A Novel Human Skin Tissue Model To Study Varicella-Zoster Virus and Human Cytomegalovirus

Varicella-zoster virus and human cytomegalovirus infect a majority of the global population. While they often cause mild disease, serious illness and complications can arise. Unfortunately, there are few effective drugs to treat these viruses, and many are toxic. To complicate this, these viruses are restricted to replication in human cells and tissues, making them difficult to study in traditional animal models. Current models rely heavily on fetal tissues, can be prohibitively expensive, and are often complicated to generate. While fetal tissue models provide helpful insights, it is necessary to study human viruses in human tissue systems to fully understand these viruses and adequately evaluate novel antivirals. Adult human skin is an appropriate model for these viruses because many target cells are present, including basal keratinocytes, fibroblasts, dendritic cells, and lymphocytes. Skin models, in culture and xenografts in immunodeficient mice, have potential for research on viral pathogenesis, tissue tropism, dissemination, and therapy.

ABSTRACT

The herpesviruses varicella-zoster virus (VZV) and human cytomegalovirus (HCMV) are endemic to humans. VZV causes varicella (chicken pox) and herpes zoster (shingles), while HCMV causes serious disease in immunocompromised patients and neonates. More effective, less toxic antivirals are needed, necessitating better models to study these viruses and evaluate antivirals. Previously, VZV and HCMV models used fetal tissue; here, we developed an adult human skin model to study VZV and HCMV in culture and in vivo. While VZV is known to grow in skin, it was unknown whether skin could support an HCMV infection. We used TB40/E HCMV and POka VZV strains to evaluate virus tropism in skin organ culture (SOC) and skin xenograft mouse models. Adult human skin from reduction mammoplasties was prepared for culture on NetWells or mouse implantation. In SOC, VZV infected the epidermis and HCMV infected the dermis. Specifically, HCMV infected fibroblasts, endothelial cells, and hematopoietic cells, with some infected cells able to transfer infection. VZV and HCMV mouse models were developed by subcutaneous transplantation of skin into SCID/beige or athymic nude mice at 2 independent sites. Viruses were inoculated directly into one xenograft, and widespread infection was observed for VZV and HCMV. Notably, we detected VZV- and HCMV-infected cells in the contralateral, uninoculated xenografts, suggesting dissemination from infected xenografts occurred. For the first time, we showed HCMV successfully grows in adult human skin, as does VZV. Thus, this novel system may provide a much-needed preclinical small-animal model for HCMV and VZV and, potentially, other human-restricted viruses.

IMPORTANCE Varicella-zoster virus and human cytomegalovirus infect a majority of the global population. While they often cause mild disease, serious illness and complications can arise. Unfortunately, there are few effective drugs to treat these viruses, and many are toxic. To complicate this, these viruses are restricted to replication in human cells and tissues, making them difficult to study in traditional animal models. Current models rely heavily on fetal tissues, can be prohibitively expensive, and are often complicated to generate. While fetal tissue models provide helpful insights, it is necessary to study human viruses in human tissue systems to fully understand these viruses and adequately evaluate novel antivirals. Adult human skin is an appropriate model for these viruses because many target cells are present, including basal keratinocytes, fibroblasts, dendritic cells, and lymphocytes. Skin models, in culture and xenografts in immunodeficient mice, have potential for research on viral pathogenesis, tissue tropism, dissemination, and therapy.

INTRODUCTION

Varicella-zoster virus (VZV) is an alphaherpesvirus that causes vesicular lesions in the skin during primary infection (varicella; chicken pox) and reactivation (herpes zoster; shingles). Most individuals are immune to VZV, either through exposure to the virus or vaccination (reviewed in references 1 and 2). Human cytomegalovirus (HCMV) is a betaherpesvirus and is the leading cause of viral congenital birth defects, such as sensorineural hearing loss (3). It is also a major cause of life-threatening posttransplant complications. HCMV causes a lifelong infection, with over half of adults in the United States infected (4); however, no vaccine is approved (5). Both viruses are restricted to humans, so research is hampered by the limited number of animal models available. Currently, both viruses are studied in cumbersome mouse models that use human fetal tissues, such as those primarily used to study latent HCMV infections (6–9). Simpler alternative models, including those that address primary and/or acute HCMV infections and primary VZV infections, are needed to advance the herpesvirus field and provide a model to evaluate viral growth kinetics and antivirals in vivo.

Adult human skin tissue is a possible alternative to the existing models. Many human herpesviruses manifest in the skin, such as herpes simplex virus cold sores and genital lesions or Kaposi's sarcoma-associated herpesvirus tumors (10). The use of human skin to study VZV is rational, since vesicular skin lesions are hallmarks of varicella and herpes zoster. In previously published reports, we used fetal skin to understand VZV growth kinetics and study the effects of antiviral drugs (11–13). Adult skin is preferable, as it is fully differentiated and constitutes the tissue environment during a VZV infection (14, 15). In contrast, the use of human skin to study HCMV is novel. Skin manifestations occur in up to a third of immunocompetent adults, with HCMV infectious mononucleosis appearing as petechial, maculopapular, or morbilliform lesions (16, 17). While vesicular skin lesions are not commonly associated with HCMV disease, many cell types that are permissive to HCMV infection are found in the skin, such as dermal fibroblasts, macrophages, dendritic cells, and endothelial cells (18, 19). For instance, congenital infections can manifest as a purpuric rash (blueberry muffin rash) in newborns (20) or ulcerative skin lesions in immunosuppressed patients (21–23), suggesting human skin can support and may be a site for HCMV infection.

