Varicella Zoster Virus Alters Expression of Cell Adhesion Proteins in Human Perineurial Cells via Interleukin 6

Abstract

Background

In temporal arteries (TAs) from patients with giant cell arteritis, varicella zoster virus (VZV) is seen in perineurial cells that surround adventitial nerve bundles and form the peripheral nerve-extrafascicular tissue barrier (perineurium). We hypothesized that during VZV reactivation from ganglia, virus travels transaxonally and disrupts the perineurium to infect surrounding cells.

Methods

Mock- and VZV-infected primary human perineurial cells (HPNCs) were examined for alterations in claudin-1, E-cadherin, and N-cadherin. Conditioned supernatant was analyzed for a soluble factor(s) mediating these alterations and for the ability to increase cell migration. To corroborate in vitro findings, a VZV-infected TA was examined.

Results

In VZV-infected HPNCs, claudin-1 redistributed to the nucleus; E-cadherin was lost and N-cadherin gained, with similar changes seen in VZV-infected perineurial cells in a TA. VZV-conditioned supernatant contained increased interleukin 6 (IL-6) that induced E-cadherin loss and N-cadherin gain and increased cell migration when added to uninfected HPNCs; anti-IL-6 receptor antibody prevented these changes.

Conclusions

IL-6 secreted from VZV-infected HPNCs facilitated changes in E- and N-cadherin expression and cell migration, reminiscent of an epithelial-to-mesenchymal cell transition, potentially contributing to loss of perineurial cell barrier integrity and viral spread. Importantly, an anti-IL-6 receptor antibody prevented virus-induced perineurial cell disruption.

VZV infection of perineurial cells that form the nerve-extrafascicular tissue barrier results in IL-6–mediated loss and gain of E- and N-cadherin, respectively, potentially facilitating neurotropic virus entry into and exit from peripheral nerves and promoting peripheral nerve injury.

Varicella zoster virus (VZV) is latent in >95% of Americans and reactivates to produce herpes zoster in >50% by age 85 [1]. VZV reactivation also produces stroke (VZV vasculopathy) due to productive virus infection of arteries [2–5]. The spectrum of VZV vasculopathy continues to expand with VZV DNA, antigen, and herpesvirus particles found in the majority of temporal arteries (TAs) from patients with giant cell arteritis (GCA) [6] and granulomatous aortitis [7].

A striking and frequent observation in GCA TAs is the presence of VZV antigen in cells surrounding adventitial nerve bundles [8, 9], consistent with viral entry to the artery along nerve fibers following reactivation (Figure 1A, left panel, representative nerve bundle surrounded by VZV-infected cells). These prior studies found that these VZV-infected cells express a tight junction protein, claudin-1, as shown in the adjacent slide section (Figure 1A, right panel), identifying them as perineurial cells. These cells form the peripheral nerve-extrafascicular tissue barrier (perineurium) through tight junctions, providing a sheath that protects interior peripheral nerves, Schwann cells, and endoneurial cells from surrounding tissue within an immunoprivileged compartment [10, 11]. Disruption of the perineurium is critical in neurotropic virus entry to and exit from peripheral nerves, antibody and immune cell access in pathological conditions, and, ultimately, nerve injury including nerve degeneration, inflammation, and demyelination. Herein, we tested whether VZV infection of primary human perineurial cells (HPNCs) disrupts tight and adherens junction proteins (claudin-1, E-cadherin, and N-cadherin), potentially facilitating infection of surrounding cells and peripheral nerve injury. To determine the relevance of our in vitro findings, we examined VZV-infected nerve bundles from a GCA TA. Subsequently, we characterized the role of interleukin-6 (IL-6) in cell adhesion protein changes and whether another alphaherpesvirus could disrupt these adhesion proteins.

Figure 1.

