were phenotyped by recombinant mutagenesis of a well-characterized bacterial artificial chromosome clone of CMV strain AD169 modified with a secreted alkaline phosphatase (SEAP) reporter gene for growth quantitation, followed by yield reduction assays to determine the drug concentration (ganciclovir or maribavir) required to reduce viral SEAP activity by 50% at 6 days (EC50) after a calibrated inoculum, all as previously published (Chou et al., 2020; Chou et al., 2019). All mutant viruses were authenticated by sequencing of the entire UL97 gene. At least 10 replicates of testing over 5 separate dates were conducted for each drug and viral mutant, with concomitant known susceptible and resistant control strains. Mean EC50 values and standard deviations were tabulated. All qualitative characterizations of increased or decreased susceptibility of mutant strains are based on an EC50 value significantly different from the wild type value with a p value of <2 x 10⁻⁶ (Student t test, 2-tailed, unequal variances). Mean achieved SEAP growth (measured as relative light units, or RLU) in the EC50 assays at 6 days without drug was used to assess the growth of new mutant strains relative to known growth-competent control strains. Growth retardation was defined as a significant reduction in mean day 6 RLU values relative to controls (p<0.05, Student t test, 2-tailed, unequal variances).

3. Results

3.1. Frequency of typical ganciclovir resistance mutations

Since this study did not have access to any clinical or treatment data about the samples examined, the extent of ganciclovir exposure is assessed by the frequency of detection of mutations at UL97 codons most commonly implicated in ganciclovir resistance, as shown in Table 1 for the 1535 samples in the set. Overall, 311 of the samples (20.3%) had one or more mutations at these codons, with the expected dominance of M460I/V, H520Q, C592G, A594V, L595S/F and C603W. As compared with the earlier tabulation from the same diagnostic laboratory (Kleiboeker et al., 2014), the frequency of C592G appears to have declined relative to A594V and L595S combined (17 vs. 138 in the current set, 16 vs. 74 in the prior set of 570). Ganciclovir under-dosing may favor the selection of C592G, which confers only low-grade resistance but well-preserved growth fitness. In both studies, the tabulation is a count of samples rather than distinct patients, and the data can be skewed by the ordering of multiple tests per patient; 35 samples had more than one of the substitutions in Table 1, such as M460I+C603W, M460V+L595S or H520Q+L595S.

Table 1.

Ganciclovir resistance mutations at canonical loci

Substitution	N	% of 1535
M460I	19	1.2%
M460T	1	0.1%
M460V	29	1.9%
H520Q	34	2.2%
C592G	17	1.1%
A594P	7	0.5%
A594S	3	0.2%
A594T	9	0.6%
A594V	78	5.1%
L595del	3	0.2%
L595F	17	1.1%
L595S	60	3.9%
L595W	10	0.7%
C603R	2	0.1%
C603W	56	3.6%
C603Y	2	0.1%
Any of the above	311	20.3%

Amino acid substitutions detected at UL97 codons 460, 520, 592, 594, 595 and 603 in 1535 specimens.

N = number of specimens with substitution.

3.2. Sequence variants detected at codons 335-440

Table 2 lists the amino acid substitutions detected, the count of samples and patients for each, including detection as a mixed mutant-wild type sequence at the same codon, and pre-existing phenotypic information for the substitution. Apart from polymorphic codons 386 and 427, those with the largest count of variant samples are codons 342, 348, 359 and 415. F342Y was detected 6 times (twice as a mixture with wild type F342) in 4 separate patients. In one patient, F342Y was detected by itself twice and in a later sample in combination with UL54 P522S; in another patient, F342Y was among a mixed UL97 sequence that included H520Q. The 10 instances of L348V were all from the same patient; early samples from this patient showed L348V by itself while later specimens showed an admixture of UL97 L595F or UL54 A809V without displacement of L348V. Mutations at codon 359 were detected in 5 samples, including the recently characterized K359E and K359Q that confer ganciclovir resistance (3.7- to 3.8-fold increased EC50)(Chou et al., 2019). K359E was detected in combination with UL97 C592G, while K359Q was detected 3 times in the same patient in combination with H520Q. One instance of a novel mutation K359N was detected at about 80% abundance with the remainder being wild type K359. H415Y was detected in 5 samples from 3 patients. The well-known maribavir resistance mutation T409M was detected in 3 patients and H411Y in 1 patient, probably reflecting experimental use of the drug.

