Age-Associated Differences in Infection of Human Skin in the SCID Mouse Model of Varicella-Zoster Virus Pathogenesis

ABSTRACT

Varicella-zoster virus (VZV) is the skin-tropic human alphaherpesvirus responsible for both varicella-zoster and herpes zoster. Varicella-zoster and herpes zoster skin lesions have similar morphologies, but herpes zoster occurs disproportionally in older individuals and is often associated with a more extensive local rash and severe zoster-related neuralgia. We hypothesized that skin aging could also influence the outcome of the anterograde axonal transport of VZV to skin. We utilized human skin xenografts maintained in immunodeficient (SCID) mice to study VZV-induced skin pathology in vivo in fetal and adult skin xenografts. Here we found that VZV replication is enhanced in skin from older compared to younger adults, correlating with clinical observations. In addition to measures of VZV infection, we examined the expression of type I interferon (IFN) pathway components in adult skin and investigated elements of the cutaneous proliferative and inflammatory response to VZV infection in vivo. Our results demonstrated that VZV infection of adult skin triggers intrinsic IFN-mediated responses such as we have described in VZV-infected fetal skin xenografts, including MxA as well as promyelocytic leukemia protein (PML), in skin cells surrounding lesions. Further, we observed that VZV elicited altered cell signaling and proliferative and inflammatory responses that are involved in wound healing, driven by follicular stem cells. These cellular changes are consistent with VZV-induced activation of STAT3 and suggest that VZV exploits the wound healing process to ensure efficient delivery of the virus to keratinocytes. Adult skin xenografts offer an approach to further investigate VZV-induced skin pathologies in vivo.

IMPORTANCE Varicella-zoster virus (VZV) is the agent responsible for both varicella-zoster and herpes zoster. Herpes zoster occurs disproportionally in older individuals and is often associated with a more extensive local rash and severe zoster-related neuralgia. To examine the effect of skin aging on VZV skin lesions, we utilized fetal and adult human skin xenografts maintained in immunodeficient (SCID) mice. We measured VZV-induced skin pathology, examined the expression of type I interferon (IFN) pathway components in adult skin, and investigated elements of the cutaneous proliferative and inflammatory response to VZV infection in vivo. Our results demonstrate that characteristics of aging skin are preserved in xenografts; that VZV replication is enhanced in skin from older compared to younger adults, correlating with clinical observations; and that VZV infection elicits altered cell signaling and inflammatory responses. Adult skin xenografts offer an approach to further investigate VZV-induced skin pathologies in vivo.

INTRODUCTION

Varicella-zoster virus (VZV) is the skin-tropic human alphaherpesvirus responsible for both varicella-zoster and herpes zoster. VZV infection begins with inoculation of upper respiratory epithelial cells, which is followed by viral amplification in pharyngeal lymphoid tissue and T-cell-mediated hematogenous spread to the skin (1–4). Skin homing T cells exit the dermal vasculature to transfer VZV particles, initiating a 10-to-21-day replicative interval that culminates in the emergence of the characteristic vesicular “chickenpox” skin rash (1, 5). Hematogenous spread by infected T cells and access to the dermal nerve plexus during skin replication are both potential routes to the sensory ganglia where lifelong VZV latency is established in nerve cell bodies (6). Reactivation of latent VZV genomes resulting in the production of infectious particles and their anterograde transport to cutaneous nerve endings is necessary to cause the zoster rash. Varicella-zoster and herpes zoster skin lesions have similar morphologies, but the zoster rash typically occurs within a dermatomal distribution and is accompanied by nerve pain.

Herpes zoster occurs disproportionally in older individuals and is often associated with a more extensive local rash and severe zoster-related neuralgia (7). The state of VZV during latency and, in particular, the molecular events that trigger reactivation are not well understood. Age-related waning of VZV-specific T-cell immunity correlates with increased frequency of zoster (8), indicating that a robust VZV-specific T-cell-mediated immune response is critical to prevent episodes of VZV reactivation from causing the clinical manifestations of zoster. Cell-mediated immune responses are detected in sensory ganglia during zoster episodes, but whether these responses can block VZV reactivations locally within the sensory ganglia or act primarily to prevent symptoms at skin sites of replication or both is not known (9). Immunization to boost VZV-specific memory T-cell populations with live attenuated, inactivated VZV or subunit glycoprotein E vaccines reduces the incidence of shingles in the elderly and in immunocompromised patients (10, 11).