The human skin is the largest organ and serves as the first line of defense for the immune system. The skin is composed of the epidermis, dermis, and subdermal layer, each with its own unique properties and different cell types. The epidermis continuously turns over as keratinocytes are regenerated from stem cells at the basement membrane (24). T lymphocytes (CD8⁺ T_RM and γδT cells), melanocytes, and Langerhans cells (LC), or epithelium-specific dendritic cells (DCs), are also found in the epidermis (24–26). The dermis is innervated and vascularized, and it contains many immune cells, including macrophages, neutrophils, mast cells, T cells [CD8⁺ T_RM, γδT cells, Tregs, and CD4⁺ T_RM and T_EM], NK cells, innate lymphoid cells (ILCs; CD3⁻ lymphocytes), and various subsets of dendritic cells [CD1c⁺, CD141⁺, and CD14⁺ DCs] (26, 27). Overall, the structure and cellular composition of the skin is designed to protect from injury and infection. The variety of cell types, particularly immune cells, in the skin makes it a strong candidate tissue for model development. In fact, prior VZV models used fetal skin (7, 8, 11–13), although it differs from adult skin tissue. For instance, fetal skin cells proliferate faster than adult skin, and adult skin is more differentiated, with a distinct epidermis and epidermal-dermal junction (14). Adult skin contains immune cells that are mostly absent from fetal skin (28). This highlights the need to distinguish and understand the differences between fetal and adult tissue models, especially since the majority of VZV cases and many HCMV cases affect individuals with a developed immune system. Thus, the use of fetal tissues, particularly skin, may not accurately reflect the environment present during an infection.

While VZV is reported to grow in skin (11–13, 15, 29), it remains unknown whether human skin could support an HCMV infection despite evidence of in vivo infection. To address this, we developed an adult human skin model and infected the tissue with VZV or HCMV. We sought to identify the location of VZV and HCMV infections in adult human skin using skin organ culture and mouse xenograft models. We also identified the cell types infected by each virus, both in the skin and in the cells that migrate out of the tissue upon infection. As expected, VZV caused lesions in the skin epidermis. We also report that HCMV infection was widespread in the skin model, infecting many known HCMV-permissive cell types. HCMV-infected cells migrated out of the skin and transferred infection in culture and in mice. This study indicates that adult human skin is a biologically relevant and novel model system for studying primary and/or acute human herpesviruses infections, particularly VZV and HCMV.

RESULTS

VZV infects the epidermis in human skin, while HCMV infects cells in the dermis.

There are a limited number of models to study human herpesviruses (reviewed in references 30 and 31). Since VZV and HCMV are restricted to replication in human cells, we aimed to develop a biological model system to study these viruses in adult human skin. To establish this model, adult human skin was obtained from breast reduction surgeries, thinned with a skin grafting knife to 700 μm thick, leaving the epidermis, dermis, and a portion of the subdermal layer, and cut into 1-cm² pieces (Fig. 1). Pilot experiments were performed to compare scarification and intradermal injection for each virus strain (not shown). HCMV replication was poor when introduced by scarification; VZV replication was similar, and mainly found in the epidermis, when either inoculation route was used. Therefore, skin was inoculated by scarification with VZV-ORF57-Luc, a robust bacterial artificial chromosome (BAC)-derived recombinant VZV based on the POka genome that contains a luciferase reporter linked to ORF57, which places the virus near basal keratinocytes, the preferred target cell in the epidermis and hair follicles (11, 15). VZV is thought to enter the skin when infected T cells extravasate from capillaries surrounding hair follicles during primary infection and from axon termini during reactivation (8). HCMV-fLuc-enhanced green fluorescent protein (eGFP), a recombinant clinical HCMV strain that contains a GFP reporter between US34 and TRS1 and a luciferase reporter replacing the nonessential UL18 glycoprotein, was inoculated by intradermal injection, which places the virus near myeloid cells and fibroblasts, the preferred target cells in the dermis (19). HCMV enters the skin when infected blood monocytes disseminate and differentiate (32). Negative-control skin was inoculated with medium (mock infected). Skin was transferred to NetWells for 7 (VZV) or 10 (HCMV) days postinoculation (dpi) and then fixed in 4% paraformaldehyde (PFA) or frozen in OCT compound. Tissue fixed in 4% PFA was sent to a commercial laboratory for hematoxylin and eosin (H&E) analysis (Fig. 2). Frozen tissue was processed and stained for immunofluorescence (IF) analysis (Fig. 3). Mock-infected skin presented with normal histopathology (Fig. 2A to C and 3A to C). The histopathology of the infected skin was examined for overt lesions and the identity of infected cells. VZV caused lesions in the epidermis (Fig. 2G to I) with the expected thickening of the epidermis, disruption of the basal cells at the epidermal-dermal junction, and formation of multinucleated cells. Lesions often formed at needle tracks where the epidermis was scratched during inoculation (Fig. 2G and H). Smaller lesions were also observed (Fig. 2I). HCMV infected the dermis, and no infected cells were found in the epidermis (Fig. 2D to F and 3D to S). In H&E sections, common HCMV histopathology was observed, including kidney-shaped nuclei (Fig. 2D and F) and “owl eye” cell morphology (Fig. 2E). As anticipated, HCMV infected multiple cell types, including fibroblasts (Fig. 3D to H), endothelial cells (Fig. 3I to N), and hematopoietic cells (Fig. 3O to S). These results are consistent with the histopathology of VZV lesions and support the use of adult skin for HCMV studies.

