Luciferase-tagged wild-type and tropism-deficient mouse cytomegaloviruses reveal early dynamics of host colonization following peripheral challenge

Abstract

Cytomegaloviruses (CMVs) establish persistent, systemic infections and cause disease by maternal–foetal transfer, suggesting that their dissemination is a key target for antiviral intervention. Late clinical presentation has meant that human CMV (HCMV) dissemination is not well understood. Murine CMV (MCMV) provides a tractable model. Whole mouse imaging of virus-expressed luciferase has proved a useful way to track systemic infections. MCMV, in which the abundant lytic gene M78 was luciferase-tagged via a self-cleaving peptide (M78-LUC), allowed serial, unbiased imaging of systemic and peripheral infection without significant virus attenuation. Ex vivo luciferase imaging showed greater sensitivity than plaque assay, and revealed both well-known infection sites (the lungs, lymph nodes, salivary glands, liver, spleen and pancreas) and less explored sites (the bone marrow and upper respiratory tract). We applied luciferase imaging to tracking MCMV lacking M33, a chemokine receptor conserved in HCMV and a proposed anti-viral drug target. M33-deficient M78-LUC colonized normally in peripheral sites and local draining lymph nodes but spread poorly to the salivary gland, suggesting a defect in vascular transport consistent with properties of a chemokine receptor.

Introduction

Human cytomegalovirus (HCMV) may be detected in a range of body fluids from infected individuals, including saliva, urine, tears, breast milk and blood (Mocarski et al., 2007). However, as HCMV infections are generally asymptomatic, understanding how the virus spreads from an initial entry site has been difficult to define. In this respect, animal models of HCMV infection provide a useful alternative, since the viruses can be quantitatively traced during experimental infection and the contribution of conserved virus genes in host colonization can be evaluated. Mouse CMV (MCMV), being a natural pathogen of mice, is a useful model for HCMV infection (Krmpotic et al., 2003). MCMV encodes numerous homologues of HCMV genes and has been shown to provide an authentic read-out during both acute and persistent infection (Rawlinson et al., 1996; Scalzo et al., 2007; Shellam et al., 2007).

Previous in vivo studies of HCMV and MCMV have identified non-lymphoid cells such as fibroblasts, endothelial and epithelial cells as key sources of lytic virus, correlating with their broad tropism in derived cell lines in vitro (Sinzger et al., 2008). HCMV and MCMV infections of myeloid cells by comparison are poorly productive (Poole et al., 2014), yet they are important for virus carriage to sites of persistence (Jarvis & Nelson, 2002; Smith et al., 2004; Stoddart et al., 1994). Virus exposure to particular cell types and the pathway(s) of virus dissemination will thus be dictated by the inoculation route. The intraperitoneal (i.p.) route of inoculation bypasses the lymphatics and delivers the virus directly to the vascular compartment and results in rapid, widespread infection of multiple tissues. By contrast, the local lymphatics provide an additional barrier to peripheral MCMV infection that restrict vascular dissemination (Farrell et al., 2015). Natural acquisition of HCMV and MCMV occurs via peripheral challenge, and thus understanding host and viral determinants of dissemination will be important for rational design of antiviral or vaccination strategies.

Tracking herpesvirus dissemination in animal models has been facilitated by the generation and implementation of marker-tagged viruses. The use of the HCMV immediate-early promoter (IEp) to mediate expression of phenotypic markers has been widely used due to its high constitutive activity. In recent years, picornavirus 2A peptides have been used in multicistronic constructs to provide stoichiometric protein expression from a single promoter with high cleavage efficiency (Szymczak et al., 2004).

In this study, we used porcine teschovirus-1 2A peptide to generate a luciferase-tagged MCMV, where the tag was inserted downstream of the early lytic MCMV M78 gene (designated M78-LUC). In vitro, our studies showed that the early kinetics of luciferase expression were consistent with authentic expression from the M78 promoter and demonstrated efficient cleavage of the luciferase tag during virus infection. M78-LUC retained replication properties in vitro and in vivo that were equivalent to wild-type MCMV. In vivo comparison of M78-LUC with a MCMV recombinant expressing luciferase driven by the HCMV IEp (herein designated IEp-LUC) showed robust temporal expression by both tagged viruses that was independent of infectious virus, allowing temporal detection of colonized sites that have been previously ignored during acute infection.

Given the high sensitivity of luciferase detection, we applied this approach to compare the early infection dynamics of M78-LUC with a MCMV M33-null derivative (ΔM33/M78-LUC). M33 encodes a chemokine receptor homologue that is essential for MCMV salivary gland tropism (Davis-Poynter et al., 1997; Cardin et al., 2009), a phenotype linked to its constitutive G protein-signalling activity (Case et al., 2008; Sherrill et al., 2009). M33 was found to be dispensable for haematogenous spread following an i.p. infection (Bittencourt et al., 2014). However, the contribution of M33 in virus spread from a peripheral site remains untested. M78-LUC and ΔM33/M78-LUC infections were traced in vivo from two peripheral entry sites, and showed a M33-dependent defect in vascular spread.

