Laboratory Strains of Murine Cytomegalovirus Are Genetically Similar to but Phenotypically Distinct from Wild Strains of Virus

Abstract

Murine cytomegalovirus (MCMV) is widely used to model human cytomegalovirus (HCMV) infection. However, it is known that serially passaged laboratory strains of HCMV differ significantly from recently isolated clinical strains of HCMV. It is therefore axiomatic that clinical models of HCMV using serially passaged strains of MCMV may not be able to fully represent the complexities of the system they are attempting to model and may not fully represent the complex biology of MCMV. To determine whether genotypic and phenotypic differences also exist between laboratory strains of MCMV and wild derived strains of MCMV, we sequenced the genomes of three low-passage strains of MCMV, plus the laboratory strain, K181. We coupled this genetic characterization to their phenotypic characteristics. In contrast to what is seen with HCMV (and rhesus CMV), there were no major genomic rearrangements in the MCMV genomes. In addition, the genome size was remarkably conserved between MCMV strains with no major insertions or deletions. There was, however, significant sequence variation between strains of MCMV, particularly at the genomic termini. These more subtle genetic differences led to considerable differences in in vivo replication with some strains of MCMV, such as WP15B, replicating preferentially in otherwise-MCMV-resistant C57BL/6 mice. CBA mice were no more resistant to MCMV than C57BL/6 mice and for some MCMV strains appeared to control infection less well than C57BL/6 mice. It is apparent that the previously described host resistance patterns of inbred mice and MCMV are not consistently applicable for all MCMV strains.

Human cytomegalovirus (HCMV) is a ubiquitous betaherpesvirus that causes life-long asymptomatic infection in the immunocompetent host. In the immunocompromised host, however, it can cause severe disease and is the leading infectious cause of congenital abnormalities in the western world. It is also one of the leading causes of posttransplant complication. It is now apparent that serially passaged laboratory strains of HCMV, such as AD169, exhibit significant biological variation compared to low-passage clinical isolates of HCMV (25). Recently, the complete genome of several different clinical strains of HCMV has been sequenced (10, 24). Levels of amino acid sequence similarity between the same gene from the different isolates range from 25 to 100%. More importantly perhaps, the gene content between clinical isolates varies (23), and it is apparent that serially passaged laboratory strains of HCMV have lost a significant number of genes (8).

The in vivo consequences of these genetic variations cannot be assessed due to the strict species specificities of cytomegaloviruses (CMVs). It is because of this specificity that in vivo studies of CMVs are typically modeled using CMVs from either rat, mouse, or guinea pig; the most commonly used is the mouse model with murine CMV (MCMV). MCMV shares many features with HCMV. The genomes of both viruses are colinear, and both viruses establish latent infection that reactivates upon immunosuppression to cause disseminated, often fatal, disease. Major cell types and organs infected by both viruses, the course of infection, and the type of pathology seen, are essentially identical, with the notable exception that MCMV does not cross the placenta (29).

The vast bulk of research on MCMV uses one of two serially passaged laboratory strain of MCMV: Smith or K181. The Smith strain of MCMV was isolated by Margaret Smith in 1954 from the salivary gland tissue of infected laboratory mice (33). The K181 strain of MCMV was later described as a more virulent Smith strain variant and was isolated by June Osborn from the salivary glands of mice after serial passage in vivo (cited in reference 22). In vitro, the K181 strain of MCMV was reported to show smaller plaques than the Smith strain of MCMV and to grow to lower titers (15). In vivo, the K181 strain of MCMV was originally reported to replicate to higher titers in the salivary glands of infected mice (22) and to demonstrate enhanced mortality in young mice (cited in reference 15). The passage histories of the Smith and K181 strains of MCMV are uncertain, and the mouse strain from which they were derived has not been recorded.

If the use of clinical isolates of HCMV is required to fully understand the complex biology of this virus, it is axiomatic that the use of serially passaged isolates of MCMV as a model of clinical HCMV infection needs to be reassessed. Between the first isolation of the Smith strain of MCMV and 2007, there have been almost 3,000 studies published on various aspects of MCMV biology. However, fewer than 1% of these (including references 4, 9, 11, 12, 14, 20, and 32) use a virus strain other than Smith or K181. In order to extend the MCMV model to other low-passage isolates of MCMV, we sequenced the viral genomes and mapped the in vivo growth capacity of three low-passage strains of MCMV in susceptible (BALB/c) and resistant (C57BL/6 and CBA) mice. These MCMV strains were isolated from wild caught mice and have defined origins and passage histories. These three isolates were compared to K181 bacterial artificial chromosome (BAC) vARK25 (27), which was also sequenced in the present study. We determined that the host resistance patterns described for “MCMV” are not constant for all strains of the virus.