While the rising incidence of zoster by decade after age 50 is related to the decline in VZV-specific memory T-cell frequencies, possibly together with other changes, such as senescence and/or exhaustion in circulating or skin-resident memory T cells (T_RMs), we hypothesized that skin aging could also influence the outcome of the anterograde axonal transport of VZV to skin. Skin aging is a complex and combinatorial process that is influenced by intrinsic genetic and extrinsic environmental factors (12–14). Skin is comprised of two morphologically distinguishable compartments separated at the dermal-epidermal junction (DEJ) by the basement membrane (BM). Basal stem cells of the epidermis attach to the upper side of the basement membrane at the basal lamina, functioning as an essential barrier through the constant renewing, stratifying, and keratinizing of the epidermis. The dermal compartment consists of the uppermost papillary dermis and the deeper reticular dermis; both are primarily comprised of a connective cross-linked extracellular matrix (ECM) of collagens, elastins, and other proteins that are produced and secreted by dermal fibroblasts. The dermis also contains the blood vessels needed to nourish the avascular epidermis by diffusion and lymphatic vessels and sensory nerve receptors, as well as epidermal appendages that project into the dermal compartment such as the rete ridges and pilosebaceous structures (hair follicles and sebaceous glands).

Intrinsic skin aging, also called chronological aging, is associated with decreased cellular metabolism, reduced rates of cell proliferation, decreased microvasculature, miniaturization of hair follicles, and a gradual decline in the production of the dermal matrix (15–20). These degenerative changes alter the composition of the dermal ECM and contribute to skin laxity and fragility. At the molecular level, age-related changes to skin include alterations in innate immune signaling, inflammation, tumorigenesis, and wound healing (21, 22) that might facilitate VZV replication in aging skin.

Human skin xenografts maintained in immunodeficient (SCID) mice offer a strategy to study VZV-induced skin pathology in vivo. Innate control of the infectious process by skin cell responses can be evaluated in this model since SCID mice lack adaptive immunity. Within 10 to 21 days after VZV inoculation of human fetal skin xenografts, vesicular skin lesions are formed that recapitulate patterns observed in biopsy specimens of varicella lesions (23). Here, we extended the model to evaluate VZV pathogenesis in adult skin. We first compared the morphological phenotypes of fetal and adult skin xenografts to assess patterns of tissue engraftment, vascularization, and cell composition. We also evaluated an alternative immunodeficient mouse strain specifically engineered to tolerate xenograft transplantation (24). Data corresponding to VZV replication and lesion formation were compared in fetal and adult skin xenografts using tissue specimens from individuals who were under or over 55 years of age. In addition to measures of VZV infection, we examined the expression of type I interferon (IFN) pathway components in adult skin, since our previous work had documented the critical role of IFN-mediated antiviral responses (3), and we investigated elements of the cutaneous proliferative and inflammatory response to VZV infection in vivo.

RESULTS

Comparison of the tissue characteristics of fetal and adult human skin xenografts.

Examined 5 weeks after implantation in the absence of VZV infection, human fetal skin grafts had the characteristics of healthy skin tissue that we have observed previously (23, 25), as shown in representative sections (Fig. 1). Examination of hematoxylin and eosin (H&E)-stained fetal skin sections at low magnification (Fig. 1A) and higher magnification (Fig. 1B) demonstrated an organized and dense dermis (Fig. 1A and B, large arrowheads) separated from the epidermis at the DEJ by a prominent epidermal rete ridge (Fig. 1A and B, small arrowheads). The epidermis was composed of a compact basal cell layer and a multilayered stratified squamous epithelium terminating with the acellular corneum. Epidermal appendages, including hair follicles, were observed extending into the dermis. While adult skin grafts retained these gross morphological features, histological changes associated with aging were also observed. Examination of H&E-stained adult skin xenograft sections at low magnification (Fig. 1C) and higher magnification (Fig. 1D) demonstrated that the dermis was less dense than fetal skin (Fig. 1C, large arrowhead) and that the stratum corneum was thicker (Fig. 1D, large arrowhead). Blood vessels and epidermal appendages were fewer and smaller in adult skin than in fetal skin (Fig. 1C, small arrowhead). Adult skin grafts contained melanocytes within the stratum basale, which is the layer of keratinocyte stem cells resting on the basal lamina (Fig. 1D, small arrowheads).

FIG 1.

Comparative tissue architectures of uninfected fetal and adult skin xenografts in SCID mice. Representative images are shown. (A) Uninfected fetal skin graft, 5 weeks after implantation, H&E staining (×20). (B) Uninfected fetal skin graft, 5 weeks after implantation, H&E staining (×200). (C) Uninfected adult skin graft, 5 weeks after implantation, H&E staining (×20). (D) Uninfected adult skin graft, 5 weeks after implantation, H&E staining (×200). (E) Uninfected fetal skin graft, anti-Ki-67 immunostaining. (F) Uninfected adult skin graft, anti-Ki-67 immunostaining. (G) Uninfected fetal skin graft, type I collagen immunostaining. (H) Uninfected adult skin graft, type I collagen immunostaining. (I) Uninfected adult skin graft, anti-PECAM/CD31 immunostaining, ×20. (J) Uninfected adult skin graft, anti-PECAM/CD31 immunostaining, ×200. For immunostaining, positive = brown (DAB staining); sections were counterstained with hematoxylin. Arrows indicate the observations described in Results.

