Frequent Coinfection of Cells Explains Functional In Vivo Complementation between Cytomegalovirus Variants in the Multiply Infected Host

Cells and mice.

NIH 3T3 (ATCC catalog no. CRL-1658) and M2-10B4 (ATCC catalog no. CRL-1972) were grown as previously described (29). Murine embryonic fibroblasts (MEFs) were prepared and maintained as described previously (36). Female BALB/c mice were purchased from Harlan-Winkelmann, infected at 7 to 8 weeks of age, and kept under specific-pathogen-free conditions throughout. Animal experiments were approved by the responsible state office (approval no. 211-2531-38/99).

Viruses and viral mutagenesis.

Wild-type MCMV (WT-MCMV) was derived from pSM3fr (52), a molecular clone with in vivo growth properties comparable to the MCMV Smith strain. All of the viruses were propagated on M2-10B4 cells, as described previously (29). Virus quantification was done by plaque assay on MEFs, essentially as previously described (36). Tn-M36.A and ΔM36-MCMV have been described previously (29).

Generation of Cre-MCMV.

A 1.2-kbp BamHI-SalI fragment of pGS403 carrying part of the Cre open reading frame interrupted by an intron (47) was inserted into BamHI/SalI-digested pMC-Cre (17). A 2.2-kbp XhoI fragment with the Cre gene was transferred into PmeI-digested pHMM5 (2) to provide the flanking sequences required for recombination into the MCMV ie2 locus, and then a 9.3-kbp HindIII fragment from the resulting plasmid was inserted into pST-TRαI, which is a pST76K-derived shuttle plasmid (37) encoding tetracycline resistance, RecA, and the lacZ α-peptide. The Cre gene was inserted into bacterial artificial chromosome (BAC) pSM3fr (52) by employing a two-step replacement mutagenesis procedure in Escherichia coli strain DH10B as previously described (30), which resulted in the formation of BAC pSM3fr-Cre. Bacterial clones carrying resolved cointegrates were identified by their white color on agar plates containing X-Gal (5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside).

Generation of loxPGFP-MCMV.

pCMV-LneoL-βGal (14) was digested with HindIII and SacI. A HindIII/SacI-cut 900-bp DNA fragment carrying the enhanced green fluorescent protein (GFP) open reading frame without the ATG start codon and followed by the simian virus 40 polyadenylation signal, which was generated by PCR using primers gfplox.for (5′-GCGAAGCTTGGTGAGCAAGGGCGAGGAGCTGT-3′) and gfplox.rev (5′-GGCGAGCTCCAGACATGATAAGATACATTGATGAG-3′), was inserted, resulting in pCMV-LneoL-gfp. A 3.1-kbp fragment from pCMV-LneoL-gfp obtained by NsiI/SacI digestion and blunt ended by Klenow treatment was inserted in PmeI-digested pHMM5-Kan, leading to pHMM-GFP-STOP. pHMM5-Kan is a derivative of pHMM5 with a kanamycin resistance gene flanked by FLP recognition target sites from pCP15 (9) inserted into the NsiI site of pHMM5. An 8.2-kbp AvrII fragment was isolated from pHMM-GFP-STOP and used for red α,β,γ-mediated recombination with the MCMV BAC pSM3fr (52) as previously described (51). The kanamycin resistance marker was subsequently removed by FLP-mediated recombination as previously described (9), which resulted in BAC pSM3fr-GFP-STOP. The BACs pSM3fr-Cre and pSM3fr-GFP-STOP were transfected into murine cells to give rise to the Cre-MCMV and loxPGFP-MCMVs, respectively.

In vivo coinfection protocols. Purified virus stocks were diluted in cold sterile phosphate-buffered saline and kept separate at all time points before infection to prevent a possible aggregation of virions. Thus, we minimized a bias to the favor of coinfection of cells. To avoid channeling of viruses towards the same cell populations at the beginning of infection, mice were infected intraperitoneally (i.p.) by two consecutive injections, with an interval of 45 min between injections or (where indicated) intravenously (i.v.), using the same time interval.

Quantification of infectious virus in organs.

