Infection of a Single Cell Line with Distinct Strains of Human Cytomegalovirus Can Result in Large Variations in Virion Production and Facilitate Efficient Screening of Virus Protein Function

ABSTRACT

Previously, we reported that the absence of the ataxia telangiectasia mutated (ATM) kinase, a critical DNA damage response (DDR) signaling component for double-strand breaks, caused no change in HCMV Towne virion production. Later, others reported decreased AD169 viral titers in the absence of ATM. To address this discrepancy, human foreskin fibroblasts (HFF) and three ATM⁻ lines (GM02530, GM05823, and GM03395) were infected with both Towne and AD169. Two additional ATM⁻ lines (GM02052 and GM03487) were infected with Towne. Remarkably, both previous studies' results were confirmed. However, the increased number of cell lines and infections with both lab-adapted strains confirmed that ATM was not necessary to produce wild-type-level titers in fibroblasts. Instead, interactions between individual virus strains and the cellular microenvironment of the individual ATM⁻ line determined efficiency of virion production. Surprisingly, these two commonly used lab-adapted strains produced drastically different titers in one ATM⁻ cell line, GM05823. The differences in titer suggested a rapid method for identifying genes involved in differential virion production. In silico comparison of the Towne and AD169 genomes determined a list of 28 probable candidates responsible for the difference. Using serial iterations of an experiment involving virion entry and input genome nuclear trafficking with a panel of related strains, we reduced this list to four (UL129, UL145, UL147, and UL148). As a proof of principle, reintroduction of UL148 largely rescued genome trafficking. Therefore, use of a battery of related strains offers an efficient method to narrow lists of candidate genes affecting various virus life cycle checkpoints.

IMPORTANCE Human cytomegalovirus (HCMV) infection of multiple cell lines lacking ataxia telangiectasia mutated (ATM) protein produced wild-type levels of infectious virus. Interactions between virus strains and the microenvironment of individual ATM⁻ lines determined the efficiency of virion production. Infection of one ATM⁻ cell line, GM05823, produced large titer differentials dependent on the strain used, Towne or AD169. This discrepancy resolved a disagreement in the literature of a requirement for ATM expression and HCMV reproduction. The titer differentials in GM08523 cells were due, in part, to a decreased capacity of AD169 virions to enter the cell and traffic genomes to the nucleus. In silico comparison of the Towne, AD169, and related variant strains' genomes was coupled with serial iterations of a virus entry experiment, narrowing 28 candidate proteins responsible for the phenotype down to 4. Reintroduction of UL148 significantly rescued genome trafficking. Differential behavior of virus strains can be exploited to elucidate gene function.

INTRODUCTION

The human cytomegalovirus (HCMV) life cycle involves a complex interplay between the virus and the host, with the virus exploiting the host cellular machinery for many of its own functions and, ultimately, releasing fully infectious virions. During a permissive HCMV infection, after virions have entered the cell, the tegument proteins and virus genome are independently trafficked to the nucleus. In fibroblasts, large bipolar viral replication centers (RCs) are formed within 48 h postinfection (hpi) and certain host cellular proteins become strongly associated with these RCs (1; reviewed in reference 2). These proteins include the regulatory protein p53 (3), as well as numerous components of the host cellular DNA damage response (DDR) and repair pathways (4–8).

Many virus infections affect the DDR. The interactions span a range of up- and downregulations and include a complex dynamic between the virus and its host's damage response (as reviewed in references 6 and 9). Some viruses appear to require DDR proteins for efficient replication (10, 11), while for other viruses an efficient DDR can be detrimental to their DNA replication (12–21). Studies from several labs, including our own, have shown that HCMV infection initiates the ataxia telangiectasia mutated (ATM)-dependent double-strand break (DSB) DDR (4–8). ATM is a key sensing protein involved in initiating DSB repair, as well as cellular growth and differentiation (22). Numerous ATM-deficient (ATM⁻) cell lines have been derived from ataxia telangiectasia (A-T) patients, and most harbor unique mutations (23, 24). HCMV infection induces ATM to phosphorylate Nbs1 and p53 (4, 5, 7, 8, 25); however, the damage-signaling cascade is defective, and damage-specific foci do not form at sites of viral deposition at early times postinfection (5).

Conflicting results regarding ATM's role in HCMV virion production have been reported. Studies from our lab performed in Towne-infected normal human foreskin fibroblasts (HFFs), an ATM⁻ cell line (GM02530) and Mre11⁻ cells found that disruption of the DSB DDR did not diminish functional virion production at either a high or a low multiplicity of infection (MOI) (5). Conversely, a study in a different ATM-deficient cell line (GM05823) infected with the HCMV strain AD169 found functional virion production was reduced by >2 logs compared to normal dermal fibroblasts (4). The present study was initiated to reconcile these results and resolve whether the presence of ATM was necessary for wild-type (wt) levels of functional virion production by HCMV. To address this discrepancy, HFFs and three ATM⁻ cell lines (GM03395, GM02530, and GM05823) were infected with both Towne and AD169. Two additional ATM⁻ cell lines (GM02052 and GM03487) and an age-matched control fibroblast line (GM07532) were also infected with Towne. As expected, both Towne and AD169 produced wt titers in HFFs. Infection of GM03395 cells with either virus strain produced negligible virus progeny. Conversely, infection of GM02530 cells with either virus strain produced wt titers, replicating our earlier finding. In addition, Towne infection of GM02052 and GM03487 also produced wt titers at late times postinfection. Perplexingly, Towne replicated at wt levels in GM05823 cells, while AD169 produced negligible virus, replicating E et al.'s earlier findings (4). Infection progressed differently for these two “sister” strains of HCMV in a single cellular environment. Despite this discrepancy, the wt-level secretion of Towne virus from four separate ATM⁻ backgrounds (GM02530, GM02052, GM03487, and GM05823) definitively established that the presence of ATM was not required for functional virion production of HCMV in fibroblasts.

Intrigued by the dichotomy of virion production in what are often considered interchangeable lab-adapted HCMV strains, we characterized several life cycle parameters of both viruses within the GM05823 cells. Initial binding, entry, and trafficking of the viral genome are essential for successful propagation of virus. The ability of the AD169 strain to enter the GM05823 cells and traffic successfully to the nucleus was significantly reduced compared to the Towne strain. Decreased trafficking efficiency of AD169 extended to an inability of this virus to robustly express the immediate early protein, IE1, a key transcriptional activator for the virus. Further inhibitions of wt-like virus behavior were noted in the capacity to establish RCs and virion assembly complexes (VACs) at late times postinfection in this AD169-GM05823 combination. Towne infections of the same cell line produced substantially wt-like behavior at all stages.