Results

Construction and characterization of M78-LUC

Plasmid pBS-M78, a derivative of pBluescriptII SK- (Stratagene), contains an EcoRI–EcoRV fragment of MCMV, comprising the MCMV M78 ORF and flanking ORFs (M75–M80) of strain K181/Perth (corresponding to nucleotides 107 316–115 082 of GenBank accession no. AM886412). pB35, a derivative of pBS-M78, has an MfeI site introduced at the M78 stop codon (CAATTGAAA). The Luc 2 ORF (Photinus pyralis) was amplified from plasmid pGL4.10 (Promega). The porcine teschovirus-1 P2A cleavage sequence was inserted via complementary oligonucleotides that were used as ‘internal’ primers for the splicing by overlap extension (SOE) amplification of M78 (STOP codon removed) and Luc, to enable in in-frame expression of luciferase downstream of the 2A motif. The primers used to generate the final fused construct are shown in Table 1. The M78-P2A-LUC cassette was cloned directly into pB35 (via AvrII/MfeI), to generate pB249, shown schematically in Fig. 1(a). Recombinant M78 P2A/Luc virus (hereafter designated M78-LUC) was prepared by homologous recombination between pB249 and purified DNA from MCMV K181/Perth that had the Escherichia coli LacZ cassette inserted into the M78 locus. Recombinant M78-LUC virus was selected by the absence of β-galactosidase as previously described (Davis-Poynter et al., 1997).

Table 1. Primers used in SOE-PCR of M78/P2A/LUC fusion.

Primer	Sequence 5′−3′
M78for	cggcgagatctATGCCGACTTCATCGTGCGC
M78rev	TCCAGCCTGCTTCAGCAGGCTGAAGTTAG
	TAGCTCCGCTTCCGGAGACAACAGAGGAGGAGGTAG
Lucfor	CTGCTGAAGCAGGCTGGAGACGTGGAGGAGAACCCT
	GGACCTACCATGGAAGATGCCAAAAACA
Lucrev	cggccaattgTTACACGGCGATCTTGCCGC

Fig. 1.

(a) Schematic outline of the ORF arrangement around the M78 region, the M78-P2A-Luc cassette introduced at the M78 locus of recombinant MCMV M78-LUC and the amino acid sequence around the P2A ‘cleavage site’. (b) Luciferase is cleaved in M78-LUC infected cells. MEF were infected with either M78-LUC (M78), IEp-LUC (IEp), wild-type K181 (+) or left uninfected (−). Forty-eight hours p.i. cells were lysed, the proteins separated by SDS-PAGE and transferred to nitrocellulose. Blots were probed with an antibody to luciferase (Promega). A 60-KDa protein, consistent with the size of luciferase, was detected (denoted by arrow). Scale to the right indicates protein mass (kDa) taken from pre-stained molecular weight markers (Bio-Rad). (c) M78-LUC growth characteristics in vitro and (d) in vivo in comparison with the wt MCMV (K181 Perth) parent and ΔM78. For in vitro infections, cells were infected at an m.o.i. of 0.01 and supernatants harvested each day (n=3 per time point) for a total of 5 days and frozen at −80 °C prior to quantification of infectious virus by plaque assay. Virus titres are expressed as p.f.u. ml⁻¹ ±sem. For in vivo infections, BALB/c mice were inoculated with 10⁶ p.f.u. MCMV (either wt, M78-LUC or ΔM78) by either the i.p. or i.n. routes. The indicated organs (n=3 per time point) were harvested during the acute (day 5) or persistent (day 12) phases of infection and the virus load quantified by plaque assay on MEF. Virus titres are expressed as p.f.u. per organ ±sem. Symbols represent individual mice; circles show M78-LUC, squares show wt MCMV, triangles show ΔM78. Significance values refer to comparisons between ΔM78, M78-LUC and wt MCMV at each time point; two-way ANOVA with Tukey’s Post-Hoc comparisons: ns, not significant; *,P<0.05; **,P<0.01; ****,P<0.0001. (e) Kinetics of luciferase expression in different cell types. Cells were infected at an m.o.i. of 1.0 with either M78-LUC or IEp-LUC. At various time points p.i., cells were lysed (n=3 per time point; SG, salivary gland) and luciferase levels quantified. Uninfected cells were included as background luminescence controls. Luminescence values of M78-LUC were normalized to IEp-LUC, arbitrarily set to 1.0.