Varicella zoster virus (VZV) in perineurial cells expressing claudin-1. A, Immunohistochemical analysis using mouse anti-VZV glycoprotein E (gE) antibody revealed VZV antigen in cells surrounding nerve bundles in the adventitia of a representative temporal artery from a patient with giant cell arteritis (left panel, pink); analysis of an adjacent slide using rabbit α-claudin-1 antibody showed that the VZV-infected cells expressed tight junction protein claudin-1 (right panel, pink), identifying them as perineurial cells that form the nerve-extrafascicular tissue barrier. Isotype control antibodies and normal rabbit serum revealed no arterial staining (not shown). B, Quiescent primary human perineurial cells (qHPNCs) were mock- and VZV-infected and analyzed at 3 days postinfection (DPI) by immunofluorescence antibody assay using the same antibodies against VZV gE and claudin-1. In mock-infected qHPNCs, claudin-1 (green) was present in the cell membrane/cytoplasm of all cells, confirming a homogeneous perineurial cell culture (left panel). In qHPNCs expressing VZV gE (right panel, red), claudin-1 was present predominantly in the nucleus. C, To test whether claudin-1 was required for VZV entry to cells, qHPNCs were pretreated with isotype control or anti-claudin-1 antibody and VZV infected. At 3 DPI, flow cytometry analyses revealed no difference in the percent of infected cells expressing VZV gE in these 2 experimental groups (79 ± 0.45 versus 81 ± 0.93, respectively, P < .17, percent expression ± SD, n = 4). A, × 600; B, × 400. Blue represents cell nuclei. A, reproduced with permission from JAMA Neurol 2015;72(11),1281–7 [9]. Copyright 2015 American Medical Association. All rights reserved

METHODS

Cells and Virus

Adult spinal nerve HPNCs were seeded at 5000 cells/cm² in basal fibroblast medium (BFM) containing 2% fetal bovine serum (FBS), 1% fibroblast growth serum, and 1% 100× penicillin-streptomycin (Sciencell, Carlsbad, CA). After 24 hours, medium was changed to BFM containing 0.1% FBS and 1% 100× penicillin-streptomycin (quiescent medium) and replenished every 48–72 hours for 7 days, establishing quiescence. At day 7, quiescent HPNCs (qHPNCs) were cocultivated with uninfected (mock-infected) or VZV-infected HPNCs (0.01 multiplicity of infection [MOI], VZV Gilden strain, GenBank No. MH379685); cells were analyzed at 1, 2, 3, and 4 days postinfection (DPI). For herpes simplex virus type 1 (HSV-1) infection, cell-free virus (KOS strain; GenBank No. JQ673480) was added to qHPNCs at 0.01 MOI and removed 1 hour later; cells were analyzed at 2 and 3 DPI. Similar experiments were completed on primary human corneal epithelial cells (HCECs; Sciencell) as described [12] using the same VZV and HSV-1 MOI and analyzed at the heights of cytopathic effect (7 and 3 DPI, respectively).

Quantitative PCR

RNA was extracted from mock- and VZV-infected qHPNCs and HCECs at 3 and 7 DPI, respectively (Direct-zol RNA miniprep kit; Zymo Research, Irvine, CA), residual DNA removed (Turbo-DNA free kit; ThermoFisher, Grand Island, NY), and first-strand cDNA synthesized (Transcriptor first strand cDNA synthesis kit; Roche, San Francisco, CA). Predesigned and validated primers and probes used for quantitative polymerase chain reaction (qPCR) analysis were: (1) claudin-1 (exon 2–3), E-cadherin (exon 2–3), and N-cadherin (exon 13–14); (2) VZV (FWD: CGAACACGTTCCCCATCAA, REV: CCCGGCTTTGTTAGTTTTGG, Probe: FAM/TCCAGGTTTTAGTTGATACCA-/BkFQ/); and (3) housekeeping gene glyceraldehyde 3-phosphate dehydrogenase (GAPDH; FWD: CACATGGCCTCCAAGGAGTAA, REV: TGAGGGTCTCTCTCTTCCTCTTGT, Probe: VIC/CTGGACCACCAGCCCCAGCAAG) (IDT Technologies, Coralville, IA). Cycling conditions were: (1) 10 minutes 95°C hold; and (2) 95°C for 30 seconds, 60°C for 1 minute, 40 cycles. Results for E- and N-cadherin were analyzed by the $\Delta$ cycle threshold (Ct) method (gene of interest, GAPDH) and results for claudin-1 by the $\Delta \Delta$Ct method.

Immunofluorescence Antibody Assay

Mock- and VZV-infected qHPNCs (3 DPI) and HCECs (7 DPI), together with mock- and HSV-1-infected HPNCs (3 DPI), were fixed for 20 minutes in 4% paraformaldehyde before dual staining with either mouse anti-VZV glycoprotein E (gE) at 1:500 (Santa Cruz Biotechnology, Santa Cruz, CA) or mouse anti-HSV-1 (1:250, Novus Biologicals, Littleton, CO) and rabbit anti-claudin-1 (1:100), anti-E-cadherin (1:50), or anti-N-cadherin (1:50) (Abcam, Cambridge, MA) as previously described [13]. Secondary antibodies were Alexa Fluor 594 donkey anti-mouse IgG and Alexa Fluor 488 donkey anti-rabbit IgG, both at 1:500 (LifeTechnologies, Grand Island, NY). Coverslips were mounted with Vectashield Hardset Mounting Medium with 4′,6-diamidino-2-phenylindole (DAPI; Vector Labs, Burlingame, CA) and visualized by fluorescence microscopy.