FIG 1.

Model system for HCMV and VZV grown in adult human skin. Human adult skin was obtained from reduction mammoplasty surgeries at SUNY Upstate Medical University. Skin was processed and used in skin organ culture (SOC) experiments or implanted in CB17 scid/beige or athymic nude mice. Readouts from SOC experiments included flow cytometry, transfer of infection from crawled-out cells, and histology of infected tissues. Readouts in SCIDhu mice included histology of xenografts, including transfer of infection from inoculated to contralateral xenografts, and infectious virus from xenografts.

FIG 2.

VZV and HCMV infections in skin result in typical disease histopathology. Skin from SOC experiments was collected and fixed in 4% PFA for H&E staining. (A to C) Representative sections of mock-infected skin at 6 dpi. Skin cell morphology and nuclei appeared normal. Boxes (dashed lines) were drawn around normal cells in the dermis (A and B) and epidermis (C), and enlargements are shown as insets. (D to F) Representative sections of HCMV-infected skin at 14 dpi. Boxes indicate the enlarged region (insets), and arrows point to single cells in the dermis with common HCMV histopathology, including kidney-shaped nuclei (D and F) or owl eye cells (E). (G to I) Representative sections of VZV-infected skin at 7 dpi. Boxes (dotted lines) highlight the area of the epidermis with VZV lesions near a needle track (arrows) (G and H) or an early lesion starting to form (I). Images are representative of 2 to 3 sections per tissue sample from 2 to 3 different tissue donors per condition.

FIG 3.

HCMV infects fibroblasts, endothelial cells, and hematopoietic cells in the dermis. Skin tissue was collected at 10 dpi, frozen in OCT on dry ice, sectioned, and stained for immunofluorescence. The antibody targets, GFP reporter, and DNA stain are indicated in their corresponding font color. Skin was stained with Hoechst (nuclei) and with antibodies to vimentin (fibroblasts [A and D to H]), CD34 (endothelial and hematopoietic progenitor cells [B and I to N]), K15 (basal cells of epidermis and hair follicles [B and I to N]), and CD45 (differentiated hematopoietic cells [C and O to S]). (A to C) Representative images of mock-infected skin. (D to S) Representative images of HCMV-infected skin. HCMV-infected cells were detected by GFP, and white arrows indicate cells with overlapping GFP and cell marker signals. Individual antibody targets are shown in each panel and then merged together, with the final panel showing an enlargement of the region indicated by the dotted square. Images are representative of multiple sections per tissue sample from 2 different tissue donors.

HCMV-infected cells migrate out of the skin and can transfer infection.

The adult human skin organ culture model was permissive for VZV and HCMV, so we further examined the infected cell types and their capacity to transfer infection. Since studying HCMV in skin is novel, analyzing the cells that migrate out of infected skin is necessary to fully understand this model. We employed crawl-out assays to characterize the immune environment and infected cell types in skin (33–36). Skin was infected by scarification (VZV-BAC-Luc) or intradermal inoculation (HCMV-fLuc-eGFP) or mock infected with medium alone. Infected or control skin was then placed in complete RPMI medium for 3 days. Cells that crawled out were collected in the supernatant and analyzed by flow cytometry. Cells were stained for CD3 (T cells), CD11c (dendritic cells), CD14 (macrophages), CD19 (B cells), CD45 (hematopoietic cells), and CD56 (NK cells). GFP expression marked infected cells. Stained cells were run through a comprehensive flow panel (Fig. 4A). There was no significance in the distribution of cell types and total number that crawled out of HCMV- versus mock-infected skin (Fig. 4B). The majority of HCMV-infected cells (GFP⁺) were CD45⁺ and CD11C⁺ (Fig. 4C; CD45, P = 0.034; CD11c, P = 0.0095; Holm-Sidak multiple t test). Additionally, HCMV also infected a large proportion of CD14⁺ cells (Fig. 4C; CD14, P = 0.016; Holm-Sidak multiple t test). Interestingly, few CD19⁺, CD3⁺, or CD56⁺ cells were infected with HCMV, and no significant difference was observed between GFP⁺ and GFP⁻ populations, suggesting that HCMV did not infect B cells, T cells, or NK cells. Infection with VZV-BAC-Luc produced a similar profile of the cells that crawled out, suggesting that HCMV was not stimulating specific infected cells to crawl out. However, in contrast to HCMV, few cells that crawled out of the skin were VZV infected (GFP⁺) and could not be identified by flow cytometry (data not shown). This may reflect the different routes of inoculation or less widespread VZV infection in the dermis at this time after inoculation.

FIG 4.

Flow cytometry analysis of HCMV-infected skin cells. Cells that crawled out of HCMV- or mock-infected skin were collected at 3 dpi and stained for flow cytometric analysis. (A) Cells were gated on forward scatter area (FSC-A) and side scatter area (SSC-A) to exclude cell debris, and then single cells were gated based on FSC-A and forward scatter height (FSC-H) to exclude doublets. Cells were analyzed for GFP expression (marker for HCMV infection) and hematopoietic cell markers CD45 (all hematopoietic cells), CD11c (myeloid lineage, dendritic cells, some macrophages/monocytes), CD14 (myeloid lineage, monocytes/macrophages, some dendritic cells, neutrophils), CD19 (B cells), CD56 (NK cells), and CD3 (T cells). CD3⁺ cells were further analyzed (bottom) to exclude T cell/monocyte complexes, which were abundant in the GFP⁺ population, by excluding CD3⁺ CD14⁺ cells. Gates were drawn based on unstained controls and marker expression on mutually exclusive populations. Compensation controls were peripheral blood mononuclear cells (PBMCs). (B) There were no statistical differences in the types of hematopoietic cells that crawled out of skin that was infected with HCMV (black bars) or mock infected (gray bars). Each bar is the mean + standard deviations (SD) from 3 separate experiments. (C) The proportion of each cell type that expressed GFP (black bars) or not (GFP⁻; gray bars) that crawled out from HCMV-infected skin. Each bar is the mean + SD from 3 separate experiments, each performed with different donor tissues (*, P < 0.05; **, P < 0.01; Holm-Sidak multiple t test).

