Human cytomegalovirus: taking the strain

Abstract

In celebrating the 60th anniversary of the first isolation of human cytomegalovirus (HCMV), we reflect on the merits and limitations of the viral strains currently being used to develop urgently needed treatments. HCMV research has been dependent for decades on the high-passage strains AD169 and Towne, heavily exploiting their capacity to replicate efficiently in fibroblasts. However, the genetic integrity of these strains is so severely compromised that great caution needs to be exercised when considering their past and future use. It is now evident that wild-type HCMV strains are not readily propagated in vitro. HCMV mutants are rapidly selected during isolation in fibroblasts, reproducibly affecting gene RL13, the UL128 locus (which includes genes UL128, UL130 and UL131A) and often the U(L)/b' region. As a result, the virus becomes less cell associated, altered in tropism and less pathogenic. This problem is not restricted to high-passage strains, as even low-passage strains can harbour biologically significant mutations. Cloning and manipulation of the HCMV genome as a bacterial artificial chromosome (BAC) offers a means of working with stable, genetically defined strains. To this end, the low-passage strain Merlin genome was cloned as a BAC and sequentially repaired to match the viral sequence in the original clinical sample from which Merlin was derived. Restoration of UL128L to wild type was detrimental to growth in fibroblasts, whereas restoration of RL13 impaired growth in all cell types tested. Stable propagation of phenotypically wild-type virus could be achieved only by placing both regions under conditional expression. In addition to the development of these tools, the Merlin transcriptome and proteome have been characterized in unparalleled detail. Although Merlin may be representative of the clinical agent, high-throughput whole-genome deep sequencing studies have highlighted the remarkable high level of interstrain variation present in circulating virus. There is a need to develop systems capable of addressing the significance of this diversity, free from the confounding effects of genetic changes associated with in vitro adaptation. The generation of a set of BAC clones, each containing the genome of a different HCMV strain repaired to match the sequence in the clinical sample, would provide a pathway to address the biological and clinical effects of natural variation in wild-type HCMV.

Fig. 1

Evolution of genetic changes that have accumulated in the most commonly used variants of strain AD169 (varUK and varATCC) and a variant obtained from the University of Chicago (varUC) that retains a part of the U_L/b′ region. Adapted from figure originally published in the J. Gen. Virol. [15], reusing the author’s own content

Fig. 2

Low-passage strain Toledo provides more effective protection against NK cells than either of the laboratory strains AD169 or Towne. An NK cytolysis assay performed in which an NK cell line (NKL) was incubated with human foetal foreskin fibroblasts infected for 72 h with the HCMV strain indicated. HCMV T/T11 1.1 is a version of Towne into which the U_L/b′ region from Toledo has been inserted. The proportion of target cells lysed by the NK cell line was measured by the release of radioactive chromium (⁵¹Cr). Adapted from a figure originally published in Nature Immunology [27], reusing the author’s own content

Fig. 3

Impact of RL13 and UL128L on HCMV replication. Fibroblasts were transfected with Merlin BAC constructs in which either (or both) RL13 and UL128 were mutated. a Plaques in fibroblast monolayers were readily visualized at 3 weeks post-transfection using an eGFP reporter function. b Areas of individual plaques measured at 3 weeks post-transfection. Cells were grown under an overlay to prevent cell-free spread of virus. Adapted from figures originally published in Journal of Clinical Investigation [44], reusing the author’s own content with permission

Fig. 4

Temporal classification of HCMV gene expression. QTV was used to track the expression of 139 canonical and 14 non-canonical genes through productive infection of fibroblasts by HCMV strain Merlin. Distinct profiles emerged when gene expression was separated into as few as five temporal classes (Tp1–5). Adapted from a figure originally published in Cell [58], reusing authors own content