The Innate Immune Response to Infection Induces Erythropoietin-Dependent Replenishment of the Dendritic Cell Compartment.

Dendritic cells (DC) play a key role in the adaptive immune response due to their ability to present antigens and stimulate naïve T cells. Many bacteria and viruses can efficiently target DC, resulting in impairment of their immunostimulatory function or elimination. Hence, the DC compartment requires replenishment following infection to ensure continued operational readiness of the adaptive immune system. Here, we investigated the molecular and cellular mechanisms of inflammation-induced DC generation. We found that infection with viral and bacterial pathogens as well as Toll-like receptor 9 (TLR9) ligation with CpG-oligodeoxynucleotide (CpG-ODN) expanded an erythropoietin (EPO)-dependent TER119+CD11a+ cell population in the spleen that had the capacity to differentiate into TER119+CD11chigh and TER119-CD11chigh cells both in vitro and in vivo. TER119+CD11chigh cells contributed to the conventional DC pool in the spleen and specifically increased in lymph nodes draining the site of local inflammation. Our results reveal a so far undescribed inflammatory EPO-dependent pathway of DC differentiation and establish a mechanistic link between innate immune recognition of potential immunosuppressive pathogens and the maintenance of the DC pool during and after infection.

Nuclear herpesvirus capsid motility is not dependent on F-actin.

A considerable part of the herpesvirus life cycle takes place in the host nucleus. While much progress has been made to understand the molecular processes required for virus replication in the nucleus, much less is known about the temporal and spatial dynamics of these events. Previous studies have suggested that nuclear capsid motility is directed and dependent on actin filaments (F-actin), possibly using a myosin-based, ATP-dependent mechanism. However, the conclusions from these studies were indirect. They either relied on the effects of F-actin depolymerizing drugs to deduce an F-actin dependency or they visualized nuclear F-actin but failed to show a direct link to capsid motility. Moreover, no direct link between nuclear capsid motility and a molecular motor has been established. In this report, we reinvestigate the involvement of F-actin in nuclear herpesvirus capsid transport. We show for representative members of all three herpesvirus subfamilies that nuclear capsid motility is not dependent on nuclear F-actin and that herpesvirus infection does not induce nuclear F-actin in primary fibroblasts. Moreover, in these cells, three F-actin-inhibiting drugs failed to effect capsid motility. Only latrunculin A treatment stalled nuclear capsids but did so by an unexpected effect: the drug induced actin rods in the nucleus. Immobile capsids accumulated around actin rods, and immunoprecipitation experiments suggested that capsid motility stopped because latrunculin-induced actin rods nonspecifically bind nuclear capsids. Interestingly, capsid motility was unaffected in cells that do not induce actin rods. Based on these data, we conclude that herpesvirus nuclear capsid motility is not dependent on F-actin. Importance: Herpesviruses are large DNA viruses whose replication is dependent on the host nucleus. However, we do not understand how key nuclear processes, including capsid assembly, genome replication, capsid packaging, and nuclear egress, are dynamically connected in space and time. Fluorescence live-cell microscopy revealed that nuclear capsids are highly mobile early in infection. Two studies suggested that this motility might be due to active myosin-based transport of capsids on nuclear F-actin. However, direct evidence for such motor-based transport is lacking. We revisited this phenomenon and found no evidence that nuclear capsid motility depended on F-actin. Our results reopen the question of how nuclear herpesvirus capsids move in the host nucleus.

A novel murine cytomegalovirus vaccine vector protects against Mycobacterium tuberculosis.

Tuberculosis remains a global health problem so that a more effective vaccine than bacillus Calmette-Guérin is urgently needed. Cytomegaloviruses persist lifelong in vivo and induce powerful immune and increasing ("inflationary") responses, making them attractive vaccine vectors. We have used an m1-m16-deleted recombinant murine CMV (MCMV) expressing Mycobacterium tuberculosis Ag 85A to show that infection of mice with this recombinant significantly reduces the mycobacterial load after challenge with M. tuberculosis, whereas control empty virus has a lesser effect. Both viruses induce immune responses to H-2(d)-restricted epitopes of MCMV pp89 and M18 Ags characteristic of infection with other MCMVs. A low frequency of 85A-specific memory cells could be revealed by in vivo or in vitro boosting or after challenge with M. tuberculosis. Kinetic analysis of M. tuberculosis growth in the lungs of CMV-infected mice shows early inhibition of M. tuberculosis growth abolished by treatment with NK-depleting anti-asialo ganglio-N-tetraosylceramide Ab. Microarray analysis of the lungs of naive and CMV-infected mice shows increased IL-21 mRNA in infected mice, whereas in vitro NK assays indicate increased levels of NK activity. These data indicate that activation of NK cells by MCMV provides early nonspecific protection against M. tuberculosis, potentiated by a weak 85A-specific T cell response, and they reinforce the view that the innate immune system plays an important role in both natural and vaccine-induced protection against M. tuberculosis.