The GFP gene in this recombinant HCMV strain is under the control of the constitutive simian virus 40 (SV40) promoter. Thus, it was possible that HCMV-infected cells detected by flow cytometry expressed GFP but were not fully permissive for virus replication. To determine whether cells were productively infected, GFP⁺ cells were sorted by fluorescence-activated cell sorting (FACS) (Fig. 5A). Single cells were first gated on CD14 (myeloid cells) and second on GFP. CD14⁺ GFP⁺ cells (HCMV-infected) and CD14⁺ GFP⁻ (bystander) cells were separated by FACS and then counted. The same number of CD14⁺ GFP⁺ or CD14⁺ GFP⁻ cells were added to cultured HEL monolayers for 3 days, and 4-fold dilution series were performed. As expected, CD14⁺ GFP⁻ cells did not transfer infection after 3 days to the fibroblasts (data not shown). CD14⁺ GFP⁺ cells efficiently transferred infection to fibroblasts after 3 days (Fig. 5C), as detected by GFP⁺ foci. There was a direct correlation between the number of CD14⁺ GFP⁺ cells added to the fibroblast culture and the number of infected fibroblasts (R² = 0.9935 by Pearson correlation) (Fig. 5B). Thus, infected cells that crawled out of the tissue can transfer infection to other cell types, suggesting that these cells can disseminate HCMV. Further studies are ongoing to address the state of infection within the infected crawled-out cells and the mechanism for the transfer of infection between these cell types.

FIG 5.

HCMV infectivity is transferred by hematopoietic cells in skin. Cells that crawled out of HCMV- or mock-infected skin were collected after 3 dpi and stained for flow cytometry analysis (see Fig. 4 for representative gating strategy). (A) Representative gating strategy for sorting CD14⁺ and GFP^+/− cells that crawled out from HCMV-infected skin. (B) The relationship between the number of CD14⁺ GFP⁺ cells added to HEL cells and the number of infected foci (GFP⁺) after 3 days of coculture was linear (R² = 0.9935, Pearson correlation). (C) Representative images of HCMV-infected HEL cells. CD14⁺ GFP⁺ sorted cells (from gating strategy used for panel A) were plated on a monolayer of HEL cells and imaged after 3 dpi. Data are representative of 2 experiments, with different donor tissues.

HCMV and VZV infect adult skin xenografts in mice and infected cells migrate.

Since both HCMV and VZV successfully infect adult human skin in organ culture, we investigated viral replication in adult skin xenografts in mice to develop an efficient small-animal model. We previously developed a SCID-hu mouse with fetal skin xenografts to study VZV growth and antiviral effects in vivo (12), which we adapted to develop a novel adult skin xenograft mouse model. Skin was prepared in the same manner as that for organ culture and implanted subcutaneously in scid-beige or athymic nude mice. Bilateral xenografts, one piece above each flank, were placed with the epidermis side against the mouse abdominal wall and then engrafted for 3 to 4 weeks. In a second survival surgery, the left xenograft was inoculated by intraxenograft injection; the right xenograft was not manipulated. The experimental groups included mock (medium injected), HCMV-fLuc-eGFP, and VZV-ORF57-Luc. After the infection period, mice were euthanized and both xenografts were removed, fixed, and prepared for immunohistochemistry (IHC). VZV-infected xenografts were harvested on day 10, when virus spread reaches maximum, and HCMV-infected xenografts were harvested on day 21 to account for the longer replication cycle of this virus. No infected cells were detected in the mock-inoculated xenografts (Fig. 6C). In the inoculated xenografts, both HCMV-infected cells (Fig. 6A, anti-GFP) and VZV-infected cells (Fig. 6B, anti-firefly luciferase) were readily detected. Areas of infection were widespread in the skin. Similar to the results in skin organ culture, VZV infection was observed primarily in the epidermis, while HCMV infection was found in the dermis. These studies demonstrate that adult human skin xenografts can provide a biologically relevant and sustainable site for active lytic HCMV and VZV infection.

FIG 6.

HCMV and VZV dissemination to the contralateral xenograft in a SCID-hu mouse model. Mice were implanted with bilateral subcutaneous xenografts on each flank. After engraftment, the left xenograft was inoculated with virus and mice were monitored for 10 to 21 dpi. The xenografts were removed, fixed, and processed for IHC. (A and A′) Representative IHC sections of the HCMV-inoculated (A) and uninoculated (A′) xenografts at 21 dpi. Tissues were stained with anti-GFP to detect HCMV. (B, B′) Representative IHC sections of the VZV-inoculated (B) and uninoculated (B′) xenografts at 10 dpi. Tissues were stained with anti-firefly luciferase to detect VZV. (C) Representative IHC section of a control-inoculated (media) xenograft at 21 dpi. Tissues were stained with anti-GFP. Insets show an enlargement of the area indicated by dotted lines. Shown are representative images of multiple experiments with multiple tissue donors and n = 4 to 6 mice per group.

