De Novo Genotypic Heterogeneity in the UL56 Region in Cytomegalovirus-Infected Tissues: Implications for Primary Letermovir Resistance

Abstract

Background

Letermovir, an inhibitor of unique long (UL)56-encoded cytomegalovirus (CMV)-terminase, shows prophylactic effects with low-grade adverse events in hematopoietic stem cell transplant recipients. Despite few case reports on acquired letermovir resistance, the frequency of de novo amino acid (A.A.) changes encoded by UL56 in CMV-infected tissues is unclear.

Methods

We analyzed CMV UL56 sequences between the conserved region IV and variable region I in 175 formalin-fixed, paraffin-embedded tissues obtained from 147 patients showing positive CMV immunochemical staining between November 2012 and October 2016. Nucleotides 552–1330 of the open reading frame of UL56 were amplified with 5 primers and sequenced by a dideoxy fluorescence-based cycle.

Results

Six (3.4%) tissues from 4 (2.7%) patients harbored A.A. substitutions. There were no known potent resistant mutations. However, we found C325Y in 2 tissues from 1 patient, along with other mutations. Four novel A.A. changes, which have not been observed in previous in vitro experiments, were identified (T244I, S301T, G312V, and M434I). Most (9 of 11, 81.8%) of the A.A. changes occurred between the codons 301 and 325 present between the conserved regions V and VI.

Conclusions

The treatment difficulties associated with letermovir resistance in a clinical setting need to be verified before its widespread use.

cytomegalovirus, genotype, letermovir, resistance, tissues

Cytopathic inflammation by cytomegalovirus (CMV) can directly result in tissue-invasive end-organ diseases [1, 2]. Active CMV replication is also associated with allograft dysfunction and higher morbidity and mortality in solid organ transplant (SOT) and hematopoietic stem cell transplant (HSCT) recipients through indirect immune dysfunction [3, 4]. These clinically significant observations have necessitated the development of exquisite prevention and management guidelines for CMV based on the risk factors and/or immune monitoring [2, 5].

Letermovir, an inhibitor of the CMV terminase complex, shows significant prophylactic effects, including reduction of mortality rate in adult CMV-seropositive allogeneic HSCT recipients [6–8]. With approval from the US Food and Drug Administration for the prophylactic use of the drug in this population, letermovir salvage therapy for CMV disease has been reported particularly in SOT recipients with valganciclovir resistance or refractoriness [9–11]. The off-label use of letermovir may be originated because of its favorable safety profile [6, 9, 12]. However, it will be necessary to evaluate the appropriate dosage and duration, as well as the precise method to measure outcomes of the letermovir treatment in CMV end-organ diseases, because a suboptimal therapy could lead to emergence of resistance [13–15].

Several CMV isolates from the blood samples of participants enrolled in the letermovir arm in a phase 3 clinical trial with allogeneic HSCT recipients have shown amino acid (A.A.) substitutions at 36 positions across the coding sequence of unique long (UL)56 upon prophylaxis or at posttreatment follow-up [16]. The majority of the A.A. substitutions in UL56 in patients with letermovir prophylaxis failure have been found to be different from the major letermovir resistance profiles observed in in vitro studies [16–18]. These findings suggest that various A.A. substitutions in UL56 can be attributed to the potential acquired resistance against letermovir [16]. Even though a few case reports have shown UL56 A.A. substitutions, including V236M and C325F/Y, during letermovir use [11, 12, 14, 19], little is known regarding the de novo genotypic nucleotide changes and primary A.A. substitutions in UL56 in CMV-infected tissues. In this study, we aimed at determining the de novo genetic variants and A.A. changes in the open reading frame (ORF) of UL56 between the conserved region IV and variable region I, which have been identified previously by the cell-culture experiments as mutation sites responsible for the letermovir phenotypic resistance.