A modified screening system for loss-of-function and dominant negative alleles of essential MCMV genes.

Inactivation of gene products by dominant negative mutants is a valuable tool to assign functions to yet uncharacterized proteins, to map protein-protein interactions or to dissect physiological pathways. Detailed functional and structural knowledge about the target protein would allow the construction of inhibitory mutants by targeted mutagenesis. Yet, such data are limited for the majority of viral proteins, so that the target gene needs to be subjected to random mutagenesis to identify suitable mutants. However, for cytomegaloviruses this requires a two-step screening approach, which is time-consuming and labor-intensive. Here, we report the establishment of a high-throughput suitable screening system for the identification of inhibitory alleles of essential genes of the murine cytomegalovirus (MCMV). In this screen, the site-specific recombination of a specifically modified MCMV genome was transferred from the bacterial background to permissive host cells, thereby combining the genetic engineering and the rescue test in one step. Using a reference set of characterized pM53 mutants it was shown that the novel system is applicable to identify non-complementing as well as inhibitory mutants in a high-throughput suitable setup. The new cis-complementation assay was also applied to a basic genetic characterization of pM99, which was identified as essential for MCMV growth. We believe that the here described novel genetic screening approach can be adapted for the genetic characterization of essential genes of any large DNA viruses.

The viral chemokine MCK-2 of murine cytomegalovirus promotes infection as part of a gH/gL/MCK-2 complex.

Human cytomegalovirus (HCMV) forms two gH/gL glycoprotein complexes, gH/gL/gO and gH/gL/pUL(128,130,131A), which determine the tropism, the entry pathways and the mode of spread of the virus. For murine cytomegalovirus (MCMV), which serves as a model for HCMV, a gH/gL/gO complex functionally homologous to the HCMV gH/gL/gO complex has been described. Knock-out of MCMV gO does impair, but not abolish, virus spread indicating that also MCMV might form an alternative gH/gL complex. Here, we show that the MCMV CC chemokine MCK-2 forms a complex with the glycoprotein gH, a complex which is incorporated into the virion. We could additionally show that mutants lacking both, gO and MCK-2 are not able to produce infectious virus. Trans-complementation of these double mutants with either gO or MCK-2 showed that both proteins can promote infection of host cells, although through different entry pathways. MCK-2 has been extensively studied in vivo by others. It has been shown to be involved in attracting cells for virus dissemination and in regulating antiviral host responses. We now show that MCK-2, by forming a complex with gH, strongly promotes infection of macrophages in vitro and in vivo. Thus, MCK-2 may play a dual role in MCMV infection, as a chemokine regulating the host response and attracting specific target cells and as part of a glycoprotein complex promoting entry into cells crucial for virus dissemination.

Metabolic labeling of newly transcribed RNA for high resolution gene expression profiling of RNA synthesis, processing and decay in cell culture.

The development of whole-transcriptome microarrays and next-generation sequencing has revolutionized our understanding of the complexity of cellular gene expression. Along with a better understanding of the involved molecular mechanisms, precise measurements of the underlying kinetics have become increasingly important. Here, these powerful methodologies face major limitations due to intrinsic properties of the template samples they study, i.e. total cellular RNA. In many cases changes in total cellular RNA occur either too slowly or too quickly to represent the underlying molecular events and their kinetics with sufficient resolution. In addition, the contribution of alterations in RNA synthesis, processing, and decay are not readily differentiated. We recently developed high-resolution gene expression profiling to overcome these limitations. Our approach is based on metabolic labeling of newly transcribed RNA with 4-thiouridine (thus also referred to as 4sU-tagging) followed by rigorous purification of newly transcribed RNA using thiol-specific biotinylation and streptavidin-coated magnetic beads. It is applicable to a broad range of organisms including vertebrates, Drosophila, and yeast. We successfully applied 4sU-tagging to study real-time kinetics of transcription factor activities, provide precise measurements of RNA half-lives, and obtain novel insights into the kinetics of RNA processing. Finally, computational modeling can be employed to generate an integrated, comprehensive analysis of the underlying molecular mechanisms.