We reasoned the differential behavior of these two closely related virus strains within a single cell type could be exploited to assess viral gene product function. Proteins absent in AD169 but present in Towne, or possibly the reverse, would likely be responsible for the differential behaviors between the strain and the host and could also, to some extent, indicate the function(s) of these viral proteins. Towne virus' greater production of functional virus appeared to commence with its increased capacity to enter GM05823 cells and traffic its genome to the nucleus; therefore, we focused our analysis on this parameter of infection. In silico comparison of the genomes of Towne and AD169 determined a list of 28 probable gene candidates responsible for the difference. Using serial iterations of an experiment involving virion entry and input genome nuclear trafficking with a panel of related HCMV strains, we reduced this list to four (UL129, UL145, UL147, and UL148). As a proof of concept, constructs expressing these proteins were nucleofected into the GM05823 cells, and one, UL148, was found to largely rescue the entry/translocation defect. We believe that this is a powerful and efficient tool to elucidate the interactions of viral proteins with the host cell.

MATERIALS AND METHODS

Cell culture.

HFFs were grown in minimal essential medium (MEM) supplemented with 10% heat-inactivated fetal bovine serum (FBS), l-glutamine (2 μM), penicillin (200 IU/ml), streptomycin (200 μg/ml), and amphotericin B (1.5 μg/ml). Five ATM⁻ fibroblast cell lines were used. GM02530, GM02052, GM03487, and GM07532 (matched wt control) cells (Coriell) were cultured in MEM supplemented with 10% heat-inactivated FBS and l-glutamine (2 μM). Telomerase(tert)-immortalized GM05823 and GM03395 (kindly provided by Lisa McDaniel, UT Southwestern Medical Center, Dallas, TX) were grown in Dulbecco modified Eagle medium supplemented with 15% heat-inactivated FBS, l-glutamine (2 μM), penicillin (200 IU/ml), and streptomycin (200 μg/ml). It should be noted that tert-immortalized and nonimmortalized GM05823 and GM03395 cells showed identical repair phenotypes (26, 27). Tert immortalization also had no effect on the permissiveness of fibroblasts for HCMV infection in previous studies (28, 29). All cells were cultured in humidified incubators at 37°C and 5% CO₂.

Background of virus strains used.

The lab-adapted Towne and AD169 HCMV strains were developed as vaccine candidates and have lost the ability to infect endothelial cells. AD169 was originally derived from adenoid tissue and cultured in fibroblasts (30, 31). Towne was isolated from the urine of an infected patient and cultured in WI-38 fibroblasts (32). Both strains were passaged extensively, resulting in genomic alterations from clinical strains, such as strain TR (33, 34). AD169 (ATCC VR-538) contains a total of 32 mutations, including frameshift mutations in genes RL5A, RL13, UL36, and UL131A, deletion of UL133 to UL150, and duplication of part of R_L (33, 35–39). Towne (ATCC VR977) is an approximately 50/50 mixture of Towne(long) and Towne(short). Towne(long) has deletions of UL143, UL146, UL147A, UL148A, and UL148C, as well as duplication of R_L, and mutations in RL13, UL1, UL40, UL130, US1, and US9 (35, 40–43). Towne(short) has the same mutations outside the UL/b′ region but is also missing most of the clinical cassette, including UL131 and UL133 through UL144 (40–44). Additional strain variants exist for both Towne and AD169, each with slightly different mutational spectra. Comparison of AD169 variant sequences found that each carries distinct mutation profiles (40, 42). The ADVarUK and ATCC VR-538 strains of AD169 are missing the UL/b′ region, but ADVarUC retains sequences from UL133 to UL139 and from UL145 on.

HCMV Towne (VR977) and AD169 (VR538) stocks were obtained from the American Type Culture Collection (ATCC). The TR clinical strain was generously provided by Jay Nelson (Oregon and Health Science University, Portland, OR [33]); Towne(short) was kindly provided by Michael McVoy (Virginia Commonwealth University, Richmond VA), and ADVarUC was kindly provided by Chris Benedict (La Jolla Allergy and Immunology, La Jolla, CA).

Virus infection conditions.

Unless otherwise stated, cells were serum starved to synchronize in G₀ before trypsinization, counting, and reseeding (28). Cells were allowed to settle for 2 h (h) before infection at an MOI of 5 (high MOI) or 0.25 (low MOI). At 2 hpi, the virus inocula were removed and replaced with fresh medium. Cells were harvested at the indicated times postinfection. Experiments were performed at least twice, with averages of the experiments shown in the figures. Error bars represent one standard deviation from the average.

For bromodeoxyuridine (BrdU) studies, viruses were labeled as previously described (45). Cells were serum starved, seeded, and infected as described above. After 1 h, the virus inocula were removed, and the cells were washed three times in phosphate-buffered saline (PBS) before adding fresh media. Cells were harvested at 4 hpi and processed as previously described for localization of BrdU-labeled genomes (45).

Virus titers.

Supernatants from cells infected as described above were harvested at the times indicated, mixed with 1% dimethyl sulfoxide, and stored at −80°C. Virus titers were determined on HFF monolayers as previously described (5).

Immunofluorescent localization of proteins.

Serum-starved cells were seeded onto glass coverslips and infected at an MOI of 5, as described above. Coverslips were harvested at the indicated times. Coverslips used in IE1 and BrdU studies were simultaneously fixed and permeabilized with 100% methanol at −20°C for 10 min. Coverslips used to detect UL57, pp65, gB, and Flag-tagged UL148 were fixed in 3% formaldehyde, followed by permeabilization with Triton X-100 as previously described (46). BrdU coverslips were further treated with 4 N HCl for 10 min at room temperature to denature the DNA, followed by extensive washes in PBS to neutralize the samples prior to detection and staining (45). Antibody incubation and mounting of coverslips were performed as previously described (46). All nuclei were counterstained with Hoechst dye. In all experiments, at least 100 cells were counted for each sample at each time point. Cells were first counted for nuclear Hoechst staining and then assessed for the presence or absence of labeled genomes or viral antigens. All fluorescence images were obtained using a Nikon E800 Eclipse microscope equipped with a Nikon Ri1 digital camera, with images captured on NIS Elements software.