Immunohistochemistry of TAs

A formalin-fixed, paraffin-embedded GCA TA that contained VZV in perineurial cells, as well as uninfected perineurial cells in a different field of the same slide, was identified from a previous study and used herein [8]. Adjacent slides were stained for E- and N-cadherin using primary antibodies as in the immunofluorescence antibody assay above, as previously described [8]. Positive controls for E- and N-cadherin were human lymph node and skin, respectively. Negative controls consisted of 2 immunostained sections, replacing each primary antibody with normal rabbit serum (Jackson ImmunoResearch, West Grove, PA) at the same dilution. Slides were analyzed by light microscopy.

Flow Cytometry

Mock- and VZV-infected HPNCs (3 DPI) were harvested and stained for VZV gE as previously described [14]. Cells were harvested, washed with fluorescence-activated cell sorting (FACS) buffer (phosphate-buffered saline containing 1% FBS) and stained with R-phycoerythrin-conjugated mouse anti-VZV-gE (MilliporeSigma, Burlington, MA) antibody for 30 minutes at 4°C, washed with FACS buffer, and fixed with 1% paraformaldehyde. Isotype controls were used in all stainings. Cells were analyzed using a Canto-II flow cytometer (BD Biosciences, San Jose, CA); >15 000 events (1 event = 1 cell) were collected for all samples. Data were analyzed using Diva (BD Biosciences) and FlowJo (Tree Star, Ashland, OR) softwares.

Conditioned Media

Supernatant was collected from mock- and VZV-infected qHPNCs (3 DPI), centrifuged, and flash frozen. The supernatant was thawed and added at 100% to uninfected qHPNCs or HCECs and replenished every 24 hours for 3 days. Coverslips were fixed and analyzed by immunofluorescence antibody assay (IFA) for adherens junction proteins.

Enzyme-Linked Immunosorbent Assay

Mock- and VZV-infected qHPNC supernatant (1, 2, 3, and 4 DPI) and mock- and HSV-1–infected qHPNCs (2 and 3 DPI) and HCECs (3 DPI), were run undiluted using the human IL-6 enzyme-linked immunosorbent assay (ELISA) MAX kit (Biolegend, San Diego, CA) per the manufacturer’s instructions. Mock- and VZV-infected HCEC supernatant (7 DPI) were analyzed using the human V-PLEX Proinflammatory Panel 1 (MesoScale Discovery, Rockville, MD) as described [12].

IL-6 Treatment and Cell Viability

IL-6–treated cells were positive controls for E-cadherin loss and N-cadherin gain. Cell viability for optimal treatment concentrations and preliminary toxicity of IL-6 were measured using qHPNCs treated with daily IL-6 (MilliporeSigma) at increasing concentrations (0–1000 ng/mL). Three days posttreatment, cells were stained with alamar blue (Invitrogen, Carlsbad, CA) for 1 hour and analyzed for viability on a plate reader at 570 nm (optimal IL-6 concentration = 350 ng/mL). For experimental conditions, HPNCs were treated with IL-6 at 350 ng/mL, daily for 3 days. Coverslips were fixed and analyzed by IFA for adherens junction proteins.

To determine if HCECs undergo an epithelial-to-mesenchymal cell transition (EMT) in an IL-6 concentration-dependent manner, HCECs were treated with VZV-infected qHPNC supernatant or with increasing concentrations of IL-6 (30, 100, and 300 pg/mL) for 3 days then fixed and stained for E- and N-cadherin.

IL-6 Receptor Blockade

Cell viability and preliminary toxicity of anti-IL-6 receptor (IL-6R) antibody (Sino Biological, North Wales, PA) were measured as above. HPNCs were grown in clear-bottom imaging plates (ibidi, Martinsried, Germany) to quiescence and treated with 100 ng/mL anti-IL-6R antibody or isotype control. Mock- and VZV-infected qHPNC supernatant was added at 100% to cells 1 hour after antibody treatment. Antibody, isotype control and supernatant were replenished daily for 3 days. Cells were fixed and stained for cadherins.

Migration/Wound Healing Assay

Proliferating HPNCs were plated into Culture-Insert 2 wells (ibidi) at 550 000 cells/chamber. Twenty-four hours later, inserts were removed and cells were treated. Positive control was IL-6 cytokine alone or in combination with anti-IL-6R antibody or isotype control. The experimental group included treatment with mock- or VZV-infected HPNC supernatant alone or in combination with either anti-IL-6R antibody or isotype control. Cells were imaged at 1 and 4 hours posttreatment and quantified using ImageJ.