Sequencing of the entire genome of three “clinical” strains of MCMV, plus the K181 laboratory strain, identified several regions with extensive sequence variation (compared to the Smith strain), which are presumably responsible for these different traits. This comparative approach has also resulted in a reannotation of the published Smith strain genome sequence. The maintenance of these fully sequenced low-passage strains of MCMV will prove invaluable in teasing out the intricacies of this animal model of an important human pathogen.

MATERIALS AND METHODS

Viruses and cells.

The BAC clone of K181, pARK25, has been described previously (27). The virus strains G4, WP15B, and C4A have also been described previously (14, 32). The BAC clone of the Smith strain of MCMV, pSM3fr, has been described previously (36) and was kindly provided by Ulrich Koszinowski (Max von Pettenkofer Institute, Munich, Germany). Viruses were inoculated onto murine embryonic fibroblasts (MEFs) from BALB/c mice from stocks as close to the original plaque-purified virus as was possible. Viruses used for sequencing were packaged in vitro a maximum of five times after plaque purification (to minimize tissue culture-derived artifacts) and are therefore described as low-passage strains of MCMV. Multistep viral growth curves were performed as previously described (18) on BALB/c MEFs. Salivary gland stocks of each virus were derived by passage in BALB/c, C57BL/6, or CBA weanling mice, as appropriate.

Viral characterization.

A total of 10⁴ PFU of salivary gland virus stocks was inoculated intraperitoneally into BALB/c, C57BL/6, or CBA mice as appropriate. Mice were sacrificed at days 3 and 18 postinoculation (n = 5 per time point), and tissues were collected for analysis. Infectious virus from organs was quantified by plaque assay on MEFs. Statistical analyses were performed by using the Kruskal-Wallis one-way analysis of variance on ranks.

DNA purification.

DNA was extracted from virus-infected cells as previously described (27) Briefly, four T80 flasks of M2-10B4 cells were infected with virus, and when the cytopathic effect reached 100% the cells were harvested by freezing and scraping. Virus was pelleted from the supernatant at 29,000 × g for 30 min at 4°C, and the pellet was resuspended in 500 μl of DNase I buffer. Extraviral DNA was digested for 1 h in 0.2 U of DNase I. Virus was treated with 500 μl of 1% sodium dodecyl sulfate and 40 μl of proteinase K (20 mg/ml) for a minimum of 4 h at 56°C. Viral DNA was purified by two phenol-chloroform extractions and precipitated in 1 volume of isopropanol. pARK25 BAC DNA was extracted from Escherichia coli strain DH10B by using the Nucleobond plasmid kit (Clontech, Palo Alto, CA).

Sequencing and sequence assembly.

Viral genomes were initially sequenced by a 454 Life Sciences sequencer (Branford, CT). Contigs were joined, and sequences were closed by PCR and dye-terminator sequencing. Anomalous sequences were confirmed by PCR and dye-terminator sequencing.

Sequence analysis.

Sequences were assembled and analyzed by using Vector NTi sequence software (Invitrogen). Amino acid similarities were calculated by using MatGat (7) Dot plot comparisons were performed by using Dottup on the Emboss server at the Centre for Comparative Genomics at Murdoch University. The moving window analysis of genomic sequences was performed by using a MLAGAN alignment (5) analyzed using the VISTA server (13) running at genome.lbl.gov.

Accession numbers.

Virus genome sequences are deposited in the EMBL database under accession numbers AM886412 and EU579859 to EU579861.

RESULTS AND DISCUSSION

The genome of the Smith strain of MCMV was first published in 1996 (26), and since this time only small sequences (full or partial genes) from other strains of MCMV have been published. These studies have shown that certain MCMV genes exist as distinct genotypes within the population of wild MCMV strains (9, 32). We have a large collection of MCMV strains that have been isolated from wild mice trapped in different locations within Australia, as well as several island locations. The virus strains described here include G4, isolated from a mouse trapped in Geraldton, Western Australia (distance from Canberra, ∼3,300 km); WP15B, isolated from a mouse trapped in Walpeup, Victoria (distance from Canberra, ∼650 km); and C4A, isolated from a mouse trapped in Canberra. We therefore consider these viruses to be geographically distinct isolates. In addition, we have recently constructed a BAC clone of the laboratory strain of MCMV, K181^Perth. This clone, designated vARK25, was shown to be identical to the parental virus by restriction fragment length polymorphism analysis and both in vivo and in vitro growth characteristics (27) and is designated K181 here.