Antibodies used.

The primary antibodies used were as follows: mouse monoclonal antibodies against IE1 (kindly provided by Bill Britt, University of Alabama, Birmingham), UL57 (clone CH167), pp65 (clone CH12) and gB (clone CH28; both from Virusys Corp.), and Flag tag (catalog no. 8146; Cell Signaling Technologies), as well as rat monoclonal antibody against BrdU (clone BU1/75 [ICR1]; Harlan Sera-Lab). The secondary antibodies used were as follows: goat anti-mouse IgG1-Alexa Fluor 488 (Molecular Probes) and -TRITC (Southern Biotechnology Associates, Inc.), goat anti-mouse IgG2A-TRITC (Southern Biotechnology Associates, Inc.), and donkey anti-rat IgG-TRITC (Jackson ImmunoResearch).

Nucleofection.

FLAG-tagged UL129, UL145, and UL148 constructs or the control pMZS3F plasmid (47) (a kind gift from Lori Frappier, University of Toronto, Toronto, Ontario, Canada) were nucleofected into GM05823 cells using an Amaxa primary fibroblast cell line nucleofector kit (Lonza) according to the manufacturer's instructions. Briefly, 5 μg of DNA was mixed with cells and nucleofected using an Amaxa Nucleofector II (program A-033). The cells were allowed to recover for 72 h prior to infection with BrdU-labeled Towne or AD169 strains, as described above. The cells were fixed at 4 hpi in formaldehyde, permeabilized with Triton X-100, and then treated as previously described to reveal BrdU residues (45). The cells were stained simultaneously with antibodies to detect incorporated BrdU and Flag-tagged viral proteins. Only UL148 nucleofections were successful enough to reliably score positive cells. The very low numbers of positive cells following nucleofection of UL129 and UL145 suggested that these proteins were toxic to the cells. Cells positive for UL148 were scored for the presence of BrdU-labeled genomes within their nuclei and compared to UL148⁻ cells.

Construction of genetic composition of different strains for in silico analysis.

The literature was searched for the genetic composition of the lab-adapted HCMV strains Towne (short and long), AD169, ADVarUC, and the clinical isolate, triple resistant (TR). Their genetic maps were constructed using the earliest references as a foundation. These maps were then modified with more recent information, the most recent information being included as the consensus. In instances of disagreement between references, the most recent reference was considered the consensus. The references for each strain are listed above in “Background of virus strains used” and also with the strain in Fig. 4. Some references were applicable to more than one strain and are listed with each.

FIG 4.

Genetic differences in a panel of HCMV virus strains. In silico analysis of four lab-adapted [Towne, Towne(short), AD169, and ADVarUC] and one clinical (TR) HCMV strains led to generation of this comparison figure. A schematic of the HCMV genome is at the top of the figure. Below is an alignment of the five genomes, showing genes with a wt sequence (green) or a mutated sequence (yellow) or genes that are deleted (red). For some regions, there was no clear consensus in the literature, and they are labeled in gray. Arrows point to the four potential candidate genes narrowed by sequential experimentation. References used to generate the consensus for each strain are listed to the right of each strain.

Statistical analysis.

Statistical significance was determined by two-tailed, unpaired t tests (using GraphPad software). A given comparison was considered statistically significant if P = 0.05 or lower. Significance comparisons of virus life cycle parameters were performed for AD169-infected GM05823 cells compared to Towne- and AD169-infected HFFs and to Towne-infected GM05823 cells. These comparisons were deemed most relevant, since each of these latter three virus strain-cell combinations produced wt levels of virions, while AD169-infected GM05823 cells did not. The significance comparisons determined which parameters were statistically affected in the AD169-infected GM05823 cells.

RESULTS

Peak virus titers were reduced in some, but not all, infected ATM⁻ cell lines and were dependent on the virus strain.

The ATM⁻ cell lines used in this study carry different mutations (see Table 1 for details). GM02530 is a compound heterozygote with three distinct mutations (24), which results in an ATM protein undetectable by Western blotting (48, 49), leading to an ineffective DNA damage response due to incomplete activation of the transcriptional coactivator Strap (50). In the GM05823 cell line, ATM is disrupted by a six-nucleotide deletion, which removes two amino acids and produces a reduced amount of full-length, nonfunctional protein compared to the wt (24, 51). These cells have a defective G₁ checkpoint and continue to replicate DNA despite exposure to ionizing radiation (IR) (26). ATM in GM03395 harbors a G→A point mutation which results in a stop codon at residue 2638 and no ATM protein detectable by Western blotting (52). These cells have severely attenuated G₁ and G₂ checkpoints and are hypersensitive to IR-induced inhibition of colony formation (27). A C→T mutation at position 103 in the GM02052 cells results in an N-terminal truncation and no expressed protein (53). Like the GM05823 cells, GM02052s have a defective G₁ checkpoint and continue to replicate DNA despite exposure to IR (26). GM03487 cells carry a compound heterozygous mutation that results in low-level expression of a truncated protein (51, 54). These cells show hypersensitivity to X-irradiation (55).

TABLE 1.

ATM cell line mutations^a

Cell line	Cell type	Functional damage response	Protein production	ATM mutation	ATM mutation
HFF	Fibroblast	+	Yes	NA	NA
GM02530	Fibroblast	– (50)	No (48, 49)	S689fsX750	2251del19
				E1892fsX	5675del88
				2191del27 (24)	6573del81
GM05823	Fibroblast	– (26)	Yes (24, 51)	2427del2 (24)	7278del6
GM03395	Fibroblast	– (27)	No (52)	W2638X (52)	7913G>A
GM02052	Fibroblast	– (26)	No (53)	R35TER (53)	103C>T
GM03487	Fibroblast	– (55)	Yes (51, 54)	K2756X	8266A>T
				S381fsX (54)	1141ins4

^aThe protein (ATM) and DNA (ATM) mutations present in the ATM⁻ cell lines used are indicated. The functionality of the cells for DNA repair is indicated when established. References used to generate the table are indicated in parentheses. NA, not applicable.

Three ATM⁻ cell lines (GM02530, GM05823, and GM03395) and HFFs were infected at an MOI of 5 with the Towne or AD169 strains of HCMV. Functional virion production was determined during peak times of virus shedding (72, 96, and 120 hpi) via plaque assay on a monolayer of HFFs. Earlier time points were not assayed in these high-MOI infections, since it has been our experience that very little functional virus is produced from fibroblasts before 48 hpi (28, 56). Towne-infected HFFs were used as our reference for all assays and serve as the wt levels reported. We noted that in almost all assays, AD169-infected HFFs exhibited minor delays in growth kinetics, small differences in protein expression, and slightly lower levels of virion production compared to the same wt cells infected with Towne.