Natural killer cells are required for extramedullary hematopoiesis following murine cytomegalovirus infection.

The immune response against a variety of pathogens can lead to activation of blood formation at ectopic sites, a process termed extramedullary hematopoiesis (EMH). The underlying mechanisms of EMH have been enigmatic. Investigating splenic EMH in mice infected with murine cytomegalovirus (MCMV), we find that, while cells of the adaptive immune system were dispensable for EMH, natural killer (NK) cells were essential. EMH required recognition of infected cells via activating NK cell receptors Ly49H or NKG2D, and correspondingly, viral interference with NK cell recognition abolished EMH. Surprisingly, development of EMH was not induced by NK cell-derived cytokines but was dependent on perforin-mediated cytotoxicity in order to control virus spread. Spreading virus reduced the numbers of F4/80(+) macrophages that were crucial for inflammatory EMH. Hence, whereas MCMV suppresses inflammation-induced EMH, NK cells confine virus spread, thereby protecting extramedullary hematopoietic niches and facilitating EMH.

Systemic and local infection routes govern different cellular dissemination pathways during gammaherpesvirus infection in vivo.

Human gammaherpesviruses cause morbidity and mortality associated with infection and transformation of lymphoid and endothelial cells. Knowledge of cell types involved in virus dissemination from primary virus entry to virus latency is fundamental for the understanding of gammaherpesvirus pathogenesis. However, the inability to directly trace cell types with respect to virus dissemination pathways has prevented definitive conclusions regarding the relative contribution of individual cell types. Here, we describe that the route of infection affects gammaherpesvirus dissemination pathways. We constructed a recombinant murine gammaherpesvirus 68 (MHV-68) variant harboring a cassette which switches fluorescent markers in a Cre-dependent manner. Since the recombinant virus which was constructed on the wild-type background was attenuated, in this study we used an M1-deleted version, which infected mice with normal kinetics. Infection of Cre-transgenic mice with this convertible virus was used to estimate the quantitative contribution of defined cell types to virus productivity and dissemination during the acute phase of MHV-68 infection. In systemic infection, we found splenic vascular endothelial cells (EC) among the first and main cells to produce virus. After local infection, the contribution of EC to splenic virus production did not represent such early kinetics. However, at later time points, B cell-derived viruses dominated splenic productivity independently of systemic or local infection. Systemic versus local infection also governed the cell types involved in loading peritoneal exudate cells, leading to latency in F4/80- and CD11b-positive target cells. Systemic infection supported EC-driven dissemination, whereas local infection supported B cell-driven dissemination.

The polyomaviruses WUPyV and KIPyV: a retrospective quantitative analysis in patients undergoing hematopoietic stem cell transplantation.

BACKGROUND: The polyomaviruses WUPyV and KIPyV have been detected in various sample types including feces indicating pathogenicity in the gastrointestinal (GI) system. However, quantitative viral load data from other simultaneously collected sample types are missing. As a consequence, primary replication in the GI system cannot be differentiated from swallowed virus from the respiratory tract. Here we present a retrospective quantitative longitudinal analysis in simultaneously harvested specimens from different organ sites of patients undergoing hematopoietic stem cell transplantation (HSCT). This allows the definition of sample types where deoxyribonucleic acid (DNA) detection can be expected and, as a consequence, the identification of their primary replication site. FINDINGS: Viral DNA loads from 37 patients undergoing HSCT were quantified in respiratory tract secretions (RTS), stool and urine samples as well as in leukocytes (n = 449). Leukocyte-associated virus could not be found. WUPyV was found in feces, RTS and urine samples of an infant, while KIPyV was repeatedly detected in RTS and stool samples of 4 adult patients.RTS and stool samples were matched to determine the viral load difference showing a mean difference of 2.3 log copies/ml (p < 0.001). CONCLUSIONS: The data collected in this study suggest that virus detection in the GI tract results from swallowed virus from the respiratory tract (RT). We conclude that shedding from the RT should be ruled out before viral DNA detection in the feces can be correlated to GI symptoms.

Characterization of conserved region 2-deficient mutants of the cytomegalovirus egress protein pM53.