Infectious virus was quantitated by plaque assay according to established procedures essentially as described previously (40) with some modification. In brief, organs were homogenized in 5 ml of Dulbecco's modified Eagle medium (supplemented with 10% fetal calf serum) and diluted in 1:10 steps. Diluted homogenates were layered on MEFs and centrifuged at 1,000 × g for 30 min for enhancement of infectivity. The centrifugation was followed by 30 min of incubation at 37°C. Finally, supernatants were discarded, and cells were overlaid with carboxymethylcellulose (Sigma; catalog no. C-4888) to prevent secondary viral spread. For quantification of GFP-expressing viruses from Cre-MCMV- and loxPGFP-MCMV-coinfected organs, homogenates were titrated in triplicate at 1:3.2 steps, which ensured an optimal number of 20 to 40 plaques/well in each sample. At this concentration, the assay allowed the detection of as few as 1 GFP-positive plaque among a total of 100 plaques. The low plaque density in combination with a carboxymethylcellulose overlay prevented the merging of plaques and minimized the possibility of in vitro recombination events. To exclude such events from the counts, only plaques in which all infected cells expressed GFP were considered to have resulted from in vivo recombination events.

Quantitative PCR assays.

Whole organs were homogenized on disposable cell strainers (100-μm pore size; Falcon BD catalog no. 352360) in a total volume of 5 ml. DNA from 200 μl of submandibular salivary gland homogenates was isolated by using the High-Pure viral nucleic acid kit (Roche), according to the manufacturer's instructions. A 1/25 aliquot of DNA was used for quantitative PCR. Thus, the theoretical detection limit was 625 MCMV genome copies for the salivary glands. Due to the larger amount of genomic DNA, homogenates of lungs and spleen were diluted 1:10 prior to DNA isolation. Accordingly, the detection limit in these organs approximates 6,250 viral genomes per organ. Quantitative PCR was done in the LightCycler apparatus using the QuantiTect SYBR Green PCR kit (QIAGEN GmbH, Hilden, Germany; catalog no. 204243). ΔM36 MCMV genomes were quantified with primers specific for the kanamycin resistance gene (kan): Kan-fw (5′-GGTTTGGTTGATGCGAGTGAT-3′) and Kan-rev. (5′-GAC GACTGAATCCGGTGAGAAT-3′). Cycling conditions were 45 cycles, with each cycle consisting of 95°C for 5 s, 61°C for 15 s, and 72°C for 10 s with signal detection at 78°C at the end of each cycle. The specificity of each PCR was monitored by melting-curve analysis.

Qualitative PCR assays.

DNA was isolated as described above. The Cre-mediated GFP recombination was detected by PCR using primers GFPloxP-fw (5′-GCGTGGATAGCGGTTTGACT-3′) and GFPloxP-rev. (5′-GTCGTGCTGCTTCATGTGGT-3′). Cycling was done for 40 cycles, each cycle consisting of 94°C for 30 s, 59°C for 30s, and 72°C for 60 s with standard PCR buffer supplemented with 8% dimethyl sulfoxide. PCR for the MCMV M36 exon 1 was done using primers M36EX1-fw (5′-TCATCACACAGACCACCGTAT C-3′) and M36EX1-rev. (5′-GCAAGAGGAACAACAACCCG-3′). Cycling was done as follows: 40 cycles, each consisting of 95°C for 30 s, 59°C for 30 s, and 72°C for 45 s. PCR specific for the MCMV M54 gene was done using primers MCMV-fw (5′-ATCATCCGTTGCATCTCGTTG-3′) and MCMV-rev. (5′-CGCCATCT GTATCCGTCCAT-3′). Cycling was done for a total of 30 cycles, with each cycle consisting of 95°C for 30 s, 58°C for 30 s, and 72°C for 30 s.

Two-color in situ hybridization (2C-ISH).

2C-ISH was used for simultaneous detection of two different viruses in serial tissue sections from coinfected mice. Procedures and applications have been described previously in greater detail (16, 36). In essence, a red label is used to visualize the viral gene M36 in intranuclear inclusion bodies of cells infected with WT-MCMV, whereas a black label is assigned to the kan gene sequence present in mutant ΔM36-MCMV (29). DNA hybridization was performed with deparaffinized serial sections of liver tissue as explained previously (16, 36), except that the technique was further improved by generating 1-μm sections to increase the probability for multiple sections within individual intranuclear inclusion bodies. The hybridization probe M36-P spans 1,644 bp of gene M36 from nucleotide (nt) positions 47,624 to 49,248 of the MCMV strain Smith genome (GenBank/EMBL/DDBJ, accession no. U68299; complete sequence) (39) It was synthesized by PCR using plasmid pCR3-M36 (29) as the template and oligonucleotides 5′-ATATCCCCGTGTCATCTTAA-3′ (nt 47,624 to 47,643) and 5′-ATGTATGAGCAAGAGGAACA-3′ (nt 49,267 to 49,248) as primers. M36-P was tagged during PCR by incorporation of fluorescein-conjugated dUTP (fluorescein-12-dUTP; catalog. no. 1373242, Roche Laboratories). The hybridization probe Kan-P spans 558 bp of the kanamycin resistance gene kan (transposon Tn903; EMBL accession no. V00359). It was synthesized by PCR using pDrive Cloning Vector (from the QIAGEN PCR Cloning kit) as the template and oligonucleotides 5′-TCAGGTGCGACAATCTATCG-3′ and 5′-TCCGACTCGTCCAACATCAA-3′ as primers. Kan-P was tagged during the PCR by incorporation of digoxigenin-11-dUTP (catalog no. 1093088; Roche Laboratories). Red staining was achieved by using alkaline phosphatase-conjugated anti-fluorescein antibody (catalog. no. 1426338; Roche Laboratories) with fuchsin as the chromogenic substrate. Black staining was achieved by using peroxidase-conjugated antidigoxigenin antibody (catalog no. 1207733; Roche Laboratories) with diaminobenzidine tetrahydrochloride as the substrate, followed by color enhancement with ammonium nickel sulfate hexahydrate. Quantitative microscopic analysis and the photodocumentation were performed as described previously (16).