The first focus in our assessment was whether a given virus strain-cell combination was capable of secreting the same level of productively infectious virions by 120 hpi as the Towne strain in HFFs. As expected from our earlier results (5), high-MOI experiments produced very similar titers in HFFs and GM02530 cells infected with either Towne or AD169 by 120 hpi, although AD169 infections had slower growth kinetics in both cell types (Fig. 1A, compare the solid and dashed red and green lines). In addition, two other ATM⁻ cell lines (GM02052 and GM03487) and an additional age-matched control cell line (GM07532) were infected at an MOI of 5 and harvested at 96 hpi. These three cell lines all produced wt-level titers, similar to the GM02530 cells (Fig. 1A, additional colored triangles). In contrast, peak titers at 120 hpi in GM03395 cells were reduced by 2 logs (for Towne) and 3.5 logs (for AD169) compared to Towne-infected HFFs, indicating a severe HCMV growth defect in these cells. High-MOI infections of the GM05823 cells yielded the most intriguing results. By 120 hpi, the production of functional Towne virions from infected GM05823 cells was quite comparable to that of either Towne- or AD169-infected HFFs, albeit arriving at this point with slower growth kinetics. However, surprising to us, but confirming earlier reports, titers in GM05823 cells infected with AD169 (the combination used by E et al. [4]) were reduced by 2 logs at the 120-hpi time point (Fig. 1A, compare the solid and dashed blue lines).

FIG 1.

Virus titers. HFF, GM02530, GM05823, and GM03395 cells were infected with Towne or AD169 at an MOI of 5 (A) or an MOI of 0.25 (B), and supernatants were collected at the times indicated. Serial dilutions of collected supernatants were used to infect an HFF monolayer to determine the PFU/ml. Towne-infected cells are denoted by solid lines, and AD169 infected cells are represented by dashed lines. The titers at 96 hpi for three additional cell lines (ATM⁻ GM02052 and GM03487 and the control GM07532) are shown as alternately colored triangles in panel A. In all figures, experiments were performed at least twice, with averages of the experiments shown. Error bars represent one standard deviation from the average.

HFF, GM02530, and GM05823 cells were also infected at a low MOI of 0.25 and monitored for 11 days (Fig. 1B). Because both Towne and AD169 virus infection in GM03395 cells had produced titers drastically reduced compared to the wt, these cells were not included in the low-MOI experiments, since they would likely have produced virtually undetectable numbers of infectious progeny. Low-MOI infections more clearly delineate differences in the efficiency of virion shedding and cell-to-cell spread of infection. In addition, low-MOI infection can reveal differences in cell line-virus combinations masked by high-multiplicity infection. As with the high-MOI titers, virion production in HFFs infected with AD169 lagged behind Towne infections early postinfection but reached approximately the same levels by 7 days postinfection and remained similar for the duration of the assay (Fig. 1B, compare the solid and dashed red lines). Towne-infected GM02530 titers paralleled HFF infections for the entire time course (Fig. 1B, compare the solid red and green lines). Virus titers in GM02530 AD169 infections lagged for the first several days, but day 7 peak titers differed by less than a log from Towne- or AD169-infected HFFs and Towne-infected GM02530s (Fig. 1B, compare the solid red or green and dashed red lines to the dashed green line). At early time points, Towne-infected GM05823 cells lagged more substantially behind in production of functional virions when infected at low MOIs compared to Towne-infected HFFs. However, no significant differences were observed by 9 d pi between these two cell types (Fig. 1B, compare the solid red and blue lines), indicating that the virus could replicate efficiently within these cells given enough time. As had been observed with high-MOI infection, titers for GM05823 cells infected with AD169 were dramatically decreased across the entire time course and remained reduced by almost 3.5 logs at 11 days postinfection. Low-multiplicity infections confirmed the differences in titers seen between cell and virus combinations at high MOIs.

Towne and AD169 infections produced virions efficiently in HFF cells as expected. If efficient, wt-level virion production was dependent upon the presence of a functional ATM protein, all ATM-deficient cell lines would have produced substantially fewer functional virions, particularly at times of peak virion production. In the ATM⁻ cell line GM03395, both strains produced significantly fewer functional virions, even when infected at a high MOI. However, a high-MOI Towne infection of four separate ATM⁻ cell lines (GM02530, GM02052, GM03487, and GM05823), with entirely different ATM mutations, produced wt levels of functional virions. In addition, both virus strains shed virus with equal efficiency in the GM02530 cell line during both high- and low-MOI infections. In high-MOI infections Towne and AD169 virion production levels in wt and GM02530 strains were virtually indistinguishable (compare the solid red and green lines and the dashed red and green lines in Fig. 1A). Clearly, the wt-like production of functional HCMV virions did not hinge on the presence or absence of ATM in a given cell line. Rather, given these results, it could be concluded that the unique cellular environment of these ATM⁻ cell lines determined HCMV virion production. Why, then, in parallel infections in the GM05823 cell line, did Towne produce wt levels of functional virions, whereas AD169 virion production fell by 2 to 3.5 logs (compare the solid to the dashed blue lines in Fig. 1A)?

The Towne and AD169 strains are used essentially interchangeably in labs and ostensibly should behave the same in any fibroblasts. We fully expected that they would behave very similarly in all three ATM⁻ cell lines (GM02530, GM05823, and GM03395). Dramatically altered functional virion production between virus strains in a single cell line, GM05823, prompted further investigation into these disparities by examining various checkpoints of the virus life cycle in these cells.

Trafficking of the AD169 genome to the nucleus was reduced in GM05823 cells and affected the subsequent expression of viral proteins.

