Modeling of Human Cytomegalovirus Maternal-Fetal Transmission in a Novel Decidual Organ Culture

Abstract

Human cytomegalovirus (HCMV) is the leading cause of congenital infection, associated with severe birth defects and intrauterine growth retardation. The mechanism of HCMV transmission via the maternal-fetal interface is largely unknown, and there are no animal models for HCMV. The initial stages of infection are believed to occur in the maternal decidua. Here we employed a novel decidual organ culture, using both clinically derived and laboratory-derived viral strains, for the ex vivo modeling of HCMV transmission in the maternal-fetal interface. Viral spread in the tissue was demonstrated by the progression of infected-cell foci, with a 1.3- to 2-log increase in HCMV DNA and RNA levels between days 2 and 9 postinfection, the expression of immediate-early and late proteins, the appearance of typical histopathological features of natural infection, and dose-dependent inhibition of infection by ganciclovir and acyclovir. HCMV infected a wide range of cells in the decidua, including invasive cytotrophoblasts, macrophages, and endothelial, decidual, and dendritic cells. Cell-to-cell viral spread was revealed by focal extension of infected-cell clusters, inability to recover infectious extracellular virus, and high relative proportions (88 to 93%) of cell-associated viral DNA. Intriguingly, neutralizing HCMV hyperimmune globulins exhibited inhibitory activity against viral spread in the decidua even when added at 24 h postinfection—providing a mechanistic basis for their clinical use in prenatal prevention. The ex vivo-infected decidual cultures offer unique insight into patterns of viral tropism and spread, defining initial stages of congenital HCMV transmission, and can facilitate evaluation of the effects of new antiviral interventions within the maternal-fetal interface milieu.

INTRODUCTION

Human cytomegalovirus (HCMV) is the leading cause of congenital infection in developed countries, affecting 0.5 to 2% of live births (7, 55). Primary HCMV infection during gestation poses a 30 to 40% risk of intrauterine transmission (7), whereas recurrent infection is associated with a lower transmission rate, highlighting the role of maternal immunity in fetal protection (7, 16). Congenital HCMV disease develops in ∼25% of infected children and can present as intrauterine growth retardation (IUGR), with a wide range of neurodevelopmental disabilities, leading to hearing loss, vision defects, and mental retardation (13, 65). Infection during early gestation is associated with a high risk for disease (13, 65). Despite the considerable public health burden, no established prenatal antiviral treatments are currently available, and no reliable prenatal markers for disease have been identified. While no vaccines are currently licensed for clinical use (59), prenatal treatment with HCMV-specific hyperimmune globulins (HIG) has recently been shown to significantly reduce the risk of congenital infection and disease following primary maternal infection (48). Furthermore, a recent pilot study of prenatal valacyclovir treatment has demonstrated reductions in fetal viral loads (29). These findings suggest the potential applicability of prenatal antiviral treatments for the prevention of congenital HCMV disease.

The most likely route of HCMV transmission to the fetus is via the placenta, which is seeded during maternal viremia (50, 54). Indeed, HCMV DNA and gene expression has been found in clinical samples taken from placentas early and late in gestation (45, 50, 54, 68). Accumulating experimental and clinical evidence indicates that, in addition to serving in viral transmission, the placenta may be actively injured by HCMV, conceivably contributing to the observed IUGR and fetal disease (1, 50, 59, 60). The placenta is a chimeric organ, containing both maternal and fetal structures (43) (Fig. 1). During normal placental development, invasive cytotrophoblasts (CTBs), originating from anchoring chorionic villi, invade the maternal decidua. A subset of these cells remodels the uterine vasculature in the decidua at the maternal-fetal interface (2, 50, 52). This process is controlled through the coordinated actions of multiple interrelated invasion- and angiogenesis-promoting factors (2, 32, 50, 52). The maternal decidua has a distinctive multicell nature, with invasive CTBs and uterine epithelial, stromal, and endothelial cells, as well as immune cells (Fig. 1).

Fig. 1.

Schematic presentation of the placenta and the maternal decidua. (A) Tree-like placental (fetal) floating villi, immersed in maternal blood, and anchoring villi, invading the maternal decidua (in gray). A rectangular area in the maternal-fetal interface is outlined in white. (B) Detailed presentation of the maternal aspect of the area of the interface outlined in white in panel A (maternal decidua), showing invasive cytotrophoblasts originating from anchoring villi (AV) (fetal) and invading the maternal decidua, where they partially replace the resident maternal endothelium and commingle with multiple types of maternal cells: uterine epithelial, decidual (stromal), and endothelial cells, macrophages, dendritic cells, and decidual NK cells.

