Susceptibility of Human and Murine Dermal Fibroblasts to Herpes Simplex Virus 1 in the Absence and Presence of Extracellular Matrix

ABSTRACT

Herpes simplex virus 1 (HSV-1) invades its human host via the skin and mucosa and initiates infection in the epithelium. While human and murine epidermis are highly susceptible to HSV-1, we recently observed rare infected cells in the human dermis and only minor infection efficiency in murine dermis upon ex vivo infection. Here, we investigated why cells in the dermis are so inefficiently infected and explored potential differences between murine and human dermal fibroblasts. In principle, primary fibroblasts are highly susceptible to HSV-1; however, we found a delayed infection onset in human compared to murine cells. Intriguingly, only a minor delayed onset of infection was evident in collagen-embedded compared to unembedded human fibroblasts, although expression of the receptor nectin-1 dropped after collagen embedding. This finding is in contrast to previous observations with murine fibroblasts where collagen embedding delayed infection. The application of latex beads revealed limited penetration in the dermis, which was more pronounced in the human than in the murine dermis, supporting the species-specific differences already observed for HSV-1 invasion. Our results suggest that the distinct organization of human and murine dermis contributes to the presence and accessibility of the HSV-1 receptors as well as to the variable barrier function of the extracellular matrix. These contributions, in turn, give rise to inefficient viral access to cells in the dermis while dermal fibroblasts in culture are well infected.

IMPORTANCE Dermal fibroblasts are exposed to HSV-1 upon invasion in skin during in vivo infection. Thus, fibroblasts represent a widely used experimental tool to understand virus-host cell interactions and are highly susceptible in culture. The spectrum of fibroblasts’ characteristics in their in vivo environment, however, clearly differs from the observations under cell culture conditions, implying putative variations in virus-cell interactions. This becomes evident when ex vivo infection studies in murine as well as human dermis revealed the rather inefficient penetration of HSV-1 in the tissue and uptake in the dermal fibroblasts. Here, we initiated studies to explore the contributions of receptor presence and accessibility to efficient infection of dermal fibroblasts. Our results strengthen the heterogeneity of murine and human dermis and imply that the interplay between dermal barrier function and receptor presence determine how well HSV-1 penetrates the dermis.

INTRODUCTION

The primary entry portals of herpes simplex virus 1 (HSV-1) in its human host are the cornea, mucocutaneous regions, and skin. Following primary infection of epithelial cells, HSV-1 establishes lifelong latency in neurons and can reactivate to cause recurrent infection. Cellular entry of HSV-1 requires the coordinated action of multiple viral glycoproteins with surface receptors (1, 2). Interaction of the viral glycoprotein gD with its receptors is preceded by viral attachment to heparan sulfate proteoglycans and finally results in the fusion of the viral envelope with cellular membranes (3–5). The primary gD receptor on human epithelial cells is nectin-1, a cell-cell adhesion molecule, whereas herpesvirus entry mediator (HVEM) and modified 3-O-sulfated-heparan sulfate (3-OS-HS) can serve as alternate gD receptors (6–11). Ex vivo infection of either nectin-1 or HVEM-deficient murine epidermis revealed that nectin-1 is also a major receptor in tissue, although HVEM can serve as a receptor in a greatly reduced manner (12).

During the primary infection of skin, HSV-1 replicates in epidermal keratinocytes leading to an intense inflammatory response in the dermis, which in turn exposes dermal fibroblasts, the major resident cell type of the dermis, to HSV-1. In addition, the dermis represents a target tissue in vivo during viral penetration via abraded skin and upon viral reactivation, where HSV-1 can be released from nerve endings within the dermal tissue. Previously, we explored the role of nectin-1 and HVEM as receptors for HSV-1 in dermal fibroblasts using nectin-1- and/or HVEM-deficient murine cells. Infection is not inhibited by the absence of either nectin-1 or HVEM, supporting that they act as alternative receptors, although nectin-1 functions as the more efficient receptor in primary murine dermal fibroblasts (13). In contrast to the highly susceptible fibroblasts in culture, little is known about infection efficiency of dermal fibroblasts and receptor expression in their native in vivo environment. The dermis makes up the bulk of the skin, provides elasticity and tensile strength, and is composed of distinct zones. The papillary dermis, which is the closest to the epidermis, and the underlying reticular dermis differ in cell density and connective tissue organization with collagen as the major dermal component. The papillary dermis contains a high density of fibroblasts and thin collagen fibrils, whereas the very thick reticular dermis is characterized by a densely packed extracellular matrix (ECM) including large collagen fibrils.