Virion entry into the cell and trafficking to the nucleus can be visualized using a BrdU-labeled virus (45). Cells were infected with BrdU-labeled virions, collected at 4 hpi, and scored for the presence of BrdU-labeled genome foci within the cell (representative images are shown in Fig. 2A). Cells were scored for the presence of genomes in both the nucleus and the cytoplasm (Fig. 2A, second and third rows), in the cytoplasm only (Fig. 2A, bottom row), and whether there were no detectable foci within the cell (as seen in mock-infected cells; Fig. 2A, top row). Approximately 90% of all Towne-infected GM05823 cells contained BrdU foci within their nuclei, indicating wt levels of virion trafficking compared to Towne-infected HFFs (Fig. 2B, red bars, P = 0.21). In contrast, only ∼65% of AD169-infected GM05823 cells trafficked genomes to the nucleus, which was significantly less than observed in Towne-infected (P = 0.035) and AD169-infected (P = 0.036) HFF cells (denoted by asterisks in this and all other graphs) and just slightly less than statistically significant in comparison to Towne-infected GM05823 cells (P = 0.065). Perhaps equally important, a relatively large fraction of AD169-infected GM05823 cells (∼15%) had no visible labeled virion foci within them (blue bars), which was a statistically higher percentage than that for any of the other virus/cell combinations (P = 0.01, 0.027, and 0.013 compared to Towne-infected HFFs, AD169-infected HFFs, and Towne-infected GM05823 cells, respectively). Reduced cellular entry and decreased nuclear trafficking of input AD169 virus genomes in GM05823 cells suggested that the AD169 genome lacked a protein coding sequence important for these processes in the context of this particular cellular environment. To assess whether the genomes that were successfully deposited in the nucleus were functional, we examined immediate-early 1 (IE1) protein expression.

FIG 2.

Trafficking of BrdU-labeled virions to the nucleus and subsequent IE1 expression. (A and B) HFF and GM05823 cells were infected with BrdU-labeled Towne or AD169 at an MOI of 5. Cells were harvested at 4 hpi and, after acid treatment, the cells were stained for BrdU (red) and counterstained with Hoechst (blue) as shown in panel A. The top row shows mock-infected cells. The second and third rows show cells with at least some BrdU-labeled virions successfully trafficked to the nucleus. Additional virions are present in the cytoplasm. The bottom row shows labeled virions that have trafficked into the cytoplasm but not into the nucleus in AD169-infected GM05823 cells. Scale bar, 5 μm. Quantitation of trafficking of BrdU-labeled virions into the nuclei of infected cells is shown in panel B. (C) HFF and GM05823 cells were infected with Towne or AD169 at an MOI of 5. The cells were harvested at the indicated times and stained for IE1. In this and all subsequent figures, asterisks above a bar represent a statistically significant difference from the AD169-infected GM05823 cells (*, P = 0.05; **, P = 0.005; ***, P < 0.0005).

Although there was fairly large variability in how rapidly IE1 was expressed within the first 6 hpi in all virus-cell combinations (data not shown), by 24 hpi the very large majority (>84%) of both Towne-infected populations and of AD169-infected HFFs expressed this protein (Fig. 2C, green bars). Unsurprisingly, and paralleling the entry data, IE1 was expressed in only ∼62% of AD169-infected GM05823 cells by 24 hpi, which was a statistically lower percentage (P < 0.0001, P = 0.001, and P = 0.006 compared to Towne- and AD169-infected HFFs and Towne-infected GM05823 cells, respectively). This indicated that only AD169-infected GM05823 cells that had efficiently trafficked viral genomes by 4 hpi went on to express IE1. These experiments demonstrated that in the early stages of infection, the Towne and AD169 strains could both efficiently express IE1 in the GM05823 cells if viral genomes were trafficked to the nucleus.

E et al. (4) had indicated that AD169 infection of GM05823 cells led to poor early and late viral gene expression; therefore, these stages were also examined to assess whether the lower level of IE gene expression carried forward. The efficient formation of RCs was investigated by UL57 staining. UL57 is a virus-encoded single-stranded DNA-binding protein involved in genome replication. Earlier studies have shown that UL57 localizes to RCs within an infected cell's nucleus starting as multiple small foci, progressing through intermediate-sized bipolar foci, and finally forming a single large focus (as shown in Fig. 3A) (1). Cells were scored for the presence and type of UL57 foci, indicating the formation and progression of RCs. Figure 3B represents the percentage of cells that expressed UL57 foci of any size. The distribution of these foci into size categories is shown in Table 2. HFF cells infected with either Towne or AD169 and GM05823 cells infected with Towne rapidly established RCs, as evidenced by UL57 foci present in the large majority (>72%) of cells by 48 hpi (Fig. 3B, blue, green, and red bars, respectively). By 72 hpi, >80% of all cells in these three virus-cell combinations possessed advanced stage foci (bipolar or larger), indicating a rapid progression of viral replication (see Table 2). Given the lower percentage of efficiently trafficked genomes and IE1 gene expression, it was unsurprising that RCs were assembled less efficiently in AD169-infected GM05823 cells (Fig. 3B, yellow bars). Throughout the 96-h time frame examined, no more than an average of 25% of cells contained RCs of any size. The difference between focus-containing AD169-infected GM05823 cells and the three other virus-cell type combinations was statistically significant at both 72 and 96 hpi. Interestingly, in all cells that did possess RCs, the majority (90%) did progress to advanced stages by 96 hpi (see Table 2).

FIG 3.

Replication center formation, trafficking of pp65, and expression of late protein gB. HFF and GM05823 cells were infected with Towne or AD169 at an MOI of 5. The cells were harvested at the indicated times and stained for UL57 (A), pp65 (C), and gB (E). The cells shown are Towne-infected (UL57 and gB) or AD169-infected (pp65) HFF cells at 72 hpi. Quantitation of the percentage of cells that expressed UL57 as discrete replication foci is shown in panel B. Quantitation of the percentage of cells that showed either nuclear or cytoplasmic staining of pp65 is shown in panel D. Quantitation of the percentage of cells that express gB within a cytoplasmic VAC is shown in panel F.

TABLE 2.