Comparison of the four new viral genome sequences to the published Smith strain sequence identified several genes where the four new sequences were identical, but the Smith sequence contained a 1- or 2-nucleotide (nt) indel which changed an open reading frame (ORF). Subsequent analysis of the literature showed that several of these changes had been reported on an ad hoc basis (6, 16). Accordingly, we resequenced these regions from the BAC clone of the Smith virus, pSM3fr, and found that all of these regions had been incorrectly sequenced within the original publication (Table 1). The Smith genome has thus been reannotated accordingly (NC_004065), and all of our comparisons were performed against this genome. Two genes (m30 and m45.1) have been removed from the annotation, since they are now part of surrounding coding sequences. The M59 gene was found to contain multiple frameshift errors in all five viruses, and we have therefore also removed this gene from the annotation. However, we left the gene m107 within the annotation, despite this ORF only being present in the Smith virus and there being no evidence of transcription from this region (34). In the reannotated genome, we also included the 11 new genes that have been identified subsequent to the original publication and for which there are experimental data (16, 19, 34). It appears evident due to the different genotypes seen for m05, m06, m15, and m154.3 that, while K181 is closely related to the Smith strain of MCMV, it is not simply a variant of it.

TABLE 1.

Genes identified in the Smith strain of virus that have been reannotated in this analysis

Smith gene changeda	Change(s)
M20	C removed at 20958, extends ORF 3′
M26	G inserted at 31975, extends ORF 5′
m29	G inserted at 36197, extends ORF
m29.1	G inserted at 36197, truncates ORF
m31	C insertion at 37262, creates readthrough from m30 into m31
m45	C insertion at 61919, creates readthrough into m45.1
m58	A removed at 92089, G inserted at 92178, G inserted at 92179, G removed at 92358, changes the central region of the ORF
m143*	No changes, but extends to 5′ATG
m150	C inserted at 208711, results in use of internal ATG
m163*	No changes, but extends to 5′ATG

^a*, An internal start codon may still be used. Genes m30, m45.1, and m59 were deleted.

One of the reasons MCMV is widely used in animal models of HCMV infection is because the genomes of MCMV and HCMV are colinear. Analysis of the genomic sequences of laboratory and clinical strains of HCMV has demonstrated that viral genome size is variable, with a difference of 13,598 bp between the Towne (laboratory strain) and Merlin (clinical strain) strains that results in the loss of a number of ORFs in the laboratory strains (25). Similar variation has been found in the genome sizes of the rhesus CMV strains 180.92 (215,678 bp) and 68-1 (221,454 bp). Consequently, similar differences were expected between the genome size of Smith and K181 compared to the low-passage MCMV strains. The genome size of the (reannotated) Smith strain of MCMV is 230,281 bp and that of the newly sequenced K181 is 230,251 bp. In contrast to expectations, the genome sizes of the low-passage strains was highly conserved (WP15B, 230,118 bp; C4A, 230,105 bp; G4, 230,229 bp), with a maximum size variation of only 0.087%. Dot plot comparisons of the Smith strain genome against each of the newly sequenced genomes demonstrate that, in addition to the conserved genome size, genome organization is also conserved with no obvious rearrangements or large indels of the type seen in the HCMV and RhCMV genomes (Fig. 1A). Hence, the genome of the Smith strain of MCMV appears to be highly conserved over 50 years of passage. This could be due to repeated in vivo passage since isolation (interestingly, the Vancouver strain of MCMV was shown to be a variant of Smith which had undergone significant genomic rearrangement during serial in vitro passage and was significantly attenuated for in vivo growth ([3]) or to the lack of the repeat sequences that are present in the human and rhesus CMV genomes. The conservation of genome size also suggests that 230 kb is close to the maximal size capable of being packaged within the MCMV virion, possibly due to constraints imposed by the size of the viral capsid.

FIG. 1.

Genomic analysis of MCMV strains. (A) Dot plot comparisons of newly sequenced virus strains shows no major insertions, deletions, or genomic rearrangements (*) as are seen in similar strain comparisons of HCMV (AD169 versus Merlin) and RhCMV (180.92 versus 68-1). (B) Vista plot of aligned genome sequences shows that sequence variation is predominantly toward the genomic termini.

Sequence variation.