MATERIALS AND METHODS

Study Design and Tissue Collection

We retrospectively collected 285 CMV-infected formalin-fixed, paraffin-embedded (FFPE) tissues from 237 patients, which were identified by histopathologic findings from the electronic medical system between November 2012 and October 2016 at the Severance and Gangnam Severance Hospital, a university-affiliated tertiary care center, Seoul, South Korea. Cytomegalovirus-infected FFPE tissues with the following 3 features were eligible for sequencing analyses: (1) acute or chronic inflammation with ulceration or necrosis; (2) giant cells with intracellular viral inclusion bodies; and (3) intranuclear and cytoplasmic positive findings of CMV immunohistochemical staining using an anti-CMV monoclonal antibody [20]. All of the tissue sections were kept in 1 laboratory. This study was approved by the local institutional review board at the Gangnam Severance Hospital with the waiver of informed consent.

Sample Preparation

The whole genomic deoxyribonucleic acid (gDNA) was isolated using the QIAamp DNA FFPE tissue kit (QIAGEN Korea Ltd., Seoul, Korea). The gDNA was incubated with uracil-DNA glycosylase (UDG) (New England Biolabs, Ipswich, MA) for 10 minutes at 37°C. The UL56 ORF between the conserved region IV and variable region I (552–1330 nucleotides) was amplified by polymerase chain reaction (PCR) using (1) 5 primers (same with sequencing primers) (Table 1), (2) 50 ng of gDNA, (3) DNA-free-dNTPs, and (4) DNA-free Taq (CellSafe, Gyeonggi-do, Republic of Korea) or Pfu (Bioneer, Daejeon, Republic of Korea) DNA polymerases in a thermal cycler (C1000 Touch; Bio-Rad Laboratories, Inc., Hercules, CA). The 5 amplicons from 1 sample ranging from 213 to 310 base pairs were overlapped (Table 1). The polymerase chain reactions (PCRs) were performed under the following sequential conditions: (1) 5 minutes at 95°C; (2) 40 cycles of 15 seconds at 95°C, 15 seconds at 60°C, and 30 seconds at 72°C; and (3) 10 minutes at 72°C. Next, we purified the PCR products from the bands by agarose gel electrophoresis using a PureLink PCR purification kit (Invitrogen, Thermo Fisher Scientific Corp., Foster City, CA). We used a positive control standard CMV strain from the World Health Organization (WHO) (National Institute for Biological Standards and Control [NIBSC] 09/162, Hertfordshire, UK) and negative control from normal FFPE tissues for PCRs and electrophoreses. To avoid DNA contamination, we performed DNA extraction from the tissues or PCR bands in separate clean benches and prepared the whole mixtures for PCR in a biosafety cabinet, using aerosol-resistant pipettes. We stored 1120 amplicons from 224 tissues (185 patients), which had definite bands in all electrophoresis, at −80°C until sequencing analyses (Supplementary Appendix Figure 1).

Table 1.

Primer Sets Used in Sequencing the Open Reading Frame Between the Conserved Region IV and Variable Region I of Cytomegalovirus UL56 and in Polymerase Chain Reactions of UL83

Numbers of Primers		PCR and Sequencing Primer Sequences (5′→ 3′)		Regions of Sequencing From the Amplicons or PCR		Amplicon Size (Base Pairs)
		Forward	Reverse	Nucleotidesa	Codons
UL56	1	GTACGTCGAAGGGACGACAT	GAAAGGACTCCAGCCAAGTG	552–764	184–254	213
	2	AGCTGACCATCATCCCGAAT	TGGATGTAGCTGTGGTAGGC	590–815	197–271	226
	3	GCCTACCACAGCTACATCCA	GAGCACGAAGATGTCCTCCA	796–1041	266–347	246
	4	CTTCTTCATCGGCCTACCAC	CGTACAGATGGCGACTGATG	785–1011	262–331	228
	5	GTGGAGGACATCTTCGTGCT	CCGTCATCAAAGTCGTACCC	1021–1330	341–443	310
UL83		GCAGCCACGGGATCGTACT	GGCTTTTACCTCACACGAGCATT	647–805	216–268	159

Abbreviations: PCR, polymerase chain reaction.

^aPosition in the open reading frame.