Statistical analysis.

The probability of random coinfection of liver cells was evaluated by using the Poisson distribution function F(n) = λⁿ × e^−λ/n! (factorial), which gives a probability for single hit of F(1) = λ × e^−λ and a probability for dual hit of F(2) = F(1) × λ/2 (25).

The helper effect of WT-MCMV is not based on recombination.

Thus far the data indicated that WT-MCMV supports the in vivo growth and dissemination of attenuated mutants. This helper effect could be due to trans-complementation or DNA recombination. Homologous or illegitimate DNA recombinations of mutant and WT genomes can reinsert a missing gene in the mutant genome and thus restore the gene function and the virus fitness. However, homologous recombination would have reinserted the original sequence in the mutated locus, a process during which the mutants would have lost the gene expressing GFP or the mutant DNA sequence that was used for detection and quantification in PCR. Therefore, the presence of GFP-expressing MCMV and the increase in virus genomes containing the mutant-specific DNA fragment could not be explained by homologous recombinations. Illegitimate recombinations, which reinsert the WT gene in an ectopic position of the mutant genome, remained to be considered. In that case, the mutant would reacquire WT-like virulence and contain both the WT as well as the mutated sequences.

To evaluate a possible contribution of illegitimate recombination, GFP-positive plaques derived from mice coinfected with WT-MCMV and mutant Tn-M36.A MCMV were used for the cloning of progeny virus by repeated plaque purification. A PCR was designed with primers flanking the transposon insertion in gene M36 to give a 200-bp amplificate with the WT sequence but no product in the presence of the intermitting, 3-kb-long transposon (Fig. 4A). The attenuated phenotype of the transposon mutant (Fig. 1) (29) proved that this genomic locus was essential for the function of the M36 gene. As shown in Fig. 4, DNA from representative GFP-positive clones did not yield a PCR product; that is, it did not contain a functional, uninterrupted M36 gene. While this experiment cannot exclude rare events of illegitimate recombination, it was obvious from our data that the frequency of recombination was too low to account quantitatively for the observed helper effect. We therefore conclude that the viruses interact primarily by trans-complementation.

Cre-mediated functional complementation upon coinfection in vitro.

As an approach to provide direct molecular evidence for functional complementation between two CMVs, we used a reporter system in which, by design, reporter gene expression requires functional complementation between two viruses. Specifically, one mutant, referred to as loxPGFP-MCMV, was constructed to carry an inactive GFP gene introduced into ie2, a locus of the MCMV genome that is dispensable for viral growth in vitro and in vivo (8, 31). The open reading frame of the inserted GFP gene was interrupted by a 3-kbp-long stop cassette, which was flanked by loxP sites. Therefore, the GFP is not expressed unless Cre-mediated recombination removes the stop cassette. Accordingly, mutant Cre-MCMV was constructed to carry the gene encoding Cre recombinase. Thus, loxP sites in loxPGFP-MCMV can only be recombined and GFP only be expressed if Cre recombinase is present within the same cells, accomplished by coinfection of cells with both viruses.