Cell proliferation in fetal and adult skin tissue specimens was assessed using a monoclonal antibody (MAb) to Ki-67, a nuclear antigen which is expressed at low levels in quiescent cells but shows increased expression in cycling cells. In fetal skin grafts, nuclear staining by Ki-67 was intense in all cell types in both the epidermal and dermal compartments except in the anuclear corneocytes of the stratum corneum (Fig. 1E, brown staining). In adult skin grafts, Ki-67+ proliferating cells were concentrated at the basal epidermal layer, which contains the stem cells necessary for skin homeostasis that are renewed throughout life (Fig. 1F, arrowhead), but were comparatively sparse in the suprabasal keratinocyte layers and in the papillary dermis (Fig. 1F); Ki-67+ cells were absent in the reticular dermis of adult skin grafts.

The composition of the dermal ECM was assessed by staining for type 1 collagen fibrils, which comprise the bulk of the ECM in the papillary dermis (26). In the fetal skin, type 1 collagen was evenly distributed throughout the dermal connective tissue with an intensification of expression at the DEJ and along epidermal appendages and blood vessels (Fig. 1G, arrowheads). In contrast, dermal type I collagen expression was dim and patchy and was distributed unevenly in adult skin (Fig. 1H).

The viability of skin xenografts depends on revascularization of the dermis. We assessed the integrity of the adult skin vasculature using human PECAM-1+ (CD31) as a marker for blood endothelial cells of human origin. CD31⁺ immunoreactivity was detected in blood vessels at the DEJ (Fig. 1I, arrowhead) and within the dermis (Fig. 1J, arrowhead). PECAM-1+ expression in adult skin confirmed that human endothelial cells show anastomosis with murine capillaries at adult skin xenograft margins, as observed in fetal skin xenografts. These results show that adult skin segments are well vascularized, tissue architecture is maintained, and, importantly, the characteristics of skin aging are preserved after engraftment.

Comparison of VZV lesion formation in fetal and adult human skin xenografts.

The altered composition and rigidity of engrafted adult skin required modification of our standard SCIDhu methods with respect to VZV skin inoculation and the tissue homogenization. In addition, the small size of the available adult donor tissue precluded analysis of multiple time points for many specimens. Therefore, we first performed two small-scale experiments to optimize the tissue inoculation method, skin homogenization, and to determine the optimal time to recover adult skin xenografts for analysis (data not shown). On the basis of these results, all tissue specimens were inoculated with pOka (∼10⁵ PFU) in a series of intradermal injections. Examined at 14 days after inoculation, some small areas of VZV-infected foci were observed in adult skin by H&E staining, whereas fully formed vesicular lesions were observed by 24 days. Therefore, all histological comparisons were performed at day 24.

A representative fetal skin graft, taken 24 days after VZV inoculation, was examined at low and higher magnification and stained with H&E (Fig. 2A and C) and with anti-VZV serum (Fig. 2B and D). A large epidermal vesicle containing degenerating keratinocytes was observed within a field of cellular debris (Fig. 2A and C, large arrowhead). Separation of the epidermis at the spinous layer, a process called “acantholysis” that results from a loss of intercellular connections, was observed (Fig. 2A, small arrowheads). The basement membrane was thinned and discontinuous along the DEJ. VZV protein expression was intense within hair follicles and other epidermal appendages (Fig. 2D, small arrowheads). The corneum stratum layer was thicker than in the uninfected fetal skin (Fig. 1B, white arrowhead), indicating keratin overexpression (hyperkeratosis). Enlarged (hypertrophic) cells were also observed within the spinous layer of the epidermis and in hair follicles (Fig. 2C, small arrowhead).

FIG 2.

Comparative pathological changes and lesion formation in VZV-infected fetal and adult skin xenografts 24 days after infection. Representative images are shown as follows: fetal skin tissue, panels A to D; adult skin tissue, 63-year-old subject, panels E to F; adult skin tissue, 49-year-old subject, panels G to J. (A) VZV-infected fetal skin graft, H&E staining (×20). (B) VZV-infected fetal skin graft, anti-VZV immunostaining (×20). (C) VZV-infected fetal skin graft, H&E staining (×200). (D) VZV-infected fetal skin graft, anti-VZV immunostaining (×200). (E) VZV-infected adult skin graft, H&E staining (×20). (F) VZV-infected adult skin graft, anti-VZV immunostaining (×20). (G) VZV-infected adult skin graft, H&E staining (×20). (H) VZV-infected adult skin graft, anti-VZV immunostaining (×20). (I) VZV-infected adult skin graft, H&E staining (×200). (J) VZV-infected adult skin graft, anti-VZV immunostaining (×200). For immunostaining, positive = brown (DAB staining); sections were counterstained with hematoxylin. Arrows indicate the observations described in Results.