Recently, we addressed the susceptibility of the dermis to HSV-1 using an ex vivo infection model. After preparing the dermis from murine skin by separation from the epidermis, we found single infected cells in the upper papillary dermis at 5 h postinfection (p.i.) with increasing numbers at later times (14). Interestingly, the onset of infection is delayed in aged mice, which correlates with a decrease of nectin-1 expressing fibroblasts during aging (14). Thus, we conclude that murine dermis is susceptible to HSV-1, and reduced receptor availability delays infection efficiency in the dermis from aged mice. Intriguingly, ex vivo infection of human dermis after separation from the epidermis revealed rare infected fibroblasts (15). These results are independent of donor age and correlate with the low expression of nectin-1 and HVEM (15). Our previous results indicate that fibroblasts in the dermis can be rather inefficiently ex vivo infected, which was more pronounced in human than in murine dermis (14, 15). So, the overall question is how critical receptor expression levels, their distribution and accessibility in tissue, skin barriers, or various defense mechanisms are to efficient HSV-1 invasion in the dermis.

Here, we studied how receptor presence and mechanical barriers contribute to the low infection efficiency of the dermis with HSV-1 and addressed whether species-specific differences provide additional contributions. We initially investigated whether the onset of HSV-1 infection differs in murine compared to human dermal fibroblasts. We then explored whether embedding human fibroblasts in a collagen I matrix leads to a delay in initiation of infection and analyzed how well the receptors nectin-1 and HVEM are expressed. Finally, we studied the uptake of beads in the murine and human dermis to address how efficient particles can overcome barriers provided by the ECM.

RESULTS

Infection of murine and human fibroblasts with HSV-1.

To address whether HSV-1 initiates infection in human fibroblasts as efficiently as in murine fibroblasts, we performed infection studies and determined successful viral entry by visualizing the very early expressed viral infected cell protein 0 (ICP0). Upon release of the viral genome in the nucleus, ICP0 localizes in nuclear foci and then relocalizes to the cytoplasm during viral replication (16, 17). Previous studies demonstrated comparable infection efficiency in primary murine fibroblasts isolated from newborn or adult skin (13). To directly compare primary human and murine cells, we isolated dermal fibroblasts from adult skin of both species and infected the cells for 1 to 6 h p.i. with 5 PFU/cell. In primary murine fibroblasts, nuclear ICP0 was detected in about 10% of the cells already at 1 h p.i. (Fig. 1b). At 2 h p.i., about 56% of the cells were infected, and viral replication was initiated in about 25% of the cells as indicated by cytoplasmic ICP0 (Fig. 1b). In primary human fibroblasts, only few ICP0-expressing cells were detected at 2 h p.i. with an increase to 34% infected cells at 3 h p.i., demonstrating that the initial infection phase progressed more rapidly in murine than in human cells (Fig. 1b). At 6 h p.i., the numbers of infected cells were nearly comparable although more cells with cytoplasmic ICP0 were found for murine fibroblasts (Fig. 1a and b). The relocalization of ICP0 to the cytoplasm, also in human cells, indicates viral replication at 6 h p.i. (Fig. 1a and b).

FIG 1.

Infection of murine and human dermal fibroblasts with HSV-1. (a) Primary murine fibroblasts (pmF) (passage 2), primary human fibroblasts (phF) (passage 2), and HFFF2 cells were infected with HSV-1 at 5 PFU/cell. Immunostainings show nuclear and cytoplasmic ICP0 (green) and DAPI (blue) as the nuclear counterstain. Bars, 25 μm. (b) The numbers of ICP0-expressing cells detected by immunostainings were determined after various times p.i. from at least three independent experiments and are shown as means plus standard deviations. The results indicate the significant delay of ICP0 stainings in human compared to murine fibroblasts at 1, 2, and 3 h p.i. P values of ≤0.05 (*) are shown.

The delayed initial infection was not only observed in primary human fibroblasts but also upon infection of human fetal foreskin fibroblasts (HFFF2), a nontransformed cell line, supporting that the delay is species-specific for human fibroblasts and not restricted to primary cells (Fig. 1a and b).

Taken together, we conclude that the entry process of HSV-1 is slower in human than in murine dermal fibroblasts. This delayed onset of infection suggests that multiple steps during the entry process could differ in murine and human cells, including attachment, use of receptors, receptor abundance, internalization pathways, as well as transport and delivery to the nucleus.

Entry of HSV-1 in collagen-embedded human fibroblasts.

Although delayed, infection in human fibroblasts progressed efficiently in contrast to human dermis, for which we recently observed very minor susceptibility to HSV-1 (15). To initially explore whether the ECM provides a potential barrier for the penetration of virus particles in the dermis, we embedded fibroblasts in three-dimensional (3D) floating collagen I matrices as previously described (18, 19). As collagen type 1 is the most abundant component of fibrous connective tissue and fibroblasts cause matrix contraction that directly depends on the cell density and inversely on the collagen concentration, the 3D matrices mimic a dermis-like environment (19, 20). After collagen embedding of primary murine fibroblasts, we previously observed a major delay of viral transcription with increasing concentrations of collagen up to 0.9 mg/ml, implying that the density of collagen fibrils delays infection of murine cells compared to unembedded cells (14).