Confirmation of cleavage of the LUC polypeptide

Primary mouse embryonic fibroblasts (MEF) were infected with either M78-LUC, IEp-LUC (a positive control for soluble luciferase), wild-type K181 MCMV (negative control) or left uninfected, harvested at 48 h post-infection (p.i.), and samples analysed by SDS-PAGE/immunoblotting as detailed in Methods. Results show the detection of a 60-kDa protein from both LUC⁺-MCMVs which is consistent with the size of luciferase (Fig. 1b). There was no evidence of the uncleaved M78-LUC product, which has a predicted size of 105 kDa. The level of luciferase from IEp-LUC-infected MEF was markedly higher than that detected for M78 -LUC, consistent with the HCMV IEp achieving high level expression.

Recombinant M78-LUC virus is not attenuated in vitro or in vivo

Recombinant M78-LUC MCMV was assessed for replication in a range of cell types in comparison with wild-type MCMV K181/Perth (wt MCMV) and an M78 null derivative (ΔM78). Cell types included primary mouse embryonic fibroblasts (MEF), endothelial (SVEC 4–10), macrophage (RAW264) and epithelial (NMuMG) cells following infection at an m.o.i. of 0.01. Previous studies of M78 in MCMV and homologues in rat CMV (RCMV: R78) and HCMV (UL78) have indicated that a loss of gene function resulted in replication defects in tissue culture; for the rodent counterparts a deficiency in salivary gland tropism in vivo was also observed (Beisser et al., 1998; O'Connor & Shenk, 2012; Oliveira & Shenk, 2001). Multi-step growth analysis showed that M78-LUC grew to wt MCMV levels in each cell type (Fig. 1c) and did not display the attenuated growth characteristics of the M78 null mutant. In vivo, organ titres of M78-LUC were equivalent to wt MCMV following either i.p. or intranasal (i.n.) infection taken from acute (day 5) or persistent (day 12) stages of infection (Fig. 1d). In contrast, the M78 null mutant was cleared more rapidly from the lung and showed negligible replication in the salivary glands. Taken together, these results suggest that the C-terminal P2A/Luc fusion did not impact upon M78 function.

Characterization of pM78 luciferase expression in different cell types in vitro

The kinetics and level of luciferase expression in cells infected at an m.o.i. of 1.0 with M78-LUC and IEp-LUC were compared in a range of cell types, including fibroblast (NIH 3T3), endothelial (SVEC 4-10), macrophage (RAW264) and epithelial (NMuMG) cell lines. Luciferase expression was quantified in triplicate (independent cultures) at each time point (Fig. S1, available in the online Supplementary Material). To highlight comparison of expression between the viruses, luciferase levels for M78-LUC are shown at each time point relative to IEp-LUC (which was arbitrarily set at 1.0; Fig. 1e). Consistent with previous studies of M78 expression, luciferase expression of M78-LUC exhibited early gene kinetics in all the cell types (Oliveira & Shenk, 2001). Notably, luciferase activity of M78-LUC was generally higher than that of IEp-LUC until later times (>24 h p.i.) consistent with down-regulation of early gene expression whereas IEp-driven transcription should be maintained.

Tracking MCMV dissemination with M78-LUC; comparisons with IEp-LUC

Given the high level of luciferase expression with M78-LUC in vitro, we sought to evaluate its in vivo expression in comparison with IEp-LUC and following different routes of inoculation.

Suppression of the swallowing reflex in anaesthetized mice results in a reliable infection of the lungs following an i.n. administration of a moderate inoculation volume (10⁶ p.f.u. in 30 µl; Tan et al., 2014). Comparison of virus titres in the lung between the K181 M78-LUC and Smith IEp-LUC during acute phase replication showed higher peak titres for M78-LUC (day 5 p.i.) compared with IEp-LUC, although titres quantified at earlier and later times p.i. were equivalent (Fig. 2c). In comparison, salivary gland titres of M78-LUC were significantly higher than IEp-LUC at each time point examined, consistent with known K181 virulence characteristics (Misra & Hudson, 1980). Bioluminescent imaging (BLI) in mice infected intranasally (under anaesthesia) with either M78-LUC or IEp-LUC (n=6 per group) was performed to monitor the expression of the luciferase reporters at days 1–12 p.i. The kinetics of luciferase expression was recorded in individual live mice over 12 days (Fig. 2a; a representative from each time point is shown) and quantified (Fig. 2b). Whilst live imaging revealed a similar temporal distribution of luciferase expression by IEp-LUC and M78-LUC, there were quantifiable tissue-specific differences in their levels of expression. Luciferase measurements from the thorax (identified subsequently as the lungs in dissected animals) were significantly higher in IEp-LUC-infected mice in the first few days p.i. In contrast, signals from the neck, first detected in each infected group by day 5 p.i. (virus dissemination to the salivary glands) were significantly higher in M78-LUC-infected mice. Notably, whilst differences in luciferase levels between virus strains in the salivary gland correlated with infectious titres, a similar correlate was not observed for the lung.