At day 24, VZV lesions in adult skin xenografts exhibited most of the features observed in fetal skin grafts. Two representative lesions from two different adult skin donors stained with H&E (Fig. 2E and G) and for VZV proteins (Fig. 2F and H) are shown. As with the VZV-induced lesions in fetal skin, epidermal vesicles were observed containing degenerating cells within a cellular debris field (Fig. 2E and G, large arrowheads). Acantholysis (Fig. 2G, small arrowhead), keratin overexpression (Fig. 2H, large arrowhead), multinucleated cells (2I, arrowhead), and ballooning cells (Fig. 2J, white arrowhead) were also observed. Notably, while the levels of VZV protein expression within the epidermal compartment were similar, the level of detection of cells expressing VZV proteins was very limited at the dermal side of the DEJ in adult skin (Fig. 2F and H, small white arrowheads), even within epidermal appendages (Fig. 2F, small black arrowheads). This restriction to the epidermal compartment created a flattened appearance of lesions in adult skin compared with those observed in fetal skin. Vesicles within the epidermis were variable in size and in the extent of their lateral spread, regardless of donor age.

Comparison of levels of viral replication in fetal and adult human skin xenografts.

Overall, levels of VZV replication were found to be equivalent in comparisons of viral titers from xenografts of all adult donors with those in fetal skin (Fig. 3). However, stratified by age, titers were significantly lower (P < 0.01) in skin from younger adults than in fetal skin, indicating that conditions were less permissive. Interestingly, VZV replication was increased significantly in skin xenografts from donors >55 years older compared with donors <55 years (P < 0.01). In contrast, titers did not differ in xenografts of fetal skin and skin from adults >55 years of age. Infectious virus was recovered from >80% of skin graft homogenates, regardless of donor age, and we did not observe any gross differences in tissue quality and lesion size between the two aged-skin cohorts. VZV replication in adults >55 years of age was highly variable; for example, virus yield from the same donor (age 63) ranged from 2,000 PFU/implant to 44,667 PFU/implant. Of note, recovery of infectious virus correlated with the detection of VZV lesions by immunostaining but, as we have reported previously, larger lesion size did not predict higher viral titers. As an example, the lesion shown in Fig. 2G represents a xenograft from a 49-year-old donor that yielded 4,800 PFU/graft, while the larger lesion shown in Fig. 2E represents a xenograft from a 63-year-old donor that yielded 4,067 PFU/graft. Levels of VZV titers and lesion formation did not differ between xenografts in C.B-17 SCID and NOG mice.

FIG 3.

Recovery of infectious virus from fetal and adult skin xenografts 24 days after infection. The plot represents viral titers calculated as PFU/graft. Each dot represents an individual skin graft titer; data represent averages of results of 3 replicates. Black bars, means. P values were determined by t test. ns, not significant.

Induction of antiviral responses in VZV-infected adult skin.

Our past work showed that VZV infection elicits type I IFN-mediated antiviral responses in fetal skin cells surrounding VZV lesions in vivo and that VZV blocks these responses within infected skin cells (1, 27, 28). Therefore, we investigated the patterns of host cell protein expression in adult skin infected with VZV, as shown in representative confocal images (Fig. 4). The type I IFN response is associated with STAT1 phosphorylation (pSTAT1), which is needed to activate the JAK/STAT pathway. Nuclear pSTAT1 was absent from VZV-infected cells in fetal skin lesions, whereas expression of pSTAT1 and IFN-alpha was increased in skin cells adjacent to lesions (1). Similarly, cytoplasmic retention of STAT1 was observed within infectious foci in VZV-infected adult skin (Fig. 4A, white arrowheads) whereas STAT1 exhibited nuclear localization that indicated phosphorylation in adjacent cells (Fig. 4A, orange arrowheads).

FIG 4.

Expression of cellular proteins in VZV-infected skin xenografts 24 days after inoculation. Tissue sections were stained with rabbit antibodies to cellular proteins (see Materials and Methods) and anti-VZV serum. (A) PML was detected using a mouse MAb; VZV protein was detected using a high-titer anti-VZV human serum (Ig fraction). Nuclei were stained with Hoechst stain, and duplicate sequential sections were imaged with a Zeiss LSM780 multiphoton laser scanning confocal microscope. Images were scanned at 1,024 by 1,024 with an 8-frame average and a pinhole size of 1 airy unit. Representative images are shown. For the data in all panels, VZV proteins were detected with secondary Alexa Fluor 594-labeled antibodies (red signal) and cellular proteins with Alexa Fluor 488-labeled antibodies (green signal) as follows: panel A, STAT1; panel B, PML; panel C, MxA; panel D, survivin; panel E, β-catenin; panel F, IL-1β. Arrows indicate the observations described in Results.