Distribution of UL57-positive cells by foci size^a

Cell line	Virus strain	Time point (hpi)	Percent cells with UL57 focus staining pattern (mean ± SD)
			Single large	Bipolar	Multiple small	None
HFF	Towne	48	48.2 ± 11*	51.8 ± 11	0	0*
		72	75.5 ± 6.5*	24.5 ± 6.5	0	0*
		96	95 ± 0.8*	4.7 ± 0.5	0.3 ± 0.3	0*
	AD169	48	18.4 ± 0.5**	54.2 ± 10.9	3 ± 0.6	24.4 ± 10.8
		72	54.6 ± 5*	26.3 ± 2.7	3.3 ± 2.2	15.85 ± 5.5*
		96	70.1 ± 0.1*	10.3 ± 2.6	9.2 ± 2.8	10.4 ± 0.3*
GM05823	Towne	48	12.0 ± 0.7**	56.7 ± 3.1*	15.0 ± 7.1	16.3 ± 9.5
		72	51.0 ± 0.7*	35.8 ± 3.5	6.6 ± 2.6	6.6 ± 5.5*
		96	75.7 ± 5.8*	13.2 ± 8.6	4 ± 1.7	7.1 ± 4.5*
	AD169	48	0.6 ± 0	16.1 ± 6.6	9.0 ± 6.7	74.3 ± 13.3
		72	6.4 ± 3.4	14.0 ± 4.1	5.2 ± 2.2	74.4 ± 9.65
		96	11.7 ± 8.3	4.6 ± 3.7	1.7 ± 0.7	82 ± 12.6

^aHCMV-infected cells were screened by IF for the presence or absence of UL57 foci and categorized into three different groups according to replication center size. Numbers represent means with one standard deviation. The total numbers of focus⁺ cells are graphically represented in Fig. 3B. Statistical significance is indicated by asterisks as defined in the figure legends.

To examine whether virions trafficked efficiently out of the nucleus, we determined localization of the tegument protein pp65. During capsid maturation de novo synthesized pp65 initially localizes within the nucleus and can be observed moving out of the nucleus and into the VAC if virions are trafficking properly, thus serving as a proxy for proper maturation (28, 57, 58) Figure 3C shows examples of AD169-infected HFFs displaying either nuclear or cytoplasmic VAC localization at 72 hpi. Figure 3D shows the enumeration of nuclear and cytoplasmic VAC localization from 48 to 96 hpi for the four different virus-cell combinations. As can be seen, there was large variation in the percentage of cells displaying maturation of pp65 into the VAC at 48 hpi, with the infected HFFs showing a more rapid maturation process (see the filled bars in Fig. 3D). However, by 72 hpi, Towne- and AD169-infected HFFs, as well as Towne-infected GM05823 cells, trafficked pp65 out of the nucleus efficiently, with at least 80% of cells showing this phenotype in each of these combinations (Fig. 3D). These were all statistically significantly increased above the percentage of AD169-infected GM05823 cells with pp65 VAC staining at this time point (∼20%). The percentage of pp65 VAC stained AD169-infected GM05823 cells did not increase at 96 hpi. A large percentage of these scored AD169-infected GM05823 cells never expressed pp65, even at 96 hpi (see Fig. 3D, the total of nuclear and cytoplasmic VAC staining is substantially less than 100%). In these cells, the absence of pp65 expression correlated with the decreased number of cells that efficiently trafficked virus genomes to the nucleus, expressed IE1, and established viral RCs. However, similar to the other cell types, AD169-infected GM05823 cells that had formed RCs (refer to Fig. 3B) trafficked pp65 (and, by extension, virions) out of the nucleus.

To determine whether late proteins were expressed in infected cells, we examined glycoprotein B (gB) staining (Fig. 3E). At late times postinfection, gB characteristically localizes to the VAC, signifying that the viral replication process has progressed enough for late structural proteins to be synthesized and localized for addition to the maturing virion. Figure 3E shows the localization of gB to the VAC in Towne-infected HFFs at 72 hpi. Although once again lagging slightly behind the Towne-infected HFFs, the very large majority of both AD169-infected HFFs and Towne-infected GM05823 cells (>82%) expressed VAC-localized gB by 72 hpi (Fig. 3F, compare the blue, green, and red bars). Dramatically decreased gB expression was observed in AD169-infected GM05823 cells at all time points (Fig. 3F, yellow bars). This was again statistically significantly decreased from the three other virus/cell combinations, with one exception, the AD169-infected HFFs at 48 hpi. This correlated with the large reduction in RC formation. Although gB production is not directly linked to UL57 expression, late structural proteins are expressed after adequate viral replication has occurred. We concluded that in infections with either virus strain, the cells that formed advanced stage RCs were also able to express gB and form VACs.

Compared to their Towne-infected counterparts, AD169-infected GM05823 cells were handicapped at many phases of infection, commencing with the reduced efficiency of the AD169 virus entry into these cells and lower capability of trafficking its genome to the nucleus. Clearly, the inability to enter all of the cells in a population would reduce a virus strain's capacity to produce functional progeny. Likewise, if an infecting virus' genome could not reach its replication site, the virus' replicative life cycle would be compromised. In the GM05823 cells the AD169 virus was inhibited from the outset. Further blockades occurred at later viral checkpoints. Some of these defects may have been attributable to the slower infection kinetics of the AD169 virus infection. However, there are undoubtedly additional infection parameters in which the GM05823 environment had differential effects on the two virus strains.

Exploitation of virus strain genetic variation to rapidly identify target genes.

The above-described experiments identified a variety of differential virus life cycle processes when comparing Towne and AD169 virus strain infections in the GM05823 cells. Virion entry and/or trafficking to the nucleus appeared to be one of the important factors responsible for the ultimately large disparities in virus titers. This suggested that DNA sequence variation between the two strains was responsible and proffered a useful tool for identifying genes involved in this process. Analysis of other differential blockades should be equally amenable using similar strategies.

In silico analysis of the available reported genome sequences in the literature found 28 probable variations between the Towne and AD169 strains. Performing reintroduction experiments with 28 individual constructs was impractical. Fortuitously, a number of other virus strains and variants with different combinations of these changes were available (see Materials and Methods for extensive descriptions). We expanded our strain battery to include a clinical strain, TR (33, 34), which has an essentially complete genome; Towne(short), a subvariant found in the ATCC Towne strain (VR977), with an extensive deletion of the UL/b′ region (40–44); and ADVarUC, containing a smaller deletion in the clinical cassette compared to AD169 (VR538) (40, 42). Our interpretation of the sometimes conflicting, existing literature yielded a consensus of these various strains' actual genetic content, as seen in Fig. 4. Comparison of the genetic maps and their variations allowed for design of a flow chart (Fig. 5) of serial iterations of our original BrdU-labeled genome trafficking experiments, which would reduce the number of candidate genes from 28 to between 4 and 12, depending upon the outcome of the experiments.

FIG 5.

Flow chart delineating serial experiments used to narrow down genes potentially involved in virus entry and trafficking. The pathway that was taken is indicated with bold arrows.

Restoration of viral genome trafficking in AD169-infected GM05823 cells.