To investigate whether viral entry in human fibroblasts is also influenced by the collagen environment, we embedded HFFF2 cells in the highest collagen I concentration (0.9 mg/ml). The morphology and distribution of the fibroblasts in the contracted gels were visualized at various times after embedding by staining the actin cytoskeleton. Only minor gel contraction was observed at 5 h after cell seeding, which correlated with a dispersed network of still rounded cells with higher cell densities at the gel edges (Fig. 2a). At 24 h post seeding (p.s.), loosely connected spindle-shaped cells with rather long cellular extensions were observed in contracted gels with no further contraction during the next 9 h (Fig. 2a and b). Gel contraction was remarkably different after embedding primary human fibroblasts for which we observed reduced gel contraction compared to HFFF2 cells even after 24 h (data not shown).

FIG 2.

HSV-1 infection of human and murine fibroblasts embedded in floating collagen matrices. (a) HFFF2 (1 × 10⁵ cells) were embedded in collagen I (0.9 mg/ml). After cell seeding, floating collagen matrices show minor contraction at 5 h post seeding (p.s.) and enhanced contraction at 24 h p.s. The magnifications of F-actin (red) stained HFFF2 cells with DAPI (blue) as the nuclear counterstain demonstrate rounded cells at 5 h p.s. at the edge and middle of the gel, while cell spreading is visible at 24 h p.s. Bars, 50 μm. (b) Second harmonic generation (SHG) shows unchanged collagen structure (green) when HFFF2 cells were mock infected or infected with HSV-1 at 5 PFU/cell for 9 h at 24 h p.s. Cells are labeled with propidium iodide (PI) (red). Contraction of the floating collagen matrix was unchanged after incubation or infection for 9 h compared to 24 h p.s. Bars, 50 μm. (c) At 24 h p.s., embedded HFFF2 cells or unembedded cells were infected with 5 PFU/cell. Reverse transcriptase PCR (RT-PCR) (n = 3) indicates ICP0 transcripts at 2 h p.i. with increasing transcript levels until 9 h p.i. in embedded and unembedded cells. (d) ICP0-expressing cells were quantified by flow cytometry (n = 3). The results show a significantly reduced number of cells after collagen embedding compared to unembedded cells at 6 h p.i. A P value of ≤0.05 (*) is shown. (e) At 5 h after seeding 1 × 10⁶ primary murine fibroblasts (pmF) (passage 2) in collagen I (0.9 mg/ml), cells were infected with 50 PFU/cell. After infection of unembedded pmF with 5 PFU/cell, RT-PCR (n = 3) indicates ICP0 transcripts at 1 h p.i. while ICP0 transcription in embedded cells was only detected at 6 h p.i. Contracted gels at various times p.i. are shown. (f) At 5 h p.s., spreading of embedded pmF (1 × 10⁶ cells) infected with 50 PFU/cell for 3 h is shown by staining of F-actin (red) and DAPI (blue) as the nuclear counterstain. Bars, 50 μm. (g) After dissociation of unembedded and embedded HFFF2 cells, flow cytometric analyses indicate a decrease from ca. 36% to 5% of nectin-1-positive cells after embedding for 24 h while ca. 27% and 34% of cells before and after embedding, respectively, expressed HVEM.

We next visualized the organization of collagen in gels with embedded HFFF2 cells by second harmonic generation (SHG) microscopy. The diffused signal intensities indicate the monomeric structure of collagen in the contracted gel at 24 h post seeding plus 9 h of mock infection (Fig. 2b). As expected, this collagen structure does not quite mimic the collagen network present in the human dermis as shown previously (15).

After infection of collagen-embedded HFFF2 cells with HSV-1, we surprisingly detected the onset of ICP0 transcription as early as 2 h p.i. followed by a continuous increase of viral transcription until 9 h p.i. (Fig. 2c). The onset of transcription was comparable to unembedded cells (Fig. 2c). Although infection was ongoing for several hours, the collagen structure showed no obvious difference at 9 h p.i. compared to mock-infected gels (Fig. 2b).

To further characterize the efficiency of infection in HFFF2 cells after collagen embedding, we analyzed the number of ICP0-expressing cells by flow cytometry. At 6 h p.i., ca. 23% of the collagen-embedded cells expressed ICP0 while ca. 64% of ICP0-expressing cells were already detected among unembedded cells (Fig. 2d). These results indicate some delay in the initial infection process upon collagen embedding. In unembedded fibroblasts, we noticed slightly reduced numbers of ICP0-expressing cells compared to those in the immunostaining analyses shown (Fig. 1b), which were not suitable for ICP0 detection in the collagen gels.

For direct comparison, we repeated previous experiments done with murine fibroblasts from newborn mice (14) and analyzed the effect of collagen embedding on infection in fibroblasts from adult mice. After infection with HSV-1 at 5 PFU/cell, ICP0 transcripts were observed in unembedded cells already at 1 h p.i., while viral transcription in collagen-embedded cells was only found at 6 h p.i. even after infection at 50 PFU/cell (Fig. 2e). These results confirm our previous data, which demonstrate a delayed viral transcription with increasing concentrations of collagen after infection with 50 PFU/cell (14).