Since expression of IFN-stimulated genes (ISGs) is evidence of type I IFN activity (29), we also tested for expression of the IFN-associated proteins promyelocytic leukemia protein (PML) and MxA in VZV-infected adult skin. PML is a multifunctional IFN-regulated antiviral protein that assembles into large intranuclear structures which sequester VZV capsids in infected cells (28). Several large PML nuclear structures were observed within VZV-infected adult skin cells (Fig. 4B, white arrowheads). Mx proteins are IFN-responsive cytoplasmic GTPases that are present at basal levels in skin epithelial tissue specimens (30). MxA was expressed at low levels in uninfected cells at the periphery of VZV skin lesions (Fig. 4C, orange arrowheads) and appeared as more-intense cytoplasmic expression (Fig. 4C, white arrowheads) and in a granular pattern in degenerating keratinocytes within the vesicle debris field (Fig. 4C, yellow arrowhead). Thus, adult skin cells retained the innate immune mechanisms needed to regulate VZV infection that were observed in fetal tissue, although the possibility of quantitative differences cannot be excluded.

Induction of proliferative and inflammatory responses in VZV-infected skin.

Previously, we reported that VZV has the capacity to induce or activate host cell signaling pathways that are necessary to permit or enhance replication, including CREB activation (31) and STAT3 activation and expression of the anti-apoptotic protein survivin (32). Here, we show that expression of survivin, reflecting STAT3 activation, was also enhanced within VZV lesions in adult skin (Fig. 4D, white arrowheads), whereas cells adjacent to infectious foci did not overexpress survivin (Fig. 4D, orange arrowheads).

As noted, VZV-infected skin xenografts exhibited mild hyperkeratosis and epidermal hypertrophy. Cytokeratin 17 (K17) is an intermediate keratin filament expressed in epidermal appendages, including hair follicles and sebaceous glands, but its expression is normally absent or weak in resting keratinocytes. K17 is regulated through STAT3-dependent mechanisms, as well as by extracellular signal-regulated kinase 1 (ERK1) and ERK2 (ERK1/2) (33). Consistent with VZV-induced STAT3 activation, K17 expression was intense in VZV-infected adult skin lesions, specifically within normally resting keratinocytes in the stratum spinosum (Fig. 5A [representative image shown at ×20 magnification] and B to D [areas from panel A shown at ×200]). K17 expression (green signal) was localized to the stratum spinosum along the edge of the VZV lesion (Fig. 5B [red signal]). The K17 signal intensity was maintained in keratinocytes along the edge of the lesion (Fig. 5C) and then declined precipitously to basal levels as the distance from the lesion increased (Fig. 5D). Notably, K17 expression in epidermal appendages in uninvolved skin areas (Fig. 5A, white arrowhead) was lower than that observed in keratinocytes near the VZV skin lesion. While K17 expression in uninfected adult skin grafts (Fig. 5E) was comparatively low, it was constitutively expressed in the epithelial compartment and in epidermal appendages in uninfected fetal skin grafts (Fig. 5F), consistent with the Ki-67+ marker of proliferation. Controls omitting the primary antibodies showed no signal (Fig. 5G).

FIG 5.

Expression of cytokeratin 17 and Wnt5a in VZV-infected adult skin xenografts 24 days after infection. Tissue sections from VZV-infected and uninfected adult skin grafts were stained with polyclonal antibody to VZV proteins and detected with secondary antibodies labeled with Alexa Fluor 594 (red signal), cellular proteins were detected with Alexa Fluor 488-labeled antibodies (green signal), and nuclei were counterstained with Hoechst stain (blue signal). (A to G) K17 expression. (A) VZV-infected adult skin, ×20. (B to D) Separate ×200 images were taken from regions denoted by the white boxes in panel A (marked as panels B, C, and D). (E to G) VZV expression (red) and K17 expression of uninfected adult skin graft (E) and uninfected fetal skin grafts (F) and control slide with omission of primary antibody (G). (H to J) Wnt5a expression; all panels, ×200. (H) VZV-infected adult skin. (I) Same tissue as that shown in panel H but with an uninvolved area. (J) Uninfected fetal skin graft. (K) Uninfected adult skin grafts. Arrows indicate the observations described in Results.