Our current understanding of congenital HCMV transmission and pathogenesis is largely limited by the absence of animal models for HCMV infection. While the guinea pig transplacental transmission model and the newborn mouse model have proven invaluable for the experimental evaluation of vaccines and virus-induced brain pathology (1, 8, 58), the species specificity of HCMV has precluded experimental modeling of congenital human infection. Consequently, much of what has been learned about HCMV infection of the placenta comes from in vitro studies of isolated CTB and syncytiotrophoblast (ST) cell cultures, revealing productive, albeit variable and low-efficiency, infection (15, 22, 26, 41, 50, 54, 60). The use of laboratory-adapted rather than clinical strains of HCMV, and CTBs obtained from term placentas, may have confounded the results in some of these studies. Importantly, studies in an explant model of first-trimester floating and anchoring placental villi have revealed virion transcytosis by STs and receptor-mediated patterns of infection in underlying CTBs (18, 38, 39, 50). Yet thus far, the initial stages of infection, which are believed to occur in the maternal aspect of the maternal-fetal interface, have remained unexplored. Complex interactions of the virus with uterine microvasculature, decidual lymphocytes, and invasive interstitial CTBs in the maternal decidua basalis could determine the outcome of infection.

The need to gain insight into these earliest critical events of transmission prompted us to establish an ex vivo organ culture model of the maternal decidua. Previous studies by us and by others have demonstrated the applicability of ex vivo organ cultures for the analysis of viral tropism within preserved 3-dimensional tissue structures in skin, lung, intestinal, arterial, cervical, and neuronal tissues (6, 17, 33, 35, 57). In the present study, we have employed a novel decidual organ culture for the ex vivo modeling of HCMV infection in the maternal-fetal interface. Using both clinically derived and laboratory-derived viral strains, we have defined the patterns of viral tropism and spread along with the effect of antiviral interventions within the decidual milieu.

MATERIALS AND METHODS

Cells and viruses.

Primary human foreskin fibroblasts (HFF) were used to propagate HCMV strains and the clinical isolate as described previously (74, 75). HFF were grown in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum, 2 mM glutamine, 100 IU/ml penicillin, 100 μg/ml streptomycin (Biological Industries, Beit Haemek, Israel), and 0.25 μg/ml amphotericin B (Fungizone; Invitrogen, CA). The HCMV strains used were AD169 (obtained from the American Type Culture Collection), TB40/E expressing UL32-fused green fluorescent protein (GFP) (generously provided by C. Sinzger, Germany) (63), TB40/E expressing UL83-fused GFP (strain RV1305; generously provided by M. Winkler, Germany) (12), and CMVPT30-gfp, a cell-free clinically derived HCMV strain expressing GFP (strain PT30 [17]). These viral strains were maintained as cell-free viral stocks. In addition, we used the low-passage-number clinical isolate CI851, recovered at the Hadassah Clinical Virology Laboratory from the urine of a congenitally infected newborn and propagated for 3 to 5 passages as cell-associated virus. A cell-free stock of CI851 was prepared by sonication of infected cells, followed by removal of pelleted cellular debris. The virus titers of the cleared supernatants were determined by a standard plaque assay on HFF.

Preparation and infection of decidual organ cultures.

Decidual tissues from women undergoing first-trimester elective pregnancy terminations were obtained by deep scraping to obtain maternal tissue from the basal plate and placental bed encompassing the decidua with interstitial trophoblastic invasion (Fig. 1) as described previously (73). The study was approved by the Hadassah Medical Center Institutional Review Board and was performed according to the Declaration of Helsinki, good clinical practice guidelines, and the human experimentation guidelines of the Israeli Ministry of Health. All donors gave written informed consent. Tissues, delivered within 4 h after surgery, were kept on ice until sectioning. For the preparation of decidual organ cultures, tissues were washed with phosphate-buffered saline (PBS), cut by a microtome (Tissue Sectioner, model TC-2; Sorvall Corp.) into thin slices (thickness, 250 μm) encompassing ∼10 cell layers (25), and incubated in DMEM with 25% Ham's F-12 medium, 10% fetal bovine serum, 5 mM HEPES, 2 mM glutamine, 100 IU/ml penicillin, 100 μg/ml streptomycin (Biological Industries, Beit Haemek, Israel), and 0.25 μg/ml amphotericin B at 37°C under 5% CO₂.

For infection of decidual organ cultures, the tissues were placed in 48-well plates (∼5 slices/well to maintain optimal viability) immediately after the sectioning and were inoculated with the virus (10⁴ PFU/well unless otherwise stated) for 12 h to allow effective viral adsorption. Following viral adsorption, the cultures were washed extensively and were further incubated for the duration of the experiment, with replacement of the culture medium every 2 to 3 days.

Tissue viability monitoring. (i) Mitochondrial dehydrogenase enzyme assay (MTT assay).