Measurement of Cytomegalovirus Concentrations

The CMV concentration in the tissue gDNA was measured by real-time quantitative PCR by using an in-house procedure and the standard curve from the WHO first international standards for the NIBSC 09/162 strain [21]. We amplified the CMV UL83 with forward and reverse primers on a LightCycler 480 platform (Roche Diagnostics, Gangnam-gu, Seoul, South Korea) (Table 1) [22]. We verified that all of the tissue gDNA samples had CMV concentrations of ≥10⁴ international unit (IU)/μL gDNA.

Sequencing Analyses for the Cytomegalovirus UL56 Region

The nucleotides of double-stranded DNA were automatically sequenced by dideoxy fluorescence-based cycle sequencing using a BigDye Terminator v3.1 Cycle Sequencing Kit and an ABI 3730XL DNA Analyzer (Applied Biosystems, Thermo Fisher Scientific Corp.) [23, 24]. Sequencing failures were defined as follows: any finding of dye blobs, poor or wrong mobility correction, and mixed sequence throughout, up to or after a certain point; mixed sequence due to extreme signal saturation; and poor 5′-sequencing quality in the electropherogram as well as no or a low relative fluorescence units signal in raw sequencer data [25, 26]. With the exclusion of 49 tissues from 52 patients, with sequencing failure in at least 1 of the 5 PCR products, the sequenced nucleotides of the final 175 tissues from 147 patients were compared with the UL56 coding sequences of Human herpesvirus 5 strain Merlin, complete genome (National Center for Biotechnology Information [NCBI] reference sequence; NC_006273.2) using the NCBI BLAST alignment program (Supplementary Appendix Figure 1) [16, 17]. We performed sequence alignment comparison, except for 90 nucleotides from the 5′-end in both forward and reverse sequencing, to analyze the qualified and precise sequencing data. All nucleotide mutations and A.A. changes were thoroughly verified by manually reviewing the signal changes in the electropherogram. We considered the nucleotide changes and A.A. substitutions that were equally observed in all the sense and antisense strands.

To minimize the false-positive sequence artifacts originating from DNA damage during tissue fixation and preservation processes, we used the following methods: (1) UDG treatment to remove uracil in the contaminated or damaged DNA, which mainly occurred by deamination of cytosine to uracil; and (2) use of different DNA polymerases of DNA-free Taq to avoid exogenous DNA contamination, as well as Pfu with low bypass efficiency over DNA lesions (uracil and abasic sites) and with high accuracy for DNA polymerization because of its DNA proofreading function through 3′→5′ exonuclease activity [27].

Statistical Analyses

The data were expressed as number (percentage) or median (interquartile range [IQR]). The nominal and continuous variables were compared between the 2 groups using Fisher’s exact test and Mann-Whitney U test, respectively. All 2-tailed P ≤ .05 were considered to be statistically significant. We performed statistical analyses using SPSS v25 software (SPSS, Chicago, IL).

RESULTS

Characteristics of Tissues

The majority of the CMV-infected tissues were from the gastrointestinal tract, except for 4 from the lung. The mean CMV concentration per microliter of tissue gDNA was log₁₀6.49 IU, which was similar in the tissues with and without A.A. changes (Table 2).

Table 2.

Clinical Characteristics of Tissues Infected by Cytomegalovirus and Tissues With or Without Amino Acid Substitutions Between Conserved Region IV and Variable Region I of pUL56^a

Characteristics	Total (n = 175)	Amino Acid Substitution		P Values
		Yes (n = 6)	No (n = 169)
Organs				.859b
Gastrointestinal tract	171 (97.7)	6 (100)	165 (97.6)
Upper	53 (30.3)	2 (33.3)	51 (30.2)
Esophagus	26 (14.9)	0 (0)	26 (15.4)
Stomach	24 (13.7)	2 (33.3)	22 (13.0)
Duodenum	3 (1.7)	0 (0)	3 (1.8)
Lower	118 (67.4)	4 (66.7)	114 (67.5)
Small bowel	3 (1.7)	0 (0)	3 (1.8)
Colon	77 (44.0)	4 (66.7)	73 (43.2)
Rectum	38 (21.7)	0 (0)	38 (22.5)
Lung	4 (2.3)	0 (0)	4 (2.4)
CMV concentration in tissue, log10 IU/μL gDNA	6.49 (5.43–7.52)	5.96 (4.58–7.34)	6.52 (5.63–7.38)	.183

Abbreviations: A.A., amino acid; CMV, cytomegalovirus; gDNA, genomic DNA; IU, international unit; UL, unique long; VL, viral load.