Dominant-negative (DN) mutants are powerful tools for studying essential protein-protein interactions. A systematic genetic screen of the essential murine cytomegalovirus (MCMV) protein pM53 identified the accumulation of inhibitory mutations within conserved region 2 (CR2) and CR4. The strong inhibitory potential of these CR4 mutants is characterized by a particular phenotype. The DN effect of the small insertion mutations in CR2 was too weak to analyze (M. Popa, Z. Ruzsics, M. Lötzerich, L. Dölken, C. Buser, P. Walther, and U. H. Koszinowski, J. Virol. 84:9035-9046, 2010); therefore, the present study describes the construction of M53 alleles lacking CR2 (either completely or partially) and subsequent examination of the DN effect on MCMV replication upon conditional expression. Overexpression of CR2-deficient pM53 inhibited virus production by about 10,000-fold. This was due to interference with capsid export from the nucleus and viral genome cleavage/packaging. In addition, the fate of the nuclear envelopment complex in the presence of DN pM53 overexpression was analyzed. The CR2 mutants were able to bind to pM50, albeit to a lesser extent than the wild-type protein, and relocalized the wild-type nuclear envelope complex in infected cells. Unlike the CR4 DN, the CR2 DN mutants did not affect the stability of pM50.

Protection from inflammatory organ damage in a murine model of hemophagocytic lymphohistiocytosis using treatment with IL-18 binding protein.

Hemophagocytic lymphohistiocytosis (HLH) is a life-threatening condition due to the association of an infectious agent with lymphocyte cytotoxicity defects, either of congenital genetic origin in children or presumably acquired in adults. In HLH patients, an excess of lymphocyte or macrophage cytokines, such as IFN-γ and TNFα is present in serum. In animal models of the disease, IFN-γ and TNF-α have been shown to play a central pathogenic role. In humans, unusually high concentrations of IL-18, an inducer of IFN-γ, and TNF-α have been reported, and are associated with an imbalance between IL-18 and its natural inhibitor IL-18 binding protein (IL-18BP) resulting in an excess of free IL-18. Here we studied whether IL-18BP could reduce disease severity in an animal model of HLH. Mouse cytomegalovirus infection in perforin-1 knock-out mice induced a lethal condition similar to human HLH characterized by cytopenia with marked inflammatory lesions in the liver and spleen as well as the presence of hemophagocytosis in bone marrow. IL-18BP treatment decreased hemophagocytosis and reversed liver as well as spleen damage. IL-18BP treatment also reduced both IFN-γ and TNF-α production by CD8(+) T and NK cells, as well as Fas ligand expression on NK cell surface. These data suggest that IL-18BP is beneficial in an animal model of HLH and in combination with anti-infectious therapy may be a promising strategy to treat HLH patients.

A beta-herpesvirus with fluorescent capsids to study transport in living cells.

Fluorescent tagging of viral particles by genetic means enables the study of virus dynamics in living cells. However, the study of beta-herpesvirus entry and morphogenesis by this method is currently limited. This is due to the lack of replication competent, capsid-tagged fluorescent viruses. Here, we report on viable recombinant MCMVs carrying ectopic insertions of the small capsid protein (SCP) fused to fluorescent proteins (FPs). The FPs were inserted into an internal position which allowed the production of viable, fluorescently labeled cytomegaloviruses, which replicated with wild type kinetics in cell culture. Fluorescent particles were readily detectable by several methods. Moreover, in a spread assay, labeled capsids accumulated around the nucleus of the newly infected cells without any detectable viral gene expression suggesting normal entry and particle trafficking. These recombinants were used to record particle dynamics by live-cell microscopy during MCMV egress with high spatial as well as temporal resolution. From the resulting tracks we obtained not only mean track velocities but also their mean square displacements and diffusion coefficients. With this key information, we were able to describe particle behavior at high detail and discriminate between particle tracks exhibiting directed movement and tracks in which particles exhibited free or anomalous diffusion.

Cytomegalovirus replicon-based regulation of gene expression in vitro and in vivo.