Tissue slices were incubated with the substrate MTT [3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide] (Sigma, Israel) for 1 h in 37°C and were then washed in PBS, followed by the addition of 100% ethanol to dissolve the colored crystals of the product. The absorbance of four replicate samples was read using an enzyme-linked immunosorbent assay (ELISA) plate reader (Organon Teknika, The Netherlands) at a wavelength of 540 nm in reference to 650 nm. Viability was determined by normalization of the absorbance values for the protein content of each extract as tested by the Bradford assay (33).

(ii) Glucose consumption assay.

Glucose levels in the medium of the incubated tissues were monitored after 48 h of culture until the end of the experiment by the Accu-Check blood sugar-sensing device (Roche, Germany).

Histological and immunohistochemistry analysis.

Decidual tissues were fixed in 37% formalin, embedded in paraffin, and cut into 5-μm-thick sections. The sections were deparaffinized in xylene, rehydrated, and stained with hematoxylin and eosin (H&E).

For immunohistochemical detection of HCMV antigens, tissue sections were placed in 0.01 M citrate buffer, warmed in a water bath to 90°C for 15 min, allowed to cool to room temperature, and then incubated with primary mouse monoclonal antibodies (MAbs) against HCMV immediate-early (IE), pp65, or glycoprotein B (gB) antigens (dilution, 1:1,000; Virusys Corporation, Taneytown, MD) diluted in CAS-Block (Zymed Laboratories, CA). For a negative control, tissue sections were incubated with CAS-Block containing no primary antibody. Sections were then washed and incubated with a horseradish peroxidase (HRP)-conjugated goat anti-mouse secondary antibody (Biocare Medical, CA). The sections were washed again, and HCMV antigens were detected by the HRP substrate 3,3′-diaminobenzidine (DAB), followed by counterstaining with hematoxylin.

Immunofluorescence.

Tissue specimens for immunofluorescence staining were fixed in 4% paraformaldehyde, embedded in OCT compound, flash frozen in liquid nitrogen, and cut into 10-μm-thick sections. Frozen sections were treated with CAS-Block in order to avoid nonspecific antibody binding and were incubated either with CAS-Block alone or with CAS-Block containing antibodies for specific cell markers. MAbs against cytokeratin 7 (dilution, 1:300; Dako, Glostrup, Denmark) were used for the detection of CTBs; MAbs against vimentin (dilution, 1:100; Dako) were used for the detection of decidual stromal cells; MAbs against CD11c (dilution, 1:50; BioLegend, CA) were used for the detection of dendritic cells; MAbs against CD68 (dilution, 1:200; Abcam, Cambridge, United Kingdom) were used for the detection of macrophages; and rabbit polyclonal antibodies against von Willebrand factor (dilution, 1:800; Dako) were used for the detection of endothelial cells. Sections were washed, incubated with Cy5-conjugated goat anti-mouse or Cy5-conjugated goat anti-rabbit secondary antibodies (dilution, 1:200; Jackson ImmunoResearch, PA), and mounted in Vectashield mounting medium with 4′,6-diamidino-2-phenylindole (DAPI) as a nuclear stain (Vector Laboratories, Burlingame, CA). Slides were visualized using a Zeiss LSM710 Axio Observer Z1 confocal microscope and were analyzed using Zen 2009 software.

Quantification of HCMV DNA and RNA.

Infected decidual tissues and HFF cultures were washed extensively and were stored at −70°C along with their corresponding supernatants (harvested 2 days after the last medium replacement).

DNA and RNA were extracted from the samples using the QIAamp DNA Mini kit and the RNeasy Mini kit (Qiagen, Hilden, Germany), respectively, according to the manufacturer's instructions. The purified DNA samples were subjected to quantitative real-time PCR on a 7900HT real-time PCR system (Applied Biosystems, Foster City, CA), using primers and probes derived from HCMV gB as described previously (5). The assay demonstrated a linear quantitation over a 6-log range. The purified RNA samples were subjected to reverse transcription using GoScript (Promega, Madison, WI), followed by quantitative real-time PCR of the late HCMV spliced mRNA R160461 as described previously (72). For comparative and kinetic analyses, the viral DNA copy number in tissues and HFF cultures was normalized to the cellular single-copy-number RNase P gene. RNase P was quantified using the TaqMan RNase P kit (Applied Biosystems) according to the manufacturer's instructions. The viral mRNA copy number was normalized to the cellular housekeeping gene encoding glucose-6-phosphate dehydrogenase (G6PD) (72).

Antiviral treatments and assays.

The antiviral drugs ganciclovir and acyclovir (Sigma) were used at concentrations of 25 μM, 50 μM, 250 μM, and 500 μM. The drugs were added to the culture medium after virus adsorption.

To measure neutralization by antibodies, the virus was preincubated with HCMV HIG (Megalotect; 100 mg protein [50 IU] per ml; Biotest, Germany) at dilutions of 1:10, 1:100, and 1:1,000 for 1 h at room temperature, followed by inoculation of the preincubated virus-antibody mixture onto the tissues.