A rolling-window comparison of the genomic sequences of the low-passage strains against the Smith sequence (Fig. 1B) demonstrates that, as expected, the most variable regions of the genome are at the left (m01 to m19) and right (m144 to m170) termini. However, individual genes elsewhere in the genome may also be highly variable (for example, M55 and m124). Within the rest of the genome there is remarkable conservation of sequence such that, of the 190 coding regions within the MCMV genome, more than 78% (151 ORFs) have >98% amino acid similarity to the coding potential of the Smith strain of virus (Fig. 2). Interesting features include the fact that both C4A and WP15B encode the variant m03.5 rather than m03 (9) and the 7.2-kb intron (17) encoded between m106 and m108 is highly conserved, although the low-passage strains contain a 3-nt insertion within the suggested splice acceptor site.

FIG. 2.

Summary of coding potential of newly sequenced strains. Genes are color coded according to their amino acid similarity to the reannotated Smith strain sequence (reannotated genes highlighted in pink). Spliced genes indicated by arrows. Variation is mostly found in the putative “immune evasion” genes. Sections denoted by “▥” indicate that no ORF present.

The variability between MCMV strains is not simply limited to changes in amino acid sequence. Based on the assumption that a coding sequence would extend at least 60 amino acids from a start codon, several previously identified genes are not present in at least one of the newly sequenced strains (Table 2). One of these genes, m154.4, was shown to be transcribed in the Smith strain with early kinetics, and its product localized in the nucleus of infected cells (34); however, it is not present in any of the low-passage strains. Indeed, it seems possible that even where transcripts have been identified from certain ORFs, for example, m01, there may not actually be functional genes present.

TABLE 2.

Comparative analysis of the MCMV genome

Category	Geneb	Virus strain(s) affected
ORFs present in lab strains but not low-passage isolates	m01	All low-passage isolates
	m12	G4
	m19	G4
	m21*	All low-passage isolates
	m107*	Only present in Smith
	m145.4	All low-passage isolates
	m154.4	C4A, WP15Bc
	m156	G4, WP15B
Changes relative to Smith
Upstream start codon (extends ORF)a	m123ex4	All low-passage strainsd
	m125	C4A
Internal start codon (shortens ORF)	m02	G4, C4A
	m03	G4
	m16	All low-passage isolates
	m23.1	WP15B
Premature stop codon (truncates ORF)	M33	G4
	m69.1	All low-passage isolates
	m108	Smith has sequence duplication, which extends the ORF compared to other viruses
	m149	C4A
	m161	C4A
	m168	All low-passage isolates
	m169	WP15B
Changed stop codon (extends ORF)	m119	All low-passage isolates
	m134	C4A, K181, WP15B
	m147	C4A, G4
	m152	C4A
	m154.3	C4A

^aAn internal start codon may still be used.

^b*, No transcript was detected by Tang et al. (34).

^cThe presence of an internal stop codon means the ORF is only 25 amino acids in length.

^dThe start codon is 86 amino acids upstream of the identified splice site.

A major type of variation between strains appears to be changes to the lengths of ORFs (Table 2). Of interest is the sequence of m123 exon 4, which in the low-passage strains extends to a start codon 86 amino acids upstream of the identified splice acceptor site (21). Although changes to the putative start codon of a gene may not be valid (as an internal ATG may still be used), a large number of genes are either 3′ truncated due to an internal stop codon or 3′ extended due to indels changing the reading frame of the gene. One such gene is the M33 from G4, which contains a single nucleotide polymorphism creating a stop codon that truncates the protein by 35 amino acids. The truncated region includes five serines and one threonine residue that are suggested to be possible phosphorylation sites involved in protein signaling (30). Further research is needed to determine the level of functional variation resulting from these changes. In addition, detailed comparative genomic analysis will undoubtedly reveal new ORFs awaiting characterization.

The in vivo and in vitro replication kinetics of these viruses was assessed to determine whether genetic differences resulted in altered phenotypes. The growth of MCMV strains in vitro is a function of the ability of the virus to enter, replicate, and exit from a cell. As far as is known for MCMV, the genes that encode these functions are located within the central region of the genome and are highly conserved between strains. These viruses should therefore have similar in vitro replication kinetics. This was indeed the case for replication in MEFs (Fig. 3A).

FIG. 3.

Phenotypic characterization of MCMV strains. (A) In vitro growth curves in MEFs. There is no inherent growth defect in any of the MCMV strains in MEFs. (B) In vivo growth of MCMV strains in inbred mice with defined resistances to Smith strain MCMV. Significant differences (bracketed) can be seen in viral tires according to tissue, mouse strain, and virus strain. (C) Resistance patterns of inbred mouse strains to MCMV. Relative resistances of inbred mouse strains differ according to the virus strain used and do not simply reiterate the known patterns for the Smith strain of virus.