After seeding murine fibroblasts, gel contraction was faster than that of gels with human fibroblasts so that embedded murine fibroblasts were already infected at 5 h post seeding. At this time point plus 3 h p.i., gels were contracted with only a minor change in diameter at 14 h p.i. (Fig. 2e). Spreading of murine fibroblasts was already visible at the early time point, which is in contrast to human fibroblasts, suggesting species-specific differences in cell-collagen interactions (Fig. 2a and f).

In summary, our results revealed that collagen embedding of human fibroblasts only had minor effects on the susceptibility to HSV-1, while infection of collagen-embedded murine fibroblasts is strongly delayed. These findings suggest that the fibrillar collagen per se does not provide a major barrier for HSV-1 to access its receptors on dermal fibroblasts.

To explore how the availability of the receptors nectin-1 and HVEM correlate with the efficiency of infection in human fibroblasts, we investigated cell surface expression of both receptors. Flow cytometry revealed nectin-1 consistently on ca. 36% of HFFF2 cells, while HVEM expression was found on ca. 27% of the cells (Fig. 2g). To study whether collagen embedding affects receptor expression, flow cytometric analyses were performed on embedded HFFF2 cells retrieved by collagenase treatment 24 h after seeding, which is the time point of infection. Surprisingly, the number of nectin-1-expressing cells decreased to ca. 5%, while collagen embedding had nearly no effect on the number of HVEM-expressing cells (Fig. 2g). Thus, we conclude that HVEM can efficiently replace nectin-1 as a receptor in collagen-embedded human fibroblasts, as the susceptibility to HSV-1 showed only a very minor delay. However, we cannot exclude that even low nectin-1 expression still allows infection.

In previous studies of murine fibroblasts from adult mice, the level of nectin-1 expression varied from 6% to 62% of cells (14). HVEM is expressed on ca. 82% of murine fibroblasts (13), which is much higher than HVEM expression on human fibroblasts (Fig. 2g; Table 1). As collagen-embedded murine fibroblasts were already infected 5 h post seeding, we assume that putative changes in surface receptor expression upon collagen embedding are not detectable.

TABLE 1.

Comparison of nectin-1 and HVEM expression^a

^aBased on flow cytometric analyses, the average percentage of nectin-1- or HVEM-expressing cells is given. When nectin-1 expression is rather variable, the range of nectin-1-expressing cells is shown.

^bSee reference 14.

^cSee reference 15.

^dSee reference 13.

Taken together, our results support that the availability of the receptors nectin-1 and HVEM is sufficient to allow efficient infection of human fibroblasts prior to and after collagen embedding. Furthermore, we assume that upon collagen embedding, it is not the receptor expression but rather a hindered accessibility of the receptors for HSV-1 that is responsible for delayed infection in murine fibroblasts. This assumption, in turn, suggests that the collagen matrix influences the accessibility and/or distribution of the receptors differently in murine and human fibroblasts.

Penetration of latex beads in murine and human dermis.

To address whether and how well particles in general can penetrate the dermis and reach fibroblasts and whether there are differences between murine and human tissue, we investigated the invasion of latex beads (500 nm). After separation from the epidermis, human dermis was incubated with beads for various times (Fig. 3a). At 16 h of incubation, some beads were present in the most apical part of the papillary dermis with an increase in the number of beads after 24 h of incubation (Fig. 3b). Beads never fully penetrated the dermis, as deepness of invasion was limited and varied in some regions (Fig. 3b). Beads appear to be trapped in the ECM, and no obvious internalization of beads was observed in fibroblasts, visualized by costaining vimentin (Fig. 3b). Only in the case of detached cells, efficient internalization of beads was found (Fig. 3c). To follow up the efficiency of bead uptake in cultured human fibroblasts, we incubated HFFF2 cells with beads for 24 h. Internalized beads were present in each cell, supporting very efficient particle uptake in human fibroblasts in general (Fig. 3f).

FIG 3.