In postadsorption treatment experiments, the culture was washed extensively to remove loosely bound virus at 24 h after viral adsorption, and then the same dilutions of HCMV HIG as those used for the measurement of neutralization by antibodies were added to the culture medium.

The tissues were further incubated in the presence of the drug or antibodies for the duration of the experiment. When the medium was replaced, it was resupplemented with drugs or antibodies.

Susceptibility to antiviral drugs was assayed by determining the drug concentration required to reduce the normalized viral DNA copy number in tissue by 50% (50% inhibitory concentration [IC₅₀]). At least 5 independent replicate experiments were carried out per drug.

RESULTS

Establishment of a decidual organ culture.

Fresh first-trimester decidual tissues were sectioned into thin slices (thickness, 250 μm) to ensure nutrient accessibility. Previous work in our laboratory indicated that a similar approach for tissues derived from different organs resulted in diverse durations of culture viability ex vivo, i.e., skin could be maintained for as long as 21 days in culture, and colon could be kept viable for 4 days only (34, 35). Based on our experience, decidual organ cultures were placed in 48-well plates (∼5 slices/well) immediately after sectioning, incubated in enriched DMEM (see Materials and Methods), and subjected to frequent medium changes to maintain optimal viability. Importantly, decidual tissue viability, as monitored by both the MTT and glucose consumption assays, was maintained for at least 12 days of incubation ex vivo. Furthermore, histological examination of sections obtained upon the initiation of culture (day 0) and at 8 days of culture demonstrated the preservation of typical decidual morphological features with no visible signs of cell death (Fig. 2A). Previous studies of HCMV infection in various tissue explants, including first-trimester floating and anchoring placental villi and arterial and cervical tissues, have revealed evidence for viral replication within 4 to 7 days (15, 17, 53). Together, these findings demonstrate that the decidual explants remain viable and retain their natural morphology over the time needed to support and monitor HCMV infection and spread.

Fig. 2.

Histopathological analysis of HCMV-infected and uninfected decidual organ cultures. Decidual tissues were prepared and maintained in culture as described in Materials and Methods. (A) H&E-stained sections (thickness, 5 μm) of uninfected decidual cultures prepared upon the initiation of culture and at 8 days of culture. Arrows point to the surface epithelium of the decidua. (B) Two different sections of cultures infected with HCMV strain PT30 (10⁴ PFU/well), obtained at 6 dpi and subjected to H&E staining. Black arrows indicate infected cells with “owl's eye” inclusion bodies; white arrows indicate granular cytoplasmic inclusions; and black arrowheads indicate irregular hyperchromatic nuclei. (C) Sections of cultures infected with HCMV strain PT30 (10⁴ PFU/well), obtained at 6 dpi and subjected to immunohistochemical analysis for viral immediate-early (IE) (showing nuclear staining) and early-late (pp65) (showing mainly cytoplasmic staining) proteins.

HCMV infection kinetics in the decidua.

To evaluate the susceptibility of decidual organ cultures to HCMV, decidual cultures were infected with the cell-free clinically derived strain PT30, expressing GFP (17). This strain was chosen for the initial development of the model to represent the wide in vivo cell tropism of HCMV and to provide visual monitoring of the infection. Preliminary experiments in our laboratory demonstrated that strain PT30 maintained a broad cell tropism to HFF, umbilical vein endothelial cells, and retinal epithelial cells (data not shown).

To evaluate the histopathological features of the ex vivo-infected decidua, histological sections of infected tissues were examined at 6 days postinfection (dpi). Histological sections of the ex vivo-infected tissues exhibited the typical histopathological characteristics of natural HCMV infection, with the appearance of cytomegalic cells with owl's eye inclusion bodies, granular cytoplasmic inclusions, and irregular hyperchromatic nuclei (Fig. 2B).

Immunohistochemical analysis of the infected sections revealed the expression of both immediate-early and pp65 early-late viral genes (Fig. 2C), as well as gB (known to be expressed late after infection) (data not shown). All control tissues, including mock-infected sections and infected sections reacted with secondary antibodies only, were negative by immunohistochemical staining. These findings indicate that HCMV undergoes a full replication cycle in the infected decidual tissues.