Given that none of these virus strains have inherent growth defects in MEFs, it must be assumed that any differences in their growth in vivo is due to either differences in tissue tropism or immune evasion. To assess this issue, three strains of mice with known resistance to MCMV—BALB/c (susceptible), C57BL/6 (moderately resistant), and CBA (resistant)—were infected with each viral strain (1). Viral titers were assessed in the spleens and livers at day 3 and in the salivary glands at day 18. Using multiple strains of MCMV in inbred mouse strains, the previous assertions about host resistance to MCMV are less obvious, since there was considerable variation in the ability of MCMV strains to replicate (Fig. 3B). Generally, there was some support for the previously identified mouse resistance to infection, with BALB/c mice typically having higher MCMV titers in the spleens, livers, and salivary glands (of infected mice) than did CBA or C57BL/6 mice. However, CBA mice appeared to be no more resistant to MCMV than C57BL/6 mice and in fact appeared to be slightly more sensitive to MCMV infection as reflected by the virus titers in the liver.

When these resistance patterns were assessed for individual virus strains, the pattern of mouse resistance is further complicated. For example, WP15B replicates to higher titers in the spleens and salivary glands of C57BL/6 mice than the titers found in BALB/c and CBA mice. Typically, Smith and K181 replicate preferentially in the spleens compared to the livers of infected mice, and it was control of viral replication in the spleens of C57BL/6 mice that led to the discovery of the Cmv1 locus and the interaction of host Ly49H and virus-encoded m157 (28). This pattern of replication was confirmed in all three strains of mice for K181 and for the low-passage isolates of MCMV in BALB/c mice. However, the low-passage isolates of MCMV consistently replicated to higher titers in the livers compared to the spleens of infected CBA mice. These data suggest a different host control mechanism in the spleens or livers of infected CBA mice for the wild viral strains than for the laboratory strains of MCMV. Hence, resistance to MCMV is complicated and is dependent on the mouse strain, tissue type, and MCMV strain (Fig. 3C). It would therefore perhaps be more appropriate to designate resistance relative to virus strain rather than to mouse strain.

When a particular resistance mechanism is known, then the genome of the virus can be used to predict in vivo replication in specific tissues in specific mouse strains. The m157 gene product of the Smith and K181 strains can directly ligate the NK cell-activating receptor, Ly49H (2). Consequently, replication of K181 and Smith strain MCMV is well controlled in the spleens of C57BL/6 mice or other Ly49H⁺ mice. The m157 gene from C4A is of the same genotype as that of K181 (92.8% amino acid similarity), and we hypothesize should ligate Ly49H. This virus should therefore replicate poorly in the spleens compared to the livers of C57BL/6 mice, and this is indeed the case. In contrast, WP15B and G4 have other genotypes (76.9 and 77.8% amino acid similarity, respectively). The m157 gene product from G4 has been shown not to interact with Ly49H from C57BL/6 (35), and presumably neither does that of WP15B. Both WP15B and G4 replicate better in the spleens than in the livers of C57BL/6 mice as predicted. Hence, the genetics of a viral strain can have a profound effect on host resistance to MCMV and, in the case of m157 (in Ly49H⁺ mice), can be used to predict in vivo replication. Subsequently, direct ligation of Ly49H by C4A-infected MEFs was demonstrated with BWZ-HD12 cells (31), kindly provided by A. A. Scalzo (data not shown).

Given that the previously described host resistance patterns to MCMV infection are not reproduced for all strains of MCMV, it may be possible to apply a comparative approach by which previously unidentified host resistance mechanisms may be determined. This may also allow for the comparison of not only mouse strain resistance to virus but also virus resistance to mouse strain control of infection. This could allow for a new “comparative resistomics” approach to MCMV biology.

In conclusion, we have demonstrated significant sequence variation between strains of MCMV and differences in the in vivo replication of these strains of virus. We have purposefully sequenced only low-passage “wild” strains of MCMV and will maintain these as reference strains. In addition, the cloning and sequencing of the K181^Perth BAC will facilitate the use of this strain of virus as a reference strain. We are currently cloning the genome of each of the sequenced viruses as BAC clones to allow the maintenance of a clonal reference genome for each virus.

Within the MCMV genome we do not find major changes due to deletions or insertions of genes, as is seen in HCMV in RhCMV. Indeed, this appears unlikely to occur in MCMV strains, given the remarkable conservation of genome size that we have observed. Differences in virus replication are likely to be due to variation in gene sequences rather than wholesale changes to gene content. These more subtle changes result in considerable changes to in vivo replication capacity. Consequently, it is not possible extrapolate results from one MCMV strain to “MCMV biology” as a whole.