Specifically, NIH 3T3 fibroblasts were infected with Cre-MCMV at a multiplicity of infection (MOI) of 0.5. Supernatant containing released Cre-MCMV virions was harvested after 48 h, a stage at which a substantial proportion of infected cells was already lysed, and was added to cultures of NIH 3T3 cells infected with loxPGFP-MCMV at an MOI of 0.5. Cells singly infected with Cre-MCMV or loxPGFP-MCMV were used as negative controls. Cells were analyzed for GFP expression by flow cytometry at 24 h postinfection. Approximately 40% of the cells that were exposed to Cre-MCMV and loxPGFP-MCMV were positive for GFP expression (Fig. 5A), whereas no GFP expression was found in cells infected with either loxPGFP-MCMV (Fig. 5B) or Cre-MCMV (not shown) alone.

These data show that Cre recombinase was active within cells that were infected with the reporter virus loxPGFP-MCMV and exposed to the supernatant of cells infected with Cre-MCMV. Clearly, expression of Cre recombinase within coinfected cells is the straightforward idea and most reasonable mechanism. However, free Cre recombinase released into the supernatant may have been taken up by the reporter cells. To evaluate the possibility of Cre recombinase release and uptake, supernatant of cells infected with Cre-MCMV was passed through 0.1-μm filters to exclude the transfer of infectious virions to the reporter cells. Reporter cells exposed to this filtered supernatant were found not to express GFP (Fig. 5C). In conclusion, reporter gene expression required coinfection of cells.

Functional complementation and single-cell coinfection in vivo.

To estimate the incidence of functional complementation in vivo, mice were i.v. coinfected with Cre- and loxPGFP-MCMV. Total and GFP-expressing viruses from organs of coinfected mice were quantified by plaque assay on day 5 postinfection. GFP-expressing viruses were detected in homogenates of spleen and lungs of all mice tested, with the median frequencies being roughly 2% in both organs (Fig. 6A).

While this finding unequivocally demonstrated the occurrence of Cre-mediated recombination, it remained a critical question whether the recombination had indeed happened already in vivo or rather during the assay in vitro (see above). Although the probability for in vitro recombination was here technically minimized by a very low MOI in the plaque assay and by the prevention of secondary plaque formation, formal proof for the occurrence of recombination in vivo is only provided by direct analysis of viral genomes in the organ homogenates. We therefore designed a recombination-specific PCR with primers binding to sequences in the GFP gene that flank the loxP sites. Thus, a positive PCR signal can be seen only upon Cre-recombination, which excises the 3-kbp stop cassette from the virus genome. For control, a group of mice was infected with the loxPGFP virus alone to monitor putative Cre-independent recombination events. All five coinfected mice showed a positive GFP PCR signal in the spleen (Fig. 6B, top) and lungs (not shown), whereas control mice did not. Presence of MCMV DNA in all tested samples was controlled by a PCR specific for gene M54 encoding the viral DNA polymerase (Fig. 6B, bottom).

In conclusion, Cre-dependent recombination had already occurred in vivo in the coinfected host. Most importantly, this set of data has shown that functional complementation occurs not only between WT virus and attenuated variants but also between two viruses that are both fully replication competent.

Functional complementation is based on single-cell coinfection.

All data presented above demonstrated functional complementation between two CMVs and provided strong but indirect evidence for in vivo coinfection of cells as the underlying mechanism. Direct evidence for simultaneous replication of two viruses in the same cell requires the detection of the two CMV genomes at the site of DNA replication, which is in the nucleus. In the late phase of the viral replicative cycle at the beginning of virion morphogenesis, viral DNA is packaged into nucleocapsids, which accumulate in a demarcated region of the nucleus that becomes histologically visible as an intranuclear inclusion body. 2C-ISH of serial tissue sections is the method of choice to verify and visualize coinfection of single cells in vivo (16, 36).