We infected GM05823 cells with the additional strains and/or variants [TR, Towne(short), and ADVarUC] we had not used in the initial life cycle experiments and visualized the input genomes using BrdU, as described previously. TR has an intact UL/b′ region and does not carry the mutations outside of the UL/b′ region seen in Towne and AD169. The TR genome trafficked to the nucleus in ~85% of cells, which was slightly lower than the lab-adapted Towne strain (Fig. 6A). This minor difference was probably due to the documented increase in efficiency of entry and replication in fibroblasts of lab-adapted strains, like Towne, that carry a mutation in the RL13 gene (59). The noticeably increased entry or trafficking efficiency of TR compared to AD169, although not quite statistically significant (P = 0.07), indicated that the gene(s) involved in efficient entry and trafficking to the nucleus was either within the UL/b′ region or was a gene(s) missing or mutated in AD169 and not in Towne (as shown in the flow diagram in Fig. 5). This experiment also determined that it was not a gene product expressed by AD169, but not by Towne, which was responsible for the entry or trafficking defect. Next, we tested Towne(short), which also efficiently trafficked into the nucleus (P = 0.008 versus AD169), further reducing the candidate gene pool both outside and within the UL/b′ region. Finally, when >95% of the cells showed efficient trafficking in ADVarUC infections (P = 0.008 versus AD169), the list was reduced to four possible genes involved in inhibiting virion entry/viral genome trafficking in the GM05823 cells: UL129, UL145, UL147, and UL148 (indicated by arrows in Fig. 4). No UL147 construct was readily available. UL147 is a secreted chemokine and unlikely to be involved in genome trafficking (35, 39). FLAG-tagged constructs of UL129, UL145, and UL148 were available (60). UL129 is uncharacterized. UL145 has a relatively conserved sequence in clinical isolates and has been predicted to be an “intranuclear regulating factor that binds directly to DNA” (61). UL148 has recently been characterized as a regulator of glycoprotein complex formation and virus/cell tropism (62).

FIG 6.

Trafficking of BrdU-labeled virions to GM05823 nuclei and reintroduction of UL148 into GM05823. (A) GM05823 cells were infected with BrdU-labeled Towne, Towne(short), AD169, ADVarUC, or TR at an MOI of 5. The cells were harvested at 4 hpi and, after acid treatment, the cells were stained for BrdU. Quantitation of trafficking of BrdU-labeled virions into the nuclei of infected cells is shown. (B) GM05823 cells were nucleofected with a UL148 construct. Cells expressing UL148 were visualized with αFlag tag (green), and trafficking genomes were visualized as previously described with αBrdU (red). Scale bar, 5 μm. The top panel shows cells just nucleofected with UL148. The middle panel shows Towne-infected cells. The bottom panel shows AD169-infected cells. (C) Quantitation of BrdU-labeled genome trafficking to the nucleus in UL148 nucleofected (UL148+) and non-nucleofected (UL148−) cells.

GM05823 cells were nucleofected with the UL129, UL145, and UL148 constructs to determine their effects on viral genome trafficking to the nucleus. Unfortunately, after repeated attempts at nucleofection with either UL129 or UL145, <1% of GM05823 cells expressed FLAG-tagged proteins, making it impossible to score enough cells to accurately assess genome trafficking to the nucleus. We suspect overexpression of these gene products may have been toxic to the cells, since parallel control nucleofections using a green fluorescent protein (GFP) reporter construct were very successful. Fortunately, nucleofection of UL148 was efficient (see Fig. 6B for examples of staining of UL148⁺ cells). This allowed observation of trafficked viral genomes after infection in the UL148⁺ environment. Figure 6B also shows examples of UL148⁺ Towne (middle panel)- and AD169 (bottom panel)-infected cells with BrdU foci within their nuclei. Cells that expressed FLAG-tagged UL148 (UL148⁺) were scored for trafficking (see Fig. 6C) and compared to cells on the same coverslip that did not express FLAG-tagged UL148 (UL148⁻ cells can also be seen in Fig. 6B). Under this protocol the UL148⁻ cells acted as controls for background trafficking and nucleofection treatment. Trafficking of both Towne and AD169 genomes in UL148⁻ cells (75 and 36% trafficking positive, respectively) was reduced compared to non-nucleofected cells (97 and 69%, respectively) (compare Fig. 6C to A). We believe the decreased trafficking was likely attributable to the somewhat harsh conditions of nucleofection treatment. There was no significant difference between trafficking of Towne and AD169 in UL148⁺ cells (74 and 62%, respectively, P = 0.15), nor was there a significant difference between trafficking of Towne in UL148⁺ compared to UL148⁻ cells (74 and 75%, respectively, P = 0.92). Trafficking of the AD169 genome into the nucleus increased significantly in UL148⁺ cells compared to UL148⁻ cells (62 and 36%, respectively, P = 0.04). The increased trafficking indicated that expression of UL148, at least partially, rescued the defective trafficking of the AD169 viral genome into the nucleus and served as proof of concept of the utility of in silico genome comparison for investigation into viral gene product function during infection.

DISCUSSION

This set of experiments was begun confident in our previous work on the role of ATM in HCMV virion production (5) but was confounded by the disparity of others' subsequent results (4). We hoped to confirm our findings, attempting to reproduce the results that conflicted with our own and expecting to uncover a discrepancy in technique, methodology, or analysis that would explain the differences. Much to our surprise, all of the earlier results were confirmed, including the conflicting findings.

The findings reported here have conclusively demonstrated and confirmed our earlier assertion that the absence of ATM in fibroblasts, in and of itself, did not affect the ability of HCMV to produce wt levels of infectious virions. This does not preclude the possibility that ATM may play a role in the kinetics of infection in certain cell types. However, in both high- and low-MOI infections Towne virus produced almost identical levels of infectious virions in wt HFFs and GM02530. In addition, high-MOI Towne infections in an additional three separate ATM⁻ cell lines (GM02052, GM03487, and GM05823), all with entirely different ATM mutations, produced wt levels of infectious virions as well. Although slightly less efficient at earlier times than Towne, AD169, at both high- and low-MOI infections, replicated to equivalently high titers in wt HFFs and GM02530 cells. Infections with other herpesviruses (VZV and HSV2) in these same GM02530 cells produced wt levels of infectious virions, demonstrating that, at least in fibroblasts, ATM was not necessary for herpesviral virion production (63). In support of this evidence, E et al. found that small interfering RNA knockdown of ATM in fibroblasts reduced functional AD169 virion production by only 10-fold versus the 1,000-fold drop found for this virus in GM05823 cells (4). In addition, shRNA experiments targeting ATM expression in fibroblasts showed no difference in functional HSV-2 virion production at either a low or a high MOI (64).