To follow the kinetics of virus spread in the decidua, PT30-infected live tissues were monitored daily for GFP-expressing cells by confocal microscopy. As shown in Fig. 3A, GFP expression was first detected at 2 dpi, appearing in individual cells. Gradual progression of infection was noted by 3 to 7 dpi, along with the formation of plaque-like clusters of infected cells (Fig. 3A). This kinetics of organ culture infection was consistently observed in more than 80 decidual tissues obtained from different subjects and could be similarly monitored following infection with the endotheliotropic strain TB40/E expressing GFP fused to the late structural protein pp150. TB40/E is known to maintain broad in vitro cell tropism to HFF and endothelial and epithelial cells (61, 64). Its importance in this study is further highlighted by the ability to visualize true-late protein expression (61, 64). It should be noted, however, that for some 30% of the decidual tissue samples obtained, no infection could be detected in any area of the sectioned tissue. The absence of infection appeared to be unrelated to the donor's HCMV serological status. Based on our previous experience and that of others (17, 33, 34), this is not an uncommon finding when one is working with live human tissues, and it may reflect variability in the quality of surgical specimens, as well as potential host/viral genetic factors determining the infection outcome.

Fig. 3.

HCMV infection kinetics in decidual cultures. Decidual organ cultures were infected with HCMV (10⁴ PFU/well). (A) Images of strain PT30-infected cells in live tissues as detected by confocal microscopy at the indicated times postinfection. (B) Viral DNA accumulation in tissue lysates following infection with the indicated viral strains, normalized by RNase P gene DNA copies.

To further quantify the progression of infection over time, we measured the accumulation of viral DNA in the infected decidual cultures by real-time PCR following infection (10⁴ PFU/well) with strain PT30, TB40/E, or AD169 or the low-passage-number clinical strain CI851. As shown in Fig. 3B, a consistent increase in tissue-associated viral DNA levels with time (above the level of the remaining input DNA detected at early times postinfection) was demonstrated for all viral strains. A 1.3- to 2-log increase in viral DNA accumulation was observed for all strains between 2 and 9 dpi, reflecting active DNA synthesis within the infected tissues. Moreover, quantitative analysis of the HCMV true-late RNA R160461 demonstrated a 1.9-log rise within 7 dpi for both TB40/E and AD169. These findings, together with the spreading pattern of GFP expression (demonstrated independently for the 2 different GFP-expressing viral strains) and the expression of late viral proteins, support active viral replication in the decidual organ cultures.

Cell-associated pattern of viral spread in the decidual organ cultures.

The consistent formation of plaque-like cell clusters in the infected tissue at late times postinfection, as revealed by GFP monitoring in live tissues and by histological analysis, suggested a cell-to-cell transmission pattern. To further define the mode of viral spread in the decidua, we employed viral titration on HFF and quantitative viral DNA assays to determine the relative levels of cell-associated and cell-free virus following infection with the different clinically derived and laboratory-derived viral strains: These included the clinically derived strains PT30 and TB40/E, the laboratory-derived, poorly endotheliotropic strain AD169 (all three strains are known to produce high titers of extracellular virus and are maintained as cell-free viral stocks in HFF culture), and the low-passage-number clinical isolate CI851 (characterized by cell-associated spread and maintained as cell-associated virus in HFF culture). Under the assay conditions used (where the detection limit was 50 PFU/ml, due to an initial 1:10 dilution of the supernatant), no infectious virus could be detected in the supernatants of infected decidual cultures at late times postinfection. Low titers could be transiently detected only within 1 dpi—representing residual input virus. This finding was common to all strains examined, including AD169, which is regularly released into the supernatant from infected fibroblasts. In contrast, the same virus strains, when grown in fibroblasts, generated high titers of extracellular infectious virus (in the case of PT30, AD169, and TB40/E) (data not shown).

While titration of decidual tissue-associated virus was technically unfeasible, we quantified the tissue-associated viral DNA levels, along with the viral DNA in the supernatant, at late times (8 to 10 days) postinfection by real-time PCR and analyzed their relative levels.

As shown in Table 1, analysis of the relative levels of tissue-associated versus supernatant HCMV DNA clearly revealed that the vast majority of viral DNA was tissue associated: 88 to 93% of the total DNA was found in tissue lysates, compared to 7 to 12% detected as cell-free DNA. Similar relative viral DNA proportions were identified for all strains examined (Table 1), despite considerable variability in absolute DNA levels. As a control, we also examined the relative proportions of cell-associated versus supernatant viral DNA in infected HFF cultures (Table 1). Although there was an excess of cell-associated DNA in both HFF and decidual cultures, significantly higher proportions of extracellular DNA were demonstrated in HFF cultures than in decidual cultures, ranging from 22% (for strains PT30, TB40/E, and CI851; P, <0.05 for strains TB40/E and CI851) to 49.5% (for strain AD169; P, <0.0001). While these differences could partially result from differences between a monolayer culture and a multilayer explant culture, the combined titration, quantitative DNA, and microscopic findings suggest a dominant cell-associated pattern of viral spread in the decidual organ culture, which is common to the different viral variants regardless of their divergent modes of spread in cell culture.

Table 1.