^aData are expressed as the number (percent) or median (interquartile range).

^bComparison of the frequencies for the upper and lower gastrointestinal tracts between the 2 groups.

Single-Nucleotide Mutations Not Leading to Amino Acid Substitutions

We did not find any difference between the sequencing using 2 different DNA polymerases (Pfu and DNA-free Taq DNA polymerase). No tissues harbored deletion or insertion mutations or changes in the stop codon. In addition, there was no consecutive alteration in 2 nucleotides. We found 10 types of single-nucleotide mutations that did not lead to A.A. substitution in 11 (6.3%) tissues. The most common mutations were TTG to CTG coding the 208th leucine (7, 4.0%) (Supplementary Appendix Table 1). There was no nucleotide mutation in the conserved regions IV, V, and VI, and 2 types of nucleotide changes in the variable region I (H427 and R431) had low frequencies (each of .6%) (Supplementary Appendix Table 1).

Amino Acid Substitutions in Cytomegalovirus UL56

Amino acid changes were observed in 6 (3.4%) tissues from 4 (2.7%) patients. We found 5 A.A. changes, which were T244I, S301T, G312V, C325Y, and M434I (Figure 2). The most frequent A.A. substitution was S301T (4, 2.3%), followed by G312Y (3, 1.7%) (Table 3). Nine (81.8%) of 11 A.A. substitutions were concentrated between the codons 301 and 325, which formed the ORF between the conserved region V and VI (Table 3 and Figure 1). Four (2.3%) tissues had ≥2 A.A. substitutions that were most commonly S301T and G312V (2 of 4, 50%). The C325Y mutation found in 1 patient, which is known as a potent resistant mutation for letermovir [9, 17, 18], was accompanied by another A.A substitution, S301T (1 of 2) or S301T + G312V (1 of 2) (Supplementary Appendix Table 2). The clinical information of patients with A.A changes is presented in Table 4.

Table 3.

Amino Acid Substitutions in the UL56 Open Reading Frame Between Conserved Region IV and Variable Region I in Cytomegalovirus-Infected Tissues

Position of A.A.	Nucleotide Mutation in Sense Strand DNA (5′→3′)	A.A. Substitution	Number (Percent)
			Tissues (N = 175)	Patients (N = 147)
244	ACA → ATA	T244I	1 (0.6)	1 (0.7)
301	TCC → ACC	S301T	4 (2.3)	2 (1.4)
312	GGC → GTC	G312V	3 (1.7)	2 (1.4)
325	TGC → TAC	C325Y	2 (1.1)	1 (0.7)
434	ATG → ATT	M434I	1 (0.6)	1 (0.7)

Abbreviations: A.A., amino acid; C, cysteine; DNA, deoxyribonucleic acid; G, glycine; I, isoleucine; K, lysine; M, methionine; Q, glutamine; S, serine; T, threonine; UL, unique long; V, valine; Y, tyrosine.

Table 4.

Clinical Information of Patients With Amino Acid Changes in Cytomegalovirus pUL56 Between Conserved Region IV and Variable Region I

Patient No.	No. of CMV-Infected Tissues Included the Study	A.A. Substitution (No.)	Type of Tissues	Age (Years)/Sex	Major Current Diseases With the Underlying Condition
1	3	S301T + G312V + C325Y (1)	Colon	68/F	ARDS, CAP, Non-immunocompromised critically ill patients in ICU care
		S301T + G312V (1)	Colon
		S301T + C325Y (1)	Colon
2	1	S301T + G312V (1)	Stomach	83/M	Diabetes mellitus, end-stage renal disease on hemodialysis, long-term glucocorticoid use
3	1	T244I (1)	Stomach	76/F	Cerebral hemorrhage, quadriplegia, enteral feeding via gastrostomy tube
4	1	M434I (1)	Colon	59/M	Liver transplant recipients

Aberrations: A.A., amino acid; ARDS, adult respiratory distress syndrome; C, cysteine; CAP, community-acquired pneumonia; CMV, cytomegalovirus; G, glycine; I, isoleucine; ICU, intensive care unit; M, methionine; S, serine; T, threonine; UL, unique long; V, valine; Y, tyrosine.