Statistical Analysis

Statistical analysis was performed using Prism (GraphPad, San Diego, CA). Statistical significance for FACS and qPCR results was determined using the Student unpaired t test, P value set at 0.05. Wound-healing assay and ELISA results used a 1-way ANOVA with a Tukey multiple comparisons test (*, P < .05; **, P < .01; ***, P < .001).

RESULTS

Claudin-1 Redistributes From Cell Membrane/Cytoplasm to Nucleus During VZV Infection and Is Not Essential for Virus Entry

At 3 DPI, IFA of mock-infected qHPNCs revealed claudin-1 in the membrane/cytoplasm of all cells (Figure 1B, left panel, green), confirming that these cells are perineurial. In VZV-infected qHPNCs (red), claudin-1 (green) redistributed to the nucleus but remained in the cytoplasm of uninfected bystander cells (Figure 1B, right panel, arrows). RT-qPCR analysis of mock- and VZV-infected cells showed no significant difference in claudin-1 transcripts (1.00 ± 0.00 versus 0.77 ± 0.11, respectively, fold difference ± SD, n = 5). To test whether claudin-1 was required for VZV entry to cells, qHPNCs were cultured, pretreated with isotype control antibody or anti-claudin-1 antibody, and VZV infected; FACS revealed no difference in the percent of cells expressing VZV gE in the isotype control and anti-claudin-1 antibody-treated groups (Figure 1C; 79 ± 0.45 versus 81 ± 0.93, respectively, P < .17, percent expression ± SD, n = 4).

VZV-Infected qHPNCs Lose E-cadherin but Gain N-Cadherin

IFA of adherens junction proteins at 3 DPI showed expression of E-cadherin (green) in mock-infected (Figure 2A, top left panel) but not VZV-infected cells (red) or uninfected bystanders (arrows, top right panel). N-cadherin (green) was absent in mock-infected (bottom left panel) but present in both VZV-infected (red) and uninfected bystander cells (arrows, bottom right panel). Mock-infected qHPNCs contained E-cadherin (14.29 ± 0.7, $\Delta$Ct ± SD, n = 3) but not N-cadherin transcripts by RT-qPCR, whereas VZV-infected HPNCs contained no E-cadherin but contained N-cadherin transcripts (0.95 ± 0.29, $\Delta$Ct ± SD, n = 3).

Figure 2.

E-cadherin and N-cadherin expression in mock- and varicella zoster virus (VZV)-infected quiescent primary human perineurial cells (qHPNCs). Quiescent HPNCs were mock- or VZV-infected with cell-associated virus. At 3 days postinfection, cells were fixed and analyzed by immunofluorescence antibody assay using mouse anti-VZV glycoprotein E (VZV gE; red) antibody and either rabbit anti-E-cadherin or rabbit anti-N-cadherin antibody (green). E-cadherin was expressed in mock-infected cells (top, left panel) but not in VZV-infected cells (top, right panel) or uninfected bystander cells (top, right panel, arrows). N-cadherin was absent in mock-infected cells (bottom, left panel) but present in VZV-infected cells (bottom, right panel) and uninfected bystander cells (bottom, right panel, arrows). All images × 400. Blue represents cell nuclei.

VZV-Infected Perineurial Cells in TAs Lose E-Cadherin but Gain N-Cadherin

A GCA TA containing VZV antigen in perineurial cells surrounding nerve bundles [15], as well as a region from the same TA that did not contain VZV antigen in perineurial cells, were studied. Immunohistochemical analysis of adjacent slides for VZV, E- and N-cadherin demonstrated that VZV-negative perineurial cells (Figure 3, top panel, left, arrows) expressed E-cadherin (top panel, center) but not N-cadherin (top panel, right), whereas perineurial cells containing VZV antigen (Figure 3, bottom panel, left) did not express E-cadherin (bottom panel, center) but expressed N-cadherin (bottom panel, right, arrows). These changes were also seen in uninfected bystander perineurial cells surrounding the same nerve bundle.

Figure 3.

Immunohistochemical analysis of uninfected and varicella zoster virus (VZV)-infected perineurial cells in temporal arteries (TAs). Adjacent slides from a TA of a patient with clinical symptoms/signs of giant cell arteritis were immunostained for the presence of VZV glycoprotein E (VZV gE), E-cadherin, and N-cadherin. Perineurial cells that did not contain VZV gE (top, left panel, arrows) expressed E-cadherin (top, middle panel, pink, arrows) but not N-cadherin (top, right panel). In contrast, perineurial cells that contained VZV gE (bottom, left panel, pink, arrows) did not express E-cadherin (bottom, middle panel) but did express N-cadherin (bottom, right panel, arrows). All images × 600.