A recent study showed that hematopoietic cells from human skin xenografts migrate out, circulate throughout the mouse, and home to distal grafts of engineered skin (37). Thus, we investigated whether virus-infected cells could migrate from the inoculated left xenograft to the uninoculated right xenograft. Using samples from the mouse experiments described above, right xenografts were removed, fixed, and prepared for IHC. In both assays, we detected individual infected cells in the uninoculated xenograft. HCMV-infected cells were rare and scattered throughout the dermis (Fig. 6A′). VZV-infected cells were more plentiful and were clustered around a putative capillary (Fig. 6B′). These results demonstrate that infected cells circulated in the mouse and disseminated the infection to human tissue.

Skin xenografts are only infected by direct intraxenograft injection.

It was possible that the virus-infected cells in the uninoculated, contralateral xenograft were due to spillover during the inoculation of the left xenograft. To address this question, we inoculated HCMV-fLuc-eGFP into mice with bilateral skin xenografts via several routes and then analyzed the xenografts for infectious virus and infected cells. The groups included mock infection (medium) by intraxenograft injection, cell-free HCMV by intraxenograft injection, cell-associated HCMV by intraperitoneal (i.p.) injection, or cell-free HCMV by subcutaneous (s.c.) injection. Mice were monitored for 14 days postinoculation and then euthanized. The xenografts were removed and divided. One-half was fixed for IHC; infectious virus was extracted from the other half and quantified by the 50% tissue culture infective dose (TCID₅₀). Virus-infected cells were not detected in the mock-infected mice (Fig. 7A), and no infectious virus was recovered (Fig. 7E). There was widespread infection in the left xenograft inoculated by direct injection of cell-free HCMV, the standard inoculation protocol (Fig. 7B), and infectious virus was measurable (Fig. 7E). Conversely, no virus-infected cells were detected in the xenograft when the mice were inoculated by i.p. or s.c. injection (Fig. 7C and D), and no measurable infectious virus was recovered (Fig. 7E). Thus, intraxenograft inoculation with cell-free HCMV resulted in significantly more infectious virus than the other routes (P < 0.01 by one-way analysis of variance [ANOVA]) (Fig. 7E). Therefore, spillover virus in the mouse body cavity or subcutaneous space did not efficiently infect the xenografts. These results suggest that transfer of infection from the inoculated to uninoculated xenograft is mediated by dissemination of infected cells.

FIG 7.

HCMV infects tissue only through direct injection of the xenograft. Mice were implanted with bilateral subcutaneous xenografts. (A and B) After engraftment, the left xenograft was inoculated with media (A) or cell-free HCMV (B). (C and D) Other mice received an i.p. injection of cell-associated HCMV (C) or s.c. injection of cell-free HCMV (D). Mice were monitored for 14 dpi and euthanized, and xenografts were harvested. Xenografts were divided and either fixed for IHC analysis using an antibody to GFP to detect HCMV-infected cells (A to D) or processed for virus titration (E). The HCMV intraxenograft-inoculated tissue was the only tissue with measurable infectious virus (*, P < 0.01, one-way ANOVA with nonparametric Kruskal-Wallis test and post hoc Dunn’s test). n = 2 to 12 mice per group.

DISCUSSION

Adult human skin is a useful tissue to study human herpesviruses, particularly VZV and HCMV. It is particularly advantageous as it fills a need for models of acute, lytic infections with HCMV and improves on VZV and HCMV models that primarily rely on fetal tissues. VZV is known for causing herpetic lesions in the skin, during both primary infection and reactivation. It was unknown whether HCMV could infect human skin. We were intrigued by several case studies that reported HCMV in ulcerative skin lesions (21–23), so we investigated the use of adult human skin as a model to study VZV and HCMV. Furthermore, HCMV is known to infect cell types found in the skin (18, 19). Using the skin model, we showed that both VZV and HCMV caused a robust infection in adult human skin in culture. Furthermore, we confirmed the location of VZV infections in the skin and, for the first time, showed that HCMV infected adult human skin under experimental conditions. The VZV lesions were typical and localized to the epidermis. Interestingly, HCMV-infected cells were restricted to the dermis. This was unexpected, since Langerhans cells (LCs) are dendritic cells specific to the epidermis and mucosal epithelium (25). HCMV is known to infect dendritic cells (reviewed in reference 38), including LCs; however, there is some evidence to suggest this is strain specific (19, 39). HCMV TB40/E infects mature LCs in culture at a higher rate than HCMV AD169 (39). Although we used a modified TB40/E strain in this study, we did not observe HCMV-infected LCs in the epidermis. A person’s age affects the density and morphological characteristics of LCs (40), yet this donor information was not available to us. Infected LCs may have been too rare in the skin to detect by immunofluorescence. Furthermore, age is already known to affect VZV infections in the skin. Another research group showed that VZV replication increased in skin from older donors, likely due to changes in the skin morphology and innate antiviral response (15). Similar to VZV, donor age may also affect HCMV infection in skin. Thus, it is possible that HCMV infects LCs of the epidermis, but we were unable to detect them in the present study due to their distribution and the age of the skin donor.