What then was the source of the widely varying titer data in the GM03395 and the AD169-infected GM05823 ATM⁻ cell lines? Several investigators have reported similar findings in other herpesviruses using multiple cell lines. Lilley et al. (65) reported that HSV-1 infections behaved differently in three ATM⁻ cell lines. Two of three infections displayed growth curves similar to that of the Towne-GM05823 combination reported here, with reduced titers at earlier times postinfection, but “catching up” to wt fibroblast titer levels at late times postinfection (36 hpi). In the third cell line, titers never recovered, remaining dramatically decreased throughout the course of infection, similar to the Towne and AD169 infections of GM03395 cells reported here. Tarakanova et al. (66) reported a clear demonstration of the importance of the cell-virus combination with the gammaherpesvirus MHV68. ATM status did not alter the titers from wt levels after either high- or low-MOI infection of matched wt and ATM knockout fibroblasts. The same was true in matched wt and KO macrophages infected at high MOIs. However, ATM⁻ macrophages infected at a low multiplicity produced a 3-log drop in titers. The authors of that study concluded that cell type-specific requirements were necessary for MHV68 replication and that perhaps fibroblasts, but not macrophages, had factors able to compensate for the lack of ATM. It is possible this same phenomenon occurred during HCMV infection in the ATM⁻ cell lines used here.

This study highlights the complex interplay between HCMV and its host. Two similar, but distinct, lab-adapted HCMV strains produced quite different titers in the same ATM⁻ cell line, GM05823. We examined, in order of occurrence during the HCMV life cycle, virus entry and input nuclear genome trafficking, RC formation, pp65/capsid trafficking, viral assembly complex (VAC) formation, and the shedding of functional virions into the supernatant. Towne-infected GM05823 cells produced wt levels of functional virions and behaved substantially like wt HFFs, whereas AD169 infection of the same cells produced at least 2 logs less virus and included multiple stages at which infection was curtailed. Clear defects in the virus life cycle were directly linked to the combination of virus strain and host cellular environment. We concluded that interactions between each strain's distinct DNA profile and the GM05823 cellular environment were responsible for the observed differences and not the presence or absence of ATM.

The regular investigational research use of the two most common lab-adapted strains of HCMV, Towne and AD169, has its pitfalls, including the occasional result from the use of one strain that cannot be reproduced in the other (67–69). However, the identified genetic variation between the strains also offers a tool to link likely gene function. The availability of sequenced clinical strains and variants of the common lab-adapted strains further enhances gene function screening. This can potentially allow for rapid screening of candidate genes.

Investigation of the virus life cycle of Towne and AD169 in GM05823 cells indicated that reduced functional virion production from AD169 infection occurred at least in part due to inhibition of viral entry and trafficking of input genomes to the nucleus. This parameter was selected to test the strain battery screening approach. Flowchart mapping of possible experimental results allowed the design of serial iterations of a single experiment (virus entry and trafficking) to rapidly narrow the number of gene candidates. In the present study, the use of five virus strains narrowed a list of 28 genes potentially responsible for the defective process exhibited by the AD169-infected GM05823 cells to just 4. Alternative outcomes would not have narrowed the list quite as extensively, with as many as 12 genes remaining, but could also have reduced the list to as few as 2 genes. Any of these possible outcomes were better than wading through 28 gene reintroduction experiments. In this study, use of a flowchart also allowed recognition that no additional experiments would have been required in the event the first iteration (TR infection) had not substantially increased the trafficking phenotype and identified a list of just seven candidate genes. This outcome would have implicated a completely different group of gene products, which did not include genes located within the clinical cassette shared by both Towne and TR. The strains used in the present study offer approximately 36 genes targetable using this approach. Additional sequenced HCMV strains are also available, e.g., Var-UK (40). Their inclusion could further expand the total number of genes and/or gene functions targetable. This technique could be used to define genes responsible for any differential result dependent on a particular strain. The assay can be optimized by utilizing strains in an order that will allow progressive narrowing of the genes involved in the differential behavior observed and the particular parameter being investigated. This approach required identification of standard metrics to make accurate comparisons, e.g., virion entry under as nearly identical conditions as possible. The absolute efficiency was less important than significant differentials produced by the virus strains.

Strain screening identified a short list of four genes likely involved in the entry/trafficking defect. Unfortunately, we were unable to successfully introduce two of the three available constructs (UL129 and UL145). We fear that expression of these proteins may have been toxic to the cells, since parallel nucleofections of UL148 and a GFP-expressing control plasmid were very successful. However, reintroduction of UL148 into GM05823 cells significantly rescued the phenotype, which further supported our finding that proper trafficking of AD169 genomes did not require ATM. Very recently published work regarding UL148 indicates that, during infection, the protein localizes to the endoplasmic reticulum and plays a role in defining the glycoprotein combinations that are displayed on the surface of the virion. In this way, UL148 may affect cell tropism of the virus, by modifying the display of proteins on the Golgi membranes, which serve as the capsids' secondary envelope, in a manner that then facilitates binding of released viral particles to the surface of the next target cell (62). An HCMV strain lacking UL148 could be expected to produce virions with greater or lesser capacity to bind to and enter certain cell types due to their differential display of glycoprotein complexes on their envelope. Perhaps more relevant to our observations on the effect of UL148 expression in the cellular environment of the target cell is a personal communication (Klaus Frueh, unpublished data) indicating that expression of the rhesus homologue of UL148 led to downregulation of NK cell ligands on the cell surface, thereby changing the topology and display of the plasma membrane of infected cells. In the cellular environment of GM05823 cells expressing UL148, cell surface expression of a protein or receptor may have been altered, thereby increasing the likelihood of AD169 binding and entering the GM05823 cells. This would indicate that the surface protein expression of the GM05823 cells was inherently less receptive to AD169 than Towne virions.

The observed partially successful rescue did not rule out a role in virus entry and trafficking for the genes which we were unable to effectively introduce but did serve as a proof of concept that the screening methodology had been effective. We believe this technique offers a highly effective tool that could prove useful to the cytomegalovirus field and may be readily adaptable to the study of other viruses and pathogens.