Fig. 2.

In vivo tracking of M78-LUC dissemination in comparison with IEp-LUC following i.n. infection. BALB/c mice were infected i.n. (10⁶ p.f.u.) with either M78-LUC or IEp-LUC (n=6 per virus group) and each of the viruses traced in vivo over a 12-day time course. (a) Visualization of MCMV spread in live animals by BLI. A representative mouse is shown for each day p.i. The radiance scale (in photons/sec/cm²/steridian) is shown to the right. The luminescence detected in (a) was quantified for the thorax and neck in (b). Significance values refer to comparisons between M78-LUC and IEp-LUC at each time point; two-way ANOVA with Sidak’s multiple comparisons: ***,P<0.001; ****,P<0.0001. Symbols denote individual infected mice; circles show M78-LUC, triangles show IEp-LUC. Results are expressed as mean radiance ±sem. As a comparison with BLI, virus titres from replicate mice infected with either IEp-LUC or M78-LUC sampled during acute and persistent phases of infection are shown in (c). The radiance of dissected organs at days 5 and 12 p.i. (n=3 per time point) of BALB/c mice infected as in (a) is shown in (d). Following i.p. administration of luciferin, lungs (Lu), heart (h), superficial cervical lymph node (SCLN), salivary gland (SG) and noses were harvested at day 5 p.i. and imaged. At day 12 p.i. the process was repeated for the salivary glands. The radiance scale is shown to the right of each group. Quantification of luminescence and infectious virus load of dissected organs is shown in (e). Significance values refer to comparisons between M78-LUC and IEp-LUC at each time point; two-tailed unpaired t-test: *,P<0.05; **,P<0.01; ***,P<0.001; ****,P<0.0001. Quantification of viral DNA in SCLN (expressed as log DNA 10⁶ cells⁻¹) from either M78-LUC or IEp-LUC-infected mice is shown in (f).

As BLI detection is restricted to signals a few millimetres below the skin and high-expressing tissues might obscure colonization of other sites, the time course experiment was repeated with half of the mice euthanized at day 5 p.i. and the remainder at day 12 p.i. Dissected organs were compared for luciferase activity and infectious virus load. In addition to the lungs and salivary glands, luciferase signals from the superficial cervical lymph nodes (SCLN) and noses were detected at day 5 p.i. (Fig. 2d) and quantified (Fig. 2e). Luciferase and infectious virus were not detected from the stomachs of mice infected with either virus (data not shown), most likely as a result of virus inactivation upon exposure to this acidic compartment. Infectious virus was recovered from dissected lungs, salivary glands and noses; quantitative PCR (qPCR) of SCLN confirmed virus colonization at this site despite the absence of productive infection (Fig. 2f). Again, luciferase levels did not always directly correlate with infectious titres since low to negligible virus titres were recovered from the nose and SCLN despite moderate to high luciferase signals.

Following i.p. infection, luciferase expression from mice infected with IEp-LUC was higher from the abdomens of live mice compared with M78-LUC at day 5 p.i. (Fig. 3a). Following dissection, enhanced expression of IEp-LUC was detected in all visceral organs (spleen, liver, pancreas) as well as in the bone marrow (Fig. 3b, d), whereas internal examination of the peritoneal cavity at day 5 p.i. showed robust luciferase expression in the omentum that was equivalent between viruses. Despite lack of infectious virus recovered from the bone marrow, qPCR of MCMV genomic DNA confirmed colonization (Fig. 3c). Similar to i.n. infections of the lung, there was no correlation between luciferase and productive infection in visceral organs during acute infection, with both Smith IEp-LUC and K181 M78-LUC returning equivalent levels of infectious virus in each tissue examined (Fig. 3e).

Fig. 3.

In vivo tracking of M78-LUC and IEp-LUC dissemination following i.p. infection. (a) BALB/c mice were infected i.p. (10⁶ p.f.u.) with either M78-LUC or IEp-LUC (n=6 per virus group) and colonization visualized by BLI during acute (day 5) and persistent (day 12) phases of infection. (b) Following i.p. administration of luciferin, liver, spleen (SP), pancreas (P), salivary glands (SG) and femurs/tibias (BM) were harvested at day 5 p.i. and imaged. At day 12 p.i. the process was repeated for the salivary glands. The radiance scale was normalized within each organ group and is shown to the right. Confirmation of MCMV colonization in the bone marrow was confirmed by qPCR (c). Luciferase expression and virus titres from dissected organs were quantified [(d) and (e), respectively]. Significance values refer to comparisons between M78-LUC and IEp-LUC at each time point; two tailed unpaired t-test: ns, not significant; *,P<0.05; **,P<0.01; ***,P<0.001; ****,P<0.0001.