The crawl-out assay was an efficient method to identify infected cell types in a human skin model. We identified hematopoietic cells that were permissive to HCMV infection, as expected. In our HCMV infection model, approximately 3 × 10⁴ to 3 × 10⁵ cells per 1-cm² piece of skin crawled out over a 3-day period. These numbers are variable and most likely due to donor variability. We observed that an average of 10.8% of the crawl-out cells were HCMV infected. Interestingly, HCMV infection did not induce the migration of any particular population of hematopoietic cells to crawl out of the skin, as we did not see a significant difference in percentages of crawled-out cell populations between mock- and HCMV-infected skin. However, we did observe that cells of the myeloid lineage were the predominant HCMV-infected hematopoietic cell type, expressing CD11c and CD14. This result is encouraging, as myeloid cells, particularly monocytes, are the cell predominantly involved in dissemination of HCMV in humans and a different humanized mouse model (41–44). Important for dissemination, in vitro HCMV infection of peripheral monocytes induces survival, differentiation, and motility of monocytes (32, 45, 46). Additionally, we also found a higher population of CD3⁺ GFP⁺ cells, which was unexpected, as HCMV is not known to infect T cells. However, upon further examination, the CD3⁺ GFP⁺ cells were also CD14⁺, suggesting that these were actually T cells responding to an infected monocyte. Such cell-cell interactions have been reported before and have been reported to increase with viral infections (47). Thus, we removed this subset of CD3⁺ cells, resulting in very few infected T cells, as was originally expected. However, more studies are necessary to fully understand the phenotype and the effects of HCMV infection on these specific cells derived from the skin, including the various interactions between cell types.

Interestingly, we were unable to analyze VZV-infected cells in the crawl-out assay. Too few VZV-infected cells migrated out of the tissue to analyze by flow cytometry or to sort cells. We were also unable to identify VZV-infected cells by immunofluorescence (data not shown). There are several possible explanations for the rarity of infected cells. First, VZV mainly infects cells of the epidermis (48–50), which are immobile. VZV lesions eventually break through the basement membrane and spread to the dermis, which takes a week or more. The crawl-out assay described here was performed over 3 days. Harvesting crawled-out cells for 14 days could accumulate enough VZV-infected cells to analyze. VZV infection can spread to the skin through T cells (49), so it is possible that T cells in the skin become infected and then migrate out of the tissue. T cells are abundant in the skin, both in the dermis and epidermis (27, 51). While CD3⁺ T cells were measurable in the samples of crawled-out cells after 3 days, very few were VZV-infected CD3⁺ cells. Again, this may be due to the length of time of the assay. Future studies are needed to fully analyze VZV-infected cells in skin and the time frame for migration of infected cells out of the tissue. Such studies would further advance the understanding of VZV biology, including dissemination of infection. Additional studies are also needed to fully characterize the effects of HCMV infection on immune cells in a relevant human tissue system. These projects are under way.

Human herpesviruses are well known for their capacity to disseminate and establish a latent reservoir that may reactivate. Here, we showed that infected cells migrated out of the tissue and transferred infection to other cells in vitro. Furthermore, this was recapitulated in a humanized mouse model. Recently, a skin immunology group found that cells from a human skin xenograft migrated out of the tissue and were found in several mouse organs, including the spleen (37). The cells from one xenograft circulated through the mouse and extravasated into another xenograft composed of engineered skin that originally lacked human hematopoietic cells (37). Their results support our findings that VZV- and HCMV-infected cells were found in some contralateral xenografts. It is important to note that infected cells were not detected in every contralateral xenograft. Only a portion of each xenograft was sectioned, so small foci of infection may have been missed. Single, HCMV-infected cells were detected in contralateral xenografts, whereas VZV foci were clusters of infected cells. This corresponds to their distribution in skin tissues that were directly inoculated; VZV lesions in skin organ culture are clusters of infected cells, while HCMV-infected cells are scattered widely. Additionally, VZV replicates faster than HCMV, which may also account for this difference. It is possible that viral foci would expand over time. This skin xenograft model will be useful to address viral dissemination in future studies. It is important to determine the characteristics of circulating human cells, both infected and uninfected. Several questions remain to fully understand this model. It is important to know how VZV affects cell migration out of the skin, as well as how both VZV and HCMV affect cell migration in a mouse model. Additionally, understanding the timing of cell migration events will help to fully validate this model system.

The existing systems to study VZV and HCMV are complicated, expensive, and/or use fetal tissues. Some of the current mouse models for HCMV include mice engrafted with fetal lung or thymus-liver, gelfoam plugs of fibroblasts, or immunocompromised mice engrafted with hematopoietic cells (reviewed in reference 6), and the models are designed for long-lasting, latent infections. Alternatively, MCMV (murine cytomegalovirus) models fill gaps in HCMV mouse models. Current VZV models have similar drawbacks (7, 8). The advantages of the model presented here are that it is straightforward, yields quick results, is relatively inexpensive, and uses adult tissue. For HCMV, this is especially advantageous, as it fills a need for an acute, lytic infection model. Recently, another laboratory showed that adult human skin can be infected with VZV in a humanized mouse model (15). The push to develop new models using only adult tissue is an important and necessary move for the field. The continued development of current and novel model systems is necessary to understand viral pathogenesis and the continued development of new antiviral drugs.

MATERIALS AND METHODS

Virus and cell propagation.