Penetration of latex beads in human and murine dermis. (a) Schematic illustrating the incubation and analyses of adult human dermis with beads. Hematoxylin and eosin (HE)-stained section visualizes the dermis after separation from the epidermis by dispase II prior to incubation in medium. Papillary and reticular layers are distinguished by the dashed line. (b) After incubation of abdominal dermis with 2 × 10⁹ beads per sample, cross sections visualize vimentin-positive (red) fibroblasts with DAPI (blue) as the nuclear counterstain. Transmission light (TL) visualizes the morphology of the papillary layer and part of the underlying reticular layer with single beads (green) penetrating in the most upper part of the papillary dermis at 16 h of incubation. Increased number of beads is visible after incubation for 24 h, as depicted in the magnification. The dashed line indicates the sample border. (c) After 24 h of incubation, the cross section shows a detached vimentin-positive (red) fibroblast with internalized beads (green) as depicted in the magnification. (d) Schematic illustrating the incubation and analyses of human skin shaves. (e) After incubation of abdominal skin shaves with 2 × 10⁹ beads per sample for 24 h, cross sections visualize beads (green) in the basal part of the papillary dermis as depicted in the magnification. The dashed line indicates the border of the dermal layer. Collagen VII (colVII) (red) indicates the basement membrane, and DAPI (blue) serves as the nuclear counterstain. (f) After incubation of 4 × 10⁴ HFFF2 cells with 2 × 10⁹ beads for 24 h, confocal sections show internalized beads (green) in each F-actin (red) stained cell with DAPI (blue) as the nuclear counterstain. (g) Schematic illustrating the incubation and analyses of dermis from skin of 5-week-old mice with beads. HE-stained section visualizes the dermis after separation from the epidermis by dispase II prior to incubation in medium. Papillary and reticular layers are distinguished by the dashed line. (h) After incubation of murine dermis with 2 × 10⁹ beads per sample for 24 h, transmission light (TL) visualizes tissue morphology and shows accumulated beads (green) at the most upper part of the papillary dermis with DAPI (blue) as the nuclear counterstain. As depicted in the magnification, vimentin-positive (red) cells with internalized beads (green) are visible close to the surface of the dermis, which is indicated by the dashed line. (i) After incubation with beads, only single beads (green) are visible in the part of the reticular dermis that is close to the sample border depicted by the dashed line. Transmission light (TL) visualizes the morphology of the reticular dermis. Bars, 25 μm.

When complete human skin was incubated with beads, only single beads were observed in parts of the reticular dermis (data not shown). To explore whether removal of the reticular dermis facilitates penetration in the remaining papillary dermis, we studied penetration of beads in skin shaves, which include the epidermis, the basement membrane, and the upper part of the papillary dermis (Fig. 3d). At the edges of the skin shaves, we rarely observed dermal cells with internalized beads (data not shown). Depending on the thickness of the papillary dermis, the beads were dispersed mainly in the basal part of the dermal layer (Fig. 3e). These results indicated that the beads could not penetrate the complete papillary dermis of the skin shaves up to the basement membrane, visualized by costaining collagen VII (colVII) (Fig. 3e). Overall, our data demonstrate that the beads invade the ECM but have insufficient access to the dermal cells to be internalized. Penetration in the papillary dermis may be easier than in the reticular dermis, which could be related to the less densely packed ECM in the papillary dermis. The results reflect our previous infection studies of human dermis where only single infected cells were found in the most apical part of the papillary dermis (15). To visualize the penetrating virus particles, we immunostained the major capsid protein VP5 after ex vivo infection of human dermis with HSV-1 at 50 PFU/cell (Fig. 4a). At 6 h p.i., some areas of the apical papillary dermis showed virus particles next to a fibroblast, and some of the detached fibroblasts were observed with VP5 expression at 16 h p.i. (Fig. 4b). Although the visualization of the virus particles was much more demanding compared to the bigger latex beads, our results support a very limited penetration for both the virus and the beads.

FIG 4.

Penetration of HSV-1 in human dermis. (a) Schematic illustrating the ex vivo infection and analyses of human dermis. (b) After infection of abdominal dermis with HSV-1 at 50 PFU/cell, cross sections visualize vimentin-positive (green) fibroblasts with DAPI as the nuclear counterstain. Transmission light (TL) shows the morphology of the papillary layer and part of the underlying reticular layer. The dashed line indicates the border of the dermis. As depicted in the magnifications, VP5-positive virus particles are visible next to a fibroblast located in the most upper part of the dermis at 6 h p.i., and a detached fibroblast with VP5 expression as wells as virus particles are shown at 16 h p.i. The mock-infected control is shown after incubation in medium for 24 h. Bar, 25 μm.

For comparison, we explored the penetration of beads in murine dermis (Fig. 3g). After incubation with beads for 5 h, murine dermis already showed single beads in the most apical part of the papillary layer (data not shown). At 16 h of incubation, beads formed clusters at the periphery of the papillary dermis (data not shown) with no obvious changes at 24 h and did not penetrate deeper in the tissue (Fig. 3h). In areas with accumulated beads, internalized clusters of beads were present in vimentin-positive cells, demonstrating uptake in fibroblasts, which were close to the surface of the murine dermis (Fig. 3h). The enhanced penetration of beads in murine compared to that in human dermis may be related to organizational and structural differences. The papillary dermis of mice is mostly very thin and differs from the complex undulating interdigitations with the epidermis in humans, which may impact the behavior of fibroblasts (21). Beads also penetrated via the reticular layer of murine dermis; however, only single beads were visible, and dermal cells with internalized beads were rarely found (Fig. 3i), showing only minor differences to human reticular dermis.