Relative levels of cell-associated and supernatant HCMV DNA in infected decidual and HFF cultures at late times postinfection

HCMV strain	Copy no. (%)a of HCMV DNA in:				Pb
	Decidual cultures		HFF cultures
	Tissue associated	Supernatant	Cell associated	Supernatant
PT30	4.7 × 104 (88)	5.2 × 103 (12)	4.8 × 107 (74)	1.5 × 107 (26)	0.095
TB40/E	1.1 × 106 (88)	1.3 × 105 (12)	5.5 × 108 (77)	1.6 × 108 (23)	0.037
AD169	1.4 × 105 (93)	1.3 × 104 (7)	7 × 108 (50.5)	6.8 × 108 (49.5)	<0.0001
CI851	3.9 × 106 (89.5)	3 × 105 (10.5)	6.9 × 108 (78)	1.9 × 108 (22)	0.046

^aPercentage of total (cell associated plus supernatant or tissue associated plus supernatant) HCMV DNA copies.

^bFor differences in the relative proportions of cell-associated versus supernatant viral DNA between decidual and HFF cultures.

HCMV infects a wide range of cells in the decidua.

While HCMV is characterized by broad cell tropism in vivo, studies in vitro are generally limited to single-cell-type cultures. To define the cellular tropism of HCMV within the decidua, we sought to identify the types of infected cells. To this end, we examined decidual tissues infected by the clinically derived strains TB40/E and PT30 (characterized by broad cell tropism [see above]) and the poorly endotheliotropic strain AD169. Frozen sections of HCMV-infected tissues were analyzed by confocal immunofluorescence using specific cell markers. As shown in Fig. 4, we found colocalization of viral strain TB40/E with markers of invasive cytotrophoblasts (cytokeratin 7), endothelial cells (von Willebrand factor), decidual cells (vimentin), dendritic cells (CD11c), and macrophages (CD68). Similar cell tropism was shown for PT30 (data not shown), indicating ex vivo infection of all major decidual cell types by HCMV. In parallel experiments following infection with strain AD169, we could demonstrate viral colocalization only with vimentin-positive stromal decidual cells.

Fig. 4.

Cellular tropism of HCMV in infected decidual cultures. At 10 dpi, frozen sections were prepared from decidual cultures infected with HCMV strain TB40/E expressing UL83 (pp65)-fused GFP, stained with monoclonal antibodies against the indicated cell types, and counterstained with DAPI to visualize the cell nuclei (blue). Colocalization of the virus (marked by GFP) with specific cell markers (shown in red) was analyzed by confocal microscopy. Yellow arrows point to cells exhibiting colocalization; green arrows point to infected cells with no colocalization with the specific cellular marker; and red arrows point to stained uninfected cells. CK7, cytokeratin 7; vWF, von Willebrand factor; Vim, vimentin.

Inhibition of HCMV infection of decidual organ cultures.

Recent studies have suggested the potential applicability of prenatal antiviral interventions in preventing maternal-fetal transmission and congenital disease (29, 48). While all currently available anti-HCMV drugs (i.e., ganciclovir, foscarnet, and cidofovir) are considered teratogenic, new and alternative antiviral options, including acyclovir and HCMV HIG, have been explored (29, 48).

To assess the usefulness of the decidual organ culture for the evaluation of antiviral activity, and to examine the ex vivo effects of relevant antiviral drugs on HCMV replication in the decidua, infected decidual cultures were incubated with different concentrations of ganciclovir and acyclovir.

As shown in Fig. 5, both antiviral drugs inhibited HCMV replication in the decidual cultures, as measured by dose-dependent reduction of viral DNA accumulation in the infected tissues. As expected, ganciclovir exhibited greater antiviral efficacy than acyclovir (IC₅₀, 1.5 μM versus 18.3 μM). These findings demonstrate that ex vivo-infected decidual cultures can be used to study the effects of antiviral interventions.

Fig. 5.

Inhibition of HCMV infection in decidual organ cultures by antiviral drugs. Decidual cultures were infected with HCMV strain TB40/E (10⁵ PFU/well) and were incubated with increasing concentrations of ganciclovir (GCV) or acyclovir (ACV). Viral DNA in the treated and untreated decidual tissue lysates was quantitated at 8 dpi, and values were normalized by RNase P DNA copies. The results are expressed as a percentage of the amount of normalized HCMV DNA present in untreated cultures ± standard error. The error bars are not readily visible due to the low values for standard error. Significant differences were found between treated and untreated decidual cultures (P, <0.05 by the two-tailed paired t test).