Human foreskin fibroblasts (HFFs; CCD-1137Sk; American Type Culture Collection [ATCC], Manassas, Virginia), retinal pigment epithelial cells (ARPE-19; CRL-2302; ATCC), and human lung fibroblasts (HEL-299s [HEL]; CCL-137; ATCC) were used. All cells were grown in Dulbecco’s modified Eagle medium (DMEM) with 4.5 g/liter glucose, l-glutamine, and sodium pyruvate (1× DMEM; Corning, Manassas, VA), supplemented with 10% heat-inactivated fetal bovine serum (Benchmark FBS; Gemini Bio Products, West Sacramento, CA), penicillin-streptomycin (5,000 IU/ml), and amphotericin B (250 μg/ml). HFFs and HELs were used prior to passage 20 and 18, respectively. ARPE-19 cells can be grown for 36+ passages.

HCMV-fLuc-eGFP was engineered by Christine O’Connor (Cleveland Clinic, Cleveland, OH) from the TB40/E strain, a variant of the clinical isolate TB40 (52, 53), and generously provided by Eain Murphy (SUNY Upstate Medical University, Syracuse, NY). To construct HCMV-fLuc-eGFP, eGFP was inserted between US34 and TRS1 under the control of the SV40 promoter. Firefly luciferase replaced the nonessential glycoprotein open reading frame (ORF) UL18 and is driven by the UL18 true late promoter. The eGFP and fLuc insertions were validated by PCR amplification and sequencing. HCMV-fLuc-eGFP was originally grown from a BAC, further propagated on HELs for up to 5 passages, and then stored at –80°C. VZV-BAC-Luc, from the parental Oka strain (54), was kindly provided by Hua Zhu (University of Medicine and Dentistry of New Jersey, Newark, NJ). VZV-ORF57-Luc was also derived from a parental Oka Strain using a different BAC (55) that was corrected for several mutations in the BAC DNA to the parental Oka genome sequence. Firefly luciferase was inserted after the ORF57 coding sequence and in frame with ORF57, separated by a T2A ribosomal skip sequence. The BAC DNA insertion was confirmed by PCR amplification across the T2A-fLuc insertion and ORF57 and sequenced. The fragments in the BAC and virus were confirmed by restriction fragment length polymorphism analysis. All genetic changes were validated, and the entire sequence of the VZV-ORF57-Luc BAC was obtained and is available from P. R. Kinchington upon request. No differences from the POka corrected BAC DNA were noted except for the T2A-fLuc insertion (P. R. Kinchington and J. F. Moffat, unpublished data). VZV-BAC-Luc and VZV-ORF57-Luc were both grown on HFFs for up to 10 passages and stored at –80°C.

Preparation of skin.

Adult human breast skin was obtained from reduction mammoplasty surgeries in accordance with approved and established Institutional Review Board protocols and procedures at SUNY Upstate Medical University in Syracuse, NY. All individuals donating tissue were healthy adults over 18 years old and gave consent per approved protocol (SUNY Upstate, Institutional Review Board number 1140572). For each experiment, 1 or 2 skin specimens were collected in saline, obtained within 2 h of surgery, and immediately processed for use (56). Whole-thickness skin tissue was cleaned with povidone-iodine and 70% ethanol to remove excess debris and washed with tissue culture media. Skin was stretched tight across a Teflon-covered foam cylinder, secured with pins, and thinned using a Weck, or skin grafting, knife and a Goulian guard to cut skin to 0.028 inches thick (approximately 700 μm) (56). The process of thinning the skin removes much of the underlying adipose and subdermal layer, leaving the epidermis and dermis intact. Thinned skin was cut into approximately 1-cm² pieces and cultured on NetWells for skin organ culture studies (Corning, NY), in 6-well plates for crawl-out assays or implanted into mice.

Crawl-out assay.

Skin was inoculated with VZV-BAC-Luc (1 × 10⁴ to 1 × 10⁵ PFU/ml), HCMV-fLuc-eGFP (5 × 10⁶ PFU/piece), or medium by intradermal injection with a 27-gauge needle. Skin was cultured in RPMI 1640 supplemented with 4% heat-inactivated fetal bovine serum, penicillin-streptomycin, and amphotericin B. Skin was cut into approximately 2- by 2-mm² pieces 24 h postinoculation. Supernatant was collected 3 days postinoculation (33). Crawled-out cells were analyzed by flow cytometry. Antibodies used in the flow panel were all obtained from BioLegend (San Diego, CA) and include CD19-allophycocyanin (clone HIB19), CD3-peridinin chlorophyll protein (clone SK7), CD11c-phycoerythrin (clone 3.9), CD14-BV711 (clone M5E2), CD45-BV605 (clone HI30), and CD56-PacBlue (clone MEM-188). Infected cells were detected by GFP expression. CD14⁺ GFP⁺ and CD14⁺ GFP⁻ cells were sorted from the crawled-out cells and added to uninfected HEL cell monolayers. Cells were added to the monolayers in a 4-fold dilution series, starting with 7,577 infected cells in the first well. At 3 dpi, the cells were analyzed for GFP expression on a Nikon Eclipse Ti-U fluorescence microscope (Melville, NY). The number of infected cells was counted at each dilution.

Skin organ culture.

Skin was inoculated with cell-free VZV-BAC-Luc (1 × 10⁴ to 1 × 10⁵ PFU/ml; 30-μl inoculum; cell associated in HFFs) by scarification or HCMV-fLuc-eGFP by intradermal injection (5 × 10⁶ PFU/piece) with a 27-gauge needle. Mock-infected skin was inoculated with medium by intradermal injection. Skin was cultured on NetWells with DMEM supplemented with 4% heat-inactivated fetal bovine serum, penicillin-streptomycin, and amphotericin B. Skin was collected 7 to 10 days postinoculation and either fixed in 4% paraformaldehyde and stored at 4°C or frozen in OCT compound embedding medium (Fisher Healthcare, Fisher Scientific Co., LLC, Pittsburgh, PA) on dry ice and stored at –80°C.