VZV-Infected qHPNC Supernatant Induced Loss of E-Cadherin and Gain of N-Cadherin, Indicating a Soluble Factor(s) Contributed to These Alterations

Mock- and VZV-infected qHPNC supernatant was added to uninfected qHPNCs for 3 days, replenished daily. IFA of mock- and VZV-infected qHPNCs showed no difference in claudin-1 localization in the cell membrane/cytoplasm. E-cadherin (green) was seen in qHPNCs treated with mock-infected supernatant, but not in qHPNCs treated with VZV-infected supernatant (Figure 4). N-cadherin (green) was not seen in qHPNCs treated with mock-infected supernatant, but was seen in qHPNCs treated with VZV-infected supernatant (Figure 4).

Figure 4.

Treatment of uninfected quiescent primary human perineurial cells (qHPNCs) with conditioned supernatant from mock- or varicella zoster virus (VZV)-infected qHPNCs. Conditioned supernatant from mock- and VZV-infected qHPNCs were added to naive, uninfected qHPNCs for 3 days, replenished every 24 hours, and analyzed by immunofluorescence antibody assay using rabbit anti-claudin-1 antibody, rabbit anti-E-cadherin, or rabbit anti-N-cadherin antibody, as well as mouse anti-VZV glycoprotein E (VZV gE) antibody. As expected, none of the uninfected qHPNCs exposed to VZV conditioned supernatant expressed VZV gE because conditioned supernatant does not contain cell-free virus (not shown). Both mock- and VZV-infected qHPNCs showed no difference in claudin-1 localization in the cell membrane/cytoplasm (top panels, respectively, green). E-cadherin was seen in the cell membrane/cytoplasm of qHPNCs treated with mock-infected supernatant but not in qHPNCs treated with VZV-infected supernatant (middle panels, respectively, green). N-cadherin was not seen in qHPNCs treated with mock-infected supernatant, but was seen in cell membrane/cytoplasm of qHPNCs treated with VZV-infected supernatant (bottom panels, respectively, green). All images × 400. Blue represents cell nuclei.

VZV-Induced IL-6 Promotes E- and N-Cadherin Changes in qHPNCs

Mock- and VZV-infected qHPNC supernatant was collected (1, 2, 3, and 4 DPI) and analyzed by ELISA for IL-6 (Figure 5A). Compared to mock-infected cells, VZV-infected qHPNCs had significantly increased levels of IL-6 at 1, 2, 3, and 4 DPI (Table 1).

Figure 5.

The role of interleukin 6 (IL-6) in E- cadherin and N-cadherin expression in mock- and varicella zoster virus (VZV)-infected quiescent primary human perineurial cells (qHPNCs). A, Conditioned supernatant from mock- and VZV-infected qHPNCs was collected at 1, 2, 3, and 4 days postinfection (DPI). IL-6 was significantly elevated at each time point in VZV-infected qHPNCs compared to mock-infected qHPNCs (223 ± 0.04 versus 56 ± 0.03; 263 ± 0.29 versus 79 ± 0.01; 332 ± 0.26 versus 79 ± 0.02; and 302 ± 0.04 versus 153 ± 0.03, respectively, mean concentration pg/mL ± SD; ***, P < .001, n = 3). B, Conditioned supernatant from mock- and VZV-infected qHPNCs was applied to uninfected qHPNCs in the presence of anti-IL-6 receptor (anti-IL-6R) monoclonal antibody or isotype control and analyzed by immunofluorescence antibody assay using rabbit anti-E-cadherin or rabbit anti-N-cadherin antibody. As a positive control, IL-6 at the same concentrations as that detected in the conditioned supernatant of VZV-infected qHPNCs at 3 DPI was added to uninfected qHPNCs at 350 pg/mL; E-cadherin expression was not detected but N-cadherin expression was induced (column 1, top and bottom panels, respectively). Quiescent HPNCs treated with supernatant from mock-infected qHPNCs and with anti-IL-6R antibody (column 2) or isotype control (iso ctrl; column 3) expressed E-cadherin but not N-cadherin (top and bottom panels, respectively). Quiescent HPNCs treated with VZV-infected qHPNC supernatant containing elevated levels of IL-6 with either anti-IL-6R antibody or isotype control showed that the VZV-induced loss of E-cadherin and gain of N-cadherin when treated with isotype control (column 5); this change was prevented by blockade of the IL-6 receptor (column 4). All images × 400. Blue represents cell nuclei.