Table 2.

Sequence variants detected at UL97 codons 335-440

Substitution	N	% of 1535	Mutant-WT mix	Distinct patients	Phenotype status
Y335H	1	0.07			unknown
F342Y	6	0.39	2	4	resistant (GCV, MBV)
V345I	1	0.07			unknown
L348V	10	0.65		1	unknown
K359E	1	0.07			resistant (GCV)
K359N	1	0.07	1		unknown
K359Q	3	0.20		1	resistant (GCV)
E362D	2	0.13	1	2	unknown
A378T	1	0.07			unknown
P385S	1	0.07	1		unknown
S386L	5	0.33		3	polymorphism
V388M	1	0.07			unknown
H393Y	2	0.13		1	polymorphism
T398A	1	0.07	1		unknown
T409M	3	0.20	2	3	resistant (MBV)
H411Y	1	0.07	1		resistant (MBV)
H415Y	5	0.33	2	3	unknown
A427G	1	0.07	1		unknown
A427S	1	0.07	1		unknown
A427T	5	0.33		4	polymorphism
A435V	1	0.07			unknown
C437Y	2	0.13		1	unknown

N = Count of samples with indicated substitution

Mixed = Both mutant and wild type sequence at indicated codon in sample Phenotype status prior to this study is listed for each substitution

GCV = ganciclovir; MBV = maribavir

3.3. Phenotypic analysis of novel mutations

Most of the unknown variants in Table 2 were selected for recombinant phenotyping, excluding those encountered as minority subpopulations in single samples (P385S, T398A, A427S) or mixed readouts at polymorphic codons (A427G), without available re-sequencing data. The polymorphism S386L was previously published as maribavir susceptible (Chou et al., 2012) and the S386L variant (T3424) was also ganciclovir susceptible (EC50=1.0 ± 0.2 μM, same as WT control). Polymorphism A427T has been noted in diagnostic samples from congenitally infected infants (Genbank AHJ86183 and AKI25072). The 10 mutations that were phenotyped after construction of corresponding recombinant viruses are shown in Table 3 with their ganciclovir and maribavir susceptibilities. Control wild type (WT) and resistant strains C592G and F342Y showed the previously published EC50 values and fold-changes (Chou et al., 2019). Among the newly phenotyped strains, notable findings were the low-grade ganciclovir resistance of K359N and E362D, the slightly decreased maribavir susceptibility of L348V, and the increased maribavir susceptibility of V345I and E362D, with the latter showing a remarkable log₁₀ EC50 decrease from baseline.

Table 3.

Genotypes and phenotypes of UL97 recombinant viruses

Strain1	Genotype2	Maribavir EC50, μM				Ganciclovir EC50, μM
		Mean	SD	N	Ratio	Mean	SD	N	Ratio
Control strains
4200	UL97 WT	0.11	0.03	39		1.23	0.19	38
4207	UL97 C592G					3.82	0.65	24	3.1
4338	UL97 F342Y	0.46	0.11	25	4.2	7.34	1.5	16	5.9
New recombinants
4520	UL97 Y335H	0.11	0.028	11	1.0	1.14	0.15	12	0.9
4514	UL97 V345I	0.046	0.017	10	0.4	1.63	0.35	12	1.3
4481	UL97 L348V	0.17	0.034	19	1.6	1.14	0.23	12	0.9
4476	UL97 K359N	0.075	0.018	12	0.7	2.63	0.64	14	2.1
4477	UL97 E362D	0.010	0.003	11	0.1	4.74	1.03	15	3.8
4515	UL97 A378T	0.13	0.029	12	1.1	1.20	0.22	11	1.0
4519	UL97 V388M	0.12	0.033	11	1.1	1.42	0.21	11	1.2
4482	UL97 H415Y	0.069	0.020	13	0.6	1.35	0.22	15	1.1
4516	UL97 A435V	0.064	0.019	10	0.6	1.23	0.20	13	1.0
4518	UL97 C437Y	0.10	0.020	12	0.9	1.17	0.23	13	0.9

^1.Serial number of recombinant CMV strain

^2.Gene and amino acid substitution

WT = wild type control, including silent Frt motif upstream of indicated gene in all strains

SD = standard deviation; N = number of replicates

Ratio = EC50 of mutant virus/EC50 of wild type control.