Wnt5a is a member of the wingless-related/mouse mammary tumor virus integration family and can initiate proinflammatory responses through activation of the canonical Wnt/beta-catenin and noncanonical Wnt/Ca(2+) signaling pathways. Wnt5a is restricted primarily to the basal layer in adult skin, with limited expression in the upper spinous and granular layers (34, 35). VZV-infected adult skin exhibited increased Wnt5a expression (green signal) adjacent to VZV skin lesions (red signal) in the epidermal compartment (Fig. 5H, white arrowheads), in contrast to uninvolved areas within the same xenograft (Fig. 5I). Wnt5a expression was also intense throughout the dermal compartment, presumably in dermal fibroblasts, in VZV-infected adult skin grafts, including uninvolved regions of the same tissue (Fig. 5I). Constitutive Wnt5a expression was absent in uninfected adult skin grafts (Fig. 5I) and was very limited in uninfected fetal skin grafts (Fig. 5J). Since canonical Wnt/beta-catenin signaling and noncanonical Wnt/Ca(2+) signaling are mutually antagonistic, we also examined the nuclear translocation of the Wnt regulator beta-catenin. Lack of nuclear beta-catenin (Fig. 4E) suggested that Wnt5a induction in VZV-infected skin occurs through a noncanonical beta-catenin-independent mechanism. Wnt5a also induces expression of interleukin-1 beta (IL-1beta), a key mediator of skin inflammation and proliferation. IL-1beta was significantly upregulated in VZV-infected adult skin, within both infectious foci and bystander cells (Fig. 4F).

DISCUSSION

This study demonstrated that the typical tissue architecture of adult skin is preserved after xenografting and that the pathogenesis of VZV skin infection can be investigated in adult skin xenografts as well as in fetal skin xenografts in the SCID mouse model. In previous work, the model was found to reproduce the diminished virulence of live varicella vaccine (vOka) compared to wild-type VZV (pOka) in skin, consistent with the clinical attenuation of varicella-zoster and zoster vaccines (4, 23). Here we found that VZV replication was enhanced in skin from older compared to younger adults, again correlating with clinical observations. In addition, we documented that innate antiviral cellular responses are triggered by VZV infection of adult as well as fetal skin, and we extended these observations to show that infection induces proliferative and inflammatory responses in the absence of VZV-specific adaptive immunity. Of interest, VZV infection elicits altered cell signaling and inflammatory responses with similarities to those identified in lesioned skin biopsy specimens from individuals with psoriasis (36, 37).

The preservation of aging skin characteristics in xenografts was evident as a marked acellularity of the dermis, decreased type I collagen synthesis in the papillary dermis, and reduced proliferation of dermal fibroblasts and keratinocytes in their respective skin compartments (38). Other morphological characteristics of skin xenografts from older adults included thickening of the corneum stratum, fewer and smaller epidermal appendages in the dermis, and fewer and smaller dermal microvessels.

Lesions in adult skin showed intraepithelial vesicles containing degenerating and ballooning keratinocytes, multinucleated polykaryon formation, and suprabasal acantholysis. VZV protein expression was more intense in infected fetal skin xenografts, especially within pilosebaceous structures, which were less prominent in adult skin. Suprabasal vesicles in the epidermis have been shown to contain the most intact infectious particles, whereas VZ particles are degraded in the endosomes of epidermal basal cells during virion assembly (39). Lesions in adult skin were located within the epithelial compartment, consistent with the capacity of zoster episodes to result in VZV transmission to naive contacts.

We have observed that VZV protein expression is prominent in hair follicles at an early stage in infection of skin xenografts. Follicular involvement in zoster lesions, particularly around the isthmus of hair follicles and sebaceous glands, suggests that these are among the first skin cells to be encountered by reactivating virus particles delivered by retrograde axonal transport (40). The dense network of nerve endings around hair follicles should facilitate the infectious process during zoster just as the capillary network is likely to enhance the T-cell transfer of VZV to follicular epithelial cells during varicella (3). Neuronal axon termini are enriched at the base of the pilosebaceous structures where VZV infection was most extensive. Clinical studies show extensive folliculitis in zoster lesions, consistent with a role in VZV spread to keratinocytes during reactivation (41, 42).

VZV activation of STAT3 is critical for skin replication and lesion formation, demonstrating that pSTAT3 regulation is as important for lytic infection by a nononcogenic herpesvirus as it is for such infections by oncogenic herpesviruses (32). Here, we show that survivin, which is regulated by phosphorylated STAT3, was upregulated in hair follicle cells within VZV-infected adult skin, as well as in fetal tissue. STAT3 is known to regulate genes common to wound healing, the anagen (growth) phase of the hair cycle, and the migration of keratinocytes (43–45). By activating STAT3 in newly infected follicular stem cells, VZV recapitulates the cutaneous tissue response to injury in which follicular stem cells rapidly switch their homeostasis programming to a wound healing program. In this process, LGR6+ stem cells in the upper isthmus migrate from the hair follicle to the interfollicular epidermis (46). Thus, STAT3 activation by VZV would alter the basal programming of infected follicular stem cells, driving their migration to the interfollicular epidermis and thereby accelerating the delivery of VZV to epidermal sites where lesion formation is initiated. This hijacking of the wound healing program of follicular stem cells by VZV would therefore reinforce the exploitation by VZV of the keratinocyte maturation pathway to traverse the epidermal strata (39). Hyperkeratosis due to corneocyte overgrowth is another consequence of STAT3 signaling that is likely to enhance VZV skin pathogenesis because the corneum stratum barrier is thought to preserve the infectivity of virion-rich intravesicular fluid (39). These pathological changes regarding hyperkeratosis and epidermal hypertrophy are observed in clinical zoster lesions (47–49).