Table 1.

Interleukin-6 Levels in qHPNCs and HCECs Infected With VZV and HSV-1

		IL-6
Virus	Days Postinfection	qHPNCs Mock, Mean ± SD, pg/mL	qHPNCs VZV, Mean ± SD, pg/mL
VZV	1	56.72 ± 0.03	223.59 ± 0.04a
	2	79.87 ± 0.01	263.92 ± 0.29a
	3	79.97 ± 0.02	332.43 ± 0.27a
	4	153.3 ± 0.03	303 ± 0.04a
HSV-1	2	19.13 ± 3	6.2 ± 1.2b
	3	31.26 ± 6	8.5 ± 0.98a

Abbreviations: HCECs, corneal epithelial cells; HSV-1, herpes simplex virus-1; IL, interleukin; qHPNCs, quiescent primary human perineurial cells; VZV, varicella zoster virus.

^a P < .001; significant change compared to mock.

^b P < .01; significant change compared to mock.

To test whether IL-6 in VZV-infected qHPNC supernatant mediated E- and N-cadherin alterations, mock- or VZV-infected qHPNC supernatant was applied to uninfected qHPNCs with anti-IL-6R antibody or isotype control and analyzed by IFA for E- and N-cadherin (Figure 5B). As a positive control, IL-6 at similar concentrations to that detected in VZV-infected qHPNC supernatant at 3 DPI was added to uninfected qHPNCs (350 pg/mL); E-cadherin was not detected but N-cadherin was increased. Uninfected qHPNCs treated with mock-infected HPNC supernatant and either anti-IL-6R antibody or isotype control expressed E-cadherin but not N-cadherin. qHPNCs treated with VZV-infected qHPNC supernatant and either anti-IL-6R antibody or isotype control showed VZV-induced E-cadherin loss and N-cadherin gain when treated with isotype control; this change was prevented by blockade of the IL-6R.

Conditioned Supernatant Increases Migration in HPNCs Via IL-6

A wound-healing assay of HPNCs treated with qHPNC supernatant was used to measure cell migration (Figure 6 and Supplementary Figure 1). Values for HPNCs exposed to mock-infected supernatant without or with anti-IL-6R antagonist 4 hours posttreatment were 18.8% ± 3% and 10.3% ± 5%, respectively (mean percent closure ± SD, n = 3). Compared to treatment with mock-infected supernatant above, uninfected HPNCs treated with VZV-infected supernatant significantly increased wound closure (72.6% ± 18.4%, P < .05, n = 3). Compared to treatment with VZV-infected supernatant, HPNCs treated with this same supernatant plus an anti-IL-6R antagonist had significantly less wound closure (22.9% ± 11%, P < .05, n = 3). Application of IL-6 alone resulted in a significant wound closure increase compared to mock-treated HPNCs (92% ± 4.2% P < .01, n = 3) that was inhibited by an anti-IL-6R antagonist (49.3% ± 16.2%). Figure 6B shows the wound closure in HPNCs treated with mock-infected qHPNC supernatant (negative control; top panel) and in IL-6–treated uninfected HPNCs (positive control; bottom panel).

Figure 6.

Effects of varicella zoster virus (VZV) infection and interleukin-6 (IL-6) on migration of primary human perineurial cells (HPNCs). A, A wound-healing assay of HPNCs treated with conditioned supernatant was used to measure cell migration under different conditions. HPNCs exposed to mock-infected supernatant without or with anti-IL-6R antagonist 4 hours posttreatment had wound closure of 18.8% ± 3% and 10.3% ± 5%, respectively (mean percent closure ± SD, n = 3). Compared to treatment with mock-infected supernatant above, uninfected HPNCs treated with VZV-infected supernatant significantly increased wound closure (72.6% ± 18.4%, P < .05, n = 3). Compared to treatment with VZV-infected supernatant, HPNCs treated with VZV-infected supernatant plus an anti-IL-6R antagonist had significantly less wound closure (22.9% ± 11%, P < .05, n = 3). Application of IL-6 alone resulted in a significant wound closure increase compared to mock-treated HPNCs (92% ± 4.2%, P < .01, n = 3) that was inhibited by an anti-IL-6R antagonist (49.3% ± 16.2%). One-way ANOVA with a Tukey multiple comparisons test; *, P < .05; **, P < .01. B, An example of quantification showing minimal wound closure in cells treated with mock supernatant (negative control, top panel) and wound closure in cells treated with IL-6 (positive control, bottom panel) that conferred increased cell migration into the gap.