Bold and underlined values denote EC50 significantly different from WT at p<2x10⁻⁶ (Bold = decreased susceptibility, underlined = increased susceptibility)

The newly constructed UL97 mutants (Table 3) grew normally in cell culture, similar to the control strains (Chou et al., 2019). The mean SEAP activity at 6 days among control strains (Table 3) was 3.9 x 10⁴ RLU (482-fold increase over day 1), and among the new mutants ranged from 3.5 x 10⁴ to 4.9 x 10⁴ RLU (415-fold to 557-fold increase over day 1). No mutant showed a significant decrease in SEAP RLU at 6 days when compared with control strains (p>0.2).

4. Discussion

This analysis of ATP-binding region sequence variants among 1535 samples submitted from diverse sites across the United States indicates that new and recently published amino acid substitutions in a region not usually covered in routine diagnostic practice can confer significant changes in maribavir and/or ganciclovir susceptibility, justifying the need for expanded testing.

The mutations identified in this expanded UL97 genotypic testing (Table 2) appear infrequent relative to the most commonly encountered ganciclovir resistance mutations (Table 1), but some were detected repeatedly. Notably, additional instances of F342Y, K359E and K359Q corroborate their recent description (Chou et al., 2019). The P-loop mutation F342Y was found in 4 separate patients illustrating the potential for selection after prior ganciclovir or maribavir exposure to confer cross resistance to the other drug. Different substitutions at codon 359 confer low-grade ganciclovir resistance without maribavir cross-resistance. A nearby mutation E362D confers ganciclovir resistance and marked hypersensitivity to maribavir. Other mutations including V345I (Table 3) and the previously reported M460V and M460I (Chou et al., 2007; Shannon-Lowe and Emery, 2010) also confer maribavir hypersensitivity, presumably by increasing the affinity of drug binding. Maribavir may be a good treatment option if such mutants are selected after prior ganciclovir therapy. The clinical significance of the L348V variant awaits correlation with treatment histories, but it shows the unpredictable maribavir susceptibility of variants in and near the P-loop as shown by differences among Y335H, F342Y, V345I and L348V (Table 3).

Phenotyped variants in the codon range 378-437 did not affect drug susceptibility and may be unusual polymorphisms. Since the L397R mutant was selected in vitro to confer high-grade maribavir resistance (Biron et al., 2002), continued assessment of mutations in this codon range is needed. No UL97 resistance mutations have been confirmed at codons below 335, where conserved kinase structure domains have not been defined.

Expanded genotypic testing of UL97 is especially warranted with the prospective therapeutic use of maribavir after successful Phase 3 trial for CMV infection refractory or resistant to conventional therapy. Resistance analysis of Phase 2 trials suggests that established maribavir resistance mutations T409M and H411Y, as encountered here in 4 patients, will often emerge in cases of viral rebound after an initial response to maribavir therapy (Chou et al., 2020). Early recognition of these mutations and the recently phenotyped C480F will facilitate therapeutic adjustment.

Results of this study suggest that coverage of codons 335-440 should be added to the routine genotypic testing of CMV UL97 going forward.

Highlights.

• Expanded resistance genotyping of CMV UL97 codons 335-440 was studied in 1535 clinical diagnostic samples

• The variant F342Y that confers ganciclovir and maribavir resistance was detected in 4 patients

• Newly found variants showed ganciclovir resistance for K359N and E362D, and maribavir hypersensitivity for E362D and V345I

• Canonical maribavir resistance mutations T409M and H411Y were also detected

• Genotypic testing to include UL97 codons 335-440 should now be routine, especially with increased use of maribavir