VZV infection of adult skin grafts resulted in upregulation of cytokeratin 17 and Wnt5a, whose expression is also induced as part of the cutaneous wound healing response. In psoriatic skin, expression of these cell proteins coincides with keratinocyte hyperproliferation, as we also observed in VZV-infected skin, and is associated with STAT3 activation (34, 36, 50). Migrating keratinocytes exhibit cytokeratin 17 upregulation during wound healing (51). Notably, fetal skin cells express cytokeratin 17 constitutively, consistent with the higher titers of VZV recovered from these xenografts than from those from younger adults. Wnt5a is expressed in basal stem cells, including hair follicles, in injured skin, but its specific role in wound repair is not well understood. As in VZV-infected adult skin, dermal fibroblasts are strongly positive for Wnt5a in psoriatic skin (34, 36). Since keratinocyte overgrowth and expression of cytokeratin 17 and Wnt5a are upregulated in VZV-infected adult skin xenografts, our observations suggest that VZV activation of STAT3 may also accelerate skin infection by exploiting components of the wound healing process, in addition to the induction of survivin as a mechanism to block apoptosis of infected skin cells.

Upregulation of IL-1beta, a key mediator of skin inflammation and cell proliferation, was observed in VZV-infected adult skin and is also characteristic of psoriasis (52). Since IL-1beta induces K17 overexpression when added to skin cultures (53), its induction is consistent with the detection of K17 and proliferation in VZV-infected skin. Because IL-1beta functions require its processing by activated caspase-1 (54), it is important that VZV infection of skin tissue specimens induces the formation of the NLRP3 inflammasome and activation of caspase-1 through a type I IFN-independent mechanism (55).

We observed decreased VZV replication in skin xenografts from younger adults (<55 years of age) than in those from fetal skin. We hypothesize that the increased level of VZV replication in fetal skin may reflect the reduced dermal cell density observed in adult skin. Decreased dermal cell density likely impairs VZV spread by reducing opportunities for cell-cell contact, which VZV requires. This may explain the “flattened” appearance of VZV lesions in adult skin tissue specimens. Alternatively, the higher rate of cell proliferation in fetal tissue, demonstrated by Ki-67 staining, may explain the increased permissiveness of fetal skin. However, work by others has demonstrated that VZV replicates efficiently in quiescent (Ki-67-negative) cells by manipulation of cell cycle regulators and transcription factors, which enhance its replication (56).

Further, we hypothesize that age-related changes to VZV-specific T-cell-mediated immunity may coexist with an age-associated decline in type I IFN-mediated innate responses in skin. The observation of increased VZV replication in skin xenografts from older adults suggests an age-associated impairment of host defenses which may facilitate the progression of VZV reactivation to clinical zoster and contribute to the extensive dermatomal involvement typical of zoster in the elderly (4). Our results demonstrate that VZV infection of adult skin triggers intrinsic IFN-mediated responses such as we have described in VZV-infected fetal skin xenografts. Effector proteins, such as PML and MxA, produce an antiviral state within the host cell. In turn, VZV has evolved strategies to counter these intrinsic responses, as was evident from the differential expression and intracellular localization of STAT1, PML, and MxA in keratinocytes within foci of VZV infection compared with uninvolved keratinocytes within the same adult skin xenograft. Our previous studies showed that the VZV IE62 protein interferes with IFN pathway signaling by phosphorylation of IFN regulatory factor 3 (27) and that the VZV ORF61 protein disrupts the PML nuclear body architecture via its three SUMO-interacting motifs, an activity that was required for VZV skin pathogenesis (57). This work demonstrates the utility of the adult skin xenograft model for investigating specific age-related proviral factors that may enhance VZV replication in aged skin.

From the perspective of VZV persistence in the human population, episodes of VZV reactivation that progress to the production of infectious virus in adult skin are important because of fluctuations in the number of susceptible hosts, requiring reintroduction of the virus to initiate a varicella epidemic (58). Adult skin xenografts offer an approach for further investigation of VZV-induced skin pathologies in vivo.

MATERIALS AND METHODS

Human tissue xenografts in SCID mice.