Previous studies of murine beta- and gamma-herpesvirus infections have shown that application of small volumes of virus (≤5 µl) to the nares of alert mice results in exclusive infection of the upper airway neuroepithelia (Tan et al., 2014; Farrell et al., 2016). Larger inoculum volumes under anaesthesia have also demonstrated involvement of the nasal-associated lymphoid tissue (Zhang et al., 2016). Although much of the inoculum is swallowed, negligible amounts are aspirated to the lungs. Following nasal challenge (5 µl) of alert mice, strong luciferase signals were observed from the majority of live mice challenged with M78-LUC by day 3 p.i. (Fig. 4a), with signals confirmed following dissection (Fig. 4b). In contrast, we were unable to detect luciferase⁺ nose signals from IEp-LUC infected mice in this, or subsequent experiments by either live imaging or following dissection. Productive infection at this site was negligible for both viruses at this early time point (data not shown). Neither luciferase activity nor virus was detected from the lungs from both virus groups (data not shown), confirming that the virus had not been aspirated to the lungs.

Fig. 4.

Detection of early colonization of the nose following M78-LUC infection. BALB/c mice were infected with either M78-LUC or IEp-LUC (10⁶ p.f.u. in 5 µl to alert mice, n=3 per group). At 3 days p.i. colonization of the nose was observed in two out of three animals infected with M78-LUC using BLI and following dissection (a). The radiance scale (in photons/sec/cm²/steridian) is shown to the right. The luminescence of live and dissected noses was quantified (b). Results are expressed as mean radiance ±sem. Significance values refer to comparisons between M78-LUC and IEp-LUC; two tailed unpaired t-test: ****,P<0.0001.

Taken together, luciferase expression for both LUC⁺ viruses was not always predictive of infectious load, and revealed sites of significant colonization in the absence of productive virus such as the nose, SCLN and bone marrow that are not normally associated with acute infection. M78-LUC gave consistently higher luciferase readings at the natural sites of MCMV entry (nose) and exit (salivary glands), but was markedly weaker than IEp-LUC in most tissues following i.p. infection.

Application of M78-LUC: tracking a MCMV mutant with an attenuated in vivo phenotype

Given the sensitivity of M78-LUC in detecting virus colonization, we sought to use the M78-LUC phenotypic marker to characterize the dissemination profile of a MCMV K181 strain mutant with a known and highly specific in vivo attenuation. The MCMV M33 ORF, a chemokine receptor homologue, is not essential for virus replication in vitro, but is critical for salivary gland tropism (Davis-Poynter et al., 1997; Case et al., 2008; Cardin et al., 2009). Recent studies have suggested that when inoculated i.p., M33 is dispensable for haematogenous dissemination to the salivary gland, but is required for infection of salivary gland acinar cells (Bittencourt et al., 2014). Following i.p. infection, MCMV seeds directly to the spleen and liver and thus bypasses the normal lymphatic pathways that are encountered following a peripheral infection. We used luciferase-tagged wild-type and M33-null viruses to determine if M33 also contributed to early virus colonization following two independent peripheral routes of inoculation: a combined nasal and intranasal infection, and a footpad infection. To facilitate these studies, we generated an M33 null (LacZ)/M78-LUC recombinant (hereafter designated ΔM33/M78-LUC) that was used in comparison with wild-type M78-LUC. In vitro growth profiles of ΔM33/M78-LUC showed that it was indistinguishable to wild-type MCMV (Fig. S2), confirming previous studies that M33 is not required for MCMV replication in vitro. For in vivo comparisons, mice were infected with either wild-type M78-LUC or ΔM33/M78-LUC. A nasal/intranasal infection with ΔM33/M78-LUC resulted in luciferase signals in the noses and lungs of live mice at day 5 that were equivalent to mice infected with wt-M78-LUC (Fig. 5a). Luciferase expression was also detected from the neck at this early time p.i. in each group of mice. Upon dissection, it was apparent that while both the SCLN and salivary glands contributed to the luciferase output of the neck in mice infected with wild-type M78-LUC, only the SCLN were positive in the ΔM33/M78-LUC group (Fig. 5b, c). For both viruses, acutely infected noses and SCLN yielded low or negligible levels of infectious virus, with the lungs being the main site of productive infection, consistent with previous studies (Fig. 5c; Farrell et al., 2016). qPCR of SCLN from M78-LUC-infected or ΔM33/M78-LUC-infected mice at day 5 p.i. confirmed colonization (Fig. 5d). These results demonstrate that M33 is not required for nose, SCLN and lung colonization following intranasal infection. However, we were not able to detect dissemination of ΔM33/M78-LUC to the salivary gland.

Fig. 5.