In summary, studies in both human and murine dermis demonstrate the preferred penetration of beads in the papillary layer. As particle penetration mimics the efficiency of HSV-1 invasion in the upper dermis, these observations suggest that the different organization of the ECM provides a barrier to differentially access the fibroblasts. The finding that penetration of beads as well as HSV-1 invasion was more efficient in murine dermis most likely reflects differences in the ECM organization of murine and human dermis.

DISCUSSION

Fibroblasts in culture are extensively used as an experimental tool to study various functions of HSV-1 during infection. How efficient dermal fibroblasts are infected under in vivo conditions, however, is still open. Here, we addressed the putative differences between murine and human dermal fibroblasts and explored to what extent the onset of HSV-1 infection is related to receptor availability and/or poor accessibility of the receptors on fibroblasts in tissue. Our results indicated that the initiation of infection was delayed in human compared to murine dermal fibroblasts in culture, suggesting a less efficient entry pathway of HSV-1 in human fibroblasts. In previous studies, we focused on the impact of receptors and internalization pathways in murine fibroblasts. Infection studies in murine fibroblasts support the impact of dynamin for HSV-1 entry and the internalization via endocytosis as well as nonendocytic pathways (13, 22). Interestingly, nectin-1 deficiency led, although delayed, to viral entry, suggesting that HVEM, which is expressed on about 82% of the primary murine fibroblasts, acts as an alternative receptor and slows down the entry process (13) (Table 1). Here, we found nectin-1 on about 36% and HVEM on about 27% of HFFF2 cells, which indicated lower levels of receptor expression in human than in murine fibroblasts (Table 1). Future work is needed to explore whether the differences in the availability of nectin-1 and HVEM contribute to the efficiency of initial viral infection in human fibroblasts and what role the internalization pathways could play.

As the high susceptibility to HSV-1 of human fibroblasts in culture is in conflict with the rare infected fibroblasts in human dermis (15), we studied how efficient the fibroblasts are accessible when embedded in collagen matrices. Unexpectedly, HSV-1 infection of embedded human fibroblasts revealed only a very minor delay of the infection onset compared to unembedded cells, which is in contrast to studies in murine fibroblasts. The studies also support that the fibroblast-collagen interactions, resulting in cell spreading and contraction of the collagen matrices, differ for the two species. These differences might also reflect distinct ECM-fibroblast interactions in the dermis, which in turn contribute to the very low infection efficiency of HSV-1 in human dermis versus the slightly higher susceptibility of murine dermis (14, 15). Fibroblasts represent a diverse population of cells that produce and remodel the collagen and elastic fibers of the ECM. Within the tissue, fibroblasts exhibit considerable functional diversity reflecting the heterogeneous architecture of the dermis (21). Studies in mouse skin revealed fibroblast lineages that clearly differ in morphology and molecular markers in human dermis for which we are just beginning to understand the heterogeneity and functional role of the fibroblast populations (23, 24). Thus, the differential susceptibility to HSV-1 in human and murine dermis emphasizes the morphological and functional differences of the skin from both species.

To follow up on the efficiency of invasion both in human and murine dermis, we explored the penetration of latex beads in tissue. Of note, beads penetrated less efficiently in human dermis than in murine dermis; however, in both tissues, invasion was restricted to the layers underneath the sample surface. The internalization of beads by murine fibroblasts located close to the surface of the papillary dermis was comparable to HSV-1-infected cells accumulating over time in the most upper papillary layer (15). The visualization of virus particles in human dermis further supports that HSV-1 can only invade the most upper part of the papillary layer. Ultrastructural analysis revealed single cells producing viral progeny at later times of infection (data not shown). Taken together, these observations strengthen the barrier function of the ECM not only for virus particles but also for 500-nm beads, implying that the accessibility of the dermal cells is hindered. The next step is to reveal how the distinct features of ECM integrity and organization in human and murine dermis contribute to the restriction of successful viral invasion.

To understand the efficiency of HSV-1 infection in tissue, the challenge is to dissect the contribution of receptor accessibility versus the impact of receptor expression. Intriguingly, nectin-1 expression dropped to ca. 5% of cells in the human fibroblast cell line HFFF2 after collagen embedding, while previous experiments revealed variable nectin-1 expression in human dermis ranging from undetectable to 30% of cells, which most likely is attributed to the heterogeneity of skin samples (15) (Table 1). These findings suggest that nectin-1 expression tends to be lower on fibroblasts in tissue than in culture. Furthermore, HVEM expression in human dermis is comparable to that in HFFF2 cells, in which the low HVEM level did not change after collagen embedding (Table 1). In contrast to the human dermis, the fibroblasts in culture are highly susceptible irrespective of very low nectin-1 expression, supporting the impact of receptor distribution and accessibility on fibroblasts in tissue. In murine dermis as well as primary murine fibroblasts, nectin-1 expression is also variable, although the average expression level is higher than in human dermis (14, 15) (Table 1). In contrast to human tissue and cells, HVEM expression in murine fibroblasts and dermis is rather high, which might enhance susceptibility to HSV-1 (Table 1). The strong delay of infection in murine fibroblasts after collagen embedding, however, also supports the barrier function of the ECM most likely acting as in human dermis.