There is increasing evidence for a connection between DNA replication and the expression of adjacent genes. Therefore, this study addressed the question of whether a herpesvirus origin of replication can be used to activate or increase the expression of adjacent genes. Cell lines carrying an episomal vector, in which reporter genes are linked to the murine cytomegalovirus (MCMV) origin of lytic replication (oriLyt), were constructed. Reporter gene expression was silenced by a histone-deacetylase-dependent mechanism, but was resolved upon lytic infection with MCMV. Replication of the episome was observed subsequent to infection, leading to the induction of gene expression by more than 1000-fold. oriLyt-based regulation thus provided a unique opportunity for virus-induced conditional gene expression without the need for an additional induction mechanism. This principle was exploited to show effective late trans-complementation of the toxic viral protein M50 and the glycoprotein gO of MCMV. Moreover, the application of this principle for intracellular immunization against herpesvirus infection was demonstrated. The results of the present study show that viral infection specifically activated the expression of a dominant-negative transgene, which inhibited viral growth. This conditional system was operative in explant cultures of transgenic mice, but not in vivo. Several applications are discussed.

A cell free protein fragment complementation assay for monitoring the core interaction of the human cytomegalovirus nuclear egress complex.

Certain viral protein-protein interactions provide attractive targets for antiviral drug development. Recently, we described a ß-lactamase based protein fragment complementation assay (PCA) to study the core interaction of the nuclear egress complex (NEC) of different herpesviruses in cells. Now, to have a cell free assay for inhibitor screens, we expressed split ß-lactamase tagged interaction domains of the viral pUL50 and pUL53 proteins representing the NEC of human cytomegalovirus (HCMV) in bacteria. After validation and basic characterization of this NEC-PCA, we tested peptide inhibitors of the pUL50-pUL53 complex. We show that peptides resembling sequences of the first conserved region of pUL53 can inhibit the NEC-PCA. This, on one hand, indicated that the core interaction in the HCMV NEC is mediated by a linear motif. On the other hand it proved that this new pUL50-pUL53 interaction assay allows a simple cell free test for small molecular inhibitors.

Degradation of cellular mir-27 by a novel, highly abundant viral transcript is important for efficient virus replication in vivo.

Cytomegaloviruses express large amounts of viral miRNAs during lytic infection, yet, they only modestly alter the cellular miRNA profile. The most prominent alteration upon lytic murine cytomegalovirus (MCMV) infection is the rapid degradation of the cellular miR-27a and miR-27b. Here, we report that this regulation is mediated by the ∼1.7 kb spliced and highly abundant MCMV m169 transcript. Specificity to miR-27a/b is mediated by a single, apparently optimized, miRNA binding site located in its 3'-UTR. This site is easily and efficiently retargeted to other cellular and viral miRNAs by target site replacement. Expression of the 3'-UTR of m169 by an adenoviral vector was sufficient to mediate its function, indicating that no other viral factors are essential in this process. Degradation of miR-27a/b was found to be accompanied by 3'-tailing and -trimming. Despite its dramatic effect on miRNA stability, we found this interaction to be mutual, indicating potential regulation of m169 by miR-27a/b. Most interestingly, three mutant viruses no longer able to target miR-27a/b, either due to miRNA target site disruption or target site replacement, showed significant attenuation in multiple organs as early as 4 days post infection, indicating that degradation of miR-27a/b is important for efficient MCMV replication in vivo.

The role of cell types in cytomegalovirus infection in vivo.

Human cytomegalovirus (HCMV) is the major viral cause of morbidity in immune compromised patients and of pre- and perinatal pathology in newborns. The clinical manifestations are highly variable and the principles which govern these differences cannot be understood without detailed knowledge on tissue specific aspects of HCMV infection. For decades the role of individual cell types during cytomegalovirus infection and disease has been discussed. The pathogenesis of mouse cytomegalovirus (MCMV) mirrors the human infection in many aspects. Although only MCMV infection is studied extensively at the level of organs, the relative contribution of specific cell types to viral pathogenesis in vivo has remained enigmatic. Here we discuss new approaches based on the cre/loxP-system to label nascent virus progeny or to lift a replication block. The salient aspect of this technique is the change of viral genome properties selectively in cells that express cre during infection in vivo. This allowed us to study the role of endothelial cells and hepatocytes for virus dissemination and will permit detailed studies on innate and adaptive immune responses to CMV.

Shedding light on the elusive role of endothelial cells in cytomegalovirus dissemination.