In vivo tracking of ΔM33/M78-LUC dissemination, in comparison with M78-LUC, following i.n. infection. BALB/c mice were infected with either M78-LUC or ΔM33/M78-LUC (10⁶ p.f.u., n=6 per group). Visualization of virus colonization at days 5 and 12 p.i. by whole body imaging of three representative live mice is shown in (a) and in dissected lungs (Lu), hearts (h), superficial cervical lymph nodes (SCLN) and salivary glands (SG) is shown in (b). The luminescence scale (in photons/sec/cm²/steridian) for both live mice and dissected organs is shown to the right of each group. The luminescence of dissected organs and virus titres was quantified (c). Quantification of viral DNA in SCLN (expressed as log DNA 10⁶ cells^−1) from either M78-LUC or ΔM33/78-LUC-infected animals is shown in (d). Significance values refer to comparisons between M78-LUC and ΔM33/M78-LUC at each time point; two tailed unpaired t-test: ns, not significant; *,P<0.05; **,P<0.01; ***,P<0.001, ****,P<0.0001.

A footpad infection resulted in high luciferase levels in live mice for wild-type M78-LUC and ΔM33/M78-LUC viruses at the inoculation site at day 3 p.i. which was reduced by day 6 p.i. (Fig. 6a). Luciferase signals from dissected organs confirmed spread of both viruses from the footpad to the draining popliteal lymph node (pLN) by day 3 p.i. (Fig. 6a, b) Virus spread to the spleen and salivary glands was detected for wild-type M78-LUC by day 6 p.i.; in contrast, luciferase expression for ΔM33/M78-LUC was not detected beyond the pLN at either time point. High levels of infectious virus were recovered from the footpads of both wild-type M78-LUC-infected and ΔM33/M78-LUC-infected mice confirming equivalent colonization of the entry site. Consistent with a lack of luciferase expression, infectious ΔM33/M78-LUC was not detected in either the spleen or salivary gland following footpad infection (Fig. 6c).

Fig. 6.

In vivo tracking of ΔM33 and wild-type M78-LUC dissemination following intra-footpad (i.f.) infection. BALB/c mice were infected with either M78-LUC or ΔM33/M78-LUC (10⁶ p.f.u., n=6 per group). Visualization of virus colonization at days 3 and 6 p.i. by BLI of live mice and in dissected footpads, popliteal lymph node (pLN) and spleens is shown in (a). The luminescence scale (in photons/sec/cm²/steridian) is shown to the right of each group. Quantification of luminescence from dissected organs is shown in (b) and corresponding virus titres in (c). Significance values refer to comparisons between M78-LUC and ΔM33/M78-LUC at each time point; two tailed unpaired t-test: ns, not significant; **,P<0.01; ***,P<0.001, ****,P<0.0001.

Discussion

BLI of LUC-tagged viruses allows sensitive, serial monitoring of infections in live hosts (Coleman & McGregor, 2015). Coupling LUC to an endogenous viral promoter provides an alternative tool to the insertion of heterologous promoters such as HCMV IEp insertion. It reduces transcriptional disruption of the viral genome and ties luciferase expression to a clearly defined phase of the viral life cycle. The essentially wild-type growth of M78-LUC in vitro and in vivo, with no sign of known M78 disruption phenotypes (Oliveira & Shenk, 2001; O'Connor & Shenk, 2012) established the M78/M79 intergenic region as a viable site for heterologous DNA insertion.

In vitro, LUC expression by the M78-LUC virus corresponded to early lytic gene expression, consistent with M78 expression (Oliveira & Shenk, 2001). HCMV IEp may be less restricted. IEp driven EGFP seems not to be expressed in latently infected tissue explant cultures prior to reactivation, suggesting that it too is suppressed once latency is established (H.E. Farrell, unpublished). Experience with gamma-herpesvirus infection has been that the expression of foreign genes generally does not cause significant virus attenuation if confined to the lytic cycle (Milho et al., 2009). However, latent expression can cause marked attenuation (Bennett et al., 2005), and inserted foreign promoters can follow this pattern (El-Gogo et al., 2008), raising concerns about reporter viral fitness. Therefore the M78-LUC approach of coupling reporter gene expression to an endogenous lytic transcript may prove a more certain way to track normal long-term infection.

M78 is not reported to be transcribed in latency, but detection of EGFP driven by HCMV IEp and to a lesser extent by early lytic promoters in apparently non-productive infections (Wang et al., 2003), raises the possibility that HCMV IEp-LUC and M78-LUC MCMV express luciferase in cells not supporting the full lytic cycle, which may have contributed to the lack of correlation between luciferase output and infectious virus load at particular sites such as the bone marrow, lymph node and nose. Latent and abortive infections can be hard to distinguish in vivo, as there is no certain way to achieve reactivation. Whilst an abortive infection by definition represents a ‘dead end’ with respect to dissemination, it may not be entirely innocuous, given the range of bioactive molecules produced in herpesvirus-infected cells prior to viral DNA replication (Amsler et al., 2013; Stevenson et al., 2014), luciferase expression here cannot be considered a guarantee of ‘normal’ infection, but must be used in an appropriate context of minimally invasive inoculation if tracking the breadth of host colonization is the aim.