Taken together, our findings revealed some of the parameters that contribute to the high susceptibility of fibroblasts in culture versus the inefficient infection of dermis with HSV-1 and illustrated the challenges of future experimentation.

MATERIALS AND METHODS

Cells, preparation of skin, and isolation of dermal fibroblasts.

HFFF2 (ECACC 86031405) cells were maintained in Dulbecco modified Eagle medium (DMEM)/high glucose/GlutaMAX (Life Technologies) containing 10% fetal calf serum (FCS), penicillin (100 IU/ml), and streptomycin (100 μg/ml). All experiments were performed at comparable passage numbers of the cells.

Murine skin samples were taken from the tails of 5-week-old mice (C57BL/6). Human abdominal skin was received from patients undergoing plastic surgery. Dermal sheets were prepared from murine or human skin by the removal of the epidermis with dispase II treatment as described previously (14, 15). Shave biopsies were taken from human skin samples with a scalpel and included the epidermis as well as a thin dermal layer with an average thickness of 0.4 mm (skin shaves). All skin samples were incubated in DMEM/high glucose/GlutaMAX (Life Technologies) containing 10% FCS, penicillin (100 IU/ml), and streptomycin (100 μg/ml).

Primary murine fibroblasts were isolated from the dermis of 3-week-old mouse tails as described (14). Primary human fibroblasts were isolated from human skin samples received from small skin biopsies or patients undergoing breast and plastic surgery. Skin samples were stored in DMEM/high glucose/GlutaMAX (Life Technologies), and subcutaneous fat was removed. After removal of the epidermis by dispase II treatment, the dermis was digested with a whole skin dissociation kit (Miltenyi) for 2.5 h with shaking (180 rpm) at 37°C before filtering through 40-μm cell strainers. After seeding, primary fibroblasts were maintained in DMEM/high glucose/GlutaMAX (Life Technologies) containing 10% FCS, penicillin (100 IU/ml), and streptomycin (100 μg/ml), and experiments were performed at passages 2 to 3 (murine fibroblasts) or at passages 2 to 5 (human fibroblasts) after isolation.

Ethics statement.

The preparation of dermal cells from sacrificed animals was carried out in strict accordance with the recommendations of the Guide of Landesamt für Natur, Umwelt und Verbraucherschutz, Northrhine-Westphalia (Germany). The study was approved by LANUV NRW (8.84-02.05.20.13.018). Human abdominal skin specimens were obtained after informed consent from all patients. The study was approved by the Ethics Commission of the Medical Faculty, University of Cologne (approval no. 17-481).

Collagen matrices.

HFFF2 cells or primary murine fibroblasts were embedded in floating collagen I lattices (19). In brief, the lattice solution was prepared by adding rat tail collagen I (Corning) to DMEM buffered by the addition of 0.1 M NaOH and FCS followed by the supply of 1 × 10⁶ murine fibroblasts per gel. To achieve a cell density in the collagen gels comparable to murine fibroblasts at the time of infection, we added 1 × 10⁵ HFFF2 cells per gel. The lattice solution was dispensed into uncoated 6-cm dishes and incubated at 37°C. Floating gels detached after 1 to 2 h (murine fibroblasts) or 5 h (HFFF2) of polymerization.

Virus.

Infection studies were performed with purified preparations of HSV-1 wild-type (Glasgow) strain 17+ as described (25). Virus inoculum was added to cells at 37°C, defining time point zero, and was replaced by medium at 1 h p.i. Fibroblasts were infected at 5 PFU/cell and floating collagen matrices at 5 or 50 PFU/cell by floating on virus-containing medium.

Human dermal sheets were submerged in virus-containing medium (50 PFU/cell) for various times as described (15).

Marker uptake.

Sulfate-modified polystyrene, fluorescent orange latex beads (500 nm) (Sigma) were used as a marker for penetration of particles in murine and human dermis. Dermal sheets and skin shaves (ca. 4- by 4-mm surface area) as well as HFFF2 cells (4 × 10⁴) were incubated with latex beads (2 × 10⁹ beads/sample) for various times at 37°C. Samples were thoroughly washed three times, and skin samples were immediately embedded for preparation of cryosections while HFFF2 cells were fixed and stained as described below.

RNA preparation and reverse transcriptase PCR.

RNA was isolated from cells by use of NucleoZol reagents (Macherey-Nagel). cDNA was synthesized using the SuperScript II reverse transcriptase (Life Technologies), and PCR was performed with Taq DNA polymerase (Life Technologies). The ICP0 primers (forward, 5′-ATGTCTGGGTGTTTCCCTGC-3′; reverse, 5′-TCTCGAACAGTTCCGTGTCC-3′) generated a 147-bp fragment and the GAPDH (glyceraldehyde-3-phosphate dehydrogenase) primers (forward, 5′-TGATGACATCAAGAAGGTGGTGAAG-3′; reverse, 5′- TCCTTGGAGGCCATGTGGGCCAT-3′) a 240-bp fragment.