Figure 1.

Representation of the sequencing area and positions of amino acid substitutions in the open reading frame of UL56 from conserved region I to variable region I in cytomegalovirus-infected tissues [17, 18]. Conserved regions I and VI as well as variable region I are indicated as gray boxes. The bold black box and light blue box indicate the sequencing areas in this study and the region with commonly occurring amino acid changes, respectively. C, cysteine; D, aspartic acid; E, glutamic acid; G, glycine; I, isoleucine; K, lysine; M, methionine; L, leucine; Q, glutamine; S, serine; T, threonine; UL, unique long; V, valine; Y, tyrosine, VRI, variable region I.

Figure 2.

Electropherograms showing single-nucleotide point mutations leading to relatively frequent amino acid substitutions in the open reading frame of UL56 in cytomegalovirus-infected tissues. (A) T244I, (B) S301T, (C) G312V, (D) C325Y, and (E) M434I. The nucleotide configuration indicates the sequences in the sense strand DNA (5′→3′). The arrows and numbers indicate the point mutations and codons in the electropherogram. The top and bottom letters in the electropherogram indicate the nucleotides and amino acids, respectively. C, cysteine; G, glycine; I, isoleucine; M, methionine; S, serine; T, threonine; UL, unique long; V, valine; Y, tyrosine.

Coexistence of Two Amino Acids in pUL56

Several electropherograms showed dual nucleotide signals at the codons 301 and 312. At the codon 301, T and A signals were mixed with a similar signal density in 12 (7.0%) of 171 tissues, except in 4, which had definite S301T A.A. changes. This indicates the possibility of a different subpopulation with S301 or S301T in the same tissue. At the codon 312, G and T in the second nucleotide had comparable signals in 7 (4.1%) of 172 tissues without a definite G312V substitution (Figure 3).

Figure 3.

Mixed variabilities and coexistence of 2 amino acids at codons 301 and 312 within pUL56 of cytomegalovirus-infected tissues. (A) Serine and threonine at codon 301. (B) Glycine and valine at codon 312. The nucleotide configuration indicates the sequences in the sense strand DNA (5′→3′). The top/bottom letters and numbers in the electropherogram show the nucleotides/amino acids and the position of codons, respectively.

Discussion

Given the direct interaction between letermovir and pUL56 and the higher resistance potential of letermovir than that of ganciclovir, it is important to investigate the genetic variabilities in CMV terminase complex subunits and phenotypic-resistant mutations of letermovir in various clinical isolates [17, 18]. Therefore, we may need to explore the possibility of intrinsic resistance, as well as the development of acquired or breakthrough resistance through rapid mutant selection for letermovir during the prevention and treatment of CMV diseases. To the best of our knowledge, this is the first study to evaluate the de novo genetic mutations and A.A. changes in UL56 in the end organs infected by CMV. We identified several new types of A.A. substitutions along with single-nucleotide mutations between 184 and 443 codons of UL56. The findings of previous cell culture experiments have revealed that this ORF region is the major area on which letermovir resistance mutations are clustered [16–18, 28].

Although the present study samples had never been exposed to letermovir because of the current unavailability of letermovir in our country, the potent C325Y resistance mutation was surprisingly detected in 2 tissues. It is interesting to note that all C325Y mutations were accompanied with other A.A. substitutions. Various combinations of double and triple A.A. changes in UL56 have been reported by previous studies using recombinant CMV [29–31]. However, there is no report describing multiple A.A. changes in CMV UL56 obtained from clinical isolates.