VZV-Infected Corneal Epithelial Cells do not Lose E-Cadherin or Gain N-Cadherin due to Lower Levels of VZV-Induced IL-6

To determine if VZV induced E- and N-cadherin changes in another epithelial cell type, VZV-infected HCECs were examined. E-cadherin loss and N-cadherin gain were not seen at the transcript or protein levels (data not shown). Compared to levels of IL-6 secreted by VZV-infected qHPNCs (3 DPI; 332.45 ± 0.27 pg/mL), VZV-infected HCECs only secreted 23.93 ± 4.12 pg/mL compared to mock at 15.73 ± 5 pg/mL (P < .001) (measured in parallel experiments and reported by Como and colleagues [12]). To determine if IL-6 levels in VZV-induced HCECs were too low to induce an EMT, we explored whether HCECs have the capacity to undergo an EMT in an IL-6 concentration-dependent manner. HCECs treated with 30 pg/mL of IL-6 did not exhibit E- or N-cadherin alterations, but when IL-6 concentrations were increased (100 and 300 pg/mL), HCECs lost E-cadherin and gained N-cadherin (Supplementary Figure 2); treatment of HCECs with VZV-infected qHPNC supernatant containing 332 pg/mL IL-6 also induced E- and N-cadherin changes. Thus, HCECs have the capacity to undergo a partial EMT with IL-6 concentrations of 100 pg/mL or higher but VZV infection of HCECs induced lower concentrations of IL-6 that were insufficient to induce an EMT.

Claudin-1, E-Cadherin, and N-Cadherin Expression in Mock- and HSV-1–Infected qHPNCs

To determine if another alphaherpesvirus produced similar changes in cell adhesion proteins, qHPNCs were mock- or HSV-1–infected. At 3 DPI, qHPNCs were fixed and stained for claudin-1, E-cadherin, or N-cadherin, which revealed no change in the presence or distribution in mock- or HSV-1–infected cells (Supplementary Figure 3A). Compared to mock-infected qHPNCs, HSV-1–infected qHPNCs significantly reduced IL-6 at 2 and 3 DPI (Table 1, Supplementary Figure 3B). Because prior reports show that HSV-1 infection increases IL-6 in HCECs and murine cornea [16, 17] and in murine trigeminal ganglia [18], we tested whether HSV-1 could induce IL-6 in primary HCECs. Compared to mock-infected HCECs, HSV-1–infected HCECs had significantly elevated levels of IL-6 (22.15 ± 11.1 and 140.57 ± 0.38, respectively, P < .001, n = 3) consistent with prior publications, demonstrating that HSV-1–induced IL-6 alterations are cell type dependent.

DISCUSSION

Herein, we found that compared to mock-infected qHPNCs, VZV-infected qHPNCs showed disruption of multiple cell adhesion proteins. Specifically, claudin-1 redistributed to the nucleus. There was also down- and upregulation of E- and N-cadherin transcripts and proteins, respectively, and enhanced cell migration mediated by VZV-induced IL-6 secretion; an anti-IL-6R antagonist prevented these cadherin and migration alterations indicating that IL-6 alone is sufficient to induce these changes reminiscent of a partial EMT (working model shown in Supplementary Figure 4). Subsequent examination of another epithelial cell type (HCECs) that demonstrated similar loss of E-cadherin and gain of N-cadherin with IL-6 levels ≥100 pg/mL, but not IL-6 levels of 30 pg/mL, showed that while VZV upregulated secreted IL-6 to approximately 24 pg/mL, this was not sufficient to induce a partial EMT. Interestingly, HSV-1 infection showed decreased IL-6 in qHPNCs and no cell adhesion protein alterations, but increased IL-6 in HCECs. Thus, the degree of IL-6 induction during infection is virus specific and cell type dependent, and the ability of epithelial cells to exhibit changes in E- and N-cadherin is IL-6 concentration dependent.