Cytomegalovirus (CMV) is frequently transmitted by solid organ transplantation and is associated with graft failure. By forming the boundary between circulation and organ parenchyma, endothelial cells (EC) are suited for bidirectional virus spread from and to the transplant. We applied Cre/loxP-mediated green-fluorescence-tagging of EC-derived murine CMV (MCMV) to quantify the role of infected EC in transplantation-associated CMV dissemination in the mouse model. Both EC- and non-EC-derived virus originating from infected Tie2-cre(+) heart and kidney transplants were readily transmitted to MCMV-naïve recipients by primary viremia. In contrast, when a Tie2-cre(+) transplant was infected by primary viremia in an infected recipient, the recombined EC-derived virus poorly spread to recipient tissues. Similarly, in reverse direction, EC-derived virus from infected Tie2-cre(+) recipient tissues poorly spread to the transplant. These data contradict any privileged role of EC in CMV dissemination and challenge an indiscriminate applicability of the primary and secondary viremia concept of virus dissemination.

Virus progeny of murine cytomegalovirus bacterial artificial chromosome pSM3fr show reduced growth in salivary Glands due to a fixed mutation of MCK-2.

Murine cytomegalovirus (MCMV) Smith strain has been cloned as a bacterial artificial chromosome (BAC) named pSM3fr and used for analysis of virus gene functions in vitro and in vivo. When sequencing the complete BAC genome, we identified a frameshift mutation within the open reading frame (ORF) encoding MCMV chemokine homologue MCK-2. This mutation would result in a truncated MCK-2 protein. When mice were infected with pSM3fr-derived virus, we observed reduced virus production in salivary glands, which could be reverted by repair of the frameshift mutation. When looking for the source of the mutation, we consistently found that virus stocks of cell culture-passaged MCMV Smith strain are mixtures of viruses with or without the MCK-2 mutation. We conclude that the MCK-2 mutation in the pSM3fr BAC is the result of clonal selection during the BAC cloning procedure.

M94 is essential for the secondary envelopment of murine cytomegalovirus.

The gene M94 of murine cytomegalovirus (MCMV) as well as its homologues UL16 in alphaherpesviruses is involved in viral morphogenesis. For a better understanding of its role in the viral life cycle, a library of random M94 mutants was generated by modified transposon-based linker scanning mutagenesis. A comprehensive set of M94 mutants was reinserted into the MCMV genome and tested for their capacity to complement the M94 null mutant. Thereby, 34 loss-of-function mutants of M94 were identified, which were tested in a second screen for their capacity to inhibit virus replication. This analysis identified two N-terminal insertion mutants of M94 with a dominant negative effect. We compared phenotypes induced by the conditional expression of these dominant negative M94 alleles with the null phenotype of the M94 deletion. The viral gene expression cascade and the nuclear morphogenesis steps were not affected in either setting. In both cases, however, secondary envelopment did not proceed in the absence of functional M94, and capsids subsequently accumulated in the center of the cytoplasmic assembly complex. In addition, deletion of M94 resulted in a block of cell-to-cell spread. Moreover, the dominant negative mutant of M94 demonstrated a defect in interacting with M99, the UL11 homologue of MCMV.

HCMV spread and cell tropism are determined by distinct virus populations.

Human cytomegalovirus (HCMV) can infect many different cell types in vivo. Two gH/gL complexes are used for entry into cells. gH/gL/pUL(128,130,131A) shows no selectivity for its host cell, whereas formation of a gH/gL/gO complex only restricts the tropism mainly to fibroblasts. Here, we describe that depending on the cell type in which virus replication takes place, virus carrying the gH/gL/pUL(128,130,131A) complex is either released or retained cell-associated. We observed that virus spread in fibroblast cultures was predominantly supernatant-driven, whereas spread in endothelial cell (EC) cultures was predominantly focal. This was due to properties of virus released from fibroblasts and EC. Fibroblasts released virus which could infect both fibroblasts and EC. In contrast, EC released virus which readily infected fibroblasts, but was barely able to infect EC. The EC infection capacities of virus released from fibroblasts or EC correlated with respectively high or low amounts of gH/gL/pUL(128,130,131A) in virus particles. Moreover, we found that focal spread in EC cultures could be attributed to EC-tropic virus tightly associated with EC and not released into the supernatant. Preincubation of fibroblast-derived virus progeny with EC or beads coated with pUL131A-specific antibodies depleted the fraction that could infect EC, and left a fraction that could predominantly infect fibroblasts. These data strongly suggest that HCMV progeny is composed of distinct virus populations. EC specifically retain the EC-tropic population, whereas fibroblasts release EC-tropic and non EC-tropic virus. Our findings offer completely new views on how HCMV spread may be controlled by its host cells.