LUC-tagged MCMVs revealed both known permissive infection sites, such as the lung, liver, spleen and salivary glands, and additional, early infection sites with low or negative plaque titres, such as the upper respiratory tract, bone marrow and lymph nodes. Several studies have shown quantitative correlations between in vivo/ex vivo luminescence and infectious virus load (Burke et al., 2011; Cook & Griffin, 2003; Heaton et al., 2013; Luker et al., 2002, 2005; Raaben et al., 2009). CMVs have broad cellular tropisms, although not all infected cells are equally permissive (Sinzger et al., 2008). In addition, the genome to p.f.u. ratio between tissues during MCMV infection is variable; in visceral organs this ratio is considerably higher than that detected in the salivary glands, a feature which may be attributed to relatively inefficient virus maturation, evident as the production of multicapsid virions (Chong & Mims, 1981; Papadimitriou et al., 1984). As genomes serve as templates for luciferase production these inter-organ differences will likely impact upon the correlative readout of BLI with infectious virus load.

The pattern of MCMV spread after peripheral challenge – local amplification, then spread via draining lymph nodes – is consistent with descriptions of other herpesviruses (Frederico et al., 2014) and even unrelated viruses showing systemic spread (Fenner, 1948). A study of MCMV dissemination from the footpad implicated CX3CR1⁺ monocytes in vascular spread (Daley-Bauer et al., 2014). However, infection was not visualized, and CX3CR1 is not a reliable myeloid lineage marker (Epelman et al., 2014) so which monocytes contributed was unclear. Direct visualization has shown that MCMV spreads from the footpad to systemic sites via the popliteal lymph nodes (Farrell et al., 2015). Finding SCLN and pLN M78-LUC signals after peripheral inoculation further supported the idea of lymphatic spread.

While differentiated myeloid cells are generally thought not to rejoin the circulation from lymph nodes, it cannot be excluded that CMVs mobilize them, to explain their key role in systemic murine and human infections (Daley-Bauer et al., 2014; Hertel, 2014; Stevenson et al., 2014). As cell migration is controlled by chemokines, viral chemokine receptor homologues are possible contributors (Case et al., 2008; Noriega et al., 2014). Studies by Bittencourt et al. (2014) showed that M33 was dispensable for haematogenous dissemination to the salivary gland following i.p. MCMV challenge. However, unlike peripheral infection, the i.p. route delivers virus direct to the bloodstream via the thoracic duct. Here, we demonstrated that M33-deficient MCMV-LUC seemed to infect peripheral sites and lymph nodes but not systemic sites. This included poor spread to the spleen, a site that is colonized by M33-deficient MCMV after i.p. inoculation (Davis-Poynter et al., 1997; Cardin et al., 2009), suggesting that M33 might contribute to MCMV dissemination via lymph nodes. However, this remains unproven. M33 contributes to evade immune clearance from visceral organs (Sherrill et al., 2009; Farrell et al., 2011), viral amplification in the salivary gland acinar cells (Bittencourt et al., 2014) and reactivation from latency (Cardin et al., 2009; Farrell et al., 2013). A possible impact on vascular spread further highlights its potential broad functionality, through exploitation of a key system of cell communication. As M33 signals constitutively via the Gq/11-dependent/phospholipase C pathway (Waldhoer et al., 2002), it may bypass the need for chemokine ligands.

The capacity of luciferase imaging to reveal infection serially and directly provided new clues as to where M33 acts in systemic MCMV spread. We anticipate that it can provide similar insights for many viral and host genes and so establish mechanistically how this complex process works.

Methods

Mice.

Female BALB/c mice (Animal Resources Center, Western Australia) were maintained at the University of Queensland Centre for Clinical Research and used when 6 weeks old. Animal experiments were approved by the University of Queensland Animal Ethics Committee in accordance with the Australian Animal Care and Protection Act (2001) and the Australian Code for the Care and Use of Animals for Scientific Purposes. Anaesthesia, where indicated, was achieved by isoflurane inhalation. Viruses were administered either intraperitoneally (i.p.; 10⁶ p.f.u. in 0.1 ml), intra-footpad (i.f.; 10⁶ p.f.u. in 50 µl), intranasally (i.n.; 5×10⁵ p.f.u. in 30 µl to anaesthetized mice), directly on the nares (nasal infection; 10⁶ p.f.u. in 5 µl to non-anaesthetized mice) or combined i.n. and nasal infections. For luciferase imaging, mice were injected i.p. with 2 mg d-luciferin, anaesthetized with isoflurane and monitored for light emission by CCD camera scanning (IVIS Lumina; Xenogen). Following live imaging, mice were euthanized (where indicated) by exposure to a rising concentration of CO₂. Organs were quickly dissected and re-monitored for light emission. The organs were then chilled at 4 °C, homogenized, and aliquots were stored at −80 °C prior to quantification of infectious virus by plaque assay on MEF.