We subsequently examined the antiviral effect of HCMV HIG in the infected decidual cultures. Studies of neutralization activity showed that an HCMV HIG preparation, when preincubated with the virus, was capable of neutralizing HCMV infection of the decidua (Fig. 6A). The calculated HIG concentration for 50% viral neutralization was 36 μg/ml. No antiviral effect was observed with HCMV IgG-negative serum. This finding could correlate with the prophylactic effect of HIG in the prevention of maternal-fetal transmission in vivo. It should be noted that the neutralizing activity of HCMV HIG in the decidual cultures was higher than their neutralizing activity in fibroblast culture (11); in comparative experiments, the neutralization titer of HCMV HIG was more than 10-fold lower in decidual cultures than in HFF cultures. To further examine the potential therapeutic effect of HIG, we investigated its antiviral activity when it was added after viral adsorption; at 24 h postadsorption, tissues were extensively washed before the addition of increasing HIG dilutions to the culture medium. Importantly, significant dose-dependent inhibition of HCMV replication was clearly demonstrated (Fig. 6B), with an IC₅₀ of 652 μg/ml. The inhibitory effect of HIG when added postadsorption was also evident by the reduction in the spread of plaque-like clusters at late times postinfection (data not shown). In contrast, no postadsorption inhibitory effect was observed in HFF cultures. These studies reveal a combined neutralization and postadsorption antiviral effect of HCMV HIG in the decidua.

Fig. 6.

Neutralization and postadsorption inhibition of HCMV infection in decidual organ cultures by HCMV hyperimmune globulins (HIG). The indicated dilutions of HCMV HIG either were incubated with 10⁵ PFU of HCMV strain TB40/E for 1 h prior to infection, in order to examine viral neutralization (A), or were added to the media of infected decidual cultures (10⁵ PFU/well) following extensive washing of the infected tissues at 24 h postadsorption (B). Viral DNA in decidual tissue lysates was quantitated at 8 dpi, and values were normalized by RNase P DNA copies. The results are expressed as a percentage of the amount of normalized HCMV DNA present in untreated cultures. Asterisks indicate significant differences between treated and untreated decidual cultures (P, <0.05 with the two-tailed paired t test).

DISCUSSION

There is an increasing need to identify reliable prenatal correlates of congenital HCMV infection and disease and to establish prenatal interventions. Preventive treatments for congenital disease could realize their full potential once we learn the fundamental mechanisms of viral transmission and the potential impact of new antiviral approaches.

The initial steps of congenital infection are believed to occur in the decidua, the maternal aspect of the chimeric maternal-fetal interface structure (Fig. 1), where virus originating from the mother amplifies before further spread to the adjacent placenta (45, 49, 54). While considerable experimental data have been derived from in vitro studies in CTB and ST cell cultures, and from an explant model of first-trimester floating and anchoring villi (fetal part) (15, 18, 22, 26, 38, 39, 41, 50, 54, 60), the earliest events of HCMV transmission in the maternal uterine environment are still largely unknown. Here we have established a novel ex vivo model of HCMV infection in first-trimester decidual organ cultures and have thereby gained insight into the kinetics, cellular tropism, and mode of viral spread in the maternal decidua. Decidual organ cultures maintained their viability and normal histology for at least 12 days, the time needed to support and monitor viral infection and spread (Fig. 2). Active viral replication in the tissue was demonstrated by (i) the gradual progression of GFP-expressing cell foci following infection with GFP-expressing HCMV strains, along with a consistent increase in viral DNA loads over an interval of 2 to 9 days postinfection, (ii) the expression of both immediate-early and true-late viral RNA and proteins in the infected organ cultures, (iii) the appearance of typical histopathological features of natural infection, and (iv) dose-dependent inhibition of viral spread and DNA accumulation by the antiviral DNA polymerase inhibitor ganciclovir. These combined findings suggested the ability of the ex vivo-infected decidual organ culture to address dynamic aspects of viral tropism and spread.

Importantly, we have identified a wide range of cells that are infected by clinically derived HCMV strains in the decidua, including invasive CTBs, endothelial cells, macrophages, stromal decidual cells, and dendritic cells. This finding reflects the unique multicell nature of the decidua and is in accordance with the diversity of HCMV cell targets identified in vivo (30, 49, 62). To date, the majority of viral tropism studies rely on viral propagation in vitro in single-cell-type cultures (20, 21, 27, 28, 30, 36, 62). Yet the importance of alternate HCMV replication among different cell types in viral transmission and its contribution to the impaired placental anchoring and perfusion functions during congenital infection are increasingly recognized (15, 19, 37, 38–40, 49, 51, 54, 61, 67, 76). In this regard, it is noteworthy that the infected cells in the decidua represent the two distinct pathways of viral entry, i.e., fusion (at the cell surface) and endocytosis, as characterized in fibroblasts and in epithelial/endothelial cells, respectively (10, 20, 21, 56, 62, 71). Whereas HCMV endocytosis-mediated entry has been shown to require the viral gH/gL/UL128-UL131 (UL128-131) complexes, the UL128-131 proteins are not essential for HCMV fusion-mediated entry (20, 21, 56, 62, 71). Our finding that low-passage-number and clinically derived HCMV strains (containing the UL128-131 gene products in the viral envelope), as well as the laboratory-derived strain AD169 (lacking the UL128-131 gene products in the viral envelope) (3, 9, 21, 56, 70), were able to infect the decidua with similar kinetics further argues for the potential use of different or combined viral entry pathways during infection in the maternal-fetal interface. In this regard, we could demonstrate infection by the laboratory-derived strain AD169 only for stromal decidual cells. Further studies aimed at defining the decidual cell types infected by different viral strains are under way in our laboratory.