Histochemistry, immunocytochemistry, and antibodies.

After mock infection or at various times p.i., primary murine fibroblasts, primary human fibroblasts, and HFFF2 cells were fixed with 1% formaldehyde for 10 min, stained with mouse anti-ICP0 (monoclonal antibody 11060; 1:60) (26), and visualized with the corresponding secondary antibody and DAPI (4′,6-diamidino-2-phenylindole) as nuclear counterstain. F-actin staining of HFFF2 cells was performed with phalloidin-Atto 565 (1:2,000) (Sigma) for 20 min at room temperature (RT).

For cryosections, human dermal sheets and skin shaves were embedded in OCT compound (Sakura), frozen, and cut into 8-μm cross sections as described (14, 27). For hematoxylin and eosin (HE) staining, samples were fixed with 3.4% formaldehyde overnight at 4°C, prepared as paraffin sections (8 μm), and stained for 10 min with hemalum followed by counterstaining with eosin for 20 s. For immunofluorescence, tissue sections were fixed with 1% formaldehyde for 10 min at RT and blocked as described (14, 27). Dermal sections were incubated with rabbit anti-vimentin (1:400) (Cell Signaling) overnight at 4°C and with secondary antibodies and DAPI for 45 min at RT. Infected dermal sections were additionally incubated with mouse anti-VP5 (monoclonal antibody DM165; 1:1,000) overnight at 4°C and with secondary antibodies for 45 min at RT. Sections of skin shaves were incubated with mouse anti-collagen VII (1:500) (Santa Cruz Biotechnology) overnight at 4°C followed by incubation with secondary antibodies and DAPI for 45 min at RT.

Second harmonic generation (SHG) microscopy was performed to analyze the collagen morphology of floating collagen lattices after embedding HFFF2 cells by using an upright multiphoton microscope (TCS SP8 MP; Leica Microsystems) equipped with a Ti:Sa laser (Chameleon Vision II; Coherent), which was tuned to 880 nm as described previously (28). Cells embedded in collagen matrices were costained with propidium iodide (PI) and imaged as whole mounts.

Microscopy of tissue sections and collagen-embedded cells was performed using a Leica DM IRB/E microscope linked to a Leica TCS-SP/5 confocal unit. Images were assembled using Photoshop (Elements 2018; Adobe) and Illustrator (version CS2; Adobe). Cells were analyzed using a wide-field epifluorescence microscope (Zeiss Axiophot) equipped with a Nikon Digital Sight camera system (DS-2MV) and NIS Elements software.

Flow cytometric analysis.

HFFF2 cells were detached with 0.05% trypsin-0.02% EDTA or with enzyme-free cell dissociation solution (CDS) (Sigma) for staining with mouse anti-nectin-1 (CK41; 1:100) (29) for 30 min on ice in phosphate-buffered saline (PBS)/5% FCS or with anti-HVEM-PE (1:11) (CD270; Miltenyi; no. 130-101-599) for 10 min at 4°C. Nectin-1 was visualized with anti-mouse IgG-Cy5 (1:100) (Jackson Immunoresearch Laboratories Inc.). For nectin-1, mouse IgG1 (1:20) (Life Technologies), and for HVEM, human IgG1 (1:50) (REAfinity Clone REA293; Miltenyi) were used as isotype controls. Collagen-embedded HFFF2 were retrieved from matrices by incubating in collagenase buffer containing 400 U/ml collagenase I for 5 min at 37°C and filtering through 40-μm cell strainers. Cell suspensions were subsequently used for staining of nectin-1 and HVEM as described above. For detection of ICP0, collagen-embedded and unembedded HFFF2 were fixed after indicated times p.i. in 1% formaldehyde for 10 min on ice and permeabilized with 0.2% saponin in PBS/5% FCS for 15 min. ICP0 staining was performed with mouse anti-ICP0 (monoclonal antibody 11060; 1:30) (26) for 30 min on ice and visualized with anti-mouse IgG-Cy5 (1:100) (Jackson Immunoresearch Laboratories Inc.), all in PBS/5% FCS/0.2% saponin. Mouse IgG2b (1:20) (Life Technologies) was used as isotype control. Samples were stained with DAPI shortly before analysis, which allowed the gating of only viable cells. Samples were analyzed using a fluorescence-activated cell sorter (FACS) Canto II flow cytometer and FACSDiva (version 6.1.3; BD) and the FlowJo (version 7.6.3; Tree Star) software. The mean fluorescence intensity of the ICP0-positive cells isolated from the collagen-embedded cells was comparable to that from unembedded cells.

Statistics.

For statistical analyses, two or more groups of data were compared by student’s test (Fig. 2d) or one-way ANOVA followed by Tukey’s test (Fig. 1) to calculate P values. Differences were considered to be statistically significant with P values of ≤0.05 (*).