Cells and viruses.

NIH-3T3 (ATCC CRL-1688), SVEC4-10 (ATCC CRL-2181), nMUMG (ATCC CRL-1636), and RAW264.7 (ATCC TIB71) cell lines were grown in Dulbecco's modified Eagle's medium (DMEM). Primary mouse embryonic fibroblasts (MEF) were prepared from 15–17-day-old embryos from outbred ARC(s) mice and were maintained in minimal essential medium (MEM). Both DMEM and MEM were supplemented with 2 mM glutamine, 100 IU ml⁻¹ penicillin, 100 µg ml⁻¹ streptomycin and 10 % FCS (HyClone, Australia).

Wild-type MCMV used in these studies is derived from the K181 (Perth) strain. The construction of the M78-LUC virus (on K181 Perth background) is described herein. A M33-null derivative of M78-LUC was generated by co-infecting MEF with M78-LUC and ΔM33 MCMV; the latter possessing a LacZ disruption of the M33 ORF (Davis-Poynter et al., 1997). Following the co-infection, individual plaques that stained blue upon the addition of the β-galactosidase substrate were screened for luciferase expression. Dual recombinants were subjected to a further round of plaque-picking and the absence of contaminating parental viruses confirmed by PCR. IEp-LUC contains a HCMV IE1 promoter-driven luciferase expression cassette that was inserted into the m157 ORF of a BAC-cloned MCMV strain Smith genome that was also repaired for MCK2 expression (Sell et al., 2015). The M78 null mutant is a LacZ disruption of M78 on the K181 (Perth) strain (Davis-Poynter et al., unpublished).

Working stocks of the luciferase-tagged viruses were prepared in NIH-3T3 cells, infected at an m.o.i. of 0.01. Supernatants from virus-infected cells were clarified by centrifugation to remove cell debris (50 g, 10 min), then concentrated 20–100-fold by centrifugation at (35 000 g; 2 h). Viruses from tissue culture or organ homogenates were quantified on MEF by plaque assay.

Quantification of luciferase expression virus-infected cells in vitro.

NIH-3T3, SVEC4-10, nMUMG and RAW264.7 were seeded into 96-well trays and infected with either M78-LUC or IEp-LUC 24 h later. At indicated times p.i., the supernatant from four sample wells from each cell type were washed with PBS and lysed with an equal volume of 2× cell lysis buffer (100 mM Tris pH 8.0, 10 mM MgCl₂, and 0.2 % Tween-20) for 10 min at 37 °C. An equal volume of luciferase substrate (Perkin Elmer) was added and 100 µl of the mixture was transferred to wells of black plates (Perkin Elmer) prior to quantification of luminescence (Wallac Microbeta). Control lysates from uninfected cells were used to determine background luminescence, which was subtracted from MCMV-infected cells.

Western blot analysis of luciferase expression.

MEF were seeded into 96-well trays and were either infected at an m.o.i. of 2 with M78-LUC, IEp-LUC or wt K181 viruses or left uninfected. At 48 h p.i., cells were lysed in 60 µl RIPA buffer (50 mM Tris-HCl pH 7.4, 1 % NP-40, 0.5 % Na-deoxycholate, 0.1 % SDS, 150 mM NaCl and 2 mM EDTA), sonicated and separated by SDS-PAGE under reducing conditions. Proteins were transferred to nitrocellulose, blocked with 5 % rabbit serum in PBS for 1 h at 25 °C and then incubated with goat anti-luciferase primary antibody (Promega) overnight at 4 °C. The membrane was washed three times (10 mins each wash) to remove unbound primary antibody and incubated for 1 h at 25 °C with an alexa fluor IRDye 680-conjugated rabbit anti-goat antibody (Li-Cor; diluted 1 : 1000). After a further four washes, detection of bound antibody was visualized using an infrared imaging system (Odyssey; Licor USA).

Viral genome quantitation.

Viral genomic DNA was recovered from bone marrow cells (taken from tibias and femurs) and SCLN of infected mice; samples from uninfected mice were extracted as controls (Wizard genomic DNA kit; Promega). MCMV genomic coordinates 4166 to 4252 were amplified by PCR (LightCycler 480 SYBR green; Roche) and converted to genome copies by comparison with purified MCMV genomic DNA amplified in parallel. Cellular DNA was quantified in the same samples by PCR amplification of a β-actin gene fragment (Titan primers; Roche), again with template dilutions amplified in parallel. Viral DNA loads were normalized with cellular DNA loads, expressed as the log DNA copies per 10⁶ cells. Viral DNA was not recovered from uninfected mice.