VZV-infected qHPNCs, but not uninfected bystander cells, relocalized claudin-1 from the cell membrane/cytoplasm to the nucleus in the absence of transcriptional changes, indicating that direct virus infection was necessary for relocalization. The function of nuclear claudin-1 during VZV infection remains unclear, but does not appear to be involved in VZV replication because an anticlaudin-1 antibody did not alter the percent of cells infected. This finding contrasts with that seen in hepatitis C virus infection, where claudin-1 is required for viral entry in human hepatoma cell lines [15]. More likely, alterations in claudin-1 expression or distribution contributes to disruption of perineurial integrity, facilitating viral spread and cell migration, as seen in HSV-1 infection and certain cancers. In epidermal HSV-1 infections, tight junction proteins are implicated as critical barrier structures to restrict viral spread; reductions in claudin-1 have been proposed to enhance susceptibility to HSV-1 infections in atopic dermatitis [19]. Furthermore, Rahn and colleagues [20] showed HSV-1 infection through unstratified keratinocytes or after wounding but not in fully stratified cultures, indicating that intact tight junctions are a major physical barrier for HSV-1 tissue invasion. In human primary colon carcinoma and metastasis and in cell lines derived from primary and metastatic tumors, nuclear localization of claudin-1 has been reported. Specifically, Dhawan and colleagues [21] found that claudin-1 was involved in the regulation of cellular transformation, tumor growth, and metastasis, with nuclear claudin-1 seen in 58% of liver metastasis and in 35% of lymph node metastasis, but not in normal colonic epithelial cells. Thus, nuclear claudin-1 in VZV-infected cells may be involved in cell cycle dysregulation and disruption of perineurial barrier integrity.

The most striking finding herein was the ability of VZV to alter cadherin expression of infected qHPNCs and bystanders via secretion of IL-6. E- and N-cadherin are members of the cadherins family, which are type 1 transmembrane proteins. E-cadherin is a tumor-suppressor gene and N-cadherin is a tumor-promotor gene; loss of E cadherin is an important step in an EMT, defined as transdifferentiation of epithelial cells into mesenchymal cells during both normal and pathological processes, including cancer metastasis. The functional hallmark of EMT is loss of epithelial characteristics, including junction and apical-basal polarity, reorganization of the cytoskeleton, and multiple biochemical changes that confer a migratory, apoptosis-resistant invasive phenotype and produces matrix metalloproteinases [22–24]. The steps involved in a complete EMT are fluid and complex, with numerous proteins involved, and it has been shown that cells can undergo a partial EMT in which they retain some cell-to-cell adhesion and gain migratory traits to perform unique tasks. While we did not extensively study alterations in all the EMT effectors, core regulators, and inducers, we conclude that the VZV-infected qHPNCs underwent at least a partial EMT with altered epithelial junction proteins and enhanced migration that was further potentiated by the nuclear redistribution of claudin-1.

Several oncoviruses are involved in the EMT process, including human papilloma virus, hepatitis B and C, Epstein-Barr virus, and Kaposi sarcoma-associated herpesvirus [25, 26] with an array of viral proteins and microRNAs regulating the EMT program [25]. While as yet unidentified VZV proteins may act as EMT regulators, the VZV-induced secretion of IL-6 was sufficient to produce adherens junction alterations and migration that IL-6R blockade prevented. To our knowledge, this is the first report demonstrating a VZV-induced partial EMT in cells of epithelial origin, raising the possibility of VZV’s role in cancer progression through an indirect effect of secreted IL-6 on bystander epithelial cells.

Finally, the IL-6–mediated disruption of the perineurium may have significant implications for peripheral nerve disease and vasculitides. While IL-6 has been shown to promote an EMT in cancer cells [27–31], the concept has not been extended to the perineurium, which protects nerves in the context of virus infection or IL-6–driven diseases. Deretzi et al [32]. identified IL-6 as an essential mediator of inflammation and demyelination in the peripheral nervous system of rats after local administration, and our findings further extend the mechanism to include IL-6–mediated perineurium disruption, allowing inflammatory cells to enter the interior, which contains Schwann cells and neurons. The ability of viruses to disrupt the perineurium via IL-6 secretion may contribute to the neurotropism of specific viruses and strains, consistent with our findings that VZV is able to induce an EMT during infection of qHPNCs through secreted IL-6, while the nonneuroinvasive HSV-1 KOS strain does not induce an EMT or increase IL-6 [33, 34]. IL-6–mediated disruption of the peripheral nerves might also lead to peripheral nerve damage and aberrant proinflammatory neurotransmitter release. In the context of vasculitides, such changes could perpetuate inflammation and vascular wall dysfunction disproportionate to the extent of VZV-infected cells due to the effect of IL-6 on uninfected bystander cells.

Supplementary Data

Supplementary materials are available at The Journal of Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.

Notes

Acknowledgments. The authors thank Marina Hoffman for editorial review and Cathy Allen for manuscript preparation.

Financial support. This work was supported by the National Institute on Aging (grant number AG032958 to M.A.N.) and National Institute of Neurological Disorders and Stroke (grant number NS094758 to M.A.N.), National Institutes of Health; and the American Heart Association (grant number 17POST33661139 to D. J.).

Potential conflicts of interest. All authors: No reported conflicts of interest. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.