Since CMV disease develops only in the immunocompromised host and since trans-complementation between virus variants most likely occurs in an immunocompromised CMV-positive host after reactivation of one variant and superinfection with another variant or by simultaneous reactivation of variants acquired previously by sequential primary infections (see Discussion), we used the coinfection of immunocompromised mice here as a model for trans-complementation under conditions of CMV disease. The liver was chosen for two reasons: first, because hepatitis is a relevant manifestation of CMV disease; second, because hepatocyte nuclei are large enough for an analysis of viral replication in an individual nucleus by 2C-ISH of serial sections. Specifically, in the experiment shown in Fig. 7, gamma-irradiated BALB/c mice were coinfected subcutaneously by the intraplantar route with WT-MCMV and mutant ΔM36-MCMV, 10⁵ PFU of each. For discrimination of the two viruses in the infected liver, the red-tagged and black-tagged hybridization probes M36-P and Kan-P were used to stain intranuclear inclusion bodies containing WT-MCMV DNA and ΔM36-MCMV DNA, respectively (Fig. 7A). Infected cells were detected by 2C-ISH and quantitated (Fig. 7B). In the single-infection control groups, red-tagged WT-MCMV was found to replicate in the liver, whereas black-tagged ΔM36-MCMV failed. Upon coinfection with WT-MCMV, however, black-tagged ΔM36-MCMV disseminated to the liver and replicated in liver cells. Importantly, the deficiency of ΔM36-MCMV was complemented by WT-MCMV only after ipsilateral intraplantar coinfection, that is, if the two viruses were administered to the same footpad, whereas contralateral intraplantar coinfection failed. This finding indicated that the two viruses had to meet for complementation not within the liver but at the local site of coinfection before they jointly disseminated to the liver. Once in the liver, the two viruses replicated jointly and independently in liver tissue cells, with the hepatocyte as the main target cell type (Fig. 7C). As expected, the fully dissemination-competent and replication-competent WT-MCMV disseminated with higher efficacy, which led to a ∼10-fold median excess of red-stained over black-stained cells in liver tissue sections after ipsilateral intraplantar coinfection (Fig. 7B and C). A detailed serial section-based 2C-ISH analysis of black-stained cells revealed the existence of coinfected hepatocytes (Fig. 7C, top) as well as of hepatocytes in which the mutant virus replicated independently (Fig. 7C, bottom). Notably, although infection density in liver tissue varied considerably between individual mice, which is a usual observation with this experimental system, the proportions of coinfected cells and mutant-only infected cells were found to be remarkably constant, with ∼10 to 20% of the black-stained cells containing only the mutant (Fig. 7D). It is also important to note that cells singly infected by the mutant virus were not regularly found in close proximity to double-infected cells but were usually seen as separate foci of infection (Fig. 5C, bottom). Moreover, a quantitative histometric analysis of foci revealed a higher frequency of red-stained WT-MCMV foci, but no significant size distribution difference between red-stained and black-stained foci (not shown). So, we conclude that mutant virus ΔM36-MCMV grows like WT-MCMV once it has made its way into the liver.

The observed frequency of coinfected foci provides an additional argument against random coinfection of liver cells by dual hits of free virions within the liver. For calculating the statistically expected frequency we need to know the actual multiplicity of infection in the liver; this is difficult to predict, as it depends on unknown parameters such as the portion of coinfection inoculum that reaches the liver passively through the circulation and the amplification by local virus replication at the site of inoculation prior to passive dissemination. However, we can empirically determine the multiplicity of infection from the ratio of the number of nucleated liver cells and the number of foci of infection actually detected in the liver. For a representative liver section area of 100 mm², this ratio was found to be ∼770:1, which is equivalent to an empirical frequency of productive hits of F(>0) = ∼0.0013 = 1 − F(0). From F(0) = e^−λ, the expected frequency of single hits, F(1), of dual hits, F(2), and of multiple hits, F(>2), can be estimated by using the Poisson distribution function. Extrapolated to 2 × 10⁸ permissive cells in the liver, this calculation predicted ∼130,000 single-hit events with each of the two viruses and only ∼168 dual-hit events, with multiple-hit events of negligible frequency. Considering the probability for dual hit by any combination of the two viruses, which is 0.25, the number of liver cells coinfected randomly by WT-MCMV and mutant ΔM36-MCMV is just 84. This calculation predicts a ratio of coinfected foci to single-infected foci of either type as 84/130,000 = ∼1:1,550. By experiment, however, we found 486 red foci, 22 double-stained foci, and only 4 black foci in the representative 100-mm² area. So, with an observed ratio of 1:22, the dual-infection events were much more frequent than statistically predicted, whereas the events of single infection by mutant ΔM36-MCMV were much less frequent than statistically predicted. Although foci of infection caused by ΔM36-MCMV alone were rare in the liver after intraplantar coinfection, the mere fact that they existed at all (Fig. 7C, bottom) demonstrated that mutant virus replication in hepatocytes, as well as spread between hepatocytes, does not depend upon trans-complementation. In accordance with this notion, systemic infection by the intravenous route favors the seeding of independently growing mutant ΔM36-MCMV to the liver (data not shown).

In conclusion, the quantitative 2C-ISH analysis revealed a dissemination deficiency of mutant ΔM36-MCMV and a coinfection of liver cells by WT-MCMV and the mutant in a frequency that cannot be explained by random dual hits. From the collective evidence, we must propose that WT-MCMV trail-ropes the dissemination-deficient mutant for migration to the liver and that the viruses most likely meet each other in a migratory cell type at the local site of coinoculation.