Since the first instance of clinical letermovir resistance was reported in 2016, CMV in the peripheral blood of several patients had the V236M and C325F/Y genotypic mutations with or without phenotypic resistance to letermovir [11, 12, 14, 15, 19]. The C325F/Y mutations have been found to develop rapidly in 2 allogeneic HSCT and each of 2 lung or heart transplant recipients with high CMV viral load after 30 or 100 days of letermovir treatment for asymptomatic CMV DNAemia and/or CMV retinitis [11, 12, 15, 19]. Further genetic analysis of participants enrolled in clinical trials revealed that 1 patient receiving letermovir developed new A.A. substitutions 9 days after the exposure [16]. A rapid evolution of letermovir phenotypic resistance was also observed in in vitro studies with mutant CMV strains, showing a high level of resistance after 7 days of exposure [29]. These case reports suggest that letermovir resistance associated with clinical treatment failure could be rapidly developed.

In 2019, Komatsu et al [16] analyzed whole sequences of CMV UL56 and UL89 isolated from peripheral blood of patients who received letermovir in the phase 3 clinical trial. The study findings showed many low frequency resistance variants of unclear clinical significance, including deletion, changing of stop codon, as well as point mutations across the entire UL56. Moreover, patients with obvious failure of letermovir prophylaxis had various resistance mutations or no A.A. changes [16]. All A.A. changes or mutations, except for V236M, E237G, and C325W in 3 subjects, were different from the known resistance mutations observed in cell culture variants [16–18, 28–30].

In this study, all A.A. substitutions in pUL56 were newly identified except for C325Y. In addition, the A.A. change at the codon 244 was isoleucine instead of the previously known lysine or arginine [18]. Because we did not perform any in vitro cell culture study with recombinant laboratory CMV mutants for each single-nucleotide change leading to new A.A. substitutions, we have no information on whether these novel mutations are associated with phenotypic resistance to letermovir. This is the main limitation of our study. However, this study has the following unique characteristics: (1) this is the first sequencing analysis of a large region of the CMV ORF UL56 obtained from CMV-infected tissues; (2) new findings of single-point mutations and similarity with in vitro studies for A.A. substitution position between the conserved regions V and VI, especially between the codons 301 and 325 [9, 17, 18]; (3) features of double and triple A.A. changes in 1 sample and coexistence of a subpopulation with different A.A. at specific codons in pUL56; and (4) de novo detection of the potent mutation C325Y. In line with previous case reports, this study suggests that letermovir has a low resistance barrier or a possibility of primary resistance.

The results obtained from in vitro nonclinical cell culture and CMV recombinant mutant strain experiments may not be applied directly to clinical settings because several studies have reported somewhat different data regarding mutations and whole-genome sequences, showing large heterogeneity in patients [16, 17, 29, 31–34]. Thus, it is important to focus on the development of letermovir resistance in many clinical isolates before its broad use; in particular, it is important to identify the clinical significance of low-level resistance or the effect of multiple mutations on CMV replication fitness [13].

Conclusions

In conclusion, we, for the first time, identified the de novo C325Y mutation in CMV-infected tissue. C325Y coexisted with other mutations in a single tissue. The CMV pUL56 codons 301 and 312 had mixed nucleotides with A.A. substitutions in some tissues. These findings suggest that the intrinsic or newly developed resistance to letermovir needs to be thoroughly examined before its extensive therapeutic use.

Supplementary Data

Supplementary materials are available at The Journal of Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.

Notes

Author contributions. All authors revised the manuscript critically and approved the final version of the submitted manuscript. H. J. and D. E. K. equally contributed to the design and experiments of the study as well as the data analysis. B. J. L. and J.-H. J. selected and prepared all the tissue specimens. S. Y. M. and Y.-M. H. performed the experiments. H. J., D. E. K., and S. H. H. wrote the manuscript.

Financial support. This research was supported by the Basic Science Research Program through the National Research Foundation of Korea (NRF) grant funded by the Korean Government (Ministry of Education) (Grant No. NRF-2017R1D1A1B03032844).

Potential conflicts of interest. All authors: No reported conflicts of interest. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest.

Author notes

H. J. and D. E. K. contributed equally to this work.