In addition to solid-tissue cells, which serve as direct platforms for viral replication, the decidua is infiltrated by immune cells, in which a unique decidual NK (dNK) cell population predominates (24) (Fig. 1). Indeed, we could detect a multitude of CD56^bright dNK cells commingling with HCMV-infected cells in the decidual organ culture (data not shown). We have recently identified the function of these cells in the regulation of key developmental processes during placentation (23). We have also shown that HCMV infection modulates dNK and NK cytolytic response (4, 44, 66). Future studies will examine the effect of HCMV infection on dNK function in the tissue milieu.

Following HCMV infection, virus could spread further to uninfected cells either via the extracellular environment or by direct cell-to cell transmission (30, 46, 61). A dominant cell-associated pattern of viral spread was suggested by the focal expansion of infected-cell clusters in the decidua. Furthermore, the inability to recover infectious extracellular virus from the organ culture medium and the consistent relative predominance of cell-associated viral DNA over viral DNA recovered from the growth medium indicate cell-to-cell spread of the virus (Fig. 2 and 3; Table 1). Interestingly, the pattern of tissue-associated spread appeared common to laboratory-adapted strains (originally propagated as cell-free virus in fibroblasts) and low-passage-number clinical strains (found as cell-associated virus in fibroblasts), regardless of their divergent modes of spread in cell culture in vitro. The cell-associated propagation in the decidua closely mirrors the mode of HCMV spread in vivo (30, 62) and may reflect direct intercellular spread and/or cell surface retention within the decidual extracellular matrix. Both mechanisms could confer a growth advantage within the solid tissue, enhancing efficient spread at sites of cell-cell contacts (46) and potentially allowing immune evasion. In support of this hypothesis, we noted rapid cell-to-cell spread along with dynamic cell fusion events during close time lapse monitoring of the infected tissue (data not shown). This finding fits well with the report of direct transfer of cytoplasmic material from infected to uninfected cells during cell-associated HCMV spread (14). While the mechanism of cell-to-cell HCMV spread has remained incompletely resolved, recent studies have demonstrated the involvement of viral glycoprotein complexes in the fusion of adjacent cellular membranes in individual cell types (31, 47, 69). Studies in the decidua will serve to identify the viral molecular signals that exploit existing cell-cell interactions and promote transmission through the maternal-fetal crossing point.

Our finding that neutralizing HCMV HIG exhibit inhibitory activity against viral spread in the decidua, even when added at 24 h postinfection, is intriguing and may, at first glance, appear to be at odds with the cell-associated mode of viral spread. However, virions retained on the cell surface or within less-tight cell-cell contacts could remain temporarily exposed to neutralizing antibodies (46). Alternatively, HIG could interfere with the fusogenic activity of viral surface glycoproteins across cell junctions or could further mediate intracellular viral inhibition (42). Importantly, the combined virus neutralization and postinfection effects of HCMV HIG, revealed for the first time ex vivo, have direct clinical relevance in light of the reported impact of passive maternal HIG immunization on reducing fetal infection and disease and the continued clinical trials in this direction (1, 37, 40, 48, 59). Moreover, the antiviral effects of HCMV HIG differed in decidual organ cultures versus fibroblast cultures: HIG showed higher neutralizing activity in decidual cultures, and a postinfection effect was observed exclusively in decidual cultures. Hence, the infected decidual organ cultures could provide the mechanistic basis for current and future clinical trials assessing prenatal prophylactic and therapeutic use of anti-HCMV antibodies.

Also of therapeutic relevance is the finding that high concentrations of acyclovir inhibited viral replication in the decidua. In this regard, prenatal treatment with high-dose valacyclovir has been shown to result in the achievement of a similar concentration range in the amniotic fluid, with a reduction in the fetal viral load (29). Since acyclovir exhibits relatively weak anti-HCMV activity, the need for new, effective, nonteratogenic anti-HCMV drugs is further emphasized.

In summary, we have established ex vivo modeling of HCMV infection in a novel decidual organ culture, revealing a broad target cell range with a consistent cell-to-cell mode of spread of both clinically derived and laboratory-derived viral strains. The ex vivo-infected decidual cultures can serve as a unique surrogate human model that could potentially address viral and tissue determinants of transmission and could facilitate evaluation of the effects of new antiviral interventions in the maternal-fetal interface.