Animal procedures.

Human adult skin was introduced subcutaneously into 5- to 6-week-old, male scid-beige (CB.17; CB17/Icr-Prkdc^scid/IcrIcoCrl) or athymic nude [CR ATH Ho; Crl:NU(NCr)-Foxn1^nu] mice (Charles River, Wilmington, MA). Skin thinned with the Weck knife was introduced as full-thickness grafts as previously described with bilateral implants (29). Three to 4 weeks after implantation, only the left xenografts were inoculated by intradermal injection with cell-associated VZV-ORF57-Luc (in HFFs) (12), cell-free HCMV-fLuc-eGFP (1 × 10⁶ viral particles/xenograft), or medium. Alternatively, mice were inoculated with cell-free HCMV-fLuc-eGFP (1 × 10⁶ viral particles/xenograft) by subcutaneous injection between xenografts or with cell-associated HCMV-fLuc-eGFP by intraperitoneal injection. Virus presence was monitored by bioluminescence imaging with the IVIS 200. Mice were euthanized between 14 and 28 days postinoculation, and xenografts were harvested immediately. Xenografts were fixed in 4% paraformaldehyde for future histological analysis or minced and prepared in medium for virus titration. This protocol was reviewed and approved by the Institutional Animal Care and Use Committee at SUNY Upstate Medical University.

Histopathological analysis.

Skin fixed in paraformaldehyde was sent to a commercial laboratory (histowiz.com; New York, NY) for processing by a standard operating procedure and fully automated workflow. Samples were processed, embedded in paraffin, and sectioned at 5 μm. Immunohistochemistry was performed on a Bond Rx autostainer (Leica Biosystems) with enzyme treatment (1:1,000) using standard protocols. Antibodies used were firefly luciferase primary antibody (AB21176; 1:1,000; eBioscience) or GFP primary antibody (AB183734; 1:100; eBioscience) and a Leica DAB rabbit secondary polymer kit. Antigen retrieval was performed at pH 6 for 20 min. Bond polymer refine detection (Leica Biosystems) was used according to the manufacturer’s protocol. After staining, sections were dehydrated and film coverslipped using a TissueTek-Prisma and Coverslipper (Sakura). Whole slide scanning (40×) was performed on an Aperio AT2 (Leica Biosystems). Hematoxylin and eosin staining was also performed per standard operating procedure.

Immunofluorescence analysis.

Skin was embedded in OCT and flash frozen. Blocks were sectioned at 7 μm on a Leica cryostat. Sections were placed on plus-charged SureFrost slides (Fisher Scientific Co., LLC, Pittsburgh, PA), dried for 20 min, and stored at –80°C. Sections were thawed, fixed in a 75% acetone–25% ethanol solution for 10 min, and rinsed with phosphate-buffered saline (PBS). Tissues were blocked with either 5% normal mouse or goat serum in PBS. Antibodies used were vimentin (AB194719; Alexa Flour 647; 1:200; Abcam), cytokeratin 15 primary antibody (AB80522; 1:100; Abcam), CD45 (number 562279; 1:100; BD), CD34 (AB81289; 1:50; Abcam), and Hoechst 33342 (number A-11008; 1:200; ThermoFisher). The secondary antibody used was goat anti-mouse (number A32727; Alexa Fluor 555; 1:200; Invitrogen) (used with K15 primary) or goat anti-rabbit (number A21245; Alexa Fluor 647; 1:200; Invitrogen) (used with CD34 primary). GFP signal was used to detect the presence of HCMV-fLuc-eGFP. Primary antibodies were diluted in blocking serum and incubated with tissues for 2 h before rinsing. If applicable, secondary antibody was applied to tissue for 1 h before rinsing. After staining, sections were coverslipped using Aqua polymount aqueous mounting medium (Polysciences Inc., Warrington, PA). Slides were imaged with a Leica SP5 confocal microscope with acousto-optical beam splitter (AOBS) and excitation lasers at 405, 488, 543, 594, and 633 nm. Band pass filters were adjusted for the appropriate settings associated with each fluorochrome used. Scale bars were added to each image using ImageJ software.

Virus extraction and quantification from skin xenografts.

Xenografts were removed from mice, minced, and placed in medium in Eppendorf tubes with glass beads. The mixture was put through three rapid freeze-thaw cycles with a 30-s vortex between freezes. After the final thaw, the mixture was vortexed for an additional 2 min and then centrifuged at low speeds for 5 min to concentrate the cellular and tissue debris. The supernatant was collected, aliquoted, and stored at –80°C for future analysis. Supernatant was thawed and applied to HFFs in 96-well plates per the TCID₅₀ standard protocol. Infectious virus was determined by the presence of GFP signal 7 to 10 days postinoculation.

Statistics.

Data from virus quantification was analyzed using a one-way ANOVA with a nonparametric Kruskal-Wallis test and post hoc Dunn’s test. Data from sorted cells in the crawl-out assay were analyzed using the Holm-Sidak multiple t test method or a linear regression analysis with a Pearson correlation. Calculations and graphs were made using GraphPad Prism (GraphPad Software, San Diego, CA; www.graphpad.com). A P value of <0.01 was considered statistically significant.