Discarded deidentified adult abdominal skin tissue was obtained from healthy donors undergoing routine surgical procedures at the Stanford Medical Center. Twelve tissue specimens were collected from donors 40 to 65 years of age (median age, 53 years). Fetal skin (scalp; 6 tissue specimens; 18 to 21 gestational weeks) was obtained from Advanced Bioscience Resources (ABR, Alameda, CA) in accordance with federal and state regulations. Skin tissue specimens were trimmed at the dermal side and cut to obtain ∼12 to 20 segments (8 by 8 by 4 mm) containing the full epidermis, the papillary dermis, and part of the reticular dermis. Tissue specimens were kept at 4°C in cell culture media until implantation. Mice were anesthetized by intraperitoneal (i.p.) injection of ketamine and xylazine, and skin segments were grafted subcutaneously on both flanks. The mouse strains used were C.B-17 SCID mice (C.B-Igh-1^b/IcrTac-Prkdc^scid; Taconic), which lack T and B cells but have NK cells, and NOG mice (NOD.Cg-Prkdc^scidIl2rg^tm1Sug/JicTac; Taconic), which were developed by backcrossing the C57BL/6Jjic-Il2rg line onto the NOD mouse strain (24). Animal use was approved by the Stanford University Administrative Panel on Laboratory Animal Care. Use of discarded deidentified adult skin tissue and fetal tissue was reviewed by the Stanford University Institutional Review Board.

VZV infection of human skin xenografts.

After 5 weeks, skin xenografts were surgically exposed and inoculated intradermally (data not shown) with a clinical isolate of VZV (parental Oka virus; passage = <20) grown in human lung fibroblasts. Approximately 10⁵ PFU of cell-associated VZV was introduced in a series of 10-μl injections (100 μl/graft). Biological variability was mitigated by infecting 6 to 10 xenografts/time point/donor. All tissue specimens were examined at 24 days after inoculation. Mice were euthanized, and the grafts were removed. Half of each graft was fixed in 4% paraformaldehyde for histological analysis; the other half was used for viral titration. Tissue was homogenized by mincing by hand to a fine paste (2 min) in phosphate-buffered saline (PBS) and then further processed using a handheld device (Tissue-Tearor) for 10 s at speed 8 for determination of viral titers (data not shown). Serial dilutions of each homogenate were added in triplicate to melanoma cells in 24-well plates. Cells were fixed in 4% paraformaldehyde after 4 days, and VZV plaques were identified by immunostaining using a mouse anti-VZV polyclonal serum (C05108MA; Meridian Life Science) and visualized with FastRed substrate (Fisher Scientific). Quantitation of viral titers and a t test to determine statistical significance were performed using the statistics and graphing software program Prism (ver. 7.0c) by GraphPad.

Histological and immunostaining analysis of skin xenografts.

Paraformaldehyde-fixed tissue specimens were paraffin embedded, and tissue sections (10 μm) were cut and baked onto slides. For histological analysis by H&E staining, slides were deparaffinated and rehydrated by incubation in a xylene substitute (Safeclear II; Fisher Scientific) and a graded alcohol series. Immunoperoxidase staining of tissue sections was performed as previously described (23). Briefly, rehydrated tissue sections were subjected to an antigen retrieval step, endogenous peroxidases were destroyed by immersion in 3% hydrogen peroxide for 10 min, and tissue specimen sections were blocked and stained with primary antibodies diluted in PBS according to the manufacturer's instructions. Tissue sections were incubated with 3,3-diaminobenzidine (DAB) for 5 min to detect the streptavidin-peroxidase complex signal (brown = positive). Sections were counterstained with hematoxylin and imaged with a Keyence BZ-X700 microscope. For confocal imaging, fluorescently labeled secondary antibodies (Alexa Fluor 488 or Alexa Fluor 594 [depending on species]; Invitrogen) were applied after the primary antibody incubation step. Nuclei were stained with Hoechst stain, and duplicate sequential sections were imaged with a Zeiss LSM780 multiphoton laser scanning confocal microscope or a Keyence BZ-X700 microscope. Images were scanned at 1,024 by 1,024 pixels with an 8-frame average and a pinhole size of 1 airy unit. Control slides without the primary antibodies were used to establish background staining in all experiments.

VZV proteins were detected using a mouse anti-VZV polyclonal serum (C05108MA; Meridian Life Science) except for the experiment represented in Fig. 4A, in which a high-titer anti-VZV human serum (Ig fraction) was used. Antibodies to cellular proteins were as follows: anti-Ki-67 (ab15580; Abcam), anti-type I collagen (ab88147; Abcam), anti-CD31/PECAM (ab32457; Abcam), anti-keratin 10 clone DE-K10 (Monosan), anti-STAT1 (ab109461; Abcam), anti-MX1 (ab95926; Abcam), anti-survivin (71G4B7; Cell Signaling Technologies), anti-PML (ab72137; Abcam), anti-beta-catenin clone E247 (ab32572; Abcam), anti-Wnt5a (1-60032; Novus Biologicals), anti-cytokeratin 17 (ab53707; Abcam), and anti-IL1beta (ab2015; Abcam).